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 |
|---|---|---|---|---|---|---|
69692237 | pes2o/s2orc | v3-fos-license | A SUPPORT TOOL FOR SELECTING AND SCHEDULING PROJECT PORTFOLIOS
: Project portfolio selection is an important issue that arises repeatedly over an organization's lifetime. The problem consists of how to plan and allocate a limited set of resources among a set of candidate projects within a given time horizon, while taking into account multiple objectives and constraints that can be different at each period of the planning horizon. This article describes a software tool with a user-friendly environment called the Project Portfolio Selection Tool (PPST), which assists decision makers (DMs) to simultaneously select and schedule efficient project portfolios. This software tool incorporates a general multiobjective model and uses a metaheuristic algorithm as a search engine. By using an interactive procedure, PPST also supports DMs in exploring the solution space to find the most suitable portfolio according to their preferences. We present an application of the software in a state university setting to illustrate the rationale underlying this environment and its usefulness.
Introduction
The selection and scheduling of project portfolios is one of the most important decision-making processes in any state or private organization.In this process, the candidate projects compete for funds and other resources (such as manpower, equipment, etc) to fulfil a set of objectives and priorities of the organization.There are usually more projects to choose from than resources to carry them out.There are two negative consequences to any erroneous decision regarding the choice of selected projects.On the one hand, limited resources are spent on inappropriate projects, and on the other, there is a loss of private, social, and state benefits that could have been obtained if the resources had been applied to more useful projects (see Martino, 1995).Thus, these decisions are crucial to the success of any organization.As a consequence of their importance, this process has received significant attention from managers and researchers, as reported in several studies (Archer and Ghasemzadeh, 1999;Teller et al., 2012;Martinsuo, 2013).
Traditionally, decision makers (DMs) in organizations have made decisions based on either their experience alone or by using a mixture of their professional judgment and ranking tools (Moore and Baker, 1969;Cooper et al., 2001), such as financial methods (Silvola, 2006), scoring models (Lawson et al., 2006), analytical hierarchy process (Feng et al., 2011) or multiattribute utility (Duarte and Reis, 2006; Lopes and de Almeida, 2015).Thus, projects were selected from the highest to lowest score until the budget available for the period was spent.
Nowadays, the size and complexity of many organizations makes this process more complicated.The DMs want to adopt a global approach to the problem, and thus, they have to select a group of projects while simultaneously taking into account all the relevant features of the problem (Urli and Terrien, 2010; Li et al., 2016).
The first important issue is that there are often multiple conflicting criteria (such as benefit, cost, risk, etc.) that have to be taken into account in the decision process.Therefore, a multicriteria approach must be followed to analyze the problem (Gutjahr et al., 2010;Cruz et al., 2014;Tofighian and Naderi, 2015).
Secondly, DMs have to deal with several constraints regarding available resources (financing, workforce, equipment, etc) and strategic, technical, political or legal issues (Mavrotas et al., 2006).Consequently, the projects are not independent.There may also be complementarity, incompatibility, or synergies produced by sharing costs and/or extra benefits derived from simultaneously conducting more than one project (Chien, 2002;Dobson, 2002).For these reasons, it is insufficient to simply compare two projects; rather, groups of projects (i.e. a project portfolio) should be compared to identify the one best adapted to the needs of the organization.This issue has been addressed in several studies (see Santhanam and Kyparisis, 1996;Klapka and Piños, 2002;Stummer and Heidenberger, 2003;Li et al., 2016;Arratia et al., 2016), which have suggested that the best individual projects do not necessarily produce the best portfolio.Given this background, optimization models have received much attention as they can address different types of interdependencies, multiple objectives, and constraints.Specifically, it is customary to use multiobjective programming models with binary variables to represent the different candidate projects (Cruz et al., 2014).
A third important aspect refers to integrating the selection process within a planning horizon.Most early studies considering project scheduling have first selected the projects for inclusion in the portfolio and then scheduled the selected projects or assumed that all the projects start during period one (Stummer and Heidenberger, 2003).However, this approach may result in some projects not being implemented due to a lack of resources in a given period.This drawback can be overcome by using models that are more flexible regarding when the projects are launched.The complexity of this situation could explain why literature until the end of the 20th century only presents few models simultaneously addressing both project selection and scheduling.So, in 1996, Coffin and Tailor stated, after a review of project portfolio selection studies, that project scheduling had not been taken into consideration in most of them.
However, in recent years, the joint problem of project portfolio selection and scheduling has received considerable attention, since it allows maximizing of the organizations´ stated objectives.In this context, Ghasemzadeh et al. (1999), Medaglia et al. (2008) and Liu and Wang (2011) proposed the use of integer programming models to integrate a project selection mechanism with precedence relationships and technical interdependencies among projects.Jafarzadeh et al. (2011) considered that the time horizon was flexible and the projects were independent and, therefore, the projects could be performed in any order.Naderi (2013) and Emami et al. (2016) presented models for selecting projects and scheduling the activities of the selected projects.In all of these cases, a single function was optimized, the total benefit or the net present value of the selected projects without synergies among them.With the same objective but considering synergistic benefits, Li et al. ( 2016) formulated an integer model that incorporated the project interruptions as a strategy to improve the scheduling of the projects.Tofighian and Naderi (2015) modelled the joint problem of project selection and scheduling using a mixed integer linear model to optimize both the total benefit of selected projects and resource usage variation, but without synergetic effects.They compared different algorithms to approximate the set of efficient portfolios.In this line, Carazo el al. (2010) and Rabbani et al. (2010) also proposed multiobjective models and analyzed different strategies to determine the solution space of efficient portfolios.At this point, there is no criterion for discriminating among these efficient solutions, so we need to incorporate additional information on preferences of the decision maker(s) into the process in order to select a final solution.
Taking the foregoing into account, we used a multiobjective binary model that incorporates all the key aspects under general conditions and which can be applied in public and private settings.Also, in order to assist DMs in choosing a final efficient portfolio, his/her preferences are included in the process.
The output of a multiobjective problem is a set of efficient solutions (the Pareto Set) rather than a single optimal solution.Generally, this set is very large and it can be difficult to identify or select a final solution from all the efficient solutions.The literature is replete with methods (see Ballestero and Romero, 1998) to assist DMs in this task, according to the type of preferences and information the DMs are able to provide: weights, trade-offs, target values, etc.Thus, some studies use goal programming (Santhanam and Kyparisis, 1996), which is suitable when the DMs can provide some thresholds or target values in advance for the conflicting objectives.Other authors (Ghasemzadeh et al., 1999;Medaglia et al., 2008) combine the different objectives into a single function by assigning different weighting scores to each objective according to their importance for the DMs.Some studies use interactive techniques to produce an approximation of the set of efficient solutions and include the DMs' preferences at a later stage (Graves and Ringuest, 2003;Stummer and Heidenberger, 2003;Nowak, 2013).We chose the latter approach, which includes multiple objectives within the model and obtains the efficient portfolios.The DMs can then refine the space of efficient solutions or choose the most suitable solution, according to their own preferences.
The mathematical formulation of the model used is an NP-hard problem (Ehrgott and Gandibleux, 2000;Demeulemeester and Herroelen, 2002) and the use of an exact method makes its solution more difficult as the time horizon and the number of projects and objectives increase.For this reason, we have used a metaheuristic algorithm.This approach offers a good compromise between the quality of the solution obtained and run-time.In a multiobjective context, evolutionary algorithms are particularly suitable as they can obtain a set of efficient solutions in one run instead of separate runs, as occurs when using traditional mathematical programming techniques (Deb, 2001).We use an evolutionary metaheuristic method called Scatter Search for Project Portfolio Selection (SS-PPS), whose problemsolving abilities in this field have proven to be superior to others (Carazo et al., 2010).
A common feature of most multiobjective metaheuristics is to approximate the efficient set (Carazo et al., 2010;Rabbani et al., 2010;Yu et al., 2012;Tofighian and Naderi, 2015), regardless of the DMs' preferences.However, in a real-world case, the DMs may not be able to find their preferred solution within this set and thus be unable to make a final decision.In these situations, interactive procedures are a good alternative because they can gradually lead to the areas of the efficient set that may be more attractive to the DM.Thus, computational effort can be reduced and the drawback of having to choose a solution from a set that is too large is solved.However, few studies have combined interactive procedures with multiobjective metaheuristics (see Branke et al., 2008) compared to the extensive use of these procedures with traditional or multiobjective exact methods.Among interactive schemes, the reference point-based approach has become one of the most widely used because of the intuitive way in which the DM inputs the requested information (Stewart, 2016).Within this framework, Molina et al. (2009) used an interactive scheme based on the g-dominance concept that can be coupled to any multiobjective metaheuristic, whether evolutionary or not.
In this work, we adapted this interactive scheme based on g-dominance to be more generic and included several DMs.Also, we describe a software tool known as the Project Portfolio Selection Tool (PPST) that assists in the process of selecting and scheduling project portfolios taking the preferences of a DM or several DMs into account.This software tool uses the SS-PPS algorithm as the search engine and allows DMs (or users) to easily input and/or modify data step-by-step.PPST consists of two modules: a solution module that approximates the set of efficient portfolios, and a second module that supports DMs in exploring the solution space in order to find the preferred solution, if they so wish.Hence, our paper contributes to the current stand of literature, on the one hand, by adapting the g-dominance interactive scheme to be more generic (including several DMs) and better fit the requirements of a Project Portfolio Selection Problem.On the other hand, we have designed an interactive tool, embedded into a metaheuristic algorithm tested by Carazo et al. (2010), in order to provide the necessary tools to implement interaction with the DMs into a real case.Consequently, the advance from the previous work is to provide a tool to incorporate the decision makers´ preferences in the resolution process, helping to explore the efficient portfolios and learning about the situation under consideration.
The article is structured as follows: Section 2 describes the mathematical formulation of the model, the techniques used to obtain a final solution, and the software implemented.Section 3 presents a case study based on a real situation in which DMs from a state university decided which project portfolio would satisfy their needs for a given period, and demonstrates how the software supported the DMs in identifying the best solution.Section 4 presents the main conclusions.
Problem Formulation and Resolution Tool
In the following, the mathematical model, the solution technique and the software used are described.
Mathematical model
This section presents a multiobjective binary model for selecting and scheduling project portfolios from a set of I candidate projects within a fixed planning horizon divided into T periods.It is assumed that a project cannot be partially funded, although different versions of a single project can be addressed on the condition that each version is treated as an indivisible proposal.Thus, the model's decision variable t i x , takes the value 1 if project i (i = 1,…, I) starts at time t (t = 1,…, T), and is equal to 0 otherwise.A project portfolio (group of selected projects and their start time) will be denoted by x1,T, x2,1,…, x2,T,…, xI,1,…, xI,T).
The aim is to select the best portfolios according to a set of objectives (value/utility, risk, cash-flows, etc) that the DMs consider important and multiple constraints (e.g.available resources, strategic or political requirements), which can be different at each period of the planning horizon.S: Overall number of synergy relationships, where the first s synergies affect the objective functions and the rest, S-s, affects the resource constraints.i: Index for projects, i = 1, 2,…, I. k: Index for time periods, k = 1, 2,…, T. t: Index for time period in which a project could start, t = 1, 2,…, T. q: Index for objective functions, .u: Index for resource categories, u = 1, 2,….U.j: Index for synergy relationships, j = 1,2,…., s, s+1,…S.Aj: The subset of projects corresponding to the synergy j.Mj, mj: Maximum and minimum number of projects in Aj to be simultaneously executed to activate synergy j. di: Duration of project i. fq,k: ci,q,k+1-t : Individual contribution of project i in its execution time k+1-t in period k, for objective function q.
wq,k: Weight assigned to the q-th objective function ri,u,k+1-t : Individual consumption of project i in its execution time k+1-t in period k, for available resource category u.Ru,k: Total amount of available resource in category u for period k. aj,q,k (aj,u,k): Additional contribution to the objective function q (resource u) in period k if the synergy j is activated.vu,k: parameter that indicates if the resources not spent in a given period are transferred to the following one (vu,k= 1) or not (vu,k = 0).n: index for linear constraints on the portfolio projects in period k, n = 1, 2,….: lower and upper bounds, respectively, for linear constraint z (z = 1, 2,…).CLi: Lower bound, equal to one or zero, depending on whether the i-th project is mandatory or not.αi, i: Upper and lower bound for the starting time of certain projects belonging to the E subset of the total projects.L: Subset of projects which are affected by precedence relationships.l: Index for projects in L. Pl: set of precursor projects of a project l. hi,l, Hil: The minimum and maximum number of time periods which have to elapse before the starting time of the project l.An important issue to be taken into consideration is that we may find interdependencies between the candidate projects.For example, when projects share resources, and the execution of two or more interrelated projects require fewer resources than if implemented separately, or when the benefits of one project are partially increased due to the simultaneous execution of another.These effects mean that the best individual projects do not necessarily make the best portfolio.Consequently, portfolio performance should be calculated considering the individual project's contribution plus (or minus) the extra contribution due to the interdependence relations (synergistic effects) when they exist.
In our model, to formalize these interdependences, on the one hand, the organization has to specify: - The minimum (mj) and maximum (Mj) number of projects in Aj, which have to be active in period k for activating the synergy j. -The extra value for the q-th objective function (aj,q,k) or the u-th resource constraint (aj,u,k).
Without loss of generality, we will assume that the first s synergies affect the objective functions and the remaining synergies (S -s) the resource constraints.These extra values represent additional benefits or shared costs by the projects in Aj, and we suppose that they do not depend on the execution time in which each project is in k.This may be restrictive in certain situations, and future research may focus on this issue.
Note that the classical types of projects' interdependencies can be modelled as special cases with mj = Mj = 2 and the subset Aj formed by only two projects.
On the other hand, these interactions are linked with the binary variables yj,k.These variables must detect if the synergy j takes effect (yj,k = 1) or not (yj,k = 0) for assessing a portfolio.Moreover, technical constraints will be introduced in the model for assuring this fact.Thus, portfolio performance should be calculated by taking into account the individual contribution of each project (ci,q,k+1-t for objective function q and ri,u,k+1-t for available resource u) and extra contributions (aj,q,k for objective function q and aj,u,k for available resource u) due to potential interdependencies.It is worth noting that both terms depend on the specific period each project i is in k.If project i starts in period t and lasts di periods, then this project would be at execution time k+1-t in period k.
Given the foregoing, the general model used in our decision-support tool was based on an improved version of the model presented in Carazo et al. (2010) and features the following mathematical formulation: Subject to: The objective functions are defined by (1).There are two blocks of functions.
On the one hand, Q ˆ objective functions (sales, risk, etc…) in each period k (k = 1, 2,…, T) of the planning horizon, noted by . Thus, in this block, there can be as many objective functions as combinations of attributes and periods in the time horizon objective functions that represent the weighted aggregated value of some other attributes at different periods (cash-flow,…), noted by In this case, if an attribute has an economic value (it is expressed in monetary units), its weight could, for example, represent a discount factor if one wishes to take the time value of money into account (for example, net present value of cash-flow).In any way, we assume that these weights ( k q w , ) are given by the decision maker according to his/her preferences.Thus, in this block, there can be as many objective functions as temporally aggregate attributes ).( Q Q Note that we have used the same sub-index q in both blocks, but the variation range is different and the attributes represented are different too. The feasible portfolios are defined by equations (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12).Expressions (2-6) are temporal constraints (for each period k) and expressions (7)(8)(9)(10)(11) are global constraints (independent of the period).Constraints (2) on available resources (workforce, machine hours, funds, etc) have the same structure as the objective functions fqk (x), as mentioned previously, but the interaction sets Aj are different, and now j varies from s+1 to S. Furthermore, some unused resources (e.g., funds) in a period can be transferred to the next one, using the corresponding interest rate (Rateu(k)).Thus, the parameter vu,k takes the value (1+ Rateu(k)) for the resources that can be transferred from period k-1 and 0 otherwise.Note that vu,1 = 0.
Inequalities (3)(4)(5) are technical constraints related to the variable yj,k linked to the interaction j.According to (5), if the number of active projects of set Aj in period k is at least mj ( m k j y , = 1) and at most equal to Mj ( M k j y , = 1), the interaction j becomes activated and yj,k = 1.Constraints ( 6) and ( 7) are linear.The first block includes the limitations that the organization wishes to impose on the active projects in each period k but which do not depend on their internal execution time (timing).The second block (7) is independent of the period.For example, let us specify that different versions of the same project or similar projects cannot be part of the same portfolio.
Expression (8) establishes that each project, if selected, can only start once.Furthermore, this restriction allows the DMs to establish project i as a mandatory choice by requiring that the parameter CLi be equal to 1. Constraint (9) establishes an upper bound (i) and a lower bound (i) for the starting time of certain projects pertaining to the E subset of the total project set.
The last two inequalities formalize precedence relationships between projects.Inequality (10) specifies that a project l cannot be selected unless its precursors (Pl) have already been selected, and (11) requires that project l cannot start until a number of fixed periods of time have elapsed, at least (hi) and at most (Hi), since its precursors began.L I denotes the subset of projects that are affected by precedence relationships.Note that (11) could be transformed into the following linear constraint: It would be interesting in future research to thoroughly explore whether this change could make the entire model easier to solve.
To summarize, the model described by equations ( 1)-( 10) is a nonlinear multi-objective binary programming model that facilitates the selection of efficient portfolios in line with the set of objectives pursued by the organization, as well as their scheduling with regard to the optimum time to launch each project within the portfolio.
Solution technique
As mentioned in the introduction, the complexity of the formulated model forces the use of a metaheuristic instead of an exact method, as occurs in many real-world problems.In particular, evolutionary methods are among the most widely used methods in metaheuristics multiobjective programming because the set of efficient solutions can be computed in a single run.Nevertheless, the DMs' preferences have to be included to select a solution within the set.Few studies have addressed this issue in a setting in which metaheuristics are used.We apply an evolutionary method called Scatter Search for Project Portfolio Selection (SS-PPS) (Carazo et al., 2010) and, as an innovation, propose to combine it with an interactive scheme which gradually leads to the areas of the efficient set that are more attractive to the DMs.Thus, on the one hand, computational effort can be reduced and, on the other, the DMs are assisted in choosing the best solution according to their preferences.
We use the interactive procedure based on g-dominance proposed by Molina et al. (2009), but extend it to the case in which there are several DMs.The main aim of this method is to reduce the size of the approximation of the efficient frontier, using the information iteratively supplied by one or more DM.Each DM must provide the values they desire for each objective or criteria (i.e., a reference point g).The iterative process continues until the DMs reach the best compromise regarding the best solution according to their preferences.
The main aim in methods that use a reference point to include preferences is to project this point on the efficient frontier to obtain the efficient point closest to this reference point.However, the g-dominance scheme provides an approximation of the efficient frontier (a set of efficient solutions) in the area where this projected solution lies, rather than a single efficient solution.This is illustrated in Figures 1 and 2, in which the areas of interest on the frontier (g-efficient points) are highlighted in red, for a hypothetical biobjective minimization example.We use this g-dominance scheme in an interactive framework for one or more DMs, who will be iteratively guided to their best solution.
When there is a single DM, the DM is shown a representative subset of efficient portfolios at each iteration h, according to the offered reference point g h .This set is denoted by RS h .If the DM is not satisfied with any of these solutions, he/she can either modify the reference point in order to refine the preferences or select a solution p (x ph ) in RS h to focus the search in this area.In the latter case, a new reference point g h+1 is calculated as a convex linear combination of the selected solution and the current reference point When there are several DMs, all of their preferences have to be combined to obtain a solution acceptable to all of them.
In this case, we follow the classic scheme of using the reference point based on scalarized achievement functions (Miettinen, 1999).According to this scheme, given a reference point and a distance L, the efficient portfolio that best matches the DMs' preferences is computed.This portfolio is the efficient point (i.e. the most preferred solution [PS]) that minimizes the distance L to the reference point.However, we have several reference points and therefore several preferred portfolios.Thus, we have to modify the g-dominance scheme to find a reference point g that represents a compromise for each of the DMs.We attempt to find the ideal reference point g of all the most preferred solutions calculated.That is, if we have N DMs for a problem with Q objectives, and we have obtained N preferred solutions PS1 h , ..., PSN h at iteration h, then g h is the reference point whose set of g-efficient points is the smallest set that includes the N preferred solutions, i.e., .Once this point is calculated (see Figure 3), we solve the problem again as if there was a single DM, but using this ideal reference point g h to generate a new representative set of g-efficient solutions (RS) that will be shown to each DM.Thus, each DM can evaluate how far from the consensus area his/her preferences are and assess how to modify his/her reference point at the next iteration to achieve consensus with the other DMs or influence them.12: END DO
Project Portfolio Selection Tool (PPST)
The multiobjective binary nonlinear model presented in Section 2.1 and the interactive procedure shown in Section 2.2 have been embedded in a new software tool implemented by the authors called PPST.This tool has been designed to be used into a Windows environment and, as mentioned in the introduction, uses the SS-PPS heuristic procedure as a search engine, given its demonstrated competitive problemsolving abilities (Carazo et al., 2010).Then, regarding codification of the solutions, we followed the scheme in the original code, in which they were explicitly defined.Therefore, each solution is represented by two arrays: -] ,..., , [ The PPST starts with a Splash screen in which the user can Create a new data file or Open an existing data file, by clicking on the File menu.In the first case, the user has to input the general parameters (number of projects, the number of periods in the time horizon, the number of objectives and constraints, etc).Figure 4 * shows the general data of the empirical real case described in Section 3, and which we use to demonstrate how PPST works in practice.Then, in a sequence of screens, the user provides further parameter values, such as the duration of each project, its contribution to each objective, synergies between certain projects, and so on.For example, Figure 5 shows the duration of project 6 and the temporal contribution of this project to the objectives under consideration (coefficients ci,q,k+1-t) or constraints (coefficients ri,u,k+1-t), as shown in Section 2, which have been input by the user.Once the final file includes all the data pertinent to the problem, the user can save it as an Office Excel format file, indicating its name and location.The editing process is similar to the previous one.
Once data have been introduced, the user begins the solution process by clicking on the Instance > Solve command and a new screen appears in which the user chooses the search intensity for the solution.There are four options: Very High, High, Medium, and Low.The lower the search intensity, the shorter the computational time; however, the solutions obtained may be further from the real efficient frontier.At this point it is important to emphasize that this option is used just for a first approximation to the whole efficient frontier, so as to be able to start the interactive procedure.Later, when interactive procedure is applied, new and better solutions are found but only in areas of interest for DMs.The key issue is to deal with the computational burden of solving a complex problem whose computational time may vary from seconds to hours.
If the interactions cannot be efficiently performed because the time needed to compute the initial solutions to show to the DMs is excessive, then the problem could be solved off-line using the option Very High in the menu.With this option, a first and in-depth resolution will be implemented without the presence of the DM.After this possibly long resolution process, the solutions obtained are loaded into the module Solutions and the interaction process is restarted, but using these good solutions obtained as a starting point, as well as in a more reduced area of the Pareto set.This drastically reduces the computational time of each new interaction with the DMs, even in large problems, and allows them to interact without having to wait hours for a solution.This is, within this option the algorithm could consume time as needed to compute a set of initial solutions of high quality, which will be used as seeds for the interactive procedure, so that, once good solutions have been obtained, the subsequent iterations, focused on a precise area and already endowed with high quality solutions, will consume quite less computational time.
The user clicks on Solutions> Interactive> One decisor (or Several decisors) to include the information for the interactive process, either by changing the reference point or by selecting a representative solution from those shown to him/her.The process repeats until the DM (or DMs) is satisfied with one of the solutions.
When several DMs are involved, the software asks for the initial reference points (ge), which are used to compute the most preferred solution for each DM (PSe) and the ideal point g that contains the best value for the solutions generated for each objective (see Section 2.2).This ideal point will be used as a common reference point to conduct the interactive procedure.In addition, the DMs are provided with a sample of efficient portfolios at each iteration.This set, called RS h in Section 2.2, includes the best portfolio for each objective and the compromise portfolios, which show the range of diversity at the efficient frontier.Thus, as mentioned in Section 2.2, each DM can refine his/her preferences by modifying his/her current reference point or by choosing a solution from the offered sample.It is worth noting that after each iteration during the interactive procedure, the output can be saved in an Excel file containing the efficient portfolios and the values for each objective function.The software also provides the user with a graph of the temporal distribution of each efficient portfolio, additional information on the money spent in each period, and the active synergies.
Thus, PPST software is a user-friendly tool for project portfolio selection and scheduling problems (with multiple objectives, constraints, synergies….) that incorporates the DMs' preferences to obtain a compromise solution.It can be used as a practical tool in which the managers can make decisions rationally.
A practical approach to the problem: an empirical application
The decision-making process for selecting and scheduling a suitable project portfolio occurs in public organizations (Mavrotas et al., 2006;Medaglia et al., 2008, Arratia et al., 2016;Cruz et al., 2014) and private ones (Stummer and Heidenberger, 2003;Ding and Cao, 2008).There are some differences between them, such as the methods used for pricing and evaluating performance, and the goals and priorities to be achieved, but this does not affect the formal model.Therefore, a general framework can be used for the decision-making process, regardless of the type of organization.
In this section, we illustrate the potential of the proposed approach by applying it to a real case in a Spanish state university.The university's Department for the Strategic Planning of Infrastructure (DSPI) needs to planat the beginning of the period and for a given time horizonthe budget available to fund a project portfolio by choosing among all the candidate projects.The aim is to choose the best project portfolio to implement, given the resources available and to simultaneously decide when to start each project in the portfolio while taking into account the organization's needs, objectives and priorities, and several strategic and political constraints.The university's current selection system starts with the proposals made by the different university departments according to their needs in the areas they represent.These proposals are analyzed by the DSPI management team, which decides which project portfolio to fund and implement.This selection is based on the DMs' experience and does not use any mathematical model.We refer to this scheme as the Traditional selection system.The disadvantage of this system is that DMs have to manage a large amount of information: multiple objectives, restrictions, interdependencies between the projects, and a planning horizon.These issues make it very difficult to obtain efficient solutions without using a structured and formal tool, and can lead to suboptimal solutions.
The alternative approach involves selecting projects by using software based on a mathematical model.We refer to this method as the Technical selection system.In this case, the DMs have to specify the following data in advance: the assessment and availability of resources, the resource requirements for each candidate project, and the evaluation of each candidate project according to the objectives set.The problem with this selection system is that the DMs do not directly participate in it.They only offer technical information to solve the corresponding multiobjective problem.The DMs then have to select a final solution from the set of efficient portfolios obtained, which is not easy because this set is normally very large.
The advantages and disadvantages of these two selection systems have led us to propose a Mixed selection system that adds an interactive process to the Technical selection system.This has all the advantages of the Technical systemthe selection is supported by a mathematical model solved with a software application-while preserving the main feature of a Traditional systemit generates a final solution taking all the DMs' preferences into consideration.As mentioned, PPST software is a tool for generating an approximation of the set of efficient portfolios and assisting DMs in finding the solution yielding the best compromise among the objectives through an interactive process to reach the ideal solution for all decision-makers.We tested this system with the cooperation of three members of the DSPI management team.They were asked to indicate their preferences regarding the efficient solutions previously obtained.In order to use PPST, the DMs had to provide desirable levels for each of the objectives of the problem.
The objectives considered by the organization are as follows: -Maximize the positive impact (f1).The DSPI management team assigned a score to each project based on the percentage of individuals who would benefit from it.The scores range from 1 to 10, where 1 is the least positive impact and 10 the most positive.These scores only affect the first execution period of each project selected (see Figure 5; Section 2.3).
-Minimize the risk (f2) † .The management team assigned risk scores to each project on a scale of 0-10, where 0 represents the absence of risk, and 10 maximum risk.This estimation was based on possible factors that could affect the correct execution of each project (delays in project execution, etc).These scores only affect the first execution period of each project selected, as above (see Figure 5; Section 2.3).
-Maximize the number of active projects in at least three periods within the time horizon (f3).The aim is to avoid choosing small projects on the basis of them being shorter and less expensive.The score allotted to each project is equal to 1 if the project is active for at least three periods within the time horizon, and 0 otherwise.
• The simultaneous implementation of project 45 and project 33 yields a cost reduction of 3.125×10 4 € when they are executed in the 3 rd or 4 th period.
• The simultaneous implementation of at least two of the projects {8, 22, 25} yields a cost reduction of 2.766×10 4 € for the respective period, when they are executed in the 1 st , 2 nd , 3 rd or 4 th period.
Analysis, numerical results, and comparison of the solution reached by both selection systems
Using the Traditional selection system, the real solution adopted by the university department included 38 of the 52 candidate projects, as follows: Figure 6 shows the temporal distribution of the project portfolio chosen, the assessment of the objectives, the number of synergies, and the money spent in each period.This chart depicts time along the X-axis: the vertical dashed black line represents the time horizon in this problem (four semesters).The boxes containing the selected projects are on the Y-axis.Each box shows the projects to be carried out, and when they have to start and finish.For example, project 6 begins in the first period and finishes at the end of the fourth semester because it lasts four semesters.Similarly, projects {16, 18} begin in the third period and finish at the end of the fifth period, i.e., these projects would not finish within the planning horizon and would thus leave a committed budget for the next planning horizon.
By means of this selection system, the university's DSPI committed a total of 649.94×10 4 €. to future projects.This amount was calculated as the difference between the budget required to complete the execution of the projects that had started (689.34×10 4 €) and the resources that had not been spent in the last period of the planning horizon (39.40×10 4 €).
To use the Mixed selection system, we introduced the data of the 52 candidate projects into the PPST software.We solved the problem using the Instance>Solve command and a set of 378 efficient Pareto portfolios were identified in 7932 seconds (s.).As regards the parameters used in the algorithm, all of them were set as suggested by the authors.The size of the reference set was set to 8 points, the size of the tabu reference set was set to 50 points, the number of iterations of the tabu searches was 3900 and the maximum number of searches without improvement was set to 4. Subsequently, we had a meeting with three DMs from the DSPI management team to conduct the interactive process.As the problem was already solved, the solutions obtained were loaded in order to start the interaction with the DMs.At this stage, the procedure required information from the DMs about their desired level (i.e., reference points), for each of the objectives of the problem.In the lower right box in Figure 7, the row labelled "Optimum" shows the optimal values of each objective (ideal point) that closely match the values on the rows "ge" (reference point of DMe, e = 1, 2, 3) during this first stage and, consequently, the solution preferred by each DM (PSe, e = 1, 2, 3) is the same.In the lower left box, under the heading "Solutions", there is a sample of eight efficient solutions with their respective attribute values.This is a representative sample of the efficient frontier (set RS), where extreme and intermediate solutions are shown, as mentioned in Section 2.3.For example, the solution with the greatest value for function f1 (Impact), Sol.1, is the one with greater risk (f2).However, the solution with the least risk, Sol.2, is the one with the least impact (f1).In the third solution, the number of projects that last longer than 3 periods (objective function f3) is 5, which is the highest value for this objective because its value ranges between 0 and 5 along the efficient frontier.It is evident that there is an important trade-off between the first two objectives.Choosing, for example, Sol.2 instead of Sol.1 leads to a 67.4% deterioration in f1 and a 64.4% improvement in f2.
Using this data, the PPST software obtained a new group of 83 efficient project portfolios (81 s.).A new sample of eight solutions was shown to the DMs and they decided to continue the interaction process by determining new reference points.Table 1 contains the solutions shown on-screen after this first iteration.Once again, the solution (Sol.1) with the greatest impact (f1) involves the highest risk (f2) and the solution (Sol.2) with the least risk (f2) has the least impact (f1).However, the trade-off between these two objectives is lower.Choosing Sol.2 instead of Sol.1 leads to a 29.2% deterioration in impact and a 41.9% improvement in risk.Given that the number of solutions remained very high and a suitable solution from the sample presented was not found, the DMs decided to continue the interactions and set new values.DM1 reduced the desired value of the impact level of the projects on the organization (f1) to 295 and the risk level (f2) to 192; DM2 was more permissive with the risk level (f2) and increased its value to 135 and also increased the value of f1 to 260; and DM3 reduced f2 and f3 to 185 and 4, respectively.The distance between these points was 0.5.
After this iteration, 65 efficient portfolios (79 s.) were obtained.The DMs were still not satisfied with the solutions found, and performed two more interactions.Table 2 shows the desired values for each objective in the third and fourth iteration.The distance between reference points was 0.5 and 0.25 respectively.After the fourth iteration, the DMs analyzed the Excel file that contained the last set of efficient portfolios (Figure 8) and noticed that there were five portfolios dominating the solution shown in Figure 6.These portfolios were Sol.1, Sol.4,Sol.6, Sol.11, and Sol.16.At the end, they selected Sol.1 even though Sol.4 has higher values for all the objectives considered.Sol.1, which has the same impact (f1 = 291) as the portfolio adopted by the university using the Traditional system (Figure 6), has a lower risk and a greater number of projects that last more than 3 periods.Figure 9 shows the graph obtained by PPST for Sol.1.This graph is similar to the one shown in Figure 6, but the projects and values are different.
Unlike the Traditional selection system, no money was committed to the future planning periods using the Mixed selection system because Sol.1 leaves a surplus of 590.19×10 4 € in the fourth period.This is enough to cover the completion of the projects that had not been completed in period 4, i.e., projects 15, 40, 44, 49, 50, 51, and 52 for the fifth period, and projects 14 and 18 for the fifth and sixth periods.Thus, if the university decides to continue with these projects, the expenditures would be 314.55×10 4 € for the fifth period and 193.64×10 4 € for the sixth.Given that the remainder in the fourth period was 590.19×10 4 €, there is surplus of 82×10 4 €.
To test our approach, the result obtained (Figure 9) was compared to the solution initially adopted by DSPI (Figure 6).The following differences were found: 1.Our solution not only dominates the solution initially adopted by the university, but also presents better values in two of the three objectives under consideration: University: f1 = 291; f2 = 165; f3 = 3; Our solution: f1 = 291; f2 =161.75;f3 = 4 2. The project portfolio presented by Sol. 1 is similar to that obtained using the Traditional selection system, but it differs regarding the number of projects and the schedule.It includes additional projects (projects 2, 4, 5, 38, and 40), but excludes projects 16 and 19.Thus, with the same budget, our solution allows us to execute more projects (41 compared to 38).
3. An additional advantage: the solution reached with the Mixed selection system was more conservative than that obtained with the Traditional selection system.The Mixed selection system does not commit resources to future periods (80 ×10 4 € surplus), whereas the Traditional solution commits a budget of 558.86 ×10 4 € to future planning periods.
Thus, the PPST can help DMs to iteratively refine their preferences, explore the efficient frontier, and gain insight into the situation they are addressing.The PPST does not involve the DMs in excessive cognitive burden, because it only asks them to give the desired value for each objective, once they know the range of objective functions.Under the g-dominance approach, these individual preferences are used to establish a frontier area, rather than being aggregated to obtain a unique consensual solution.This area represents the smallest set of efficient solutions that not only includes the solution preferred by each DM, but also the intermediate solutions that represent a compromise between all the individual solutions chosen by each DM.Once the most satisfactory of the efficient frontier is located, the DMs can focus on it and more deeply analyze the alternatives to reach a joint decision.Thus, although the final set of efficient portfolios depends on the reference points established by the DMs in successive stages, minor modifications to these points will lead to very similar regions of the efficient frontier.
The analysis was presented to the real decision makers and the feedback was that they have now become more conscious of the importance of using a quantitative tool for selecting and scheduling project portfolio.They found the tool intuitive and readily applicable and highlighted that they had learned more about the problem, the possibility of knowing how their preferences affect the final set reached, and the fact that it was possible to find an efficient portfolio that was better than the one previously chosen.
In order to analyse the influence of the reference points in the final set of efficient portfolios, we considered different scenarios in the interactive process shown in this section.
Firstly, we assume that during the iterative process, the DMs adhere to their preferred value for each of the functions they are interested in.Thus, DM1 does not change the value of 300 for the impact of the portfolio (f1), DM2 does not change the value of 100 for risk (f2), and DM3 does not change the value of 5 for the number of projects in the portfolio lasting more than three periods (f3).In this hypothetical case, named "All", a total of 66 solutions are obtained and 13 of them match those obtained in the initial process (Figure 8).Furthermore, if all the other solutions obtained are carefully analysed, it can be seen that there is a great range of variations in f1 and f2, which is only to be expected since the DMs do not intend to give up their initial choices.This process leads to a final set with a greater diversity of solutions and a greater range of variations in the first two objective functions.Figure 10 shows the distribution of the values obtained in each of the two objective functions in the initial scenario and the alternative scenario.In the "All" scenario the distance between the preference points obtained in the last iteration is 0.43. ) does not make changes to his/her initial value for f1 or f2 or f3, respectively, during the entire process.The 3 potential scenarios are denominated SD1, SD2, and SD3, respectively.In SD1, a final set of 31 solutions is obtained with the first two objective functions distributed as shown in Figure 10.In this case, the new set of solutions includes 18 matches with those obtained at the end of the original process, and the rest generally have higher values in the first objective function due to the high value set by DM1 for f1.In the case of SD2, the final set obtained contains 62 solutions, of which 10 match the original final set, whereas the remainder mainly consist of solutions with a lower value for the second objective function, due to the fact that DM2 preferred to set a low value for this objective.Finally, if DM3 does not change his/her initial preference for the third objective function (SD3), then a final set consisting of 19 solutions is obtained, 17 of which are the same as those obtained in the original set.Thus, the distribution of these solutions has very similar core and dispersion values to those found in the original set (see Figure 10).
The distances obtained between the reference points in the last iteration for each of the last three scenarios are 0.25, 0.43, and 0.4, respectively.These results, when added to information previously described, show that DM2 is the DM who makes more concessions during the original process.If DM2 had been less willing to reach a consensus, he/she would have caused a major change in the set of solutions finally obtained.However, even in this case, the solution set includes solutions obtained in the original set, while providing a wider view of the approximate efficient frontier.In particular, Sol.4 (original final set), which improved all the objectives of the solution finally adopted by the university, appeared in all the final sets obtained in the different scenarios.
Given the foregoing, it follows that the zone of the efficient frontier finally obtained will vary according to the positions of the DMs in such a way that the further the DMs' preferences are from each other the greater the zone of the efficient frontier will be, without entailing a dramatic change in the results.
Thus, the PPST tool helps DMs to identify the efficient frontier area on which they should focus their attention, since it contains the solutions preferred by each DM and the other intermediate solutions.
Political and legal issues or other aspects (resources committed to subsequent periods, the number of projects that compose them, etc) can be addressed to reach a final decision.If this procedure is insufficient to reach consensus, the DMs could use any voting rule or more elaborated procedures, such as those proposed by González-Pachón and Romero (2011).
Conclusions
This article analyzed the theory and practice of selecting and scheduling a project portfolio from a group of candidate projects with a fixed planning horizon.
From the theoretical point of view, we described the three components involved in the process: the mathematical model, the solution technique, and the software.The mathematical model is a nonlinear multiobjective binary programming model that facilitates the selection of efficient portfolios according to the set of objectives pursued by the organization, as well as their scheduling regarding the optimum time to launch each project within the portfolio.The solution technique makes use of a metaheuristic approach rather than an exact method because of the complexity of the model formulated.We also described an interactive procedure based on g-dominance that is used to assist one or several DMs in the task of selecting a project portfolio from the efficient frontier.To make it easy for the DMs to find the solution, the multiobjective binary model and the interactive procedure were embedded into new software, called PPST, which uses the SS-PPS heuristic procedure as a search engine.
In practical terms, we tested an empirical application of the process of project portfolio selection and scheduling in a Spanish university.The software described in the first part of the paper provided a better solution in less time than the solution originally reached by the DSPI.In addition, it allows the DMs to add their preferences during the decision-making process.
The main contribution of this study is that it presents a decision-making model that is integrated with computer-based support.The model provides DMs with quantitative support when they have to make satisfactory decisions on project portfolio selection within a short period.The tool is based on a multi- The parameters, indices and variables used are as follows: I: Total number of available projects.(Number of projects) T: The set of time periods in the planning horizon.(Number of periods) Q: Overall number of objective functions, in which Q ˆ of them depend on the time period.U: Number of resource categories.
:
Coefficient of project i in linear constraint n, if this project is active in period k.
:
lower and upper bounds, respectively, for linear constraint n in period k. z: index for global linear constraints on the portfolio projects, z = 1, 2,… .They do not depend on period k. i z b , : Coefficient of project i in linear constraint z.
Binary variables taking value 1 if project i starts at t, and 0 otherwise.
Binary variables taking value 1 if the number of active projects of Aj, in period k, is at least mj, and 0 otherwise.
Binary variables taking value 1 if the number of active projects of Aj, in period k, is at most Mj, and 0 otherwise.
Binary variable taking value 1 if synergy j is active in k and 0 otherwise.
Figure 3 . 3 :
Figure 3. Calculation of the reference point (g) integrating the preferred solutions (PSe)
Figure 4 .
Figure 4. General Data of the empirical application
Figure 6 .
Figure 6.Selection adopted by the university using the traditional system
Figure 7 .
Figure 7. Screenshot after loading solutions to apply interactive (several dec.)
Figure 8 .
Figure 8.Output after the fourth iteration (several DMs)
Figura 9 .
Figura 9. Schedule of project portfolio Sol.1, using a mixed selection system
Figure 10 .
Figure 10.Box-and-whisker plot of the final efficient points under different scenarios
Figure 10
Figure 10 also shows the other three possible scenarios, in which it is assumed that only one of the DMs (DM1, DM2, or DM3) does not make changes to his/her initial value for f1 or f2 or f3, respectively, during the entire process.The 3 potential scenarios are denominated SD1, SD2, and SD3, respectively.In SD1, a final set of 31 solutions is obtained with the first two objective functions distributed as shown in Figure10.In this case, the new set of solutions includes 18 matches with those obtained at the end of the original process, and the rest generally have higher values in the first objective function due to the high
Table 1 .
Sample of solutions after the first iteration
Table 2 .
Reference points for each decision-maker at the third and fourth iteration | 2019-02-19T14:06:38.179Z | 2017-12-31T00:00:00.000 | {
"year": 2017,
"sha1": "134d99cc60917d9bcab2cc645b3f736a51f31d5e",
"oa_license": "CCBYNC",
"oa_url": "https://doi.org/10.24309/recta.2017.18.2.04",
"oa_status": "GOLD",
"pdf_src": "ScienceParsePlus",
"pdf_hash": "7cb004c34ed9cca9720bd52b4dbf96970be2338a",
"s2fieldsofstudy": [],
"extfieldsofstudy": [
"Computer Science"
]
} |
244451114 | pes2o/s2orc | v3-fos-license | IMPLEMENTING DIGITAL TOOLS TO MOTIVATE GRAPHIC DESIGN STUDENTS AT HVCT TO STUDY ACTIVELY IN THE PERIOD OF COVID19 PANDEMIC
In the period of the Covid 19 pandemic, because of social distancing policy, students were not allowed to go to school in person. In that situation, a Graphic Design class (C18DHO) with 21 students was selected to study online. At the beginning of the course, half of them were not ready to study only as being at home. This article describes an online teaching procedure of 4 specialisms; three of them applied at least 3 digital tools to organise online class activities; the other one only used 2 tools to organise the online class. The author interviewed teachers, conducted surveys of students’ opinions, collected results to analyse and evaluate them. The results showed that: digital tools help teachers manage their students learning effectively and attract them to get involved in the lessons; digital tools motivate Graphic Design students to study actively and gain good learning outcomes. It can be concluded that teaching Graphic Design students online can be successfully deployed if teaching procedure and suggested digital tools are implemented. This way of teaching can be applied to teaching courses that have the same characteristics as Graphic Design courses.
INTRODUCTION
Industrial Revolutions (IRs) has made big changes in human life and manufacturing activities, especially IR 4 (Diwan, 2017). The 4 th IR strongly affects all aspects of life including education, so-called Education 4.0 (Edu 4.0). It requires changes in both teaching and learning mindsets which leads to differences in teaching and learning methods in order to adapt to the new era. In Edu 4.0, digital learning tools can promote deeper learning in which learners can personalize knowledge, be motivated with new experiences, learn more hours; they can make products, improve collaboration and problem-solving skills; they can get supports from numerous teachers on the internet, have more pathways to possess knowledge and skills and everything becomes much faster due to digital technology (VanderArk, 2012). Implementing digital tools in a small group of adolescents in high school can foster literacy practices (Beach, 2002). In 2019, Mei used a qualitative research method to explore how to use digital tools effectively in teaching higher education and she concluded that digital learning tools can really motivate learners. Vocational institutions, in which task-oriented learning seems to be a popular way of learning, digital media and the internet should be taken into consideration (Falk, 2015). However, there is not much research paper on implementing digital tools in teaching vocational students in Vietnam.
College of Technology II (HVCT) -Vietnam, is a vocational institution that trains both college and technical levels. A majority of students in HVCT do not have highachievements in previous classes, are not really active in learning and their collaboration skills are not effective. Typically, vocational schools focus on training vocational skills so 70% of course duration is used for practising and in graphic design class, students mostly practice on graphic design software. In traditional
Journal of Technical Education Science No.62 (02/2021) Ho Chi Minh City University of Technology and Education
classes, it takes time for teachers to provide and collect students' test papers or homework as well as to remind them to do homework. Therefore, digital tools are considered as significant supporters for teachers to save time, effectively manage classes, facilitate teaching and they are especially useful in the period of social distancing. This paper aims to figure out (1) the effectiveness of implementing digital tools in managing students' learning; (2) digital tools motivate Graphic Design students' learning and gain better learning outcomes. It can be concluded that online learning is suitable for Graphic design students in vocational schools.
Participants and sites
Online learning becomes more and more popular and attractive to students because they have more options to choose online learning formats. There are 03 types of online learning to be considered as taking a course: asynchronous, synchronous online learning and blended learning. In asynchronous online learning, there is no time for a class meeting; every knowledge and assignments are provided with due days to be finished. Meanwhile, as taking synchronous online courses, students can participate in their classes from a distance and interact online with teachers In this study, blended learning based on collaborative web conferencing was implemented to teach C18DHO class with 21 students for the reason of social distancing during Coronavirus pandemic. They are good at computers and studying in all four classes: Multimedia technology, Interior design, Web design and Printmaking techniques. Online teaching had not been applied before at the college. Four teachers selected to teach the class online during the pandemic were experienced teachers; however, only three of them were trained to use digital tools. That was the reason why only three classes: Multimedia technology, Interior design, Web design were implemented digital tools to conduct activities in their classes. The author was one of the three trained teachers. The last teacher only used 2 digital tools to organize her online class.
Digital tools selection
Digital pedagogy is the implementation of digital tools in teaching and learning based on pedagogical perspectives. For different pedagogical purposes, different digital teaching tools will be used in order to make sure that they can motivate learners to achieve their best learning outcomes. Bloom's taxonomy is considered as a standard for teachers to design learning objectives accompanied by order thinking. For the purpose of reminding teachers about what they are doing and facilitating teachers in choosing suitable digital tools, Allan Carrington combined Bloom's taxonomy and SAMR model which is developed by Ruben Puentedura to design and develop Pedagogy Wheel. The four teachers selected suitable digital tools based on 5 levels of the cognitive domain in Bloom's Taxonomy to make a new model of digital tools and applied to their class.
Data collection
The four teachers were interviewed about the helpfulness of digital tools in online teaching and the impacts of digital tools on students' attitudes and learning outcomes. Surveys on students' levels of agreement about organizing online courses, applying digital tools in teaching, advantages and disadvantages of studying online and students' levels of satisfaction were also conducted.
❖ Step 1: Course orientation
This step is considered as the first and the most important step because it helps to motivate learning, including 3 following aspects: ⮚ Course introduction: Students were introduced to knowledge, skills and attitudes that the course requires and provides; job opportunities after the course. It helped students clearly understand the significance of the course to their future jobs as well as their lives. Course outcomes also need to be shown in the order that students are aware of how and how much effort they need to put into the course. Moreover, this activity also introduced soft skills that students could accumulate to adapt themselves to era trends.
⮚ Online classroom rules agreement: Teachers and students discussed and reached an agreement on some rules such as attending class frequently (at least 75% of the course duration), completing group and individual tasks to receive bonuses, etc. All the requirements and rules should be clarified at the beginning of the course and strictly followed during the course.
⮚ Online learning tools instructions: The ability to use digital tools is a vital skill to study online. Students need to be trained how to access and work on Google Classrooma learning management system, on which students can get learning materials, exercises and submit their work; and how to join online video conferences and activities on Google Meet.
❖ Step 2: Organizing active learning
With the purpose of improving students' critical thinking, students did not only listen and answered questions but also make questions, discuss in pairs or groups to gain some interesting experience after a class. The benefits of each online classroom activity need to be explained clearly. Selecting suitable digital tools ensured all students have opportunities to show their opinions, ideas and to get teacher's support at the right time.
Checking students' attendance frequently with various digital tools helped to attract student attention and participation.
❖ Step 3: Surveying and assessing students
The results of surveying and assessing students after a lesson or a unit reflected students' work results. Teachers gave comments, evaluations of student attitudes, collaboration, knowledge and skills, and collected students' feedback, the expectation at the same time to make appropriate adjustments for the next class activities. Two or three different digital tools should be used for each activity to create interesting interaction.
Effective learning management
In specialisms of Graphic Design course, there are plenty of tasks for students to practice in each class. In traditional classes, it took a long time to provide exercises, feedback and mark students' work. Other than that, in case students were absent from class or could not finish exercises in class, they may find it difficult to understand the lesson and could not redo the exercises. With the support of Google Classroom, all the teachers participating in teaching online agreed that it was a very convenient tool to send students exercises and lectures, to schedule sending and submitting exercises for individuals or whole class, and to manage students' learning process individually. Teachers saved much time while following and supporting students became much more effective. To students who were absent from class or could not finish exercises, they could easily follow the lessons as well as get and finish exercises via the Google Classroom Reminder system. This platform also reminded students to complete exercises on time. Therefore, the number of completed exercises were increased in comparison to the ones in traditional classes. Google Classroom stores and synthesizes students' learning results as individual profiles so that teachers can easily follow students' learning process so as to encourage and support students to get over difficulties.
Figure 3. Records of learner's performance on assignment in Google Classroom
The four teachers found that hosting an online meeting during the period of online teaching was rather easy with Google Meet. With this tool, teachers delivered their lectures, shared screens and acted as a model; in the meantime, students could also share their screens to present their products. Teachers and students discussed, made and answered questions via a conversational function. The feature of automatically taking attendance was added on Google Meet that helped teachers control students list in realtime. Therefore, with digital tools support, teachers can manage students learning effectively.
More active learning activities
Apart from Google classroom and Google meet, the teachers of Web Design, Interior Design, and Multimedia Technology used 2 to 3 other tools for active learning activities. Not only individual work like answering questions, doing exercises, students can also join group work activities and improve critical thinking such as making presentations, group discussions, critical comments and evaluating other groups' work. Padlet and Trecider are the two useful software to implement those activities easily and effectively. All students' ideas and opinions are stored on one page. Teachers can give comments, assessments on that page and save them as references for students. These activities were really attractive to students and helped them to concentrate on lectures and became more active in class.
All students have opportunities to participate in active learning activities
Kahoot, Mentimeter and PearDeck make interactions to get students involved in the lessons. In a traditional class, teachers can only call a few students to show their opinions because of time limit or students' shyness; whereas in class with digital tools, all students have more opportunities and feel more comfortable to show their opinions then their self-confidence can be built up also. It is also a chance for teachers to note students' opinions to encourage them and facilitate those who are in difficulties. It can be clearly seen that thank to digital tools all students have opportunities to participate all online class activities as well as express their opinions.
Organizing online courses
At the beginning of the course, students have clearly introduced course objectives, course syllabus, course requirement and online classroom rules that help to orient and motivate students. They were also instructed to fluently use Google Classroom and Google Meet to join online classes. Teachers'care, support and active learning activities are important factors to contribute successful online teaching. It can only be done with digital teaching tools. Course requirements and rules were clearly introduced at the beginning of the course. Figure 10. Factors lead to successful online learning
Implementing digital tools in teaching
Students admitted that plenty of active learning activities were implemented in class. Especially, implementing digital tools to make interaction activities in Multimedia Technology, Interior Design and Web Design classes makes students concentrate on lessons and learn better. Figure 11. The level of students' agreement on digital tools helps to be more concentrated on the lessons.
They also believed that they got learning materials, assignments and easily submitted their work everywhere and every time. This feature is particularly useful to those who could not join the class or complete their tasks in class. Communicating with teachers and other students is not a problem as learning online with the support of Google Meet.
Advantages and disadvantages of learning online
According to the survey, the students appreciated learning online for it advantages: saving time for commuting (85.7%); reading materials, watching lectures and doing assignments at any time (71.4%); having opportunities to participate in all online class activities (66.7%)
Figure 15. Advantages of online learning
On the other hand, the students also had to face some difficulties as learning online. Most difficulties were not caused by teachers' pedagogy but mostly by device problems or internet connection that interrupted their learning and somehow caused negative effects on learning outcomes.
Levels of student satisfaction
According to the survey, most students were satisfied with those 4 online courses. Especially the levels of student satisfaction on Multimedia Technology, Interior Design and Web Design were higher than the one of Printmaking Techniques. It could be understood that the courses implementing more digital tools to organize online class activities were appreciated and satisfied. After experiencing pilot online courses, most of C18ĐHO were ready to participate in next online courses. In comparison with half of students who did not want to join online courses at the beginning of the courses, this result is considered a big success of the teachers.
Diligence
Regarding diligence, most students attended classes with acceptable classtime (70% or over) in the three courses using more digital tools. Figure 19.The frequency of students' online class attendance
Learning outcomes
Concerning learning outcomes, analysing each student's final scores in 4 courses, the results show that they all had final scores of more than 5; there were a lot of students gaining high scores. There was one exception -student number 19-his Printmaking Techniques final score was lower than 5 and he did not have scores for Interior Design. It can be explained by his diligence. He attended only about 30% of Printmaking Techniques and Interior Design class time because he had a part-time job at that time. The statistics show that a majority of Printmaking Techniques students' final scores were lower than the ones in the other 3 courses. You are ready to participate in next online courses.
How often did you attend online classes?
Multimedia | 2021-11-21T16:21:40.117Z | 2021-02-28T00:00:00.000 | {
"year": 2021,
"sha1": "30d5bc8779f60ff81ab29e98de23f3d1b17a6bf8",
"oa_license": "CCBYNC",
"oa_url": "https://jte.hcmute.edu.vn/index.php/jte/article/download/82/66",
"oa_status": "HYBRID",
"pdf_src": "Adhoc",
"pdf_hash": "e2cd72ab656c159137228aaeacc712360b376215",
"s2fieldsofstudy": [
"Education"
],
"extfieldsofstudy": []
} |
276422 | pes2o/s2orc | v3-fos-license | Antinociceptive Activity of Aqueous and Alcohol Extract of Evodia Rutaecarpa
Water, methanol and ethanol extracts of Evodia rutaecarpa were tested for antinociceptive activity, which were correlated with the contents of evodiamine, rutaecarpine and evodine. Determination of contents was achieved by chromatographic techniques. Extracts were evaluated for antinociceptive activities using hot-plate test; acetic acid-induced writhing test and formalin test. All three extracts of Evodia rutaecarpa showed antinociceptive activities but the ethanol extract exhibited better effect. The better antinociceptive activity appeared to be related to higher contents of evodiamine, rutaecarpine and evodine in ethanol extract of Evodia rutaecarpa.
Evodia rutaecarpa is commonly used and a "pungent, bitter and hot" natured drug in traditional Chinese medicine (known in Chinese as Wu Zhu Yu).It is used as an analgesic, antiemetic, antiinflammatory, and astringent agent in the Chinese Pharmacopoeia.The main alkaloids and terpene lactones with biological activity have been identified in ER including evodiamine, rutaecarpine and evodine (fig.1).Pharmacological investigations have revealed that different extracts of E. rutaecarpa and its chemical constituents display a number of biological activities related to antinociception, antiinflammation and many more.Matsuda et al. [1] studied the antinociceptive effects of 70% methanol extract of ER, the results suggest that E. rutaecarpa possesses antinociceptive effects and proposed that it is possible to evaluate the quality of E. rutaecarpa using the antinociceptive effect and the content of evodiamine, rutaecarpine and limonin as indexes.Ko et al. [2] .revealed that the ethanol extract of E. rutaecarpa and its four bioactive components all exhibited antiinflammatory activities.The major virtue of E. rutaecarpa according to the herbal literature and books on the traditional Chinese system of medicine is its possible antinociceptive effect.Its antinociceptive effect could possibly be attributed to the presence of high concentration of alkaloids such as evodiamine, rutaecarpine and evodine.
It is normally used as aqueous extract of E. rutaecarpa in Chinese clinical application, but there have been a few reports on the aqueous extract and the antinociceptive activity difference between aqueous extract and alcohol extract remained to be clarified.In this paper we selected three classic pain models to study the antinociceptive effects of water, methanol and ethanol extracts of E. rutaecarpa and compared with contents of evodiamine, rutaecarpine and evodine in the three extracts.The hot-plate test was used to measure response latencies according to the method described by Almeida et al. [3] .with minor modifications.The mice were placed on an YLS-6B hot-plate maintained at 55±0.5° and the time between placement of the mouse on the platform and shaking or licking of the paws or jumping was recorded as the hot-plate latency.Mice with baseline latencies from 8 to 22 s were selected into the study.A significant increase of the latency was considered as indicative of analgesic activity.Twenty-four hours later and 60 min before the test, groups of animals was treated with water, methanol or ethanol extract of E. rutaecarpa 150 or 300 mg/kg p.o., respectively.Another group was given compound aminophenazone and barbital injection (150 mg/kg p.o.) while control animals received the same volume of 0.5% CMC-Na p.o. (20 ml/kg).
Antinociceptive effects by acetic acid-induced writhing test were performed by the method of Koster et al. [4] .Animals were divided into eight groups containing ten animals in each.Group I (normal) mice were fed with standard diet and were administered with an aqueous solution of 0.5% CMC-Na (20 ml/kg p.o.).In Group II, mice were treated with compound aminophenazone and barbital injection (150 mg/kg p.o.).Group III and group IV mice were treated with water extract of E. rutaecarpa 150 and 300 mg/kg p.o., respectively.Group V and VI mice were treated with methanol extract of E. rutaecarpa 150 and 300 mg/kg p.o., respectively.Group VII and VIII mice were treated with ethanol extract of E. rutaecarpa 150 and 300 mg/kg p.o., respectively.All groups were treated once daily for a period of 7 days.Mice were given an intraperitoneal injection with 1% acetic acid (10 ml/kg) 60 min after the last administration.The number of abdominal constrictions produced in these animals was counted cumulatively for 15 min after the injection.Antinociceptive activity, indicated by the reduction in the mean of the number of abdominal constrictions in the test groups compared to the control group, was calculated as percent inhibition of abdominal constrictions (percent of inhibitory level) using the following formula: (mean of (control-test group)/control group×100 %).
The formalin test was carried out as described by Hunskaar and Hole [5] but with slight modifications.The design of grouping and drug administration was the same as acetic acid induced writhing test.Pain was induced by injecting 9 μl of 2% formalin in the subplantar region of the right hind paw 60 min after the last administration.The amount of time that the animal spent licking the injected paw, considered as an indicator of pain, was recorded for duration of 30 min in two phases, known as the early (0-10 min) and late (10-30 min) phases.
Chromatograms obtained from a standard solution and different extracts (fig.2) reveal the selected marker constituents in standard and sample solutions were having same retention time.Spiking the sample solution with the standard compounds was also used to assist confirmation of peak identity.The method enables precise, sensitive and highly accurate quantification of evodiamine, rutaecarpine and evodine (method validation data not shown).
When the method was used for analysis of ER, the contents of evodiamine, rutaecarpine and evodine showed in Table 1.Results revealed that alcohol was a better solvent to extract ER than water, the contents of evodiamine and rutaecarpine were about 20-30 times better extracted by alcohol than water and for evodine was about 1.6 times, respectively.There was no difference between methanol and ethanol.In hot plate test, oral administrations of different E. rutaecarpa extract were ineffective on the reaction to thermal stimuli in mice but compound aminophenazone and barbital injection (150 mg/kg p.o.) could raise the pain threshold (data not shown).These results were consistent with the report of Matsuda et al. [1] .
In acetic acid-induced writhing test, the water extract, methanol extract and ethanol extract (150 and 300 mg/kg, p.o.) demonstrated a significant (P<0.05)antinociceptive activity in the acetic acidinduced writhing test (Table 2) with the percentage of analgesia ranging between 35 to 56%.Though there was no statistical significance among different extracts of E. rutaecarpa, the ethanol extract of
Fig. 3 :
Fig. 3: Effects E. rutaecarpa extracts and compound aminophenazone and barbital injection on paw licking in the formalin test in mice.Effects of E. rutaecarpa and compound aminophenazone and barbital injection on paw-licking divided into (a) early phase, 0-10 min after formalin injection an (b) late phase, 10-30 min after formalin injection.Each column represents the mean±SEM of 10 mice.Significant differences compared with control group, *p<0.05(One way ANOVA followed by Dunnett's multiple comparison tests).AP: Compound aminophenazone and barbital injection, WE: water extract of E. rutaecarpa, ME: methanol extract of E. rutaecarpa and EE: ethanol extract of E. rutaecarpa.b China (lot number 110801-201006, 110800-20040).Compound aminophenazone and barbital Injection was obtained from Tianjin Pharmaceutical Group, Tianjin, China (lot number 1105122) which contained 0.1 g aminophenazone per 2 ml.Methanol for HPLC was obtained from Beijing Dikma Science and Technology Limited Company, Beijing, China.Other analytical grade reagents were used for chemical analysis.
TABLE 2 : EFFECT OF DIFFERENT EXTRACT OF EVODIA RUTAECARPA IN ACETIC ACID-INDUCED WRITHING TEST IN MICE
Values expressed as mean±SEM, n=10 animals in each group.The results were analyzed using One way ANOVA followed by Dunnett's multiple comparison tests.*P<0.05whencompared with normal group.There was no statistical significance among different extracts of Evodia Rutaecarpa the others.It is likely that the E. rutaecarpa was able to produce greater activity due to the presence of higher contents of evodiamine, rutaecarpine and evodine.In conclusion, the results that the extract of E. rutaecarpa exhibited analgesic activities in acetic acid-induced pain model and the phase II of pain model mice induced by formalin, we could deduced that the type of analgesic activity of E. rutaecarpa is peripheral analgesia and the exact analgesic mechanisms of E. rutaecarpa need further study.Our observations confirm that water, methanol and ethanol extracts of E. rutaecarpa all showed antinociceptive activities and ethanol extract exhibited better effect.The better antinociceptive activity was involved with the higher contents of evodiamine, rutaecarpine and evodine in ethanol extract of Evodia rutaecarpa.
TABLE 1 : THE ACTIVE CONTENTS IN DIFFERENT EXTRACTS OF EVODIA RUTAECARPA
Values expressed as mean±SEM using three duplicated assay data.The results were analyzed using One-way ANOVA followed by Dunnett's multiple comparison tests.*P<0.05 when compared with water extract.There was no statistical significance between methanol and ethanol extracts of Evodia Rutaecarpa | 2017-06-17T16:04:19.108Z | 2014-05-01T00:00:00.000 | {
"year": 2014,
"sha1": "e4f49837734a6def2ce61bf9ddac53405847f49a",
"oa_license": "CCBYNCSA",
"oa_url": null,
"oa_status": null,
"pdf_src": "PubMedCentral",
"pdf_hash": "e4f49837734a6def2ce61bf9ddac53405847f49a",
"s2fieldsofstudy": [
"Agricultural And Food Sciences"
],
"extfieldsofstudy": [
"Medicine",
"Chemistry"
]
} |
256016045 | pes2o/s2orc | v3-fos-license | Isolation and characterization of putative functional long terminal repeat retrotransposons in the Pyrus genome
Long terminal repeat (LTR)-retrotransposons constitute 42.4 % of the genome of the ‘Suli’ pear (Pyrus pyrifolia white pear group), implying that retrotransposons have played important roles in Pyrus evolution. Therefore, further analysis of retrotransposons will enhance our understanding of the evolutionary history of Pyrus. We identified 1836 LTR-retrotransposons in the ‘Suli’ pear genome, of which 440 LTR-retrotransposons were predicted to contain at least two of three gene models (gag, integrase and reverse transcriptase). Because these were most likely to be functional transposons, we focused our analyses on this set of 440. Most of the LTR-retrotransposons were estimated to have inserted into the genome less than 2.5 million years ago. Sequence analysis showed that the reverse transcriptase component of the identified LTR-retrotransposons was highly heterogeneous. Analyses of transcripts assembled from RNA-Seq databases of two cultivars of Pyrus species showed that LTR-retrotransposons were expressed in the buds and fruit of Pyrus. A total of 734 coding sequences in the ‘Suli’ genome were disrupted by the identified LTR-retrotransposons. Five high-copy-number LTR-retrotransposon families were identified in Pyrus. These families were rarely found in the genomes of Malus and Prunus, but were distributed extensively in Pyrus and abundance varied between species. We identified potentially functional, full-length LTR-retrotransposons with three gene models in the ‘Suli’ genome. The analysis of RNA-seq data demonstrated that these retrotransposons are expressed in the organs of pears. The differential copy number of LTR-retrotransposon families between Pyrus species suggests that the transposition of retrotransposons is an important evolutionary force driving the genetic divergence of species within the genus.
Background
Repetitive sequences make up a large proportion of plant genomes. Among repetitive sequences are transposable elements [1,2], which are broken into two main classes according to their transposition intermediate: Class I retrotransposons transpose via an RNA intermediate by a "copy and paste" mechanism; and Class II transposons transpose via a DNA intermediate by a "cut and paste" mechanism [2]. LTR-retrotransposons are Class I retrotransposons that have been found in all plant species investigated to date [2][3][4]. These retrotransposons are flanked by LTRs and undergo replicative transposition; thus, their copy numbers increase and occupy a large portion of the genome, especially in higher plants [5][6][7]. For example, retrotransposons make up more than 50 % of the maize and wheat genomes [8,9]. Active LTRretrotransposons increase the size of plant genomes. In Oryza australiensis, a wild relative of rice, transposition of retrotransposons led to a rapid two-fold increase in genome size during the last 3 million years [10], suggesting that rapid amplification of LTR-retrotransposons has played a major evolutionary role in genome expansion. Environmental stress and demethylation have been hypothesized to activate retrotransposons and induce duplication events in the genome [11][12][13]. The retrotransposons isolated from plants appear to be young-less than 5 million years old [14]. Therefore, pathways must exist for the removal of retrotransposons. The rice genome has lost a large number of retrotransposons, corresponding to a rapid reduction in genome size [15].
Retrotransposons can insert within or near transcriptionally active regions and can cause mutations by disrupting genes, altering gene expression levels, or by driving genomic rearrangements [16,17]. Recent evidence indicated that a retrotransposon inserted into a myb-related gene was associated with pigmentation loss in grape [18]. In blood orange, insertion of a retrotransposon upstream of an anthocyanin biosynthesis-related gene caused color formation in its fruit to become colddependent [19]. Retrotransposons display extreme sequence diversity, and there are thousands or even tens of thousands of different retrotransposon families in plants [2,5]. An autonomous retrotransposon is composed of two nearly sister LTR sequences flanked by target site duplications of usually 4-6 bp [1]. The internal region is usually composed of two open reading frames required for replication (in some cases, LTR retrotransposons possess one unique open reading frame, such as Tnt1, Tto1, or Tos17): the pol gene encodes products with the enzymatic functions of a protease (PR), reverse transcriptase (RT) and integrase (INT); and the gag gene encodes structural proteins involved in the maturation and packaging of retrotransposon RNA. Conserved sequence motifs, for example, the primer-binding site and the polypurine tract are also essential for retrotransposon replication. LTR-retrotransposons can be subdivided into the Ty1-copia and the Ty3-gypsy groups based on the order of the domains encoded within pol genes. The order in the Ty3-gypsy group is PR-RT-INT, and that in the Ty1-copia group is PR-INT-RT [2].
The Pyrus L. (pear) is believed to have originated in the Tertiary period in the mountainous regions of western and southwestern China [20]. According to its original distribution area, Pyrus can be divided geographically into two groups: the occidental pear group and the oriental pear group [21]. The major species of oriental pear are native to China [22]. The oriental pear group contains wild pea pears and cultivated species with large fruit. Their evolutionary history is still controversial [23]. Recently, the whole genome of P. pyrifolia Chinese white pear 'Suli' was sequenced. The assembled P. pyrifolia genome consists of 2103 scaffolds with an N50 of 540.8 kb, totaling 512.0 Mb with 194× coverage. Sequencing and assembly revealed that much of the P. pyrifolia genome is retrotransposon-derived [24]; 16.9 and 25.5 % of the genome was reported to be copia and gypsy retrotransposons, respectively. A large number of retrotransposons were also found in other species in the Rosaceae family. For example, retrotransposons accounted for 37.6 and 18.6 % of the genomes of Malus and Prunus species, respectively [25,26]. Jiang et al. (2015) reported that the retrotransposon Ppcr1 was inserted in many loci in the genomes of cultivated Pyrus species, but only in a few loci in the genomes of wild Pyrus species [27]. This suggested that retrotransposons might play a major role in species evolution. Therefore, research on retrotransposons in Pyrus species will be helpful to understand the evolutionary history of Pyrus. Yin et al. (2014) reported that LTR retrotransposons in the Pyrus genome have complex structures [28], and that frequent recombination events followed by transposition of retrotransposons may have played a critical role in the evolution of Pyrus genomes. However, their study did not focus on the various retrotransposon families in Pyrus and their inner structural domains, nor did it involve the copy number of retrotransposon families in different Pyrus species.
In this study, we predicted the LTR-retrotransposons present in the 'Suli' genome, and annotated all LTRretrotransposons with three inner functional domains (RT, INT and GAG) to identify putative functional LTRretrotransposons. LTR-retrotransposons in the 'Suli' genome [24] were extremely divergent [27,28], which made it difficult to analyze every predicted LTR-retrotransposon. Therefore, we focused on conserved LTR-retrotransposon families with a high copy number in 'Suli' genome, and investigated the distribution of these families in different Pyrus species and other closely related species to evaluate the roles of LTR-retrotransposon replication and mutation in the evolution of the Pyrus genome.
Results
Annotation and structure of LTR-retrotransposons in the 'Suli' genome In previous study, a total of 1836 putative full-length LTR-retrotransposons were identified in the 'Suli' genome by LTRharvest. To determine which of these were most likely to be functional, we searched all identified LTR-retrotransposons for the conserved protein domains GAG, INT, and RT. A total of 440 putative LTRretrotransposons (24.0 %) contained at least two domains and were analyzed further. Their positions in the 'Suli' genome and annotation information are listed in Additional file 1: Table S1 and Additional file 2, respectively. According to the order of the RT and INT domains, 373 and 67 retrotransposons belonged to the copia and gypsy groups, respectively (Table 1). Copia-type retrotransposons (average length, 5448 bp) were significantly shorter than gypsy-type retrotransposons (average length, 10,742 bp) (p < 0.01 by t-Test). The average LTR length of copia and gypsy retrotransposons was 374 and 542 bp, respectively.
Transposable elements can affect gene expression by disrupting functional genes or by inserting into the upstream or downstream regulatory regions of genes. We used BLAST to align our 440 conserved domaincontaining LTR-retrotransposons to annotated introns in the 'Suli' genome, and used the Blast2GO annotation tool to assign probable gene ontology (GO) terms. A total of 734 genes aligned to LTR-retrotransposons, suggesting that they were disrupted. Of these, 531 unigenes could be annotated using GO. The unigenes were categorized into three main GO categories: biological process, cellular component, and molecular function (Fig. 1). These putatively disrupted genes were annotated using the NCBI nr database and listed in Additional file 3: Table S2. To further analyze putative retrotransposonassociated gene sequences, we searched 10,000-bp genome regions flanked by the predicted retrotransposons. A total of 2536 sequences were found, of which 1922 unigenes could be annotated using GO (data not shown).
To group the identified retrotransposons into families, we used each identified retrotransposon to conduct BLASTN searches against the whole dataset of 440 LTRretrotransposons (coverage: 80 % and e-value: 10 −5 ). In this initial effort, we identified five LTR-retrotransposon families with high-copy numbers, which we investigated further ( Table 2). BLASTN searches against the Repbase database were conducted to identify conserved repetitive elements in these five families. Similar sequences identified in Repbase and reference sequences in the Pyrus genome are listed in Table 2. The PFAM database has many gene models related to LTR-retrotransposons. In this study, three genes (gag, reverse transcriptase, and integrase) were predicted to be present in high copy numbers, while the other two genes (aspartic protease and RNase H) were infrequently identified in Pyrus Table 1 Characteristics of copia and gypsy putative full-length retrotransposons with more than two gene models identified in Pyrus genome t-Test ** ** ** **means significant difference at the p < 0.01 level (t-Test) Fig. 1 Histogram of gene ontology classifications of sequences disrupted by isolated retrotransposons. Unigenes correspond to three main categories: biological process, cellular component, and molecular function using the present gene models. Based on the predictions of three gene models, we described the structure of the five LTR-retrotransposon families isolated from Pyrus ( Table 2, Additional file 4: Figure S1).
Putative insertion time of LTR-retrotransposons
The insertion time of LTR-retrotransposons was estimated by analyzing the divergence of sister LTRs. We used the molecular clock rate of 1.3 × 10 −8 substitutions per site per year [29]. The insertion time can only be considered as a rough estimate, and only large differences should be considered significant. The divergence between sister LTRs ranged from 0 to 0.076, representing a maximum insertion time of 2.93 MYA. The predicted mean insertion time of the 440 LTRretrotransposons analyzed in this study was 0.42 MYA. The predicted mean insertion time of copia-type LTRretrotransposons was 0.35 MYA, significantly shorter than the predicted insertion time of 0.81 MYA (p < 0.01 by t-Test) for gypsy-type LTR-retrotransposons. Most of the retrotransposons were estimated to have inserted into the genome during the last 2.5 million years (Fig. 2). The peak of retrotransposon mobilization was observed at 0-0.5 MYA, indicating that our predicted retrotransposons were inserted relatively recently.
The mean insertion time of each member of the five LTR-retrotransposon families was estimated to be within the last 1 million years (Additional file 5: Figure S2). One member from Family I was inserted 1.75 MYA. In Families II, III, IV, and V, some members did not show LTR variations, indicating that they were inserted into the genome recently.
Phylogenetic relationships among isolated LTRretrotransposons
The LTR-retrotransposons showed wide variations in their full-length sequences and could not be clustered. To evaluate the relationship among predicted LTRretrotransposons, we used the neighbor-joining method to cluster the translated nucleotide sequences of reverse transcriptase (rt) in our identified LTR-retrotransposons with known TE families (Fig. 3). Both translated copiaand gypsy-type RT sequences clustered into many groups (Fig. 3). Although there was wide divergence among RT sequences, five and three conserved clades of RT sequences were identified among copia and gypsy retrotransposons, respectively. The average divergence of untranslated copia-and gypsy-type rt sequences was 0.64 and 0.55, respectively, indicating high heterogeneities among rt sequences (data not shown). Five rt sequences from each conserved clade of copia retrotransposons were aligned (Additional file 6: Figure S3), and the sequence divergence ranged from 0.068 to 0.691. rt4 and rt5 were similar. For the gypsy retrotransposons, the sequences of rt6, rt7, and rt8 were aligned (Additional file 6: Figure S3), and their sequence divergences were 0.775, 0.898, and 0.98, respectively.
Transcriptional analysis of LTR-retrotransposons in various organs in Pyrus
Two transcriptomes assembled from RNA-Seq datasets were used in this study. A total of 116,182 sequences (62.6 Mb) assembled from 19,878,957 reads collected from buds of 'Suli' (SRX147917) and 36,495 sequences (15.8 Mb) assembled from 452,428,795 reads collected from fruit of P. pyrifolia 'Meirensu' (SAMN03857509-SAMN03857515) were aligned using BLAST to the 440 LTR-retrotransposons that we identified. LTRretrotransposons were transcriptionally active in both (Fig. 4). A total of 266 copia-type and 66 gypsy-type LTR-retrotransposons aligned with transcripts from the bud of 'Suli' and 146 copia-type and 55 gypsy-type LTR-retrotransposons aligned with transcripts from the fruit of 'Meirensu' , indicating that these retrotransposons were expressed (Fig. 4). Because the normalized expression values of individual retrotransposons were very low (data not shown), we only showed the reads per kilobase of gene model per million reads values of eight RT families (rt1-rt8). In fruit of 'Meirensu' , the high transcription level of rt3 were represented.
Distribution of LTR-retrotransposon families among Pyrus species
To determine the exact copy number of LTRretrotransposons, we used the reverse transcriptase gene model to search the database of protein sequences translated from 'Suli' genome data with Hmmer3.0. A total of 8144 copia-type RTs and 3748 gypsy-type RTs were Table S3). We could not calculate the exact copy number of retrotransposons in Pyrus, but the relative copy number could be measured by real-time quantitative PCR (Q-PCR). Analyses of the LTR and inner sequences of five LTR-retrotransposon families showed that all LTR-retrotransposon families were present in all Pyrus species and Malus × domestica, but not in Prunus persica (Fig. 5). Families I and II were found infrequently in Malus genomes and two cultivated pear species (P. pyrifolia and P. ussuriensis), but they were abundant in the genomes of three wild pear species (P. pashia, P. betulaefolia, and P. nivalis). Interestingly, families II, III, and IV in P. elaeagrifolia and P. nivalis, exhibited increased copy number of the inner sequence relative to LTRs of retrotransposons. The copy numbers of family III and V retrotransposons were higher in oriental pears than in occidental pears.
Distribution and duplication of copia and gypsy retrotransposons in Pyrus
Recent evidence showed that a large proportion of retrotransposons were non-functional because of mutations in their protein-coding domains [30]. In this study, we identified predicted LTR-retrotransposons in the 'Suli' genome, and focused on LTR-retrotransposons that had the highest likelihood of being functional based on the presence of annotated inner protein domains. Previously, we identified 1836 retrotransposons by running LTRharvest based on two nearly sister LTR flanking sequences and some conserved sequence motifs [27]. However, the current study showed that only 440 retrotransposons had at least two inner protein domains. This finding suggests that there are very few full-length retrotransposons, and even fewer potentially functional LTRretrotransposons in the Pyrus genome.
In a previous study, copia and gypsy retrotransposons were reported to account for 16.9 and 25.5 % (ratio, 0.66) of the genome of the 'Suli' pear, respectively [24]. However, in the present study, copia and gypsy retrotransposons were estimated to account for 8.8 and 8.0 % (ratio, 1.1) of the genome of the 'Suli' pear, respectively, based on RT gene models. Our predictions focused on the existence of rt gene in LTR retrotransposons, which is essential for retrotransposon transposition. Therefore, the retrotransposons predicted in this study may be functional, suggesting that at least 60 % of retrotransposons in the 'Suli' pear genome lack rt genes, and are therefore unable to replicate. Previous studies have established that lacking rt genes causes many LTR retrotransposons to be non-functional entities within host genomes [30].
High heterogeneity of LTR-retrotransposons in 'Suli' genome
The sequences and sequence length differed significantly among the full-length LTR-retrotransposons from the 'Suli' genome. We analyzed rt sequences to evaluate the diversity of retrotransposons. Our data showed that the average divergence of rt sequences in copia-and gypsyfamily retrotransposons was 0.64 and 0.55, respectively. These findings indicate that the rt sequences from pear are highly heterogeneous (Fig. 3), like those in rice [31], strawberry [32] and masson pine [33]. There could be several reasons for the observed high sequence heterogeneity. First, gene mutation is the major cause of heterogeneity. In recent reports, many retrotransposons were existed in the genome for a long time [31,34]. In this study, some retrotransposons were predicted to exist before the speciation of Pyrus and Malus based on sequence divergence (Fig. 5). The long period since the first retrotransposon insertion events is one potential source of variation. Both active and non-functional retrotransposons would have accumulated mutations over time, giving rise to a highly heterogeneous population [1]. Second, all transposons are integrated into chromosomal DNA. Therefore, mutated retrotransposon sequences, carrying mainly nonsense mutations are heritable, permitting a high degree of heterogeneity of retrotransposons between generations. Third, the high divergence between rt sequences of the LTR-retrotransposons we identified suggests a complex origin. For example, the divergence between rt6 and rt7 and between rt6 and rt8 was 0.898 and 0.98, respectively, suggesting that the origin of these related retrotransposons was complex, rather than from a single source. High sequence heterogeneity is the main obstacle that makes it difficult to classify retrotransposons as copia-or gypsytypes. In this study, we identified five related families of LTR-retrotransposons ( Table 2). The members of each family showed high similarity and were strongly conserved, suggesting that these families have duplicated many times in recent years.
The insertion time of LTR-retrotransposon in 'Suli' genome
The divergence of sister LTR sequences was used to estimate the insertion time of retrotransposons. When an LTR-retrotransposon is inserted into the genome, the similarity of LTR sequences is 100 %. As time passes, mutations occur within the two LTRs, resulting in a larger genetic distance between them. In this study, only putative full-length LTR-retrotransposons were analyzed, and annotation of LTRs was performed by LTRharvest, which is known to be biased toward recent insertions of LTR-retrotransposons. Therefore, only recently inserted LTR-retrotransposons might be identified in our study. Our data showed that the majority of the retrotransposons we identified in the 'Suli' genome were inserted into the genome over the last 2.5 million years (Fig. 2). It was estimated that Pyrus and Malus diverged from each other between 5.4 and 21.5 MYA [24], suggesting that mobilization of these retrotransposons occurred frequently in the evolution of Pyrus species after the divergence of Malus and Pyrus. Within the retrotransposon families, the majority of members of families I-IV were estimated to have inserted into the genome over the last 1 million years (Additional file 5: Figure S2), confirming that these retrotransposons in Pyrus were inserted into the genome only recently.
Transcription of LTR-retrotransposons in pear organs
The expression of LTR-retrotransposons is likely to be silent in plant tissue during normal development. Many retrotransposons are expressed and transposed in protoplasts [35], and some are activated by abiotic stresses [11,36]. In our study, the isolated retrotransposon sequences were aligned against the assembled transcriptomes of 'Suli' pear buds (SRX147917) and 'Meirensu' pear fruit (SAMN03857509-SAMN03857515) using BLAST. The expression of retrotransposons was detected in the fruit and buds of Pyrus cultivars (Fig. 4), which suggested that retrotransposons are expressed in Pyrus organs under normal conditions of growth and development. The expression of retrotransposons is advantageous for replication of retrotransposons, and retrotransposon transposition commonly results in mutation [18,19]. In pear fruit and buds, retrotransposons showed transcriptional activity, which could increase their copy number in the genome. The mutations in buds and seeds could be transmitted to the next generation. The high rates of retrotransposon expression and transposition may contribute to the large proportion of retrotransposons in the Pyrus genome (as high as 42.4 %) [24].
Genetic diversity of LTR-retrotransposons in Pyrus and other close-related genera
Multiple studies support the hypothesis that retrotransposons might be associated with the evolution of plant genomes [7,15]. In Pyrus, we identified 440 full-length LTR-retrotransposons that differed significantly from each other (Fig. 3). Five high copy-number retrotransposon families (four from the copia group and one from the gypsy group) were identified to further analyze the diversity of retrotransposons in Pyrus and other closely related genera. All five LTR-retrotransposon families were detected in six Pyrus species (Fig. 5), among which P. betulaefolia and P. pashia are believed to be the ancestral species in the genus Pyrus [23,37]. The detection of a large number of retrotransposons indicates that these retrotransposons have widely existed in pear species for a long time. However, these five LTRretrotransposon families were rare in Malus, and absent from Prunus (Fig. 5), indicating that they were duplicated and increased their copy number in Pyrus genomes after the differentiation of Pyrus and Malus. Both Malus and Prunus genomes contain a large number of retrotransposons [25], which are likely descended from different families than those found in Pyrus. These results suggest that the evolution of retrotransposons has varied among the different genera in the Rosaceae family.
Retrotransposons have played a major role in changing the size of genomes by either increasing genome size [10] or promoting rapid genomic DNA loss [15]. In Pyrus, the genome size does not vary greatly among species (Additional file 7: Table S3). Therefore, we can estimate the relative copy number of retrotransposon families in different Pyrus species. Our result shows that the copy number of retrotransposon families differs in Pyrus species. For example, P. nivalis, P. pashia and P. betulaefolia have a higher copy number of family I and II LTR-retrotransposons than P. pyrifolia, P. ussuriensis, and P. elaeagrifolia. In addition, P. nivalis has a low copy number of family III and V, implying these families were lost in P. nivalis evolution. The changes in the number of retrotransposon families might cause genetic divergence in Pyrus species. In P. betulaefolia, all five LTRretrotransposon families showed high copy numbers in the genome, indicating that this species has a larger proportion of retrotransposons in the genome than other Pyrus species. Pyrus nivalis and P. elaeagrifolia have a low copy number of the LTR regions of retrotransposons in families II, III and IV. The LTR region of these families might be lost and formed solo LTRs, or this region might have mutated. We inferred that the retrotransposon families have mutated and duplicated highly during the evolution of Pyrus.
Conclusions
We predicted 440 full-length LTR-retrotransposons from the 'Suli' pear genome, and annotated three inner protein domain sequences (GAG, INT, and RT) in retrotransposons, suggesting that the isolated retrotransposons might be functional. The analysis of three RNA-Seq databases of buds and fruit in different Pyrus cultivars showed retrotransposons were still active in pear organs. The isolated retrotransposons were highly heterogeneous. They had existed in Pyrus species for a long time, but have rapidly expanded during the last 2.5 million years after the divergence of Malus and Pyrus. Our results showed that the copy number of retrotransposon families varied among Pyrus species. To our knowledge, this is the first investigation of genetic variation of retrotransposons within the genus Pyrus. These findings support that retrotransposon transposition is an important evolutionary force driving the genetic divergence of species within the genus Pyrus.
Plant materials and DNA extraction
The plant materials used in this study consisted of six Pyrus accessions (two oriental cultivars: P. pyrifolia Chinese white pear 'Suli' and P. ussuriensis 'Balixiang' , two oriental wild species: P. pashia and P. betulaefolia, and two occidental wild species: P. nivalis and P. elaeagrifolia), Malus × domestica 'Fuji' , and Prunus persica 'Hujingmilu'. Genomic DNA was extracted from the young leaves of each specimen using the modified CTAB protocol described by JJ Doyle and JL Doyle [38] The precise concentration of DNA was detected using DNAQF-1KT (Sigma, St Louis, MO, USA). The DNA concentration of each sample was diluted to 1 ng · μl −1 , and 1 μl was used as a template for real-time quantitative PCR analysis.
Identification and annotation of LTR-retrotransposons
In a previous study, 1836 full-length LTR-retrotransposons were mined from the whole-genome data of Pyrus (AJSU00000000) [27]. The details of each retrotransposon were obtained from the output of LTRharvest. All retrotransposons were translated into proteins in all six possible reading frames using an in-house Perl script. All of the copia and gypsy gene models were downloaded from the PFAM database (gag, PF03732; integrase, PF00665; reverse transcriptase, PF00078 and PF07727). Each gene model was used to search all of the proteins translated from retrotransposons with Hmmer3.0 software. To describe the genes around retrotransposons, 10,000 bp upstream and downstream of each LTR-retrotransposon were annotated with the BLAST algorithm using Blas-t2GO, and the results were visualized using the WEGO tool [39]. In the 'Suli' genome, a total of 42,812 coding genes were identified [24], and we searched gene introns isolated from the Pyrus genome to detect genes that were disrupted by retrotransposons.
Phylogenetic analyzes
According to the position of rt in the Hmmer3.0 results, we calculated the start and end of the rt sequences in the assembled 'Suli' genome. An in-house Perl script was used to extract nucleotide sequences from the wholegenome data, and translated them to amino acid sequences. The amino acid sequences of RT in copia and gypsy retrotransposons were aligned with known TE families, including Maximus, Ivana, Ale, Angela, TAR, Bianca in copia elements and Athila, Tat, Tekay, CRM, Reina, Galadriel in gypsy elements separately using ClustalW, and a neighbor-joining tree was constructed based on their genetic distance using Mega 5.2 software [40].
Estimation of insertion time of full-length LTR-retrotransposons
Bioperl scripts were used to automate the process of estimating the time of retrotransposon insertion. The two LTRs of each isolated retrotransposon were first aligned using ClustalW 2.0 [41], and genetic divergence between the two LTRs was estimated using the baseml module of PAML4 [42]. The insertion time (T) was estimated for each LTR-retrotransposon using the formula T = k / 2r, where k is the divergence between two LTRs and r is the substitution rate of 1.3 × 10 −8 substitutions/site/year [29].
Estimation of LTR-retrotransposon copy number by Q-PCR Q-PCR was used to estimate the copy number of retrotransposons in the genome [43]. We aligned five retrotransposon families with the Malus and Prunus genomes using BLAST, and designed Q-PCR primers (Additional file 8: Table S4) in the conserved region of LTR and inner domain using Primer 3 software (http://primer3.ut.ee/). The reaction solution (total volume, 20 μl) consisted of 10.0 μl SYBR Premix Ex Taq (Takara, Shiga, Japan), 0.4 μl each primer (10 μM), 1 μl DNA (1 ng · μl −1 ), and 7.2 μl double distilled water. The reaction, performed on a LightCycler 1.5 instrument (Roche, Mannheim, Germany), started with a preliminary step of 95°C for 30 s followed by 40 cycles of 95°C for 5 s and 60°C for 20 s. A template-free control for each primer pair was set for each run. Three biological replicates were used and three measurements were performed on each replicate. The relative copy number of each sample was calculated using the Ct value [43].
Transcriptional analysis of retrotransposons in various organs/tissues of Pyrus | 2023-01-20T14:15:48.881Z | 2016-01-15T00:00:00.000 | {
"year": 2016,
"sha1": "c34e87f1f2d8fa1734a61484eb4521d9cf32701f",
"oa_license": "CCBY",
"oa_url": "https://doi.org/10.1186/s13100-016-0058-8",
"oa_status": "GOLD",
"pdf_src": "SpringerNature",
"pdf_hash": "c34e87f1f2d8fa1734a61484eb4521d9cf32701f",
"s2fieldsofstudy": [
"Biology"
],
"extfieldsofstudy": []
} |
258875362 | pes2o/s2orc | v3-fos-license | Mindful Coping Power Effects on Children’s Autonomic Nervous System Functioning and Long-Term Behavioral Outcomes
Mindful Coping Power (MCP) was developed to enhance the effects of the Coping Power (CP) preventive intervention on children’s reactive aggression by integrating mindfulness training into CP. In prior pre–post analyses in a randomized trial of 102 children, MCP improved children’s self-reported anger modulation, self-regulation, and embodied awareness relative to CP but had fewer comparative effects on parent- and teacher-reported observable behavioral outcomes, including reactive aggression. It was hypothesized that MCP-produced improvements in children’s internal awareness and self-regulation, if maintained or strengthened over time with ongoing mindfulness practice, would yield improvements in children’s observable prosocial and reactive aggressive behavior at later time points. To appraise this hypothesis, the current study examined teacher-reported child behavioral outcomes at a one-year follow-up. In the current subsample of 80 children with one-year follow-up data, MCP produced a significant improvement in children’s social skills and a statistical trend for a reduction in reactive aggression compared with CP. Further, MCP produced improvements in children’s autonomic nervous system functioning compared with CP from pre- to post-intervention, with a significant effect on children’s skin conductance reactivity during an arousal task. Mediation analyses found that MCP-produced improvements in inhibitory control at post-intervention mediated program effects on reactive aggression at the one-year follow-up. Within-person analyses with the full sample (MCP and CP) found that improvements in respiratory sinus arrhythmia reactivity were associated with improvements in reactive aggression at the one-year follow-up. Together, these findings indicate that MCP is an important new preventive tool to improve embodied awareness, self-regulation, stress physiology, and observable long-term behavioral outcomes in at-risk youth. Further, children’s inhibitory control and autonomic nervous system functioning emerged as key targets for preventive intervention.
Introduction
Coping Power (CP) is a cognitive behavioral preventive intervention for children at risk for substance use and delinquency [1]. CP reduces aggressive and externalizing behavior of children prior to the onset of adolescence by effectively targeting the mediating processes of the child (hostile attributional biases, emotional dysregulation, social perspective-taking, problem-solving) and of the family (parenting) [2,3]. CP is conceptually framed by the contextual social-cognitive model of aggressive behavior [1,2], and multiple RCTs demonstrated beneficial CP prevention and treatment effects on children's externalizing behavior problems, even at a follow-up years later [4][5][6]. Preventive effects of CP at follow-up include increased child prosocial and academic outcomes [7] and reduced rates of adolescent substance use and antisocial behavior [1,8,9].
Proactive and Reactive Aggression
Childhood aggression, as a predictor of later problem behaviors, such as substance use and delinquency, has been categorized into two independent dimensions: proactive and reactive [10]. While these two types of aggression often co-occur, they differ in etiology, with independent genetic influences [11] and social-cognitive patterns [12,13]. Proactive aggression has been associated with violence and antisocial personality in young adulthood, whereas reactive aggression has been associated with greater emotional lability and impulsivity, with an increased risk of substance use in adolescence [14][15][16]. Proactive aggression is instrumental and uses aggressive behaviors to obtain deliberate outcomes. Proactive aggression has been characterized as "cold-blooded," that is, predatory, organized, and unprovoked, with a greater instrumentality orientation, while reactive aggression is explosive and "hot-blooded." Brain imaging studies revealed that children with heightened levels of proactive aggression are marked by differing neural correlates and prefrontal and ventromedial prefrontal cortex activation [17][18][19] compared with children with reactive aggression.
CP has been effective in reducing children's proactive aggression at post-intervention in two school-based samples of children screened for moderate-to-high rates of aggressive behavior. There was a pre-post reduction in parent-rated proactive aggression in the small effect size range (0.28) in a study of 245 elementary school boys and girls in the United States [20]. In a second study, there were small parent-rated (0.29) and moderate teacherrated (0.57) reductions in proactive aggression for elementary school boys screened for aggression in Pakistan [21]. Moderate-sized (0.47) reductions in teacher-rated proactive aggression over a long-term three-year follow-up were found in a study that compared CP with typical school services with 241 elementary school boys and girls. Intervention effects in this follow-up study and another 6-year follow-up study [8], were also found on stable callous-unemotional traits, which were associated with proactive aggression in longitudinal research [22].
While CP was effective in consistently reducing proactive aggression in this set of randomized trials, CP's effects on reactive aggression were more mixed in these same studies. CP had positive moderate-level effects on reactive aggression in a three-year follow-up study [4], likely due to children perceiving fewer possible provocations from others as their externalizing behavior decreased. However, there were conflicting results from the immediate post-intervention, which is when the intervention-related mechanisms should have been activated. CP produced strong reductions in reactive aggression with the boys in the Pakistan study [21], but there were no significant effects on reactive aggression in the earlier US study [20]. In this latter study, it is notable that a targeted mechanism associated with proactive aggression (outcome expectation) was influenced by the intervention but targeted mechanisms typically associated with reactive aggression, such as hostile attributions and affective dysregulation [12], were not. Thus, it appears that the targeted mechanisms associated with reactive aggression may be more easily affected by differences in the training and implementation of CP and by cultural context, and are vulnerable to not being addressed as deeply or as well as the mechanisms associated with proactive aggression. As a result, the active mechanisms of reactive aggression may need to be further targeted in more intensive ways to enhance the CP intervention effects.
Active Mechanisms of Reactive Aggression
Behavior dysregulation. Children high on reactive aggression have been associated with volatility, inattentiveness, and emotional/fear responsiveness, followed by a sense of relief [10]. Reactive aggression is also independently related to a variety of other indices, including internalizing problems, poor prosocial outcomes, and attentional problems [16,23]. Children with increased levels of reactive aggression are also marked by hyperresponsivity in the midbrain and limbic systems rather than the cortex, as children with heightened proactive aggression demonstrate [24].
Emotional dysregulation. Children with reactive aggression evidence high levels of anger [25,26], emotional arousal [27], and negative emotionality [25]. From a physiological perspective, reactive aggression is associated with autonomic over-arousal and increased amygdala response to social threats [28]. During an experimental anger induction task, reactive aggression (but not proactive aggression) was associated with children's skin conductance reactivity and angry nonverbal behaviors [29]. Reactive aggression was associated with decreased SCL reactivity at the beginning of the social challenge task, then with increased SCL reactivity as the task continued. In another study, children with reactive aggression demonstrated higher skin conductance and mild cortisol decline compared with children with proactive aggression in stress-inducing tasks [30].
Cognitive and attention dysregulation. Children with reactive aggression often interpret social cues as hostile provocations or threats, which consequently results in heightened anger responses [31], and anger rumination has been linked with higher levels of reactive aggression in young adults [32]. In adults, anger rumination can lead to increased aggressive behavior [33], and children with high levels of reactive aggression are also associated with a reduction in executive function [34].
Polyvagal Theory and Reactive Aggression
Since reactive aggression is associated with greater dysregulation and differs etiologically from proactive aggression, recent studies suggest that polyvagal theory may provide a hypothesis for why reactive aggression shows resistance to intervention: reactive aggression relates to the autonomic nervous system differently than proactive aggression [35]. For children with elevated levels of reactive aggression, their dysregulated responses may be linked to poor regulation of arousal or hyperarousal of the autonomic nervous system [36].
According to polyvagal theory [37], the autonomic nervous system (ANS) automatically engages when an individual experiences threats and uncertainty, and self-regulation is linked to neurophysiological responses of the two ANS branches of the vagus nerve, namely, the sympathetic (SNS) and parasympathetic nervous systems (PNS). Polyvagal theory hypothesizes that increasing arousal is linked to the SNS, and decreasing arousal is linked to the PNS [36,38]. The SNS is responsible for increasing the metabolic responses associated with preparing for "flight or fight", such as increased blood pressure, higher heart rate, and cardiac output, while the PNS physiologically restores and conserves energy by regulating the blood pressure, heart rate, and stress [39]. Together, these two branches (SNS and PNS) regulate arousal responses. The ANS has been considered a predictor of aggressive behavior in youth [40,41]. Impaired autonomic system reactivity (as measured by sympathetic and parasympathetic reactivity) has been correlated with increasing levels of aggressive and disruptive behavior [42][43][44].
Skin conductance level (SCL) as a measure of increasing arousal. The SNS prepares the body for "fight or flight" by increasing physiological arousal through increased electrodermal activity, or the skin conductance level (SCL). Thus, the SCL is a physiological signal for the stress response. Increased SCL activation has also been conceptualized as a marker for behavioral inhibition in aversive contexts (activated as fearfulness in aversive contexts) [45], whereas reduced levels of SCL reactivity are associated with higher levels of externalizing behaviors in youth [44,46]. While a low SCL has been consistently associated with aggression, this relationship is not always straightforward. Some studies found that SCL reactivity varies in children with reactive and proactive aggression [29,38]. In one study, reactive aggression was associated with decreased SCL reactivity at the beginning of a social challenge task, then with increased SCL reactivity as the task continued. Meanwhile, SCL reactivity was not related to proactive aggression [29].
Respiratory sinus arrhythmia (RSA) as a measure of decreasing arousal. In contrast, the activation of the PNS conserves and restores energy by reducing physiological arousal. The PNS is indexed by respiratory sinus arrhythmia (RSA), or heart rate variation, which is considered a measure of vagal tone according to polyvagal theory [37,47]. RSA functions as an index for the PNS, specifically for cardiac activity [48]. Lower RSA at baseline and higher RSA reactivity following arousal (i.e., vagal withdrawal) has been associated with greater externalizing behaviors and greater increases in delinquency over time, as well as increased prefrontal cortex dysfunction [49,50]. From the framework of polyvagal theory, higher RSA reactivity indicates the vagal brake is not engaged, and higher reactivity has been associated with anxiety, social difficulty, and problem behavior [51], while higher RSA and lower RSA reactivity are associated with prosocial behavior and emotion regulation [52,53].
Mindfulness Training Effects
Recent studies indicate that mind-body therapies have a wide range of positive impacts, including on the physiological response to stress [54]. Mindfulness training is one form of mind-body practice [55,56]. Mindfulness is described as the practice of bringing non-judgmental awareness to the present moment [56]. Mindfulness training practices focus on noticing one's breathing, bodily sensations, thoughts, and moment-to-moment experiences through an accepting, non-judgmental lens. Paying attention in this way can help to increase one's focus on the present moment (rather than ruminating on the past or future), reduce negative thought and emotion patterns, and create space between emotional reactions and behavioral responses [55,56].
Mindfulness training in the school setting was shown to improve child behavior outcomes, such as attentional control, emotion regulation, and impulse control [57][58][59]. These studies indicate that integrating mindfulness training into Coping Power can improve the active mechanisms of reactive aggression and strengthen program effects on children's behavior and stress physiology.
Mindfulness effects on behavior dysregulation. Mindfulness training improves behavior outcomes for the social development of elementary students [60] and adolescent impulsivity [61]. Mindfulness training also showed improved behavioral outcomes for children with externalizing behavior disorders [62].
Mindfulness effects on emotion dysregulation. Mindfulness training improves emotional outcomes by affecting externalizing anger in adolescents [63] and self-regulation in young adults [64]. Brain imaging studies indicate that mindfulness training is linked to improvements in the density of brain regions linked to emotion regulation [65].
Mindfulness effects on cognitive and attention dysregulation. Mindfulness training improves cognitive and attention outcomes and was shown to decrease rumination [59,62]; increase cognitive outcomes, such as cognitive flexibility [60] and attention [66]; and improve attention outcomes in children with attention disorders [67].
Targeting Active Mechanisms of Dysregulation with Mindful Coping Power
Given the demonstrated effects of mindfulness training on the active mechanisms of reactive aggression, it was hypothesized that incorporating mindfulness practices into Coping Power would strengthen the program's effects on reactive aggression and improve children's ANS functioning. The Mindful Coping Power (MCP) program, which integrates mindfulness into the cognitive behavioral Coping Power program, was developed to test this hypothesis [3].
A recent randomized comparative effectiveness study of MCP versus CP found that for children with high levels of reactive aggression, MCP yielded significantly stronger effects than CP on children's self-reported dysregulation [68]. By targeting the active mechanisms of attentional, cognitive, behavioral, and emotional dysregulation present in high levels of reactive aggression, MCP optimized the benefits of the cognitive-behavioral intervention with anticipated effects on the underlying mechanisms of reactive aggression [3]. Of note, the strongest effects were on children's self-reported internal experiences, with fewer effects on children's overt behavior, as observed by teachers. Based on these comparative findings, the authors hypothesized that children's improved sense of internal self-regulation and embodied awareness in MCP versus CP at post-intervention may lead to greater improvements in the observable behavior over time for children in the MCP group. This was hypothesized because children's increased internal awareness and self-regulation at post-intervention, if maintained or strengthened with ongoing mindfulness practice, may lead to more prosocial and fewer hostile-aggressive observable behaviors over time. It was also hypothesized that MCP improved children's ANS functioning at post-intervention, corresponding with their perceived improvements in internal self-regulation. If found, improvements in children's ANS functioning at post-intervention are also expected to contribute to improved observable child behavior over time.
The Current Study
The current study examined the above hypotheses. Specifically, the following research questions guided this study: (1) Does Mindful Coping Power (MCP) produce better child behavioral outcomes at a 1-year follow-up than Coping Power (CP), as measured by teacher-rated child reactive aggression, externalizing behavior problems, proactive aggression, and social skills? (2) Does the Coping Power preventive intervention (full sample with both MCP and CP) have a beneficial impact on children's physiological stress reactivity (SCL and RSA reactivities) from pre-to post-intervention? If so, does MCP produce stronger beneficial effects on children's stress physiology than CP? (3) Do intervention-produced improvements in children's stress physiology (SCL and RSA reactivities) and perceived self-regulation and mindful awareness (total dysregulation, inhibitory control, breath awareness) from pre-to post-intervention mediate improvements in child behavioral outcomes at a 1-year follow-up (as measured by child reactive aggression, externalizing behavior problems, proactive aggression, and social skills)?
Participants
Participating schools. This study is a follow-up from a previously published study [68], which reported the results of a preventive trial that compared standard CP versus MCP child behavioral outcomes at pre-and post-intervention. Five elementary schools in Tuscaloosa, Alabama, were recruited for this study (four participated in the first annual cohort and a fifth was added in the second annual cohort). The schools varied in terms of sociodemographic characteristics, including families enrolled in free or reduced lunch (ranging from 32% to 76%) and each child's race (with Black or African American the most prevalent, ranging from 92% to 32%). To control for sociodemographic characteristics, participants were randomly assigned to MCP or CP within each school.
Child participants. The full recruited sample included 102 children with elevated levels of reactive aggression, as well as their parents and teachers. Child participants were identified through teacher screener ratings completed at the end of the fourth grade so that recruited children could participate in the intervention during their fifth-grade year. The CP preventive intervention was designed to target preadolescent children immediately prior to the middle school transition when substance use initiation risk increases [69].
To identify children with elevated reactive aggression, fourth-grade teachers completed the Teacher Report of Reactive and Proactive Aggression [70] on all classroom students. Ratings were compiled across the participating schools and divided into quartiles to identify an empirical cut-off score that reflected the top quartile of reactive aggression in fourth-grade students. An empirical cut-off score of 8 was used, which was consistent with previous CP studies in which children with teacher-rated reactive aggression scores at or above this level had elevated parent-rated externalizing problems (at-risk or clinical range) on the Behavior Assessment Scale for Children [71].
All enrolled child participants were offered an opportunity to provide stress physiology data and to have their sixth-grade teacher provide one-year follow-up data (which was not part of the initial study consent), with parent consent. The current analyses included the subset of participants who provided either or both forms of additional data. Table 1 summarizes the characteristics of the children with complete one-year follow-up data (n = 80), SCL data (n = 46), and RSA data (n = 45) compared with the full sample.
The current sample was predominantly Black/African American (88%, n = 70), male (64%, n = 51), and the majority of families were from a lower-income household (66% had a household income of less than USD 30,000). No significant differences were observed between participants in the intent-to-treat sample (n = 102) and those in the current analyses on key baseline or demographic variables. Compared with the intent-to-treat sample, there were trend effects (p < 0.10) for children from higher-income families to be less likely to have one-year follow-up data, and for children in CP to be less likely to have SCL data.
Procedure
All study procedures were approved by the University of Alabama Institutional Review Board. Data on the participants was collected in multiple waves, with participant data gathered at three time points. Additional details about full study procedures can be found in a prior publication [68].
Participant recruitment and retention. Rising fifth-grade students were recruited from five schools across two annual cohorts. A total of 638 fourth-grade students were screened for study participation. Of these, 428 scored below the empirical cut-off for teacher-rated reactive aggression. One child with an eligible screener score was excluded due to a language barrier that could not be addressed with local resources. The remaining families were contacted in random order until the total number of intervention slots at each school had been filled. Six families declined study participation (the most common reasons were that the child already received services elsewhere or a lack of perceived need). The children who declined did not significantly differ regarding baseline characteristics from those enrolled.
One hundred and eight children were initially enrolled in the study. Five of these children moved to different schools prior to starting fifth grade. The remaining 103 participants were randomly assigned to either the MCP or CP group at their school in yoked pairs, stratified by reactive aggression screener score and demographic characteristics (to create equivalent MCP and CP groups at each school). One child withdrew from the study after participating in one session (due to a perceived lack of need and social concerns), which resulted in a total sample of 102 child participants, as well as their parents and teachers. This resulted in a sample of 52 children randomly assigned to MCP and 50 to CP.
Four schools participated in both cohorts and a fifth school was added for cohort two. This led to a total of 44 participants in cohort 1 and 58 participants in cohort 2.
Pre-intervention data were collected from parents and children at the start of fifth grade and from teachers four to six weeks into the school year (to allow teachers to gain familiarity with the children's behavior). The intervention began following the completion of the time 1 assessments. Post-intervention data were collected at the end of fifth grade for teacher data and the month following fifth grade for parents and children.
Child's age 10 Several months after completing the original pre-post study, families were contacted and offered an opportunity to participate in an additional one-year follow-up assessment. For eighty participants, one-year follow-up data were collected at the end of sixth grade. Reasons for non-participation given by the remaining 23 participants included the following: 8 did not respond to multiple attempts to reach them, 5 declined, 5 indicated a willingness to participate but did not follow through, 3 offered to complete a mailed packet but did not return it, and 2 moved out of the area.
Child's gender
The administration of parent, child, and teacher assessments was blinded to the condition. For each child assessed, teachers received USD 10, and for each assessment time point, parents received USD 50 and the children received USD 10.
Intervention. This study compared two active preventive interventions, namely, MCP and CP. Both interventions included the same number of sessions (25 child group and 10 parent sessions) and utilized curricula with specified objectives and activities at each session. The child group sessions lasted 45 min, met weekly, and were held during the school day in a private meeting space. The parent group sessions lasted 60 to 90 min, met biweekly or monthly, and were held in the mornings and evenings in a central location near the school.
Coping Power. CP is an evidence-based cognitive behavioral preventive intervention for youth at risk for substance use, delinquency, and other disruptive behavior disorders [72,73]. Drawing from a cognitive-behavioral framework, CP teaches emotion coping and social-cognitive skills to children and positive parenting and self-care skills to parents. The child program topics include setting personal goals, emotional awareness, coping with anger, problem-solving, perspective-taking, prosocial peer affiliations, and resisting peer pressure [72]. The parent program topics include supporting children's academic learning, strengthening parent-child and family relationships, managing parenting stress, setting household expectations and rules, praising, ignoring, developing effective discipline techniques, problem-solving as a family, and future planning [73].
Mindful Coping Power. MCP is a novel adaption of CP that integrates mindfulness into an evidence-based intervention [3]. MCP was shown to enhance the effects on children's anger modulation and perceived self-regulation [68]. All of the core content of CP is included in MCP. Three distinct modifications were made to distinguish MCP from CP. First, MCP included additional mindfulness-only sessions, in which both the child and parent MCP programs introduced the theory and practice of mindfulness. For example, mindfulness was defined as "training your brain" to pay attention to the present moment more fully, which was called "noticing right now." Second, MCP incorporated mindfulness in every session. Each session started and ended with 1-2 mindfulness practices (e.g., the ringing of a chime, breath awareness, yoga poses, body scan, mindful eating, mindful listening, thought awareness, or "Feel and Spread the Good Vibes" compassion practices). Children and parents also completed these mindfulness practices at home between sessions. Third, MCP integrated mindfulness into current CP activities, which included identifying early physiological cues of anger with body awareness practice, perspective-taking situations with compassion practice, and coping with anger with thought awareness practices for both the parent and child. For greater detail about how MCP incorporates mindfulness into CP, see Miller and colleagues [3].
CP did not have any intervention content that was not also in MCP. Each condition had the same number of sessions. An effort was made to keep the length of sessions consistent across the two conditions. Due to the added mindfulness practices in MCP and mindfulness-only sessions, more time was spent on some standard program elements in CP than in MCP (e.g., reviewing progress toward weekly personal goals and setting new goals, personal sharing related to intervention topics, and opportunities to practice new skills in session). Some topics were extended across more than one session in CP to account for the mindfulness-only sessions in MCP without having any content in CP that was not in MCP.
Interventionists. There were five primary group leaders. Each implemented both versions of the intervention at their assigned school (MCP and CP). These leaders were doctoral (n = 2) or master's (n = 3) level clinicians with considerable experience implementing Coping Power, as well as training in Mindfulness-Based Stress Reduction (MBSR) and maintenance of regular mindfulness practice. MBSR is a standardized mindfulness training program that incorporates mindfulness meditation, body scanning, and simple yoga postures to obtain depth practice in moment-to-moment, non-judgmental awareness [74]. Four of the leaders were female and one was male. Four of the leaders were Caucasian and one identified as more than one race. Master's and advanced undergraduate students served as co-leaders to provide additional group oversight and behavior support. They received CP and mindfulness training prior to co-leading the intervention groups.
Behavioral Measures.
The behavioral measures in this study were the same as the previously published study, which provides greater detail on each measure [68].
Measures used in the current study included the Teacher Report of Reactive and Proactive Aggression (TRRPA; [70]), which was used as a screener to identify students with elevated reactive aggression and as an outcome measure; the Behavior Assessment System for Children (BASC; [71,75]), which is a broadband teacher-reported measure with composite scores for externalizing behavior problems and social skills; the Abbreviated Dysregulation Inventory (ADI; [76]), which measures children's perceived affective, behavioral, and cognitive dysregulation; the Early Adolescent Temperament Questionnaire (EATQ; [77]), which includes a child self-reported measure of inhibitory control; and the Scale of Body Connection (SBC; [78]), which measures breath and body awareness. Additionally, as described below, the current analyses utilized physiological data gathered to measure the skin conductance level (SCL) and cardiac respiratory sinus arrhythmia (RSA) using data recorders during a decision-making card task, namely, the Iowa Gambling Task.
Physiological Measures. RSA data. RSA data were obtained using a BioLog™ physiological data recorder, which measures the participant's heart rate and interbeat intervals with the placement of an electrode each at the collarbone and behind the left knee, with a reference electrode placed on the right side of the neck. The RSA was derived using the Porges-Bohrer method using the CardioBatch and CardioEdit computer programs [79,80].
First, interbeat interval data were cleaned by hand-editing participants' heart rate data in the CardioEdit program. Second, outliers or artifacts were reviewed as data points (i.e., errors in the heart rate data that were likely due to the digitizing process or physiological anomalies). Outliers were identified and edited (using standard methods, such as averaging, combining, and dividing). Files in which 5% or more of the data points required editing were removed from the sample. After cleaning, the CardioBatch program was used to extract the RSA from one of the predominant rhythms exhibited in the participants' heart period series. The mean RSA was calculated for each of the five blocks. SCL data. SCL data were obtained by placing electrodes on the volar surface of the distal phalanges (first and third fingers) of the non-dominant hand. The SCL data were processed and artifact removal was completed through the use of Ledalab [81], which is a Matlab-based software for the analysis of the SCL data. Each SCL file was scanned for outlier data points relative to adjacent data points. These outlier points were replaced by data that were more consistent with the surrounding data.
Iowa Gambling Task (IGT). The IGT is a psychological task used to measure decisionmaking and to elicit an affective response by varying the rewards and penalties for selecting certain card decks. The IGT has been used to examine other physiological and disorder data, including brain lesions, alcohol disorders, violence and aggression, and gambling and impulse control disorders [82][83][84][85][86]. In the task, participants are not instructed about the uncertain nature of the task and are unlikely to recognize the probability of receiving positive and negative reinforcement across the different card decks; thus, participants make decisions under ambiguous conditions, with the participants' responsiveness to emotional cues from previous card draws contributing to later advantageous task decisions [87][88][89].
Prior to the IGT, children first viewed a neutral stimulus (a two-minute scenery video) to permit the necessary time to acclimatize to the physiological equipment with minimal emotional response, language, or movement. Values for the baseline RSA and SCL were obtained during this neutral stimulus period. Following this neutral stimulus period, participants received instruction for and played the Iowa Gambling Task on a computer screen (IGT; [82]). In the task, participants are presented with four decks (varying in penalty and reward levels) and told to win as much game money as possible. Participants complete five, 5 min blocks with 20 cards, for a total of 100 cards. The SCL and RSA reactivities were assessed during the first block for this study, given that the first blocks have more uncertainty and ambiguity [90,91]. The SCL and RSA reactivities were computed as the change from baseline (neutral stimulus) to block 1 of the IGT.
Statistical Approach
Statistical Analysis Software (SAS/STAT 15.1) [92] was utilized to conduct the analyses for this study.
Research questions 1 and 2 were examined using mixed-effects models. For research question 1, the outcomes of interest were the teacher-rated child social skills, externalizing behavior, reactive aggression, and proactive aggression, measured at baseline and the 1year follow-up. The SCL and RSA reactivities, measured at baseline and post-intervention, were the dependent variables of interest for research question 2. The fixed effects included the time, treatment assignment (CP = 0, MCP = 1), and interaction between these two. The presence of the interaction allowed for the estimation of the time slopes for each condition separately, as well as testing whether the time slope in the MCP condition significantly differed from that in the CP condition on average. Both CP and MCP were delivered in group settings. Likelihood ratio tests examined whether it was necessary to account for nesting at the intervention-group level. The tests were insignificant. Therefore, the random effect included a random intercept only at the individual level. Models were estimated with a restricted maximum likelihood. A Kenward-Roger adjustment was used to mitigate the small-sample bias. The effect size was estimated using Equation (8) in Morris [93]. All analyses included dummy variables to control for school differences. The models for SCL and RSA also controlled for humidity and temperature.
For research question 3, mediation analyses were conducted using the ANCOVA approach, which involved two regression equations [94]. In the first equation, the mediator measured at post-intervention was regressed on the treatment assignment, adjusting for the outcome and the mediator measured at baseline. In the second equation, the outcome measured at the 1-year follow-up was regressed on the mediator measured at post-intervention, adjusting for the treatment assignment, and the outcome and the mediator measured at baseline. The significance of the indirect effect was determined using a confidence interval generated with PRODCLIN [95]. This was repeated for each potential mediator and outcome variable when the following conditions were satisfied: the treatment condition had a significant effect on the potential mediator at post-intervention, and the potential mediator had a significant effect on the outcome variable at the one-year follow-up, controlling for the treatment condition. The SCL and RSA reactivities were explored as potential mediators, as well as the child's self-reported measures of self-regulation and mindful awareness, which showed greater improvement with MCP in prior pre-post analyses [68]. Outcome variables included externalizing behavior problems, reactive aggression, and social skills.
Mixed-effects models were also used to examine the within-person association between the SCL and RSA reactivities and each behavioral outcome variable (reactive aggression, externalizing behavior problems, proactive aggression, and social skills). For example, it was hypothesized that an increase in the RSA reactivity between baseline and postintervention would be associated with an increase in externalizing behavior problems between baseline and the 1-year follow-up. Other time-varying variables included reactive aggression, proactive aggression, social skills, and a variable time to capture a secular trend. All time-varying independent variables were person-mean-centered to estimate their within-subject association with the behavioral outcome variable (e.g., externalizing behavior). In addition to race and gender, the model incorporated individual-specific means of the time-varying independent variables as (time-invariant) individual-level variables. Individual-level variables captured between-person associations with the child's behavioral outcome variable (e.g., externalizing behavior).
Research Question 1
Does Mindful Coping Power (MCP) produce better child behavioral outcomes at a 1-year follow-up than Coping Power (CP), as measured by teacher-rated child reactive aggression, externalizing behavior, proactive aggression, and social skills? Table 2 summarizes the mean scores for each teacher-reported child behavioral outcome at pre-intervention and the 1-year follow-up, as well as the effect size estimates comparing outcomes from MCP and CP. The effect sizes were computed using the pooled pretest standard deviation [93]. The effect sizes observed from pre-intervention to the 1-year follow-up each favored MCP relative to CP and were as follows: 0.57 for social skills, −0.46 for reactive aggression, −0.36 for externalizing behavior problems, and −0.26 for proactive aggression. Table 3 summarizes the results of the mixed-effects models comparing the MCP and CP interventions on teacher-rated child behavioral outcomes from pre-intervention to the one-year follow-up. Figures 1-3 illustrate these findings. A significant time × condition effect indicated that MCP produced significantly stronger improvement than CP in children's social skills (p = 0.037). The time × condition effect was marginally significant for reactive aggression (p = 0.057), with more improvement in the MCP group than in the CP group. Children in the MCP group showed a significant increase in social skills (p < 0.01) on average, and a marginally significant decrease in reactive aggression (p = 0.055). Meanwhile, the average change in the CP group was non-significant for both social skills and reactive aggression (p > 0.20). Children in the CP group showed a significant worsening of externalizing problems (p =.041) on average, while the change in the MCP group was non-significant (p = 0.884). However, the time × condition interaction did not attain statistical significance at 5% for externalizing problems (p = 0.109). In terms of proactive aggression, children in the MCP group showed a significant decline (p =.002) on average, while the change in the CP group was non-significant (p = 0.120). The time × condition interaction, however, did not attain statistical significance for proactive aggression (p = 0.294).
Research Question 2
Does Coping Power (full sample with both MCP and CP) have a beneficial impact on children's physiological stress reactivity (SCL and RSA reactivities) from pre-to postintervention? If so, does MCP produce stronger beneficial effects on children's stress physiology than CP? SCL and RSA Reactivity. For the stress reactivity measures, the SCL and RSA reactivities were computed as the change from baseline (neutral stimulus) to block 1 of the IGT. The difference between the two time points, which reflects reactivity, was used to determine a change in children's stress physiology, either increasing or decreasing depending on the measure (e.g., skin conductance or cardiac respiratory sinus arrhythmia). Since the physiological assessment was optional and the authors employed stringent cut-off scores for the usable data (i.e., data with 5% or greater data editing was removed), the number of cases was reduced, with SCL and RSA reactivity scores for 46 and 45 participants, respectively. Table 4 summarizes the mean scores for the SCL and RSA reactivities at pre-and post-intervention, as well as effect size estimates comparing outcomes in the MCP and CP conditions. The effect sizes were 0.66 for the SCL reactivity and −0.25 for the RSA reactivity, each favoring MCP. Table 5 summarizes the results of the mixed-effects models comparing the MCP and CP intervention conditions on the SCL and RSA reactivities from pre-to post-intervention. There was a significant difference between MCP and CP for the SCL reactivity (p = 0.027). On average, children in the CP condition showed a decline in SCL reactivity at postintervention (p = 0.005), while the change in the MCP condition was not significant (p = 0.830). The effects on RSA reactivity were not significant.
Research Question 3
Do intervention-produced improvements in children's stress physiology (SCL and RSA reactivities) and perceived self-regulation and mindful awareness (total dysregulation, inhibitory control, breath awareness) from pre-to post-intervention mediate improvements in child behavioral outcomes at a 1-year follow-up (as measured by reactive aggression, child externalizing behavior problems, proactive aggression, and social skills)?
Inhibitory control emerged as the primary mediator of interest. SCL reactivity, total dysregulation, and breath awareness did not show significant mediation effects on reactive aggression or social skills. RSA reactivity was not assessed as a mediator, and externalizing behavior problems and proactive aggression were not included as outcome measures due to not meeting the criteria for inclusion in mediation analyses. Table 6 presents results from the mediation analysis with inhibitory control as the mediator, reactive aggression as the outcome, and treatment assignment as the independent variable. The effect of MCP (vs. CP) on inhibitory control at post-intervention was marginally significant at 10% (b = 0.217, p = 0.070). The effect of inhibitory control at postintervention on reactive aggression at the 1-year follow-up was significant at 5% (b = −1.614, p = 0.024). The indirect effect of MCP (vs. CP) on reactive aggression at the 1-year follow-up through inhibitory control at post-intervention was estimated by the product of 0.217 and −1.614. Children in the MCP condition had higher inhibitory control at post-intervention on average, which, in turn, was associated with lower reactive aggression at the 1-year follow-up. Using PRODCLIN, it was suggested that the indirect effect was significant at 10% (90% CI = [−0.827, −0.009]). The models controlled for child demographics, including age, sex, and race. Table 6. Inhibitory control mediates reactive aggression outcomes at the one-year follow-up. For the analysis of within-person associations, the main finding of interest was found for the RSA reactivity and externalizing behavior problems. As shown in Table 7, the mixedeffects analyses indicated a within-person association between the RSA reactivity and teacher-reported externalizing behavior problems. An increase in RSA reactivity between baseline and post-intervention was associated with an increase in externalizing behavior problems between baseline and the 1-year follow-up (p < 0.01). This model was estimated with an interaction between the RSA reactivity and MCP to test for differential intervention effects. It was not significant, and thus, the model was re-estimated without the interaction, as reported in Table 7 below. Although it is not of main interest, there was a significant within-person association between reactive aggression and externalizing behavior problems (p < 0.01). An increase in reactive aggression from baseline was associated with an increase in externalizing behavior problems from baseline (p < 0.01). On the other hand, proactive aggression and social skills were not within-person-associated with externalizing behavior problems (p = 0.549 and 0.533, respectively).
MCP Effects on Children's Long-Term Behavioral Outcomes
MCP was developed to enhance CP effects on reactive aggression by incorporating mindfulness training into the intervention to target the active mechanisms of reactive aggression more directly. At post-intervention, Boxmeyer and colleagues [68] found that MCP improved children's self-reported embodied awareness, self-regulation, and anger modulation relative to CP, but had fewer comparative effects on parent-and teacher-reported behavioral outcomes, including reactive aggression. The current study examined teacherreported child behavioral outcomes one year after the intervention ended to appraise the hypothesis that MCP-produced improvements in children's internal experiences at post-intervention would yield improvements in children's observable behavior later in development. Indeed, the results of the current study partially support this hypothesis. One year after the intervention, children in the MCP group showed improvements in teacher-reported prosocial behavior and a statistical trend for the reduction in reactive aggression relative to children in the CP group.
As a preventive intervention, one of the main goals of CP is to alter the developmental trajectory of at-risk youth as they transition to adolescence. Preadolescents who exhibit high levels of aggression are at risk for a range of negative outcomes across the developmental cascade, including violence, delinquency, substance abuse, and academic difficulties [16,96]. Positive program impact has not been defined solely as an overall net reduction in aggression and externalizing behavior, especially when fall-to-spring teacher ratings are used and teachers have greater exposure to children's behavioral problems throughout the school year. Positive program impact has also been defined as an attenuation in the slope of the progression toward negative adolescent outcomes [68,97]. Thus, it is even more noteworthy that children in the MCP group exhibited higher levels of social skills, lower proactive aggression, and a trend for lower levels of reactive aggression at the end of their first year of middle school, one year after completing the intervention, compared with pre-intervention levels early in fifth grade. Moreover, MCP appears to have added a benefit beyond the already established, evidence-based CP in terms of improving children's long-term social skills and the tendency to reduce reactive aggression. The hypothesis was upheld that the MCP-produced improvements in children's internal awareness and perceived selfregulation at post-intervention would eventually lead to comparative improvements in their observable behavior, even though the MCP versus CP impacts on observable behavior were not consistent immediately post-intervention [68]. This additional one-year follow-up data underscores the idea that MCP is an important enhancement of CP that can maximize the program's downstream effects.
Additional analyses examined whether these long-term improvements were attributable to MCP-produced changes in children's stress physiology and children's self-reported embodied awareness, anger modulation, and self-regulation.
MCP Effects on Children's Autonomic Nervous System Functioning
MCP produced stronger effects than CP on children's autonomic nervous system functioning from pre-to post-intervention. MCP had a significant effect on the children's SCL reactivity (ES = 0.66) and a smaller, non-significant effect on the RSA reactivity (ES = −0.25).
Children's poor emotional regulation can result from overactivity of the sympathetic nervous system (SNS; prepares for "fight or flight") and a diminished ability of the parasympathetic nervous system (PNS) to dampen that arousal and bring children back to their initial level of arousal. Extreme difficulties regulating arousal can cause aggressive children to experience intense anger and "hot" angry cognitions, which then interfere with their adaptive information processing and leads to reactive aggression [31,98,99].
Higher SCL activation has been conceptualized as facilitating behavioral inhibition through the production of fear and anxiety. Lower levels of SCL reactivity have been associated with higher levels of youth problem behaviors [44,46,100]. Some studies found that SCL reactivity relates differently to reactive and proactive aggression [29,38] and at different phases of an anger-inducing game [29]. The present study examined the SCL reactivity in the transition from baseline to the first block of an arousal task and found that, on average, children in the CP condition showed a decline in SCL reactivity from pre-to post-intervention, while the change due to the MCP condition was not significant. This indicates that children in the MCP group maintained better regulation of arousal during an emotionally arousing decision-making task from pre-to post-intervention than those in CP. Improving the regulation of physiological arousal, or at least attenuating declines in arousal regulation, has the potential to have a lasting impact on functioning across the developmental cascade.
RSA is a measure of the parasympathetic nervous system, which conserves and restores energy. Higher RSA at rest indicates the vagal brake is engaged and has been associated with prosocial behavior, while higher RSA reactivity indicates the vagal brake is not engaged and has been associated with anxiety, social difficulty, and behavior problems [37,41,49,[101][102][103]. In the current study, there was a small but non-significant effect favoring MCP on the RSA reactivity. Even small improvements in the children's RSA reactivity may have lasting benefits on their emotional regulation and behavior, which was supported by the within-person analyses in the full sample discussed below.
Mediational Analyses
Inhibitory control emerged as a primary mediator of interest. MCP-produced improvements in inhibitory control at post-intervention mediated the program's trend for effects on reactive aggression at the one-year follow-up. Inhibitory control is an important aspect of executive functioning, which entails the effortful or willful control of behavior, including both approach and avoidance behavior [104]. Problems with inhibitory control are common in youth with aggressive behavior [105,106], predict later externalizing problems and substance use [107][108][109], and may moderate the relationship between negative emotionality and externalizing behavior problems [110]. Boxmeyer and colleagues [68] found that children in the MCP group showed greater improvements in inhibitory control than children in CP, making MCP an important tool for preventing and reducing a range of negative adolescent outcomes in at-risk youth.
Within-person analyses found that improvements in RSA reactivity were associated with improvements in reactive aggression at the one-year follow-up in the full sample (both treatment conditions combined). As described above, RSA is increasingly seen as an important biomarker of emotion dysregulation and various forms of externalizing and internalizing psychopathology [49]. These findings indicate that both inhibitory control and RSA are important targets for preventive intervention to improve children's physiological, behavioral, and emotional regulation. MCP is a valuable new prevention tool, particularly to improve inhibitory control in children exhibiting reactive aggression.
Other potential mediators did not meet analytic requirements or show significant effects, including the SCL reactivity, perceived self-regulation measured using the ADI, or body awareness measured using the SBC. In the original pre-post sample [68], the Child and Adolescent Mindfulness Measure [111] was also collected but was not as sensitive to differences between MCP and CP as was the SBC measure of body and breath awareness. For this reason, the SBC was included as a potential mediator in the current analyses rather than the CAMM. The differences in the performance of the CAMM and SBC with this sample may have been because both CP and MCP have significant components on thought awareness, while MCP invests much more heavily than CP on present moment breath and body awareness.
Study Limitations
This study employed a rigorous design in that it directly compared MCP with standard CP (which already has an extensive evidence base) in a randomized controlled trial. Teachers were blind to the condition and the same clinicians implemented both the MCP and CP groups at each school, thus controlling for clinician characteristics. One limitation of the study, however, was that it did not include a no-treatment control group. As described above, previous studies found that without a preventive intervention, such as Coping Power, youth with elevated anger and aggression are at risk for an increasing trajectory of negative outcomes into adolescence and adulthood [16,96]. Attenuating this trajectory toward negative outcomes is seen as a success (i.e., reducing the slope of increases in externalizing behavior, reactive aggression, social skill deficits, and associated behaviors such as substance use), as well as net improvements in these outcomes from baseline. If this study had included a no-treatment control group, the full preventive benefits of MCP and CP may have been characterized more comprehensively.
Another study limitation was attrition from the intent-to-treat sample. Collection of the teacher data at the one-year follow-up was not part of the initial study design, thus participants had to consent to this separately. About 78% percent of the original intent-to-treat sample provided one-year follow-up data. Overall, the follow-up sample was reflective of the intent-to-treat sample, except for a trend toward less retention of higher-income families (in an already lower-income sample, with the majority of families having an annual income of less than USD 30,000 per year). Participants also consented to the collection of physiological data separately and high standards were set for the quality of the SCL and RSA data to be included in analyses. Thus, only 44% of the initial sample was included in the analyses of intervention effects on the SCL and RSA reactivities. These children were generally representative of the full sample, except for a trend effect for children in the CP group to be less likely to have usable SCL data than those in the MCP group.
The original sample of 102 participants was powered as a pilot and feasibility study to estimate the comparative effects of MCP versus CP in preparation for a large-scale trial. Since fewer participants provided stress physiology and one-year follow-up data in the current analyses, the power to detect small effect sizes was even more limited. For example, the trend effect favoring MCP on children's reactive aggressive behavior (ES = −0.46) and externalizing behavior at the one-year follow-up (ES = −0.36), and other outcomes with smaller effect sizes, may have been statistically significant with a larger sample. Thus, an important future direction is to explore the effects of MCP in a larger, more adequately powered sample.
A final limitation was the use of a nonsocial stressor procedure to evoke uncertainty and emotional arousal, namely, the IGT, and to examine the change in children's physiological functioning. The IGT assesses participants' affective decision-making under conditions of uncertainty and ambiguity about the consequences of choices on a card selection task and produces resultant physiological arousal evident in SCL and RSA reactivities [100]. Other paradigms are available that measure vagal withdrawal and skin conductance reactivity during simulated social interactions designed to elicit social appraisals and emotional arousal similar to real-life rejecting or conflictual interpersonal interactions, e.g., [112]. It would be beneficial to study the effects of MCP using such a typical social stressor paradigm in the future.
Conclusions
This paper provides further evidence that targeted preventive interventions delivered in elementary school can significantly alter the developmental trajectory of at-risk youth, impacting not only the children's behavior but also their autonomic nervous system functioning. CP is one of the leading evidence-based preventive interventions for at-risk aggressive youth. MCP is a novel adaptation of CP that incorporates mindfulness training to maximize the program's effects on reactive aggression. The current study built upon prior findings stating that MCP improves children's self-reported anger modulation, self-regulation, and embodied awareness compared with CP at post-intervention. The current study found that these post-intervention improvements were accompanied by improvements in children's autonomic nervous system functioning and that MCP produced long-term benefits to children's social skills and trends for the reduction in reactive aggression (as rated by teachers) one year after the intervention. Mediational analyses indicated that MCP produced long-term effects on children's reactive aggression by improving children's inhibitory control. Based on within-subjects analyses, improving children's RSA reactivity is also an important mechanism for improving children's long-term externalizing behavior outcomes. Overall, these findings highlight that MCP is an important new preventive intervention for at-risk youth, in particular, those with emotional and behavioral dysregulation. Future studies are needed to examine the effects of MCP in large-scale trials. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement:
The data presented in this study are available upon request from the corresponding author. The data will be made publicly available after all planned grant analyses have been published.
Acknowledgments:
We would like to acknowledge the schools, youth, parents, and teachers who participated in this research.
Conflicts of Interest:
John E. Lochman is the co-developer of the Coping Power program and receives royalties from Oxford University Press for the Coping Power child component implementation guidebook and workbook. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results. | 2023-05-25T15:05:01.186Z | 2023-05-23T00:00:00.000 | {
"year": 2023,
"sha1": "db12fa637ffecd00fc82d5b78c03f9d44c1503ee",
"oa_license": "CCBY",
"oa_url": "https://doi.org/10.3390/jcm12113621",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "f5074805bcc0a624c19a153f6626d17374b96960",
"s2fieldsofstudy": [
"Psychology"
],
"extfieldsofstudy": []
} |
125177395 | pes2o/s2orc | v3-fos-license | Measurements of Neutrons In A Mixed Gamma-Neutron Field Using Three Different Types of Detectors
,
INTRODUCTION
Neutrons are indirectly ionizing radiation that causes, at the same absorbed dose, biological effects more important than those produced by photons. Therefore, they are particles that must be taken into account in the dosimetry of personnel exposed to radiation. One of the facilities where the most important neutron fields are located is the interior of the containment buildings of nuclear power plants and nuclear fuel treatment units. Also found are neutron fields in cosmic radiation and in the vicinity of particle accelerators, such as those used for medical applications.
While it is true that in installations such as nuclear power plants, thecontribution of neutrons to the personal dose is in most cases negligible,there are cases where there is a significant risk of neutron irradiation, such as an intervention in the containment area during reactor operation.
Another field where it is important to have very well determined the neutron field is in medicine, for example, machines like the LINAC. On the one hand, the radioprotection of the personnel working in the installation requires that the radiation fields, including the neutrons, present are perfectly characterized. In addition, the recommendations of the International Commission on Radiological Protection, published in its publication ICRP103 [1], take into account for the first time the radioprotection of patients undergoing treatments with ionizing radiation, and in particular that of patients submitted to radiotherapy treatments in linear electron accelerators (LINACs). Among the new recommendations is that of limiting the doses due to the radiation that can receive those healthy organs next to the treated tumor.
The neutron photo-production reactions that occur in the heavy materials (iron, lead, tungsten, ...) that constitute the gantry and collimators of the accelerator itself, give rise to a neutron field that occupies the entire irradiation room and which consequently contributes to the radiation dose to the tissues and organs of the patient outside the treated tumor. Most radiotherapy treatments are currently performed with energy potentialsranging from 15 to 18 MV, so this neutron dose must be taken into account. In addition, the survival of patients undergoing radiotherapy has increased in recent years and Measurements of Neutrons In A Mixed Gamma-Neutron Field Using Three Different Types of Detectors it is not negligible that a treated patient will, in the long term, appear to have a new tumor in an organ other than that originally treated, which could be radio -induced. Other particle accelerators for medical, industrial or research applications (cyclotrons for the production of radiopharmaceuticals used in nuclear medicine, synchrotrons used for the modification of structures of materials or for medical applications, high energy accelerators for researchcan also originate neutron fields by mechanisms similar to those of LINAC, so that in all this type of installations the dosimetric control of the neutron component of the radiation must be correctly carried out.
One of the problems associated with neutron dosimetry is that neutron fields are usually accompanied by other components of radiation, often much more intense than the neutron component although their contribution to the total dose may be of the same order of magnitude. The main problem with neutron dosimetry is that the neutron weighting factors, which allow the calculation of the equivalent dose from the absorbed dose, depend strongly on the energy of the neutrons, and therefore it is not advisable to carry out its dosimetry without knowing the energy spectrum of the neutron field to be detected, at least approximately. This leads to the fact that to perform the dosimetry correctly it is necessary to have spectrometric information. In addition, the detection of neutrons encounters the added problem that, in the case of neutral particles, they do not directly produce a measurable signal in the detectors, but the detection mechanisms are based on the secondary charged particles originated in the interaction of the same with the different components of the absorbents.
EXPERIMENTAL SETUP
In this work,we usethe electronical system mounted as in different laboratoriesto obtainthe neutron spectrum from the Am241/Be source [2, 3, and 4].The scintillator-photomultiplier system consists of a cylindrical Bicron© BC-400 scintillator coupled to Hamamatsu© R1250, No. RA2457 photomultiplier, powered by two C4840 and three HTV C762-01 Hamamatsu© power supplies, through Ortec© 269 photomultiplier bases. The scintillator has a 120mm diameter, and is 50mm high, and he is coupled to the photomultiplier by a silicon interface. The BC-400 plastic scintillator is known to produce a light output in the range of 400-500 nm, with a peak about 425 nm. The R1250 Hamamatsu© photomultipliers, on the other hand, they are sensitive between 300nm and 650nm, respectively, with peak sensitivity at 420nm, so they are an ideal match for the BC-400 series.
The first step is to verify the scintillation spectrometer using an oscilloscope to check the type and form of the pulses. The system was mounted as can be seenin the Figure 1, where at the bottom the Visiflux contains the source in the middle of the water and in front it is shown the detector. For comparison two crystals are proposed, the BC-400 and the NE213 scintillator systems. The distance from source to detector were 100 cm collimating the beam and directed to the face of the detector, then between the source and the detectors will be used plates of lead and polyethylene to compare the signal. Due to the intensity of the source and the physics in the neutron behavior, the contribution of the neutron source in ambient background was present all the time in the laboratory.
The Bubble Detector Spectrometer (BDS) have six different energy thresholds ranging from 10 keV to 10 MeV [5, 6, 7, 8, and 9]. The number of bubbles obtained in each measurement is related to the dose (standardized response R) equivalent neutrons through sensitivity (b/µSv) and with the neutron flux (neutrons per unit area) through a relationship that is provided by the manufacturer. Bubble detectors were used with six different answers (0.11 b/ µSv, 0.14 b/µSv, 0.17 b/µSv, 0.093 b/µSv, 0.051 b/µSv) [6]. As can be seen in Figure 2, the detectors were placed inside the water container at 4 cm distance. We used several times to achieve low statistical error and to compare the response during the measurements. It was also measure the bare-naked beam from the source to analyze the contributions of thermal neutron. Following the equations1 and 2, it is possible to construct the spectra thru neutron fluence. Where Ni is the neutron fluence for each detector, R i is the quotient of bubble's number and the sensitivity and σ ij is the neutron cross section. The arrangement is the same for both BDS and DLD, using different times depending on whether they are thermal or gamma ray detectors, the objective isto obtain the linear response, the reproducibility and the sensibility of the detectors. The distance to be used are in the range from 2 to 14 cm and times from 8 to 240 min. In every measure the charge was set to 0 to evaluate the response.
The direct reading dosimeter (DLD) is the size of a pocket pen, has a carbon fiber electroscope with an iron chamber for the detection and indication of integrated exposure to gamma radiation and neutrons. It has a thin wall that allows penetration and the detection of radiation. The dosimeters used are:Thermal neutron dosimeter, model 609, Equivalent dosimeter (gammas + rapid neutrons) model 884 and Gammas dosimeter model 5415 (Victoreen).
To obtain the sensitivity of the DLDs it is necessary to calculate the specific gamma constant for the radionuclides used in the experiment by equation 3 Where Γ is the specific constant, K i is the emission's probability, E i is the photon´s energy, µ ρ em is the mass energy-absorption coefficient. To obtain the calibration factor it is necessary to calculate the exposure rate at different distances by equation 4 where X the exposure rate is Γ is the specific constant and d is the source to detector distance.
In Table 1 are shown the DLDs used in the experiment, where two of them are for thermal neutrons and three for gamma rays. Time of measure needs to guarantee that the fiber in the DLDs are in the middle of the range, so the uncertainty in all the dosimeters is the same.
For the last part, we used a single channel analyzer (Berthold Nuclear Spectrometer LB 2040) with a BF3 tube (Berthold LB6400) and a preamplifier (Charge sensitive LB 2008) to measure the spectra of thermal neutrons [10,11]. Figure 3 shows the BF3 detector and the polyethylene moderator. The measures were done at 1 m, in three conditions bare-
Measurements of Neutrons In A Mixed Gamma-Neutron Field
Using Three Different Types of Detectors naked detector, detector with polyethylene moderator and a measure with the source in its shield.
RESULTS
From the modules used, no defective signals were observed on the oscilloscope and cables and connectors were checked to avoid false contacts. Figure 4 and 5 shows the spectra obtained with the two detectors, BC-400 and NE-213, in the case where the detectors were placed in front of the source in the collimated beam and without attenuators the spectra are similar, the first peak corresponds to the signal of the photons and from there the neutron spectrum is shown. To verify the behavior of the spectra two plates, one of lead and another of borated polyethylene, were placed between the source and the detector. The decrease due to neutron attenuation and moderation by the plates in the contribution of both gamma rays and neutrons is observed. For the bubble detectors, we used 18 tubes and count the bubbles manually, for each set of thresholds the equations 1 and 2were used.In Figure 6 we present the BDS before and after the irradiation with neutrons from the Am-241/Be source, the color in the background helped us to count the bubbles using a two-dimension photographic image, some tubes did not present bubbles at all.
The neutron spectra measured at 4 and 26 cm from the source are shown in figure 7. As expected the spectra as a function of water thickness changes not only in altitude but in shapebecause the neutron fluence change due to moderation and absorption inside the water. One parameter that cannot be controlled was the temperature in the laboratory, the manual specified that all the measures had to be done at 20°C, the spectra at 26 cm was obtained at 25°C. This could lead to a change in the response of bubbles detectors.
With the sensitivity of the BDS, we obtained the dose equivalent Hnear the source, then by dividing H per time we get the dose equivalent rate, which can be extrapolated to a different distance. Table 2 shows the data with which H was calculated, the number of bubbles is the average of three measurements performed under identical conditions, it should be noted that the measurements were made 4 cm apart from the source for a time of 45 minutes. According to current Mexican regulations [12], the limit of the dose equivalent for public is set at 5 mSv per year.
If we add the contributions of each energy threshold and then sum the average dose equivalent of the complete spectrum, we obtained a value of2.16mSv/h. Whereas for a distance of 26 cm that corresponds to the outer part of the shield, the dose equivalent rate of the complete spectrum is 61µSv/h. Using a Bonner sphere detector we get a reading of 70 μSv/h value similar of that obtained by the bubble detectors. Table 3 shows the values obtained for the distance of 26 cm and 17 h of irradiation. The gamma exposure rates measured as a function of distance with DLDs are presented in figure 8, as well as a theoretical calculation with eq. 4, as can be seen, the exposure decreases rapidly as function of thickness. The exposure rate measured compared with the calculation has a maximum difference of about 9% for detector D1, detector D2 has a difference of 16% at 2 cm and around 9% for distances greater than 2 cm for detector D2 and between 15 and 28% for detector D3. Figure 9 shows the behavior of DLD4 and DLD5 for thermal neutrons, for comparison we fit an exponential decay equation, showing that no points match the attenuation.
In the case of thermal neutrons, the laboratory background was measured with the source inside its shield but without a polyethylene cover, then the collimated beam was left directly on the naked detector and finally a polyethylene cover was placed on the detector, all measurements were made one meter from the source. Figure 10 shows that both background and bare Figure 8: Exposure rate as a function of distance measured for photons using DLDs. Also shown a theoretical calculation using eq. 4.
Measurements of Neutrons In A Mixed Gamma-Neutron Field
Using Three Different Types of Detectors detector readings coincide, since fast neutrons are thermalized and increase after the polyethylene shielding, for the case of the detector with polyethylene the values are higher because the cover that surrounds the detector moderates more neutrons.
CONCLUSIONS
In the case of the neutron spectrometry system, it has the disadvantage of not being calibrated with another neutron source; in addition, the background of a 5 Ci source is intense, obscuring the signal of interest. DLDs provide fast information, but in the case of thermal neutron detectors cannot be compared with some analytical calculation. The mono channel system will be useful when the counts can be compared with some dosimetric magnitude. The system that provided most information was the bubble detectors, in addition to obtaining a neutron spectrum, the dose equivalent was determined near the source and outside its shield. It was found that it is safe to use the Am241/Be9 source if it is in the water container. | 2019-04-22T13:07:48.291Z | 2017-08-07T00:00:00.000 | {
"year": 2017,
"sha1": "133615eadee6a2edcb6e7819baaed4443b6dcfa0",
"oa_license": "CCBY",
"oa_url": "https://jnp.chitkara.edu.in/index.php/jnp/article/download/80/54",
"oa_status": "HYBRID",
"pdf_src": "ScienceParsePlus",
"pdf_hash": "a306fbfa779115f15e69431f29881efc1c01e0e1",
"s2fieldsofstudy": [
"Physics",
"Medicine"
],
"extfieldsofstudy": [
"Physics"
]
} |
14373730 | pes2o/s2orc | v3-fos-license | Kink estimation in stochastic regression with dependent errors and predictors
In this article we study the estimation of the location of jump points in the first derivative (referred to as kinks) of a regression function \mu in two random design models with different long-range dependent (LRD) structures. The method is based on the zero-crossing technique and makes use of high-order kernels. The rate of convergence of the estimator is contingent on the level of dependence and the smoothness of the regression function \mu. In one of the models, the convergence rate is the same as the minimax rate for kink estimation in the fixed design scenario with i.i.d. errors which suggests that the method is optimal in the minimax sense.
Introduction
Assume that we observe a bivariate dataset {X i , Y i } n i=1 that follows the regression model, where µ is the regression function and σ is the scale function. Also, ε i and X i are the error and random design variables respectively (both being possibly longrange dependent) and X i has cumulative distribution function F = F X : R −→ [0, 1] that is strictly increasing. We are interested in testing the presence of a change point in the slope of a regression function µ and if one exists, estimating its location. We describe this jump in the first derivative of µ as a kink and denote the change point by θ. Knowledge of this change point will allow us to identify change in trends in the underlying regression function of a non-parametric model. This could explain the change in qualitative or quantitative behaviour of an underlying process.
Existing results
Before examining the kink estimation under the random design regression model (1), we first look at other non-parametric and parametric models and their link to the existing theory for kink point estimation. A change point estimation technique was pioneered by Goldenshluger, Tsybakov and Zeevi (2006) for estimating change points in the regression function itself, not the kink scenario. The underlying model assumed for their framework was the indirect model with fixed design. The indirect model assumes that the regression function is not observed in practice but a so called 'blurred' version of the regression function is observed whereby the regression function has been transformed by a convolution operator. More specifically, the indirect model assumes that observations are realisations of the asymptotic model, dY (x) = K µ(x) dx + ǫdB(x). ( In the above model the function K µ(x) = R K(t − x)µ(x) dx represents the convolution of µ and K and the noise is driven by a regular Brownian motion, B(x) and controlled by ǫ ≍ n − 1 2 where the statement a n ≍ b n means that the ratio a n /b n is bounded above and below by positive constants. The fixed design implies that the design variables x i = i n are equally spaced points on the unit interval. The asymptotic model (2) is considered due to a result by Brown and Low (1996) that shows (2) is asymptotically equivalent to the model, where z i is an i.i.d. sequence of error variables. The specific estimation technique that Goldenshluger, Tsybakov and Zeevi (2006) formulated was the zero-crossing technique and it used a particular class of kernel functions to identify the change point. Their technique will be adapted for use in this article and is pursued in further detail in Section 4.1. At this stage it will suffice to say that the main result of their paper established that the zerocrossing technique is optimal in the minimax sense under the framework given in (2).
The zero-crossing technique has been applied by Cheng and Raimondo (2008) to estimate a kink instead of a jump point and was done in the direct model in the fixed design setting. In this framework the observations are assumed to follow a fixed design and realisations derived from the following asymptotic model, Model (3) and its asymptotic equivalent is usually appropriate in practice when a variable is observed at regular intervals indexed by time and the errors are i.i.d. homoscedastic random variables. More recently, Wishart (2009) extended the technique further to include longrange dependent (LRD) noise observations instead of independent noise. The kink estimation technique was extended to include the model, where B H (x) is a fractional Brownian motion with self-similarity index H ∈ [ 1 2 , 1). The noise process was normalised by ǫ α where α = 2 − 2H. Wang (1996) has shown that Model (5) is the asymptotic equivalent to the discrete model, where e i is a LRD sequence of random variables.
In this paper we are interested in model (1), which extends the fixed design cases given in models (3) and (6) above. They are extended in the sense that the design points are no longer restricted to a uniform grid of points and the scale function σ(·) allows heteroscedasticity for the error terms in the regression model. The analysis of this random design model needs to be considered quite carefully, since the asymptotic behaviour of the estimators will depend on the behaviour of the scale function and on the level of dependence present in the design variables and errors themselves. It has been shown by Reiß (2008) that there exists an asymptotic equivalence between model (1) and (4) when σ(·) ≡ constant, and the design variables are independent uniform random variables. However, this is not the case in general. As noted in Kulik and Raimondo (2009a), with LRD design variables, model (1) cannot be equivalent to any asymptotic model, which is in contrast to model (5) being the asymptotic equivalent to model (6) in the fixed design case.
There is an extensive treatment in the literature on both parametric and nonparametric methods for regression models with a random design framework that assume i.i.d. design and error variables. The methodologies used include, but are not limited to, kernel smoothing, wavelet decompositions and orthogonal series. The methods of change point estimation for the random design case have been considered in Gijbels, Hall and Kneip (1999); Huh and Park (2004); Korostelëv and Tsybakov (1993).
There is also literature on the fixed design scenario in the presence of longrange dependent errors and the introduction of dependence in the errors always has a detrimental effect on estimation in this scenario. In the context of function estimation some recent treatments of this topic include Cavalier (2004); Csörgő and Mielniczuk (1995); Johnstone (1999); Johnstone and Silverman (1997); Kulik and Raimondo (2009a); Wang (1996). For change point estimation work has been done by Wang (1999); Wishart (2009).
There is a new emerging literature that attempts to combine the two scenarios with random design regression models where the design variables and/or the error variables are LRD. When the framework includes a random design and possibly LRD variables there is a more subtle asymptotic theory that is based on a delicate balance between the behaviour of the σ function and the level of dependence present. This is evident in a number of papers in the area and will be the case here as well. The interested reader is referred to work by Guo and Koul (2008); Robinson and Hidalgo (1997) for a parametric linear model approach in this context and to Csörgő and Mielniczuk (1999); Kulik and Raimondo (2009b); Mielniczuk and Wu (2004); Yang (2001) for regression estimation in a non-parametric framework. Finally some studies to estimate change points in the non-parametric context include Lin, Li and Chen (2008); Wang (2008).
Article outline
Some preliminary framework is outlined in Section 2, setting up the class of functions that are considered and specific dependence assumptions made in the random design model. The main result of the paper is described in Section 3, along with a brief discussion. The estimation method is explained in detail in Section 4, with a brief outline of the zero-crossing technique in the fixed design and its extension to the random design case. All the necessary proofs of the results are given in Section 5.
Smoothness assumptions and kernels
First we look at the smoothness of the regression function µ and the properties of the kernel function that was constructed to use the zero-crossing technique by Cheng and Raimondo (2008). We define a class of functions that have domain X ⊆ R, a kink at θ ∈ X and s ≥ 3 derivatives that exist in the neighbourhood of θ.
2. µ has a kink, that is, there exists a θ ∈ X and a µ ∈ R with a µ = 0 such that, where µ (1) (θ + ) and µ (1) (θ − ) are the right and left first derivatives of µ respectively. 3. The higher order derivatives µ (i) exist and are finite everywhere and satisfy, 4. For all x + ∈ (0, sup X − θ) and x − ∈ (inf X − θ, 0), Condition 4. should be interpreted in the sense that µ (1) has a separate Taylor expansion for points to the left and right of θ respectively. Condition 3. of Definition 1 might seem overly restrictive but is required to exploit the class of kernel functions that are introduced later in this Section. We will also denote F s (θ) = F s (R, θ). For completeness and comparison purposes we will also introduce another smoothness class G s to denote the class of functions that do not have a kink. This class is identical to F s (θ) except conditions 2. and 3. are relaxed in Definition 1 in the sense that there does not exist a θ ∈ X such that, [µ (1) ](θ) = 0.
In the fixed design setting, we can assume that the domain of the regression function is [0, 1] since any finite interval, [a, b] can be mapped to the [0, 1] interval by an affine transformation. However this assumption is not always valid in the general random design case. In particular, if the design variables are LRD then it is required that they have a domain across the whole real line.
To use the zero-crossing technique for this class of regression functions Cheng and Raimondo (2008) constructed a class of kernel functions via Legendre polynomials and we will denote this class of functions by K s . The full description of the zero-crossing technique and the consequent technical details required of the kernel functions are not covered here and the reader is referred to Goldenshluger, Tsybakov and Zeevi (2006) and Cheng and Raimondo (2008) respectively for full treatment. However, some key aspects will be given and for our case we will say K ∈ K s , where s = 2k + 1 and k ∈ Z + if, where the polynomial coefficients are defined by This class of kernel functions is indexed by the level of smoothness s and is constructed to exploit the extra smoothness of the class F s (θ). To save on notation we denote K i = K (i) , to represent the i th order derivative of K. The kernels have the following properties: Property (10) of K s ensures that the smoothness of F s (θ) can be exploited to obtain faster rates of convergence of the estimator θ in estimating θ. For our purposes of estimation assume that µ ∈ F s (θ) and σ ∈ G r where s ∧ r ≥ 3.
Dependence assumptions
Throughout the paper there will be a dependence assumption either among the design random variables or in the error random variables. In particular, the assumed dependence structure is a causal LRD linear process that is defined below.
Definition 2. Let c i be a set of square summable constant coefficients that are defined, where L : R + −→ R + is a slowly varying function and 0 < α ≤ 1. Then, a random variable ξ i , is said to be a causal LRD linear process if, where |µ ξ | < ∞ and η i are i.i.d. random variables with density f η and moments Eη t = 0 and Furthermore, a random variable ξ i is said to be a causal LRD Gaussian linear process if ξ i satisfies Definition 2 and {. . . , η i−1 , η i } are i.i.d N 0, σ 2 η . The case of α = 1 is to be interpreted as a short range dependent case and by the construction the random variable has Eξ i = µ ξ and Varξ i = 1. Moreover, it can be shown that ξ i is a second-order stationary process and has asymptotic covariance Therefore the process exhibits long-range dependence and a consequence of this asymptotic covariance structure is that, where C 2 1 := 2C 2 0 /((1−α)(2−α)), C 2 2 := 4C 4 0 /((1−2α)(2−2α)) and when 1/2 < α < 1, the sequence Cov ξ 2 0 , ξ 2 i is summable and C 2 3 = 1 + 2 ∞ i=0 Cov ξ 2 0 , ξ 2 i . Also, when α = 1/2, Var n i=1 ξ 2 i is asymptotically proportional to a term of order n times another term involving slowly-varying functions. Now throughout the paper, the design variables and error variables are assumed to follow one of the following dependence conditions: random variables with domain X ⊆ R and common density f such that f (x) > 0 for all x ∈ X and sup x∈X |f (s∧r) (x)| < ∞. The error variables {ε i } n i=1 are a causal LRD process with parameter α ε . Furthermore, the random variables {ε i } n i=1 are assumed to be independent of {X i } n i=1 . Define the associated set of σ-fields, G i := σ(. . . , η i−1 , η i ; X 1 , X 2 , . . . , X i ).
(B) The design variables, {X i } are a causal LRD linear process with parameter α x where f (j) η is a Lipschitz continuous function for j = 0, 1, . . . , s + 2 with f X (x) > 0 for all x ∈ R. The error variables {ε i } n i=1 , are centred and i.i.d., with a finite variance, independent of {X i } n i=1 . Similarly, define the associated set of σ-fields, In both cases, the support of the design variables will be denoted X . Let F = F X be the cumulative distribution function of X which is strictly increasing and denote by F n (x) = n −1 n i=1 ½ {Xi≤x} the empirical distribution function of X. Also let Q = F −1 and Q n = F −1 n be the quantile and empirical quantile functions respectively. We require that Q is Lipschitz, that is, there exists an Finally, we need to impose some mild restrictions on σ. We assume σ is bounded away from 0 and ∞ in the sense that, and that σ ∈ G r where r ≥ 3. Throughout the article we denote by C a general constant that is assumed to be positive and finite but which possibly changes from line to line.
Main result
The main result of the paper is concerned with the construction and analysis of an estimator, θ, of the kink location θ. The analysis of the estimator is given in Theorem 1 and concerns the rate of convergence of θ to the true the kink location θ. The estimator, θ, will be constructed in Section 4 along with the motivations and analysis.
Theorem 1. Suppose a bivariate sequence of observations {X i , Y i } that follow model (1) are observed such that µ ∈ F s (θ) with s ≥ 3. Then an estimator, θ of the change point, θ, can be constructed such that, where C is an arbitrary positive constant and s ∧ r ≥ 3.
It is worth noting at this stage that the further restriction that σ ≡ C under Assumption (A) is unnecessary for the specific estimation technique and is only required in the maximal deviation result to ensure that a kink can be detected in practice. Further detailed discussion of this matter along with the proof of Theorem 1 is given at the end of Section 4.
The minimax optimality of this result is not pursued in this paper since the lower bounds on the convergence rate of θ for the functional class F s (θ) are not determined in the framework of random design. However, it is worth making the specific point that the obtained rate of convergence under Assumption (A) is the same as the minimax rates for the fixed design case with i.i.d. errors (see Cheng and Raimondo (2008)). Consequently, it seems reasonable to conjecture that the rates of our estimator are optimal in the minimax sense.
Kink estimation method
In this section, the basis of the zero-crossing technique is studied and a brief overview given. Firstly, the zero-crossing technique pioneered by Goldenshluger, Tsybakov and Zeevi (2006) and applied by Cheng and Raimondo (2008); Wishart (2009) will be described briefly in Section 4.1 and then an adaptation for the random design case constructed in Sections 4.2-4.7.
Approximation of the third derivative for the fixed design model
In the fixed design setting (cf. model (6)) it can be assumed without loss of generality that the regression function µ has domain [0, 1]. More specifically, assume that µ ∈ F s ([0, 1], λ) and estimate µ (3) (t) by, where h = h(n) is the bandwidth that depends on n. Throughout the article it will be assumed that the bandwidth satsifies, at the very least, h + 1 nh → 0, as n → ∞. This is a standard regularity condition for kernel smoothing techniques and additional conditions on the bandwidth will be stated as needed. Using the functional class F s (θ) and the properties of the kernel function it can be shown where L h (t) is the localisation term. Indeed, by exploiting the conditions of K 3 we can express κ h (t) as follows. Change variable of integration to obtain, The last equality follows because the domain of K is [−1, 1] and the values of t are restricted to t ∈ (h, 1 − h). This restriction is used to avoid possible edge bias effects from the two sided kernel function. Using integration by parts and exploiting the boundary condition (9), Let D = {t : |λ − t| < h} and τ = (λ − t)/h. Then |τ | < 1 for all t ∈ D. We now split (13) into two integrals, The order bound follows by using (7) and (8) in combination with (10). Therefore, this allows us to express κ h (t) in the following way, Since the method is based on estimating a smoothed third derivative of µ, it is assumed that s ≥ 3. This will guarantee that µ (3) exists, is finite and the method makes sense. Then the expansion in the above equation ensures that . More specifically we have the following, As seen in all three of the aforementioned papers that use the zero-crossing technique, the δ−separation rate Lemma given below is the technical result that explains why the above representation is effective.
The proof of this Lemma is given in Cheng and Raimondo (2008). Their proof requires a minor correction as the extra regularity condition 3. is needed in the smoothness class F s (θ).
The main idea of Lemma 1 allows us to exploit the expansion given in (12) and focus in on the location of the kink. The kernel function has specific properties to guarantee that a unique global maximum and minimum occurs within order h of the kink point. Furthermore, the estimator was constructed so that the rate of convergence of kink location estimation is minimax for model (4). We will seek to adapt these results to the random design setting.
Adapted random design estimator of the third derivative
Now consider µ ∈ F s (X , θ) in model (1). An estimator is constructed to exploit the smoothed third derivative of µ and the argument built around Lemma 1 discussed in Section 4.1. The most natural extension would be to use the estimator, where f X (t) is the estimate for the density of X i at the point t given by, Unfortunately, from a brief computational investigation, the estimator given in (15) appears to suffer from poor numerical performance. Instead of using (15), another estimator with better numerical performance is constructed by rescaling the design variables by the distribution function F and defining a rescaled regression function µ F (·) = µ(Q(·)). This new estimator of κ h (t) in the random design setting is given by, Apart from the gain in numerical computation benefits, the estimator κ h (t) has some properties that can be exploited. It reduces the general random design problem into a somewhat simpler framework. To see this, consider F (X i ) to be the new random design variables of the regression problem. These new design variables follow the theoretically easier uniform distribution on [0, 1]. The price paid for this simplification is that the regression function that corresponds to the new design variables is now µ F . This simpler framework is useful for a couple of reasons. Firstly, (16) is an unbiased estimate of the smoothed third derivative of the rescaled regression function µ F , . So, the smoothed third derivative of µ F given in (17) can be exploited by Lemma 1 and the argument shown in Section 4.1. Then the problem is equivalent to estimating a kink location λ for the function µ F in the fixed design setting.
With the previous argument in mind, an estimator θ of a kink location of the regression function µ in the random design setting is constructed that is approximately the same as the estimator for kink location λ of µ F in the fixed design setting. This is done by estimating the value of λ by λ using the established zero-crossing technique in the fixed design setting and then rescaling λ back by the quantile function to obtain an estimate of θ. Thus to assess the performance of our estimator we need to check that the convergence of κ h (t) to κ h (t) is sufficiently fast. To do this consider the two following processes, With these definitions, the overall accuracy of the estimator can be decomposed into, where b h (t) and Z h (t) represent respectively the stochastic bias and stochastic error contributions to the estimator and are given by, The analysis of the above terms are given in the next subsection.
Probabilistic behaviour for the adapted estimator
In this section the analysis of the stochastic bias and stochastic error terms are considered before proceeding to the next stage of the zero-crossing technique to ensure that the stochastic contributions do not overwhelm the signal generated by the κ h (t) term. The proofs of the claims in this section will be deferred to Section 5. The first term to be considered is the stochastic bias term which did not appear in previous kink analyses pursued by Cheng and Raimondo (2008); Wishart (2009) since there is some stochastic contribution by adapting the fixed design estimator to the the random design framework. Therefore, this term needs to be appropriately dealt with and the next Lemma is a useful tool that considers this term.
Lemma 2. Consider a function µ : X −→ R such that µ ′ exists and is bounded. Then define the function If the design variables follow Assumption (A) then, If the design variables follow Assumption (B) then, Note that the two claims in given in Lemma 2 are respectively a uniform law of iterated logarithms for independent variables and a similar type of iterated logarithm result for martingale difference sequences.
We now state some central and non-central limit theorems for the estimator, κ h (t). The convergence of the estimator κ h (t) under both Assumption (A) and (B) is contingent on the size of the bandwidth relative to the level of dependence α. The specific details of this relationship between h and n α will be shown in detail inside the Theorems. Roughly speaking, if the bandwidth is too 'large' compared to α then the dependence of the random variables dominate and the estimator converges to a process that needs to be normed by a sequence that relies on α. Conversely, if the bandwidth is 'small' compared to α then the dependence of the random variables is negligible and a regular central limit theorem holds with a norming sequence that is not reliant on α. In the forthcoming Theorems the extra smoothness of the regression and variance functions are exploited to be able to obtain an estimator that is not as sensitive to the level of dependence. In practice, this extra level of smoothness will most likely be unknown. Due to its common occurrence in the subsequent Theorems, define the asymptotic variance term, υ 2 (t) := σ 2 The following Theorem deals with the case of Assumption (A).
Also if the design variables and error random variables follow Assumption (A) and the bandwidth h = h(n) also satisfies, then the following convergence result holds, Conversely, if the bandwidth h = h(n) satisfies, then, Theorem 3 and Theorem 4 deal with case under Assumption (B) and give the central limit theorems when there is a 'small' or 'large' bandwidth respectively. In the 'large' bandwidth scenario a stronger assumption is used whereby the design variables are a causal LRD Gaussian linear process.
If the design variables and error random variables follow Assumption (B) and the bandwidth h = h(n) satisfies, then the estimator obeys the following law, Assume the design variables and error random variables follow Assumption (B) and that the design variables are a causal LRD Gaussion linear process. If the bandwidth h = h(n) satisfies, and the estimator κ h (t) has a Hermite rank of 1 then the the estimator obeys the following law, where s 2 X = 1 − σ 2 η and φ and Φ are the standard normal density and cumulative distribution functions respectively. Remark 1. If the estimator κ h (t) has Hermite rank q for some q ∈ {2, 3, . . .} then the asymptotic distribution depends on the size of the bandwidth relative to qα. Firstly, if n 1−qαx h 7 L 2q (n) → ∞ then it can be shown using a similar argument used in the Proof of Theorem 4 with the result of Theorem 2 of Avram and Taqqu (1987) that the normed process n qαx/2 L and H q (x) is the Hermite polynomial of degree q and H q is the Hermite-Rosenblatt process, notes a standard Brownian motion. In Avram and Taqqu (1987), they considered Appell polynomials for a generalised sequence of stationary LRD random variables. In our case the LRD variables are Gaussian and consequently the Appell polynomials reduce to the Hermite polynomials. On the other hand, if the bandwidth satisfies n 1−qαx h 7 L 2q (n) → 0 then (21) holds.
As will be seen in Section 4.5, some large deviations results are needed to be able to distinguish between the signal generated by the κ h (t) term and the stochastic bias and noise contributions. Unfortunately, a slightly weaker large deviations result is proved under Assumption (A) in Theorem 5. In particular we assume that the scale function, σ(·) ≡ σ, is constant however this restriction could possibly be relaxed by using a different method. The large deviations result for Assumption (B) in Theorem 6 does not carry this restriction and the scale function need not be constant.
Theorem 5. Let K ∈ K s∧r and the design and error variables satisfy Assumption (A). Further assume that the bandwidth h = h(n) also satisfies, Define, and a partition of [0, 1], where m n = ⌈ 1 2h ⌉. Then, for all x ∈ R.
Theorem 6. Let K ∈ K s∧r , σ ∈ G r with s ∧ r ≥ 3 and the design and error variables satisfy Assumption (B) and assume that the bandwidth h = h(n) also satisfies, Then define, then for T n defined in (24) with m n = 1 2h , lim n→∞ P sup t∈Tn S B n (t) ≤ 2 log m n = 0.
Localisation step
Recall from (12), that the probe function given by κ h (t) gives a signal from the localisation term, L h (t) with some approximation error and the estimator adds a stochastic bias and error term, Clearly, h −2 > h s−3 , since s ≥ 3. So to be able to discern the signal generated from L h (t) = O(h −2 ), it is required that L h (t) dominates the stochastic terms, Z h (t) and b h (t). By construction of the kernel function, (cf. Cheng and Raimondo (2008)), K 1 (·) has two unique extrema in the form of a unique global minimum and maximum in the interval [−1, 1]. This implies that K 1 (·/h) has the same unique extrema in an interval of a length O(h). Consequently, L h (·) has two unique global extrema near t * = λ + O(h) and t * = λ − O(h). As in the fixed design scenario considered by Cheng and Raimondo (2008); Wishart (2009) However, in practice the location of t * and t * are not known and estimated using κ h (t) with, t * = arg min There are two respective bandwidth restrictions, ((A1), (A2); (B1), (B2)) for the asymptotic behaviour of the estimator under Assumption (A) and Assumption (B) respectively. Starting with (A1) and (B1), to have a well defined signal, it is required that, h −2 ≥ Cn − 1 2 h − 7 2 ⇒ h ≥ Cn − 1 3 . Furthermore, since it is assumed that s ∧ r ≥ 3, to ensure that (20) and (21) always hold it suffices to choose h such that h ≤ Cn − 1 7 +(αx∨αε)/7−δ , for some δ > 0 or, for some δ > 0. With this choice, the bandwidth restrictions given by (A1) and (B1) will always hold. It is worth noting that under this choice, the order of the stochastic terms does not involve α x or α ε , the level of dependence. Note that h is chosen in a very similar manner if ε i and X i , i ≥ 1, were i.i.d. Consequently, there will be no influence of the (long range) dependence on the change point estimation. The influence of the long range dependence will only affect testing purposes of the threshold used to determine if a signal is genuine and this will be discussed in the next subsection.
Kink detection step
For simplicity in notation, assume that [µ F ] (1) (λ) > 0, which means, t * < t * (a similar argument follows if [µ F ] (1) (λ) < 0 ⇒ t * > t * ). To detect a kink, first standardise the statistic κ h (t) to have unit variance. This will allow us to appropriately notice if there is a change-point present when the observed extrema of κ h (t) exceed the threshold for the noise process. Define this standardised process as, Then by (26) the T κ (t) process has expansion, As seen earlier, the information regarding a kink is generated by the L h (t) process. A thresholding regime will be considered to be able to distinguish between the signal generated by L h (t) against the noise signal generated by the Z h (t) and b h (t) terms. This thresholding will be split into the two scenarios for Assumptions (A) and (B).
Begin by giving a general decomposition of the estimator for both cases by using, γ * i (t) : and using (18) and (19). So, Focus on Assumption (A) and assume σ(·) ≡ C, constant. The assumption that the scale function is constant is required in the proof of the maximal deviation result in Theorem 5. It may be possible to relax this condition and have the same result for σ ∈ G r under Assumption (A) but remains a conjecture at this stage. Nevertheless, to control the stochastic terms in (30) first apply Lemma 2 and use (10), Then consider the values of t on the initial coarse grid T n (see (24)) where the increments are of size 2h. The grid values will be refined later in Section 4.6. From Theorem 5, it is known that sup t∈Tn S A n (t) will diverge to infinity no faster than 2 |log 2h|. Also, if µ ∈ G s , then from (14), κ h (t) = O(h s−3 ) and However, if µ ∈ F s (θ), then (27) holds and by (29), max t∈(t * ,t * ) T κ (t) ≥ Cn 1 2 h 3 2 > 2 |log 2h| and a kink is detected when, A very similar argument holds for Assumption (B). In this case assume that the scale function σ ∈ G r with r ≥ 3 and proceed as before. In conjunction with (30) and (10) apply Lemma 2, where the extra term D n (t) is defined, Using Lemma 3 (see Appendix in Section 5), with the bandwidth condition (25), sup t∈(h,1−h) |D n (t)| = o p |log h| . Also, the bandwidth restriction (28) guarantees that (25) and consequently Theorem 6 holds. Then for Assumption (B) the same argument applies that was used to show (31) for Assumption (A).
This thresholding technique does raise some restrictions that could possibly be removed by another technique. Recall from (28), that h > Cn − 1 3 +δ for some δ > 0 is required to be able to distinguish the signal from the stochastic terms. Also, (22) and (25) are required to be able to apply Theorem 5 and Theorem 6 respectively and obtain a large deviation result for the process. Therefore to ensure both conditions are satisfied, it is sufficient to consider α x > 8 9 or α ε > 4 9 .
Zero-crossing technique
If a kink is detected (when (32) is satisfied) then the method can proceed to the zero-crossing step. This step considers the interval A h := t * , t * , which will contain λ and t * − t * = O(h). The main idea behind the zero-crossing technique is that for t ∈ A h , κ h (t) ≈ κ h (t). Using Lemma 1 we can locate the zero-crossing-time of κ h (t) which occurs at t = λ with an accuracy of order δ, δ < h. This is done by minimising | κ h (t)| within the interval A h : By comparing (12) with the bounds in Lemma 1 we see that the minimum is well defined if, We will obtain the best possible accuracy if we choose δ as small as possible, as long as both inequalities of (33) still hold. The left hand expression of (33) implies that δ ≍ h s and substituting this into the right hand expression of (33) we derive the order of the smallest possible bandwidth h * ≍ n − 1 2s+1 .
We now apply Lemma 1 with δ * = h s * to locate the change point λ in µ F with an accuracy of order, Remark 2. There are some limitations to the procedure presented thus far in terms of detection. More specifically, dependent on the location of λ relative to the grid values in T n , the detection phase may fail. Indeed, define the closest grid value λ * := arg min t∈Tn |λ − t|. If λ * is too close to λ, that is, |λ − λ * | < δ then the procedure will not detect a kink since L h (t i ) = O h s−3 for t i ∈ {λ * − 2h, λ * , λ * + 2h} and consequently κ h (t i ) = O |log h| and (32) will not hold. However, if δ < |λ − λ * | < h then a kink will be detected since (32) holds and the aforementioned procedures in Sections 4.5-4.6 will follow. Furthermore, the limitations imposed by the coarse T n grid affect only the kink detection step and will not influence on the zero-crossing step.
Modified estimator of kink
Recall that θ = Q(λ). In practice the true distribution function F is unknown, so it is estimated in the usual manner by the empirical distribution function F n (x) = n −1 n i=1 ½ {Xi≤x} and consequently can obtain an estimator of Q via the empirical quantile function Q n (·). Estimate θ by, θ = Q n ( λ). The rate of convergence of this estimator is evaluated below, The rate of convergence in (34) is therefore contingent on the maximum of the rate from the generalised quantile process for the design variables or the rate from the initial unscaled kink estimator. Under Assumption (A), the quantile process involves independent and identically distributed design variables and for all t ∈ (0, 1), (see Csörgő (1983) and references therein for a detailed treatment). For Assumption (B), the rate is dependent on α x and for all t ∈ (0, 1), (see Theorem 5.1 of Ho and Hsing (1996)). Therefore, using (35) and (36) in (34), where s ∧ r ≥ 3 which proves Theorem 1.
Remark 3. The method can be extended to the multiple kink scenario by observing multiple instances of (32). For each instance of (32) there is a corresponding interval A h and the localisation and zero-crossing-time steps are executed on each of those intervals to produce an estimate for each kink location. The interested reader is referred to Cheng and Raimondo (2008); Wishart (2009) for a more detailed treatment of the method in the multiple kink scenario with numerical examples. However, it is worth pointing out that there are some limitations to the accuracy of this method in this situation. Problems will arise if the multiple change-points are not well spaced apart in the sense that they are within order h of each other. To see this, let λ 1 and λ 2 be two such changepoints. When t is within order h of both the change points, the localisation term, L h (t) will not produce two unique disjoint signals for the kinks. Instead the signals generated by K 1 λi−t h for i = 1, 2 will interact and be confounded in one overlapping signal.
Mathematical Appendix
Before giving the proofs, some notation is described. Let X denote a random variable and denote the L p -norm X p p = E |X| p and · = · 2 . For a function f : X −→ R denote the sup-norm |f | ∞ = sup x∈X |f (x)|. Throughout this Section a Taylor expansion of composite functions will be used to exploit the vanishing moment condition of K 3 . For the Taylor expansion to be well defined, the derivatives of the composite functions need to exist. A generalised chain rule for composite functions exists (see the Faà di Bruno formula from Hernández Encinas, Martín del Rey and Muñoz Masqué (2005) and references therein), and these are of the form, (37) where K n = {k i ∈ {Z + ∪ 0} : k 1 + 2k 2 + · · · + nk n = n} and k = n i=1 k i . Also, through tedious but elementary calculus it can be shown that, the n th derivative of Q = F −1 will exist, and the Taylor expansions of µ F and σ F up to order n will exist if f (n) exists.
Proof of Lemma 2. Begin with the proof of the first claim under Assumption (A). Since γ * i (t) will be non-zero only if F (X i ) ∈ (t − h, t + h), there exists a τ i ∈ (−1, 1) that depends on X i such that, and ξ i depends on τ i . The ν i (t) terms are independent random variables, each of which have variance that is of order h. Therefore by the Law of Iterated Logarithm (see Bingham (1986)) we have the following result, which proves the first claim of the Lemma. Now to concentrate on the claim for Assumption (B), a proof of a similar claim in Lemma 4 of Zhao and Wu (2006) is adapted to our framework. This technique bounds the martingale difference sequence γ * i (t) − E [γ * i (t)| F i−1 ] above and below by two discretised martingale difference sequences and uses an exponential martingale inequality to gain the required probabilistic bounds. To do this, again exploit the Taylor expansion of µ in Definition 1 and use the fact that Support(K 3 ) = [−1, 1], which means that there exists a τ i dependent on X i with |τ i | ≤ 1 such that F (X i ) = t + τ i h and, where |ξ| ≤ 1. Then split the function in (38) into its positive and negative parts by defining ξ i := t + ξ |τ i | h and τ i µ denote the respective positive and negative parts of f . Then, By the linearity of the conditional expectation operator and (39) we can decompose the martingale difference sequence into parts, To begin with we will concentrate on the first martingale difference term on the RHS of (40) and bound it above and below by a discretised version that does not depend on t directly. For this discretization let N = ⌈ nh −3 1 2 ⌉ and t j = j N where 0 ≤ j ≤ N. Then for any t ∈ [0, 1] there exists a j such that t ∈ [t j , t j+1 ) and the distance |t j+1 − t j | = O(N −1 ). Define the two new tweaked martingale difference sequences versions of ς ++ i (t), It can be shown that, the martingale difference sequence ς ++ can be bounded uniformly in t above and below by, We have the following result, where for each fixed j, S n (j) and S n (j) are martingales with respect to the filtration F n and are defined, These martingales will be bounded by an exponential martingale inequality. Consider firstly the martingale S n (j), its martingale differences are bounded Also using the Lipschitz property of Q and the bounded domain of K 3 , Then, a martingale inequality for bounded differences given by Theorem 1.5A of de la Peña (1999) can be used to yield, where a = C b h and y = C cv nh 3 . Furthermore if ax/2y = o(1) then using a Taylor expansion of sinh −1 , Now consider the chance that max 1≤j≤n S n (j) exceeds the threshold with order x = C T nh 3 |log h| for some C T > 0 which combined with a = C b h and y = C cv nh 3 implies, ax/2y = O |log h| /nh = o(1) and by (41) and (42), So, fix ǫ > 0 and use (43), By choosing C T large enough will ensure that Cn The similar conclusion can be reached that for any ǫ > 0 there exists a finite constant C such that, Therefore, (44) and (45) Using a comparable argument, the same conclusion can be reached for the S n (j), Also, a similar technique can be used to bound the other martingale difference terms given in (40), details omitted.
Proof of Theorem 2. To prove the Theorem we appeal to similar results that were shown by Kulik (2008); Wu and Mielniczuk (2002) by decomposing the stochastic terms into two parts, a martingale part and a LRD part. This is done by defining, and then decomposing the standardised estimator κ h (t) into two terms, The Theorem will follow by showing that either the first or last term on the RHS of (46) dominates under the bandwidth conditions (A1) or (A2) respectively.
More specifically, it will be shown that the dominating term will follow a CLT and the other term converges to zero in probability; then Slutsky's Theorem completes the proof. Firstly consider the case where (A1) holds, then apply the martingale CLT of Brown (1971) Note that {χ i (t), G i } form a martingale difference sequence. So it remains to check that the sum of the conditional variances converge in probability to the unconditional sum and the Lindeberg condition holds. Before we prove the Lindeberg condition note that for t ∈ (h, 1 − h), Exploiting (10) and the assumption that σ ∈ G r , where τ ∈ (0, 1). Therefore, using (48) and (49), Due to the fact that the bandwidth is assumed to follow h ∈ (0, 1), there exists a h 0 such that for all 0 < h ≤ h 0 < 1, From (48), it follows h −1 Eζ 2 1 (t) → σ 2 F (t) 1 −1 K 2 3 (u) du and similarly from (49), Also, the same argument applies for the γ i (t) term to yield, Now the Lindeberg condition is shown to hold. Let ǫ > 0 be arbitrary, where A n = {|χ 1 (t)| > ǫ}. The size of this set can be maximised using (50), Using the fact that nh → ∞ and h → 0 as n → ∞ we see that A n → ∅, the empty set. Consequently with (51), (52) and nEχ 2 1 (t) < ∞ imply that, and the Lindeberg condition holds. By a consequence of (11), let ǫ > 0 be arbitrary, with both of the above equations being o(1). So, the sum of the conditional variances to converge in probability to one: and by the martingale CLT, (47) follows. Now we show that the last term on the RHS of (46) converges in probability to zero. Consider an arbitrary ǫ > 0, then using (49) and (11), and the last line follows by the bandwidth restriction given in (A1). Thus, the proof of the first claim under the 'small' bandwidth scenario holds.
Consider now the 'large' bandwidth scenario. Using (46), (47) and (49), Also, from Ho and Hsing (1997), it is known that Therefore, normalising the expression on (53), and the result follows from (A2) and (54) with Slutsky's Theorem Proof of Theorem 3. First break down the estimator into its separate martingale and LRD part in a similar fashion to the method employed in the proof of Theorem 2. Using (30), apply Lemma 2, Define the standardised stochastic terms, Then in a similar fashion to the Proof of Theorem 2 it will be shown by the martingale CLT of Brown (1971) Indeed, ∆ i (t) is a martingale difference sequence with respect to the σ-fields {F i }. Thus we need to check that the Lindeberg condition holds and that the sum of the conditional variances converge in probability to 1. First, focus on the convergence of the conditional variances. The conditional variances can be broken into two parts, Dealing with the second term on the RHS of (57), use Lemma 1 of Zhao and Wu (2008), A bound is required for E K 3 to deal with the first term of (57). Define X i,i−1 := X i − η i = µ X + ∞ j=1 c j η i−j and Z i := s −1 X (X i,i−1 − µ X ) and define f η (x) := f X x F i−1 = f η (x − X i,i−1 ) and g(x) = 1/x. Then X i,i−1 and Z i are F i−1 -measurable and for all t ∈ (h, 1−h) the conditional expectation can be evaluated as follows.
Use a Taylor expansion of the composite functions, p(t) := f η • Q (t) and q(t) := (g • f X • Q) (t) by using the Faà di Bruno chain rule given in (37); starting with the latter Taylor expansion, where |τ | < 1. The intermediate derivatives for j = 0, 1, . . . , s ∧ r are given by due to restrictions imposed in Assumption (B). Similarly, (61) where |δ| ≤ 1. Therefore, using (61) and (60) in (59) with the vanishing moment condition (10) implies that, However, by Assumption (B), f (j) η and Q are Lipschitz continuous for j = 0, . . . , s and therefore bounded. Consequently p (j) and q (j) are also bounded which means that uniformly in t, then Eg(X i,i−1 , t) = 0 and by Jensen's Inequality E K 3 (X i,i−1 , t) 2 < ∞. It will be shown by an application of Theorem 1 of Wu (2007) ] to measure the physical dependence. To bound ϑ i , let η ′ 0 be an i.i.d. copy of η 0 and define X * i,i−1 = X i,i−1 − c i η 0 + c i η ′ 0 with the associated sigma field F * i = σ (η i , η i−1 , . . . , η 1 , η ′ 0 , η 1 , . . . ; ε 1 , . . . , ε i ). Then by Theorem 1 of Wu (2005) it was shown that there is a bound ϑ i ≤ sup t∈(h,1−h) g(X i,i−1 , t) − g(X * i,i−1 , t) . Using this, (62) and the Lipschitz property of f η it will be shown that ϑ i < Ch s∧r+2 i −β L(i), where the last line follows due to the Lipschitz property of Q and the bounded domain of K 3 . Then by Theorem 1 of Wu (2007) and Karamata's Theorem, . Using this and (62), Then the first term on the RHS of (57) can be bounded by (63) and a similar application of Lemma 1 of Zhao and Wu (2008), Substituting (64) and (58) into (57) implies that, For the Lindeberg condition, let ǫ > 0 and define A n = {|∆ 1 (t)| > ǫ}, then similar to the procedure used in the Proof of Theorem 2, it can be shown that A n → ∅ and the Lindeberg condition holds. Thus by the martingale CLT, (56) holds and by using (B1) in the decomposition given in (55) the result follows by Slutsky's Theorem.
Proof of Theorem 4. Again, use the decomposition (55) used in the Proof of Theorem 3. Then, define the standardised process, .
It will be shown via use of a Hermite expansion of the LRD variables that, To do this, split the LRD variable X i into two parts, . Then clearly, EG(Z i , t) = 0 and by Jensen's inequality, EG(Z i , t) 2 < ∞. So by Taqqu (1975), G(Z i , t) can be re-expressed by its Hermite expansion, is the m th Hermite coefficient. For our case it is assumed that a 1 = 0. Evaluating a 1 , By exploiting the Faà di Bruno formula further, it can be shown via Taylor expansions that the asymptotic behaviour of a 1 satisfies, From Corollary 5.1 of Taqqu (1975), Therefore (65) holds by Slutsky's Theorem in the decomposition given in (55) in conjuction with (56), (65) and (B2).
Proof of Theorem 6. The proof of the Theorem uses similar moderate deviation inequalities from Grama and Haeusler (2006) that were used in the proof of Theorem 5. However, slight modification is needed. Firstly fix k ∈ N and choose distinct integers 1 ≤ j 1 , j 2 , . . . , j k < m n then modify S B n (t) to obtain a martingale by adding and subtracting the conditional expectation. Define, With this definition, S B * n (t), F n n∈Z + is a martingale and The proof of the result will follow if it can be shown that the supremum over T n for both terms of (66) are o p |log h| . Starting with the latter term, from (62) Also, since k is fixed E sup tj r R B * n (t) 2 = O nh 2(s∧r)+1 = o(1) so by the Chebyshev inequality, P sup tj r R B * n (t jr ) ≥ 2 |log h| = o(1).
Now turn attention to the first term on the RHS of (66) and apply a moderate deviation martingale result from Corollary 2 of Grama and Haeusler (2006). To be able to use their Corollary a bound is needed on the trace norm of the quadratic characteristic matrix of the martingale and a bound on the Euclidean norm of the martingale difference sequence. These will be investigated, starting with the former. For a symmetric k × k matrix U define the trace norm U tr := p i=1 |e i | where e i are the eigenvalues of U . For a sequence x = {x 1 , x 2 , . . . , x n } define the usual Euclidean norm of a sequence |x| 2 := i x 2 i 1 2 . Now let Q be the quadratic characteristic matrix of S B * n,k (t), that is, By a similar domain argument that was presented in the proof of Theorem 5, if r = r ′ , the first and second terms on the RHS of (68) are zero. Using this fact with (62) it follows that, for r = r ′ . On the other hand, if r = r ′ , then, Define the eigenvalues of Q to be e 1 ≤ e 2 ≤ · · · ≤ e k . To evaluate the trace norm of Q consider, Using (58) and (64) in (70), Then consider the third power of the above statement and expand, To give a bound on the expectation of the above term, it is sufficient to look at the higher order expectation terms and apply the Lyapunov inequality. Starting with the expected value of the suprema of the kernel function, using the fact that the kernel function has support on [−1, 1] and t jr − t j r ′ ≥ 2h for r = r ′ , | 2010-03-08T00:38:36.000Z | 2010-03-01T00:00:00.000 | {
"year": 2010,
"sha1": "69605c4557d51554e6daa7aa9f39509730ffe04d",
"oa_license": "CCBY",
"oa_url": "https://doi.org/10.1214/10-ejs571",
"oa_status": "GOLD",
"pdf_src": "Adhoc",
"pdf_hash": "6424a97f69404af87208cf46bdbaf3cc1f538f13",
"s2fieldsofstudy": [
"Mathematics"
],
"extfieldsofstudy": [
"Mathematics"
]
} |
10548711 | pes2o/s2orc | v3-fos-license | Immunometabolic Regulation of Interleukin-17-Producing T Helper Cells: Uncoupling New Targets for Autoimmunity
Interleukin-17-producing T helper (Th17) cells are critical for the host defense of bacterial and fungal pathogens and also play a major role in driving pathogenic autoimmune responses. Recent studies have indicated that the generation of Th17 cells from naïve CD4+ T cells is coupled with massive cellular metabolic adaptations, necessary to cope with different energy and metabolite requirements associated with switching from a resting to proliferative state. Furthermore, Th17 cells have to secure these metabolic adaptations when facing nutrient-limiting environments, such as at the sites of inflammation. Accumulating data indicates that this metabolic reprogramming is significantly linked to the differentiation of T helper cells and, particularly, that the metabolic changes of Th17 cells and anti-inflammatory Forkhead box P3+ regulatory T cells are tightly and reciprocally regulated. Thus, a better understanding of these processes could offer potential new targets for therapeutic interventions for autoimmune diseases. In this mini-review, we will highlight some of the recent advances and discoveries in the field, with a particular focus on metabolic demands of Th17 cells and their implications for autoimmunity.
Interleukin-17-producing T helper (Th17) cells are critical for the host defense of bacterial and fungal pathogens and also play a major role in driving pathogenic autoimmune responses. Recent studies have indicated that the generation of Th17 cells from naïve CD4 + T cells is coupled with massive cellular metabolic adaptations, necessary to cope with different energy and metabolite requirements associated with switching from a resting to proliferative state. Furthermore, Th17 cells have to secure these metabolic adaptations when facing nutrient-limiting environments, such as at the sites of inflammation. Accumulating data indicates that this metabolic reprogramming is significantly linked to the differentiation of T helper cells and, particularly, that the metabolic changes of Th17 cells and anti-inflammatory Forkhead box P3 + regulatory T cells are tightly and reciprocally regulated. Thus, a better understanding of these processes could offer potential new targets for therapeutic interventions for autoimmune diseases. In this mini-review, we will highlight some of the recent advances and discoveries in the field, with a particular focus on metabolic demands of Th17 cells and their implications for autoimmunity.
Keywords: glycolysis, oxidative phosphorylation, immunometabolism, interleukin-17-producing T helper cells, regulatory T cells, autoimmune diseases inTRODUCTiOn T cells respond to alterations in the host environment via multiple steps of activation, proliferation, and finally differentiation into specialized subsets, which are exquisitely programmed to deal with the challenge at hand; whether it is the presence of an intracellular or extracellular pathogen, or other alterations in tissue homeostasis, which may require either inflammatory or regulatory responses. These various responses are tightly regulated by the interplay of specialized CD4 + effector T helper (Th) cell subsets and antiinflammatory Forkhead box P3 + (FOXP3 + ) regulatory T cells (Tregs) (1).
One such type of a specialized T effector cell subset consists of interleukin (IL)-17-producing T helper (Th17) cells. In mice, Th17 cells could be differentiated from stimulated naive CD4 + T cells in the presence of the pleiotropic cytokine transforming growth factor (TGF)-β1 in combination with IL-6. Th17 cells are specialized to respond against certain bacterial and fungal pathogens in the tissue sites in which they are located, namely the mucosal linings of the gut and airway epithelia (2). However, Th17 cells are also known for their pathogenic potential against the host, due to their association with several autoimmune diseases such as multiple sclerosis (MS), psoriasis, and rheumatoid arthritis (RA) (1)(2)(3)(4). More recent studies have demonstrated that Th17 cells can be heterogeneous in phenotype and function and can even show antiinflammatory properties. This pro-inflammatory versus antiinflammatory/ homeostatic phenotype of Th17 cells seems to be determined by a set of specific signaling modules, where pathogenicity is critically influenced by a high level of expression of IL-23 receptor, granulocyte-macrophage colony-stimulating factor, and Th1-like transcripts [e.g., interferon-γ, T-box transcription factor 21 (TBX21/Tbet)] and by the absence of the antiinflammatory cytokine IL-10 (5). The induction of these different Th17 phenotypes can be mimicked in vitro by varying the combination of stimulatory triggers and cytokines. For example, the stimulation of naive T cells with a combination of IL-1β, IL-6, and IL-23 in the absence of TGF-β1 induces the differentiation of Th17 cells that exhibit a highly pro-inflammatory and pathogenic phenotype, compared to "classically" TGF-β1 + IL-6 differentiated Th17 cells (6)(7)(8). Together, these studies illustrate that the pro-inflammatory potential of Th17 cells is extremely sensitive to the presence and combinations of stimulatory cues within the local microenvironment.
Of course in vivo, the tissue site in which Th17 cells are generated does not only consist of cytokines but is also packed with a variety of small molecules, including metabolites. This "metabolomic" milieu is dynamically changing in both its composition and concentration, which may have various effects on Th17 cells, particularly through their cellular metabolism (9)(10)(11)(12). This is highlighted by many recent studies showing even further variation in Th17 phenotype can been fashioned in vitro by the addition of e.g., fatty acids (13,14), phospholipids (15), cholesterol intermediates (16), oxysterols (17,18), and even electrolytes such as sodium or potassium (19,20). In this mini-review, we will highlight some recent advances in understanding the metabolic adaptations and mechanisms employed by T cells when they undergo activation and differentiate into specialized subsets, focusing on Th17 cells.
MeTABOLiC ADAPTATiOnS OF Th CeLLS
The overall aim of cellular metabolism, independently of the cell type, is to generate energy [adenosine triphosphate (ATP)] and metabolites, which are essential for cells to perform various functions, sustain life and growth. Glucose is the major cellular fuel source, and it is broken down into ATP by two separate, but connected pathways: glycolysis and oxidative phosphorylation (OXPHOS). In glycolysis, glucose is broken via 10 enzymatic steps down to pyruvate, yielding two ATP molecules; a process that does not require oxygen. Most cells go on to oxidize pyruvate in the tricarboxylic acid cycle, subsequently fueling mitochondrial OXPHOS, which, in an oxygen-dependent process, yields more than 30 ATP. Alternatively, pyruvate can be converted to lactate, which ultimately feeds back into glycolysis. While glycolysis produces far less ATP, it has several advantages in that it is fast and generates metabolites, so under oxygen-poor conditions, it can suffice to provide energy and necessary biomolecules for the cell (11,12,21).
A metabolic phenomena first observed in cancerous cells by Otto Warburg in 1924 is when proliferating cells preferentially utilize glycolytic metabolism even when oxygen is abundant (22). This aerobic glycolysis, often called "Warburg" metabolism, while bioenergetically less efficient, results in an additional benefit from increased flux into new molecules (nucleotides, amino acids, and fatty acids), as well as pathways involved in generation of redox-protective metabolites. It is now apparent that the activation of T cells induces this same metabolic switch to aerobic glycolysis (Figure 1). In fact, this was first observed more than 50 years ago that the activation of T cells, such as with mitogens like concanavalin A, results in immense changes in glycolytic cellular metabolism (23)(24)(25)(26)(27). More recently, Frauwirth et al. demonstrated the importance of costimulation in initiating these metabolic adaptations (28). Costimulation of CD3 and CD28 induced increased expression of the glucose transporter 1 (Glut1), subsequently leading to augmented glucose uptake and flux through glycolysis. The same has been reported for the expression of amino acid transporters and amino acid uptake (29,30). Functionally, this provides the activated T cell with ATP, but more importantly, the accumulation of nutrients and biomass permits the activated T cell to expand and rapidly proliferate. Adopting an aerobic glycolytic metabolism is now appreciated to be a feature of all Th cell populations (9)(10)(11)21). Furthermore, this switch to aerobic glycolysis is not only limited to proliferating immune cells, as the activation of myeloid cells like dendritic cells (31) and macrophages (32) with toll-like receptor agonists also induces these same metabolic adaptations, indicating that this mechanism may infer a general requirement for cells of the immune system.
Th17 CeLLS veRSUS TReG: A CHOiCe in CeLLULAR MeTABOLiSM
For differentiation into special T effector cell subsets, cellular metabolism remains a process that is distinct and necessary for the acquisition of downstream specialized functions. This is none the more apparent when comparing the metabolic configurations of Th17 cells and Tregs. As a consequence of costimulation in combination with TGF-β and IL-6, the transcription factor RARrelated orphan receptor gamma t (RORγt) is activated leading to the generation of Th17 cells. Conversely, while maintaining a requirement for TGF-β in the absence of other cytokines, with the exception of IL-2, induced Tregs (iTregs) are generated via the activation of the transcription factor FOXP3. Metabolically, these two subsets are incredibly distinct (13,33,34). In vitro stimulation of naïve T cells with Th17-stimulating conditions induces robust glucose uptake and a shift to aerobic glycolysis. By contrast, Tregs instead increase their lipid uptake and utilize energy-efficient pathways such as fatty acid oxidation (FAO) and mitochondrial OXPHOS (Figure 1) (10)(11)(12)21).
The key factors essential for the regulation of Treg and Th17 cell metabolic reprogramming have now been identified as a network of metabolic kinases that function as nutrient/energy sensors and metabolic transcription factors, in line with their regulation of FiGURe 1 | interleukin-17-producing T helper (Th17) cells and regulatory T cells (Tregs) have distinct metabolic requirements. The presentation of antigen and costimulation by dendritic cells induce the reprogramming of T cells into an aerobic glycolytic "Warburg" metabolism, characterized by the increased expression of glucose transporter 1 (Glut1) and augmented glucose uptake. Generally, in the presence of transforming growth factor (TGF)-β with the combination of other cytokines such as interleukin (IL)-6, activated T cells are polarized into specialized Th17 cells, which have an even further enhanced glycolytic metabolism and flux into intermediate pathways, required for the generation of amino acids, nucleotides, and fatty acids essential for Th17 effector function. Conversely, the induction of Tregs from activated T cells by TGF-β and the combination of other cytokines, such as IL-2, results in an increased uptake of fatty acids (FA) and an elevated lipid and mitochondrial oxidative metabolism. Within each subset, there exists further heterogeneity, depending on the presence of further local stimulatory cues, such as other cytokines and small metabolites, thus generating pathogenic or non-pathogenic Th17 cells and various polarized regulatory Treg cells; all of which exhibit further metabolic complexity. Th17 cells are dependent on aerobic glycolytic metabolism, as inhibiting glycolysis with 2-deoxyglucose (2-DG) or dichloroacetate (DCA) prevents Th17 cell generation even in the presence of Th17 cell-promoting conditions. Likewise, the addition of exogenous fatty acids or soraphen A (SorA), which enhances lipid oxidative metabolism, similarly inhibits Th17 cell generation. These metabolic interventions have the opposite effect on the generation of Tregs, which are conversely enhanced by treatments that augment lipid oxidative metabolism and blunted by inhibitors of lipid transport such as etomoxir. Cholesterol derivatives are essential for Th17 cell differentiation and blockade of cholesterol biosynthesis, for example, with ketoconazole, inhibits the generation of Th17 cells but has no effect on Tregs. nutrient transporter expression (9-12, 21, 35). Major regulatory checkpoints are mechanistic/mammalian target of rapamycin (mTOR), the kinase complex that is a promoter of aerobic glycolysis and anabolic metabolism after stimulation of T cells and the AMP-activated protein kinase (AMPK) complex that promotes FAO and catabolic metabolism [reviewed in Ref. (9-12, 21, 35)]. Particularly for Th17 cells, the metabolic transcription factor, hypoxia-inducible factor 1 (HIF-1α), seems to play a special role (36). Under Th17-promoting conditions, HIF-1α expression is rapidly increased in an mTOR-dependent manner, and its deletion prevents the generation of Th17 cells both in vitro and in Th17-promoting disease models in vivo (34,37). As HIF-1α is a transcription factor regulating the expression of metabolic enzymes (36), Shi et al. hypothesized its importance in regulating the cellular metabolic reprogramming of Th17 cells (34). Deletion of HIF-1α under Th17-promoting conditions results in a blunted upregulation of Glut1 and the reduced expression of crucial glycolytic enzymes such as hexokinase 2, phosphofructokinase 1, and lactate dehydrogenase. Together, these data indicate that indeed HIF-1α is an essential facilitator of the acquisition of Th17 glycolytic metabolism (34). In line with this, a recent study identified another important regulatory checkpoint for Th17 cell or Treg-specific metabolic pathway decisions. Gerriets et al. identified pyruvate dehydrogenase (PDH) and the pyruvate metabolism as a key decisive point between T cell glycolytic and oxidative metabolism. The conversion of cytosolic pyruvate into mitochondrial acetyl-CoA is catalyzed by PDH for oxidative metabolism and is inhibited by PDH kinase (PDHK). PDHK is regulated by hypoxia and HIF-1α and promotes the generation of lactate by suppressing pyruvate oxidation (38). By a detailed metabolic analysis, they identified with PDHK1 an isoform that is predominantly expressed in Th17 cells but not in Th1 cells or Tregs. The inhibition of PDHK1 by dichloroacetate (DCA) was able to suppress glycolysis and selectively affected the generation and survival of Th17 cells in part through the generation of reactive oxygen species (ROS) (38).
By contrast, both in vitro generated iTregs and ex vivo isolated thymic-derived Tregs exhibit increased phosphorylation of AMPK (13), a broad sensor of decreased cellular nutrients and energy (39). Subsequently, treatment in vivo with an AMPK agonist, metformin, resulted in an increased generation of Tregs (13). However, conversely, AMPK-knockout CD4 + T cells did not demonstrate deficient Treg generation (40). An explanation for these discrepant results may be because AMPK has a much broader role in T cell metabolic adaptations, not only within the Treg subset. Recently, Blagih et al. showed that T cells deficient in AMPK are unable to adapt in vitro changes in the availability of nutrients such as glucose and glutamine. Subsequently, this translated to an impaired ability of both Th1 and Th17 generation in vivo (41), pointing toward multiple layers of cellular metabolic sensing and adaptation in the acquisition of T effector function, particularly for Th17 cells. In line with this, another metabolic checkpoint for fate decisions between Tregs and Th17 cells was recently described (33). De novo fatty acid synthesis (FAS), which is inhibited by AMPK (42), was shown to be essential for the generation of Th17 cells in contrast to Tregs (33). Accordingly, the inhibition or deletion of acetyl-CoA carboxylase 1 (ACC1), a key enzyme for de novo FAS, resulted in an impaired Th17 differentiation, whereas Tregs were induced (21,33,43).
These findings highlight how the generation of Tregs and Th17 cells is tightly linked to their metabolic state, offering potential new targets for the regulation of these two reciprocally regulated T cell subsets (Figure 1). Indeed, it was shown already in addition to the targeting of AMPK with metformin that the inhibition of mTOR by rapamycin could block Th17 cell generation while favoring Tregs (11). Moreover, the inhibition of lipid oxidation with drugs such as etomoxir, which inhibits the activity of carnitine palmitoyltransferase I, resulting in a reduction of the fatty acid import into the mitochondria (44), impairs the differentiation of iTregs but has no effect on the generation of Th17 cells (13,44). Conversely, inhibiting glucose metabolism with 2-deoxyglucose impairs Th17 cell differentiation while reciprocally promoting iTreg induction (34). Similarly, the addition of exogenous fatty acids (13) or the inhibition of de novo FAS with small-molecule metabolic modulators like soraphen A (SorA) (33) impairs Th17 cell differentiation and promotes Tregs, even in the presence of pro-Th17-inducing conditions. This does not mean, however, that Th17 cells do not require lipids for their metabolic remodeling. It has been recently shown that oxysterols, such as 27-dihydroxycholesterol (18), metabolites in the cholesterol biosynthetic pathway (16), and desmosterol (17) bind to, and agonize RORγt activity, subsequently promoting the differentiation of Th17 cells. Thus, inhibiting cholesterol synthesis, for example with ketoconazole, impairs Th17 differentiation and IL-17 production, but has no effect on Treg differentiation (17). Moreover, PDHK1 blockade by DCA was able to selectively impair Th17 cell function and ameliorated experimental autoimmune encephalomyelitis (EAE) without having significant effects on Tregs or Th1 cells (38).
Th17 MeTABOLiSM: A POTenTiAL THeRAPeUTiC TARGeT FOR THe TReATMenT OF AUTOiMMUne DiSeASeS?
Under homeostatic conditions, Th17 cells are easily able to undergo the critical metabolic changes required for their specialized function described above, as nutrients are plentiful. However, under altered conditions (e.g., infection, disease, diet), metabolic restrictions may occur, which could have significant bearings on Th17 cell function. These recent discoveries discussed before highlight the intimate connectivity between immunity and cellular metabolism leading to the following questions: (a) Do disturbances in T cell metabolism contribute to the development of human autoimmune diseases? (b) Can Th17 metabolism be therapeutically targeted, thereby manipulating the immune system to become less inflammatory, and thus providing protection or treatment for autoimmune diseases with strong Th17 components such as RA and MS?
Besides in animal models of disease, it has additionally been demonstrated that several autoimmune diseases go in line with metabolic alterations and that also T cells of patients have dysregulated metabolic profiles, indicating that indeed an altered T cell metabolism is associated with disease. This was for instance already shown for RA, systemic lupus erythematosus, and MS (11,12,(45)(46)(47). Albeit for most of the diseases premature and incompletely understood, the notion of altered metabolic profiles in human autoimmunity points towards intriguing new avenues for therapeutic interventions. In fact, recent studies have generated promising experimental data indicating the feasibility of this hypothesis. Most of the current reports concentrated here on the reciprocal regulation in regards of cellular metabolism between Tregs and Th17 cells, since Th17 cells primarily depend on glycolysis and Tregs seem to utilize mainly FAO. As described above, the key targets here would be the inhibition of the mTOR/ HIF-1α pathway to prevent the development of the initiation of a "pro-inflammatory" metabolic signature or the activation of AMPK to influence cellular metabolism favoring the generation of antiinflammatory Tregs. In line with this, the interference of de novo FAS by inhibiting ACC1 with the inhibitor SorA was demonstrated to be effective at promoting the generation of Tregs, as well as inhibiting Th17 cells (33). However, the major problem, as for many other currently available immunotherapies, remains specificity. The above-mentioned metabolic reactions are all important in other T cell subsets and immune cells, and therefore, unspecific targeting of these pathways bears the risk of general immunosuppression and/or the interference with the generation of memory cells, which seem to have similar metabolic demands as Tregs; not to mention effects on non-immune cells (9,10,12,21). Moreover, it remains to be seen on how other Th cell subsets react and in addition, importantly, on how these pathways interfere with pathogenic and regulatory/homeostatic subsets of Th17 cells (5). In this respect, it is of interest that it has already been demonstrated that an altered lipid biosynthesis and the fatty acid and cholesterol composition of the cell could impact Th17 cell pathogenicity by generating endogenous RORγt ligands (15)(16)(17)(18).
It is also important to note that for many of the studies described in this review, there are only sparse data on the contribution of these same metabolic pathways to the differentiation of human Tregs and Th17 cells, and the contribution of these molecules to disease in vivo is not clear. Recent studies have, for instance, shown that in contrast to findings in murine Tregs, human Tregs in fact highly depend on glycolysis (46,48). Similarly, murine Tregs have also been shown to be glycolytic in vivo (49). This obvious discrepancy of human to murine Tregs might be associated with the existence of different FOXP3 splice variants in humans, which could dictate their function in contrast to mice (50,51). It was shown that glycolysis through the glycolytic enzyme enolase-1 controls the induction of the FOXP3 splice variant containing exon 2 that is crucial for human Treg function (46). Therefore, defining metabolic targets depending only on in vitro findings and experimental animal models must be carefully examined if they can be fully translated to the setting of human disease. Promising experimental strategies to cope with these problems might be the inclusion of different mouse models expressing different FOXP3 splice variants or the use of humanized mouse models, in combination with the analysis of patient samples as recently demonstrated by Yang et al. for RA (52). This elegant study identified that patient samples exhibited a dysregulation in the pentose phosphate pathway (PPP), which was associated with a depletion in ROS. Insufficient oxidative signaling resulted in ataxia telangiectasia mutated kinase-dependent bypass of the G2/M cell cycle, consequently leading to hyperproliferation and a shift of RA patients into a highly pathogenic Th17/Th1 cell profile. Importantly, this dysregulation could be rescued by metabolic interventions that restored the ROS pool: supplementation with menadione (vitamin K3), disrupting synthesis of the ROS quencher glutathione, or by blocking glucose shunting into the PPP, which was even demonstrated in a humanized mouse model in vivo, transplanted with synovial tissue and T cells of RA patients (47,52).
COnCLUSiOn
The research on immunometabolism and its relation to disease is an extremely fast growing and promising field, which may offer novel therapeutic avenues for immunomodulation in settings of various human diseases. Particularly in autoimmune diseases like RA or MS, metabolic disturbances influencing the Treg/Th17 axis may play a role, and we are just at the beginning to understand the tight linkage between metabolic pathways and immune cell function.
While this increased knowledge of the metabolic requirements and alterations of Th17 cells and Tregs in autoimmunity provides many specific targets for intervention, it may also offer indirect dietary approaches that beneficially influence dysregulated T cell metabolism by systemically changing nutrient or metabolite availability. This is not a new concept, as it is well known that nutrition and the general metabolic condition of individuals has a profound impact on the immune system. Malnutrition, particularly prominent in developing countries, is clearly associated with reduced immune function. By contrast, obesity, which is associated with a "Western" lifestyle and a nutritional pattern high in calories, fat, and salt, goes in hand with the onset of a low-grade systemic inflammation (12,53,54). While obesity is well known to predispose individuals to diabetes and cardiovascular diseases, it additionally seems to represent a risk factor for inflammatory autoimmune diseases like MS (53,55,56). This may be driven largely by the strong effect of obesity on the Treg/ Th17 axis through various metabolites or adipokines like leptin and through pro-inflammatory cytokines like IL-6 (53, 57). An additional molecular mechanism contributing to this effect was recently described by Endo et al. Here, mice fed a high-fat diet had an elevated induction of Th17 differentiation, in a mechanism that was reliant on the expression of ACC1, which subsequently contributed to RORγt activation (43). Therefore, it is important to understand the direct and indirect influences of diet on the immune system and Treg/Th17 axis, or more broadly, the connection of systemic and cellular metabolism and its relation to human autoimmune disease (45). A variety of studies have indicated that particularly dietary interventions could have severe effects on autoimmunity by influencing metabolic parameters of immune cells. It is thus tempting to speculate that, for instance, dietary interventions like fasting mimicking diets, that were recently shown to be beneficial in EAE and MS patients by shifting the Treg/Th17 balance (58), or by lowering the sodium content in diets, that had a strong metabolic impact on macrophages (59,60), could serve as interesting alternatives or additions for drugs directly targeting cellular metabolism.
AUTHOR COnTRiBUTiOnS
All authors listed have made substantial, direct, and intellectual contribution to the work and approved it for publication. | 2017-05-03T19:07:13.098Z | 2017-03-21T00:00:00.000 | {
"year": 2017,
"sha1": "24fecbc5070f2d7d9133b85c1b840e4974d2312d",
"oa_license": "CCBY",
"oa_url": "https://www.frontiersin.org/articles/10.3389/fimmu.2017.00311/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "24fecbc5070f2d7d9133b85c1b840e4974d2312d",
"s2fieldsofstudy": [
"Biology",
"Medicine"
],
"extfieldsofstudy": [
"Biology",
"Medicine"
]
} |
228918234 | pes2o/s2orc | v3-fos-license | Intelligent Generation of Evolutionary Series in a Time‐Variant Physical System via Series Pattern Recognition
Intelligent generation of time‐variant control series remains the critical challenge for acquiring the desired system evolution, due to the difficulties in perceiving temporal correlation and conducting appropriate feedback propagation. A machine learning (ML) algorithm named time‐series generative adversarial network (TSGAN) is developed to overcome the difficulties, by incorporating a long short‐term memory (LSTM) kernel for recognizing multirange temporal patterns beyond the Markovian approximation and an adversarial training mechanism for efficient optimization. A variety of time series are examined by temperature‐control experiments, and the results demonstrate an exceptional accuracy (>95%, 35% higher than prevalent ML methods) as well as strong transferability and stability of the TSGAN algorithm. The dependence of generation performance on underlying statistical mechanisms associated with different ML algorithms, including the deep neural network (DNN), hidden Markov model (HMM), LSTM, and TSGAN, is elucidated by analyzing the generation quality of characteristic temporal patterns. The capability of generating arbitrarily complex response series opens an opportunity for inverse design of time‐variant functionals as strenuously pursued in material science and modern technology.
Such temporal features require the simultaneous perception of both short-term and long-term correlations across the entire dataset, as well as the appropriate weights of these correlations for making predictions. The high-dimensional coupling within the physical system and the limited datasets attained by experiments make accurate generation even more formidable. [26][27][28] Recently, the family of deep generative models (DGMs), including the generative adversarial networks (GANs) and variational autoencoder, has received considerable attention given its powerful capabilities on unsupervised generation, pattern extraction, and feedback propagation. [29][30][31] It learns the evolution mechanism and generates arbitrary distribution samples based on deep learning algorithms to implement a recurrent optimization routine through an intelligent feedback mechanism. In the areas of material physics, DGMs have been used to inversely design molecular structures, to predict material performance, and to detect abnormal reactions by GAN-based algorithms, with the focus on static and spatial features. [32][33][34] We speculate the integration of a temporal correlation kernel and a DGM can pave an avenue to address the critical challenges in generating time series. While time series patterns with different ranges can be characterized by the temporal kernel, DGM algorithms are devoted to reorganizing these patterns and generating time series toward objective functions. Rationally designing the weight of temporal pattern information contributed to DGM optimization is the key to realizing such promise.
In this work, we developed a new DGM to capture multirange temporal correlations and to achieve the accurate generation of time series that govern the physical system evolution. Implementing simple training data and no preset conditions, our model successfully perceived and reconstructed temporal patterns with different ranges, which achieved an average generation precision of 95.2% in our prototype systems (compared to the 59.7%, 52.3, and 66.7% precision for the traditional DNN, HMM, and LSTM, respectively). The accuracy, generality, stability, and diversity of our approach were further demonstrated via generating representative functionals observed in realistic TV-P systems.
Statistical and Physical Inference of Time Series
In contrast to the self-evolution process widely explored in traditional time series, our dynamic model of a TV-P system consists of both the control sequence P(x t ) ¼ P(x 1:tÀ1 ), x t ∈ X ⊂ ℝ n x and the observed sequence P(y t ) ¼ P(y 1:tÀ1 ), y t ∈ Y ⊂ ℝ n y , with their joint distribution satisfying Equation (1) The evolution of the physical system is projected in the observed sequence affected by the control sequence, which manifests as a high-dimensional reaction space and deep time correlation over time. [35,36] To ensure the joint optimization of prediction precision and generation diversity in the learning process, we designed a two-step architecture as shown in Figure 1a: 1) to extract a high-dimensional coupled mapping f θ ð·Þ:X 1∶T !Y T focusing on pattern evolution with temporal and spatial dependencies; 2) to build a generative model for predicting the future patterns of the TV-P system.
For the first step, different series distribution modes lead to differences in constructing methods of mapping space. In case the marginal distribution is Gaussian and its variants, we can apply the expectation maximum (EM) method to solve the maximum likelihood function of θ. This has been achieved by ML techniques such as the HMM and Kalman algorithms within the categories of the DBM. [37][38][39][40] Although the DBM approaches Figure 1. Theoretical framework of our interpretation on time series in TV-P system based on machine learning (ML). a) Schematic illustration of twostage ML algorithm for systems comprising control and observed sequences, with the generative and extractive stages highlighted in yellow and green, respectively. The red and blue circles denote the projection of sequences on high-dimensional parameter space. b) Architecture of the TSGAN algorithm, with the color scheme-blue: recurrent experimental process; green: basic frameworks of LSTM; purple: basic frameworks of neural networks; orange: optimization process. The solid and dashed lines present the generated data and the real data flows, respectively. www.advancedsciencenews.com www.advintellsyst.com have been extended to address distributions other than multi-Gaussian, they come with the price of much higher complexity and lack the capability to be applied in general situations. [41][42][43][44][45] Therefore, to extract salient features and mappings from arbitrary distribution modes of high-dimensional systems, we developed a new algorithm structure, combining a statistical mechanism and deep learning, to appropriately interpret the hidden space of the TV-P system and to mine pattern propagation features. Consider a TV-P system with a control sequence X ¼ {x 1 , …, x T } and an observed sequence Y ¼ {y 1 , …, y T }, and let Θ be the model parameter space that satisfies a general mapping y t ¼ f θ ðx t , y 1∶tÀ1 Þ. The optimal parameters can be obtained by solving the equation for the maximum likelihood function Given the infeasibility of directly solving this equation, we introduced the evidence lower bound functional Lðx, θÞ (Equation (3)) where q θ ð·Þ is the approximate posterior distribution gradually approaching p θ ð·Þ. Lðx, θÞ is equal to Θ given the optimal parameter set θ, which makes q θ ð·Þ identical to p θ ð·Þ. The optimization proceeds via either maximizing E q θ ðyÞ ½lnð·Þ or equivalently minimizing KLð·Þ.
For the TV-P system, we introduced the joint distribution (Equation (4)) of input and output instead of the margin distribution to improve solution accuracy where ∫ p θ ðyÞKLðq θ ðxjyÞjjp θ ðxjyÞÞ is non-negative, and thus a stronger bound KL (q θ ðx, yÞjjp θ ðx, yÞ) is provided. In previous studies, numerical sampling (NS) methods, such as Markov chain Monte Carlo [46] and particle filter, [47,48] have been successfully implemented to numerically calculate the multidimensional integral. These approaches, though, cannot be applied to the TV-P system because 1) physical systems comprising both control sequence and observed sequence with different pattern time-scales may not obey the Markovian property, which is a prerequisite of the NS methods; 2) the description of complicated and discontinuous time evolution patterns characterizing physical systems driven by an external control sequence requires sampling with temporal correlation over tremendous pattern space, which may not be accessed by NS methods given the status generation based on detailed balance. To overcome the first challenge, we used the LSTM algorithm, [41] within the category of a recurrent neural network, to achieve extraction of non-Markovian multirange patterns embedded in time series. The time evolution is driven by where g σ ð·Þ is a deterministic nonlinear transition function, and W o and b o are the weight and bias of the output unit, respectively. The propagation of high-dimensional space is described by the evolution of hidden state h t passing through the synergy of three elements [24] where f t , i t , and C t are composed of different neural networks. The non-Markovian property is achieved by collaboration between the forget unit rationally preserving information accumulated in the past and the input unit incorporating instantaneous information at current time in Equation (6), which efficiently captures multirange patterns of time series. Nevertheless, the network updating in LSTM algorithm solely depends on accumulation of errors at independent temporal points, which inappropriately evaluates the similarity between series patterns and leads to abnormal generation and diversity collapse during deep learning.
Design of the TSGAN Algorithm
For addressing the challenge of insufficient sampling of pattern space and substantially enlarging the scope of time series analyses, we developed the TSGAN algorithm to analyze the complex distribution patterns over different temporal ranges and to further enhance the extraction and generation abilities for time series. Our model comprises generative (G) and discriminative (D) units, both of which are independent LSTM-based modules. The generative learning establishes maximum likelihood functions to train models based on distribution differences and to recurrently proceed via adversarial optimization between G and D. [29] where min G denotes minimization of error in G, max D denotes maximization of the capability of D to distinguish generative and real data, and V (G, D) is the cost function set of G and D for GAN optimization.
It is worth noting that traditional GAN architectures, which have been widely used in qualitative recognition and manipulation, cannot be directly applied to the TV-P systems requesting quantitative sequence prediction, because 1) at the time series generation stage, more accurate measurement of distance between low-dimensional manifolds of distribution patterns in high-dimensional hidden space is demanded to drive the generative sequence toward the target function; 2) at the training stage, the highly dispersed data of training sets lead to irregular gradient propagation and consequentially instable optimization within the hyperparametric space.
Accordingly, we introduced the Wasserstein distance (WD) to characterize the similarity of series distribution driven by Equation (8) [49] Wðp, qÞ ¼ inf γ∈Γ E ðx,yÞ$γ ½jjx À yjj |fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl} distribution transport (8) where Γ ¼ Q ðp, qÞ is the set of all joint distributions γ (x, y) whose marginals are respectively p and q. Compared with the calculation strategy of the traditional KL divergence, the WD Wðp, qÞ can measure the pattern distribution differences of submanifolds between high-dimensional distributions even if no intersection exists between p and q and provide meaningful gradients for backpropagation in any situation. Additional tests have been conducted to elucidate the impact of the WD, and the results suggest that the WD mechanism is crucial for realizing the desired performance ( Figure S17 and Table S2, Supporting Information). In addition, we used an energy-based optimization method in the D unit by calculating the potential function of the TV-P system to fully integrate pattern information. Regarding algorithm stability, we took advantage of the spectral normalization regular (SNR) operation in Equation (9) that overcomes the unequal training of basic GAN and guarantees the satisfaction of the Lipschitz constraint to stably achieve the synchronized training and thus to generalize and transfer into different scenarios. [50] where W θ is the complete weights of GAN and σð·Þ indicates the largest singular value. In contrast to the prevalent methods for calculating gradients, such as weight clipping and gradient penalty, the SNR operationW SN ð·Þ performs on LSTM and full connection layers before backpropagation and intelligently adjusts the weights, which avoids the overlearning of iterative training and preserves the temporal correlation of the welldesigned weights. This results in feature protection of the highly dispersed weight matrix constructed by the G and D units, and thus ensures high stability of generating and training procedures. The workflow of our TSGAN algorithm incorporating LSTM, WD, and SNR to achieve precise and directional manipulation of time series is summarized in Figure 1b; it was developed on the platform of Python v3.6 and Tensorflow v1.12. [51] Detailed information regarding the algorithms can be found in section A of the Supporting Information.
Representative Physical System
To demonstrate our interpretation of the TV-P system based on different models, including DNN, HMM, LSTM, and TSGAN, as shown in Figure 2a Table S3, Supporting Information, we developed a temperature control system with the input power (w) serving as the control sequence and the temperature ( C) serving as the observed sequence. Given the exposure of equipment to ambient environment, the relation between temperature and power is too complicated to be analytically described. For elucidating the algorithm performances on diverse series distribution modes, we designed three complicated test sets, including multi-Gaussian data (MGD), complex Gaussian data (CGD), and complex uncertain data (CUD) as the dashed line show in Figure 2g. The accuracy of test results was directly verified through experiments without model judgment errors. Unified training sets were used in the training processes regardless of test sets and algorithms, which consist of three basic data sets attained in experiments driven by linear rising, natural decline, and linear periodic power inputs (Figure 2f ).
To optimize the performance of ML models, a recurrent experimental system (RES) was integrated with the TSGAN architecture to construct the cross-iterative optimization platform (Figure 1b), with the complete training process shown in Figure S3, Supporting Information. The reinforcement operation combining computations and experiments not only increases the training sets and reduces the dependence of the algorithm on data, but also guides the optimization algorithm toward convergence with the true generative pattern distribution. We indiscriminately implemented the RES to the four types of ML algorithms as documented in Figure S4-S9, Supporting Information. There is substantial improvement demonstrated with the TSGAN algorithm, but the DNN, HMM, and LSTM algorithms fused with the recurrent optimizations failed to converge as explained by their inappropriate evaluations of prediction errors. Therefore, the RES results associated with the first and converged steps are presented only for the TSGAN algorithm in the following discussions.
Learning, Prediction, and Generation of Time Series
The performances of the four algorithms were qualitatively assessed by comparing the target temperature series to those acquired from experiments with the input power series generated by ML calculations. Qualitative comparison among algorithms in the aspects of generation accuracy, time correlation, and pattern diversity is schematically shown in Figure 2e. As expected from their varying capabilities of manipulating temporal correlation, distinct generation results were obtained, as summarized in Figure 2g and section B, Supporting Information, with the corresponding power control sequences documented in Figure S10, Supporting Information. The DNN algorithm manages to generate temperature series with the basic trends roughly matching those of the target functions, yet it suffers from unacceptably low accuracy (34.8%, 39.4%, and 59.6% for MGD, CGD, and CUD) due to the lack of explicit description of the temporal correlation. Despite the introduction of first-order temporal correlation, the generation quality of the HMM algorithm is barely enhanced compared to that of the DNN algorithm for all samples with accuracies of 30.4%, 40.2%, and 14.9%. The LSTM algorithm shows noticeable improvement (accuracy: 41.7%, 44.1%, 62.6%) compared to HMM due to the recognition of multirange temporal patterns, yet it still suffers from the training instability and fails to quantitatively reproduce the target series. The TSGAN algorithm demonstrates its outstanding performance (accuracy: 93.3%, 93.7%, 94.9%) by accurately reproducing all the distribution patterns and periodicities of the target series, which can be attributed to its exclusive capability of extracting and generating temporal correlations. The results of www.advancedsciencenews.com www.advintellsyst.com the first generation with TSGAN are already much better than those of the other approaches, while the high stability and diversity of the algorithm enable further RES optimization with recurrent procedure to gradually approach excellent generation. To elucidate the fundamental mechanisms responsible for the distinct performances of ML algorithms, we investigated the multirange patterns embedded in the temperature series as they can reflect the range, strength, and diversity of temporal correlation. The overall quality of generating patterns with varying lengths was probed by calculating the regional accuracy (RegAcc, Equation (10)), which is defined to evaluate the accumulated deviation of pairwise temperature difference with a certain time interval referred to that of the target series.
where Gen x is the generative sequence, Y x is the target sequence, r is the time interval, ΔT(x i ,r) is the temperature fluctuation within r, i is the index of data point, N is the number of data points obtained in the experiment, and n r is the number of data points associated with the time interval (i.e., pattern length).
The RegAcc results associated with the four algorithms for the MGD, CGD, and CUD samples are shown in Figure 3a. The RegAcc of the DNN algorithm is relatively low and shows a nonmonotonic trend with increasing pattern length, consistent with the absence of temporal correlation. Such behavior can be further understood by observing the replication of patterns within the target temperature series on the generated control power series ( Figure S11, Supporting Information), which arises from the independence and commutativity between data points of an observed sequence during the training process. In contrast, both the RegAcc results of the HMM and LSTM algorithms tend to decrease with increasing pattern length, reflecting their insufficiency in recognizing a long-range pattern.
The first-order Markovian attribute of the HMM algorithm poses an intrinsic disadvantage on perceiving continuous series, given that the artificial discretization leads to inappropriate weighting assignment on time correlation at different lengths. This results in the severe decay of RegAcc for long-length patterns, and thus explains the comparable generation performances of the HMM and DNN algorithms. Much slower decay of RegAcc is observed for the LSTM algorithm, as can be ascribed to the recognition of multirange patterns beyond the Markovian assumption. The RegAcc of LSTM, though, is still unsatisfactory because the discrete accumulation of generated error is inappropriate for measuring the distribution discrepancy between the generated series and the target series, which is even worsened by the optimization routine of gradient descent that is prone to get trapped in the local optimal space. The TSGAN algorithm shows the highest RegAcc (>90%) among the studied approaches, with no attenuation regarding the pattern length. This arises from the improved optimization scheme incorporating an adversarial generation procedure based on comparisons between multirange patterns and explains the superior generation performance of the TSGAN algorithm.
In addition to pattern length, the detailed features of patterns may affect the capability of the ML algorithm to perceive and generate temporal correlation. Accordingly, we decomposed the MGD, CGD, and CUD time series into basic building blocks and then extracted four representative patterns as shown in Figure 3b. The first pattern has local characteristics of linear changes and sharp turning, the second and third are single extreme patterns of different evolution directions, and the last is a complex periodic pattern with a multimodal feature. To evaluate the quality of pattern generation, we calculated the first derivative, median, fluctuation, and integral for all patterns extracted in our target series and their counterparts in generative series, which characterize the propagation trend, the average position, the fluctuation amplitude, and the shape of patterns, respectively, as shown in Figure 3c-f.
The recognition precision of the pattern propagation trend (defined as Prec ¼ N G N T , where N G is the number of generated patterns with correct trends and N T is the number of target patterns within a specific category) with different algorithms is summarized in Figure 3c. As expected, the DNN algorithm can barely generate the correct propagation trend for any type of pattern. The Markovian property makes the HMM algorithm perform much better than the DNN algorithm for both pattern-3 and pattern-4, which resemble the composition of Gaussian distributions, whereas little improvement is attained for pattern-1 and pattern-2. This can be explained by the assumption of marginal distribution as a Gaussian function within the statistical variational framework, based on which the HMM algorithm is originally derived. The deviation of the target pattern from Gaussian distribution causes intrinsic difficulties in optimizing transition and emission probabilities. The predictability of the pattern propagation trend is substantially enhanced by the LSTM algorithm, taking advantage of the collaboration among three gates within the temporal kernel. The generation accuracy for an irregular pattern is still limited by the inefficient backpropagation. Such a deficiency is remedied by the adversarial mechanism of the TSGAN algorithm. Within each recurrent loop, the generator (G) first produces an initial sequence, the discriminator (D) then extracts temporal correlations, and afterward the features between generative and target series are compared to optimize weights of the network until achieving the Nash balance between G and D, which enables the generation of complex and diverse patterns.
The generation quality of different algorithms regarding the position and scope of patterns is shown in Figure 3d-f. The generation errors of both the DNN and HMM algorithms are much higher than those of the LSTM and TSGAN algorithms, while HMM exhibits even lower quality compared to the DNN algorithm. One reason is that the single-value mapping between the input and output of the DNN algorithm guarantees the approach of generation values toward target values despite the incorrect pattern evolution trend. The other reason is that the necessary condition of discreteness in the initial state distribution and transition probability weakens the temporal sensitivity of the HMM algorithm for continuous changeable distribution, which is further demonstrated in Figure S12, Supporting Information. This in combination with its deficiency in manipulating non-Gaussian distribution explains the outliers of pattern-1 associated with the HMM algorithm. The generation errors for patterns are noticeably reduced with the LSTM algorithm by virtue of the improved temporal correlation. The remaining discrepancies stem from the inaccurate optimization methods associated with the serious overfitting phenomenon www.advancedsciencenews.com www.advintellsyst.com and the hysteretic anomaly of temporal correlation, as shown in Figure S13, Supporting Information. It is worth noting that the error distributions on various patterns within the same category obtained by the DNN and HMM algorithms are much broader than those obtained by the LSTM and TSGAN algorithms, implying the low transferability and generalization of the DNN and HMM algorithms. The TSGAN algorithm provides the narrowest error distributions, which indicates that distinct patterns can be precisely and stably generated given simple training sets, and thus confirms that the evolution mechanism has been successfully learned by the TSGAN algorithm. The exceptional accuracy www.advancedsciencenews.com www.advintellsyst.com and transferability of pattern generation with the TSGAN algorithm can be ascribed to the excellent mapping and extraction function contributed by flexibly leveraging and reorganizing LSTM kernels. More importantly, the cross-recurrent adversarial mechanism of the TSGAN algorithm keenly perceives the local features of the D contribution while robustly providing the data information from diverse time scales, which further improves the temporal kernel.
Generation of Time Series in Physical Systems
Although the previous analyses have revealed the potential of our developed TSGAN algorithm on recognizing and generating temporal patterns, further verification of transferability is necessary for implementation in realistic systems, which typically contain complex functionals with nonlinear and non-Gaussian characteristics. To this end, four types of prevalent functionals observed in evolutions of natural and artificial systems were used as target temperature series, with the generation procedure following that of the previous section. The four representative functionals include 1) logarithmic series (Figure 4a), observed in the scrambling process of quantum systems, which describes the operator entanglement entropy and the early growth regime of the squared commutator for the mixed-field Ising model; [11] 2) sigmoid series (Figure 4b), observed in quantum control systems of oscillators, which describes the variation of cavity population associated with oscillator decoherence; [52] 3) unimodal polynomial series (Figure 4c), observed in pyroelectric sensing and light-induced systems, which describes the intensity evolution of photovoltage impulse in response to laser excitation, and the charge density wave peaks on pump laser fluence, respectively; [8,10] and 4) sinusoidal series (Figure 4 d), observed in piezoelectric mimicry, Ramsey experiments, and flexoelectric microelectromechanical systems, which describes the variation of polarization under external force, [53] cavity population stimulated by two resonant pulses, [52] and the displacement evolution in experimental characterization of flexoelectricity. [8] The TSGAN algorithm successfully generates all the evolution functionals with the precisions of 95.4%, 97.1%, 96.2%, and 91.7%, respectively, demonstrating the high accuracy and transferability of our algorithm on generating temporal series. To comprehensively verify the generation ability of TSGAN, we added more complex distribution patterns to increase the diversity of the target data, and the experimental test accuracy has reached more than 93.9% ( Figure S16, Supporting Information).
In addition to its scientific merits discussed previously, the TSGAN algorithm possesses compelling technical value that may enable significant advances in existing control systems. We unveiled such potential by comparing the performance of the TSGAN algorithm and that of a proportional-integralderivative (PID) controller ( Figure S14, Supporting Information), which has been dominantly implemented for temperature management in industry. The results indicate that much higher www.advancedsciencenews.com www.advintellsyst.com precision of generation series, lower fluctuation of control series, and stronger capability to represent series with abrupt turning have been achieved by the TSGAN algorithm compared to a PID controller manipulated by a skillful worker ( Figure S15, Supporting Information). Implementing the TSGAN to common scenarios of time series is necessary to justify the generalization and transfer learning ability of TSGAN. To this end, we have performed two standard nature language processing (NLP) tasks using TSGAN, including semantic analysis and text generation. [54] Satisfactory performances compared with prevalent algorithms have been demonstrated for both NLP tasks ( Figure S18, 19, Supporting Information), implying that TSGAN possesses strong generalization and transferability to intelligently deal with distinct series objects.
Conclusion
We developed a new DGM algorithm named TSGAN to perceive temporal correlation and to accurately generate control time series for accomplishing desired evolution of a TV-P system. With the assistance of an experimental platform wherein the temperature variation is controlled by input power, exceptionally high precision (>95%), the transferability and stability of the TSGAN algorithm have been demonstrated in a range of series that are either representative mathematical structures or prevalently observed in physical systems. By investigating the temporal pattern quality generated by DNN, HMM, LSTM, and TSGAN algorithms, we found that the excellent performance of the TSGAN algorithm originates from the recognition of multirange temporal patterns beyond the Markovian assumption, the efficient generation of samples for assessing tremendous pattern space, and the reliable optimization guaranteed by the adversarial training mechanism.
Although the current discussions are limited to a relatively simple dataset, there is no intrinsic drawback that prevents the TSGAN algorithm from being extended to more complex physical systems. The fundamental relation between the generation performance and the underlying models of ML algorithms elucidated in this study not only determines their implementation scope, but also facilitates the development of novel ML algorithms. We envision that the TSGAN algorithm may generate desired response series driven by complicated time-dependent stimulation signals, and thus pave an avenue toward the intelligent inverse design of time-variant functionals in science and technology.
Supporting Information
Supporting Information is available from the Wiley Online Library or from the author. | 2020-11-19T09:08:09.387Z | 2020-11-16T00:00:00.000 | {
"year": 2021,
"sha1": "ef453e6f847210ef7c43c3486a4b7c74229b82f7",
"oa_license": "CCBY",
"oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/aisy.202000172",
"oa_status": "GOLD",
"pdf_src": "Wiley",
"pdf_hash": "f5879a2387d7df600bb5622d5d6da8eb034e2b76",
"s2fieldsofstudy": [
"Engineering",
"Computer Science"
],
"extfieldsofstudy": [
"Computer Science"
]
} |
247874918 | pes2o/s2orc | v3-fos-license | A Magic Filter Filled with Waste Plastic Shavings, Loofah, and Iron Shavings for Wastewater Treatment
Integrated sewage treatment equipment has been widely used, but the commonly used fillers for wastewater treatment are not suitable in rural areas due to their price and performance issues. In this study, an integrated magic filter filled with waste fillers was proposed and established for wastewater treatment. The filter was composed of functional modules and an equipment room, and the fillers in each module can be taken out separately and changed arbitrarily according to the needs of specific treatment conditions. The fillers used include waste plastic shavings, loofah, and waste iron shavings, generated during the processing of plastic, crop, and steel. At the same time, a 91 d experiment was performed for real wastewater treatment, and a satisfactory removal performance was obtained, with average removal rates of COD, TP, NH4+-N, TN, and SS being 83.3%, 89.6%, 93.8%, 74.7%, and 94.0%, respectively. Through microscope observation, a large number of microorganisms were attached to the surface of the fillers, which was conducive to the simultaneous removal of nitrogen and phosphorus. The micro-electrolysis of waste iron shavings can produce Fe2+ and Fe3+, which would combine with PO43− to form Fe3(PO4)2 and FePO4 precipitates, enhancing the removal of phosphorus. In addition, the filled fillers have an excellent physical filtering effect, which can reduce the effluent SS. The magic filter achieves both the recycling of wastes and the treatment of wastewater.
Introduction
Due to the advantages of a small footprint and short construction period, integrated sewage treatment equipment has become one of the main options for rural sewage treatment [1][2][3][4]. In practice, biological fillers are often added to the integrated equipment for wastewater treatment [5][6][7]. Commonly used fillers include ceramsite, quartz sand, activated carbon, polyvinyl chloride, etc., but high costs, being easy to plug, and poor film hanging performance are the main problems for practice application [8][9][10][11][12]. In this regard, the filler with low cost, a wide source, high efficiency, and simple operation is still limited for rural sewage treatment.
Waste plastic shavings are generated during the processing of plastic, and its recycling has always been a problem. Plastic recycling plants process about 30% of the material received, while the remaining 70% is disposed of in landfills [13,14]. However, it should be noted that waste plastic shavings have many advantages, such as high strength, stable chemical properties, and easy availability [15,16], which basically meet the conditions for being used as biological fillers. In addition, waste iron shavings are produced in the process of steel processing and utilization, which is an easily available scrap metal in the form of rolled flakes, with Fe 0 as the main component. Previous studies have confirmed that waste iron shavings have an excellent effect on phosphorus removal [17,18]. Loofah, an agricultural waste, is featured by its multi-layered fibrous network structure, making it an ideal microbial carrier. Meanwhile, loofah will slowly release the carbon source during the reaction process, thereby saving costs [19][20][21].
To this end, an integrated magic filter filled with waste fillers was proposed and established for real wastewater treatment. The fillers used included waste plastic shavings, loofah, and waste iron shavings. The structure of fillers and sludge morphology were observed and analyzed, and the pollutants removal performance of the magic filter was investigated. This work presents a valuable effort in expanding the practical application of waste fillers in wastewater treatment.
Design Thought
The thought of a magic filter is inspired by the Rubik's cube, which can be combined freely and installed modularly. The magic filter includes an equipment room and functional modules. The equipment room is mainly responsible for controlling the operation of the magic filter, such as influent flow distribution, aeration, sedimentation, water flow direction, reflux, and disinfection. The functional modules are filled with different fillers according to the process and needs. A box-type aquatic plant can be placed on the top layer of the filter, which has both landscape and ecological pollutant removal effects. The functional modules are connected by connecting pipes, and the connection methods include welding, flange connection, flexible connection with rubber parts, etc.
In a practical application, the number of functional modules can be set flexibly according to the site conditions. Filler installation and replacement steps are as follows: filler is added to the filler frame according to the process requirements firstly, and then the filler frame is placed into the functional module in sequence, and finally, the landscape plant is set on the top of the filter. When the replacement of filler is required, the filler can be taken out simultaneously by taking out the filler frame, which is convenient for use.
Reactor Setup and Operation
According to the idea described above, a magic filter filled with various types of waste fillers was established and operated for real wastewater treatment, and a 91 d experiment was performed ( Figure 1). The effective volume was 12 m 3 (2.4 m × 2.4 m × 2.4 m), and the hydraulic retention time was 2 d. The reactor was composed of eight modules and an equipment room of the same size (0.8 m × 0.8 m × 2.4 m). The equipment room was located in the center of the reactor, module 8 was set for disinfection and sedimentation, and the remaining seven modules were filled with corresponding fillers according to processing requirements. A fine grille system (screen pore size of 1 cm), with the same size as the filler frame, was set in module 1.1 for eliminating the large-size particles in raw wastewater. Each module was loaded with three layers of filler frames (0.7 m × 0.7 m × 0.7 m), and each filler frame was equipped with brushes to prevent short flow. To isolate the aerator and support the filler frame, a support structure was provided at the bottom of the functional module. At the top of each module, boxed aquatic plants for landscape and biological deodorization were placed.
The following operational mode was determined via continuous optimization and debugging in the early operation ( Figure 1c). The influent water was distributed to modules 1-3 using a three-stage non-uniform distribution, according to the 3:1:1 flow rate. There were two independent reflux pipelines connected by pipeline pumps in the equipment room, and the return paths were as follows: module 5 to module 1 and module 7 to module 3. In addition, air distribution pipes were installed in the equipment room and connected to aerators at the bottom of each module through an air pump and regulating valve. During operation, module 1 was set for consuming dissolved oxygen in raw wastewater through waste iron shavings, leaving module 2 in an anaerobic state. The dissolved oxygen of modules 3, 5, and 7 was controlled at 2-3 mg/L with an aerobic state, and modules 4 and 6 were controlled at 0.2-0.5 mg/L with an anoxic state.
Filler Selection and Characteristic
The selection of filler should consider its physical filtration and biochemical effect, and from this, the following four fillers were identified: waste plastic shavings, waste iron shavings, loofah, and polyurethane ( Figure 2). Among them, waste iron shavings are mainly used for phosphorus removal, and loofah is also a solid carbon source. Waste plastic shavings and polyurethane can be used as microbial attachment carriers. In addition, the above fillers also have a physical filtering effect.
Thereinto, the waste plastic shavings are obtained from a plastic product factory, in the shape of pleated waves and long strips, with a rough surface, and can produce large gaps between individuals after stacking. Furthermore, it has the advantages of high strength, excellent chemical stability, and easy to obtain, etc. The waste iron shavings are taken from a processing steel plant, with a spiral-shaped and 98.2% Fe content. The loofah comes from agricultural waste, which is an interwoven mesh of multi-layer filamentous fibers, with a long shuttle shape, lightweight, hard texture, slightly curved, thin at both ends, yellowish-white, etc. The polyurethane is purchased from an environmental protection enterprise, with a 99% open-pore rate, and has the advantages of a large specific surface area and easy biological adhesion.
The specific filler configuration is shown in Figure 2. Each filler basically fills the entire filler frame, and the mass of waste plastic shavings, waste iron shavings, loofah, and polyurethane for a single filler frame is about 20 kg, 200 kg, 5 kg, and 10 kg, respectively, and the filling rates are about 58.3 g/L, 583.1 g/L, 14.6 g/L, and 29.2 g/L, respectively.
Wastewater Characteristics and Seed Sludge
The wastewater comes from the domestic sewage of a company's dormitory, which is collected in the septic tank and regulating tank before entering the magic filter. The raw wastewater quality varied greatly, and the main parameters of the real wastewater are as follows: COD (274.4 ± 100.0 mg/L), TN (121.9 ± 26.6 mg/L), NH 4 + -N (102.8 ± 16.1 mg/L), TP (7.9 ± 1.8 mg/L), SS (84.5 ± 38.6 mg/L). The reactor was inoculated with dewatered sludge obtained from a municipal WWTP in Hangzhou, China.
Analytical Methods
Wastewater samples were filtered through 0.45 µm filter paper before analysis. Parameters, such as COD, NH 4 + -N, TP, TN, and SS were measured using the standard methods [22]. Photographs of the fillers were taken using a stereo microscope (Olympus SZ61).
Statistical Analysis
An analysis method of the cumulative frequency with reference to the German ATV-DVWK-A 131E standard [23] was used to evaluate the reactor performance in pollutants removal.
Variation and Replacement of Fillers
The filler structure and sludge morphology were observed, as shown in Figure 3. The surface of the waste plastic shavings was wrinkled and wavy before use (Figure 3a), and a thick biofilm can be observed on the surface after use (Figure 3e), indicating that waste plastic shavings have good bioadhesive properties. In addition, the surface of waste iron shavings was shiny and spiral before use (Figure 3b), but the surface was continuously corroded due to micro-electrolysis during operation. As a result, the surface of the waste iron shavings was rough, and iron deposition and sludge could be clearly observed after use (Figure 3f). Polyurethane was a commonly used filler, and it has a porous mesh structure that makes it easy to adhere to the organism. As shown in Figure 3g, a large number of microorganisms were attached to the polyurethane from inside to outside after use. Loofah had a rough surface and porous structure, which was easily attached to by microorganisms (Figure 3d,h). In addition, carbon sources would be released slowly during operation to save carbon source input costs. With the progress of the reaction, the cellulose, and other components in the loofah were continuously decomposed, so the carbon sources released gradually decreased and the structure gradually collapsed. Similarly, the surface of waste iron shavings would be gradually corroded under the effect of micro-electrolysis. It is time to replace waste iron shavings and loofah when the phosphorus and nitrogen removal performance is poor [18]. The replacement cycle of loofah and waste iron shavings is about 80 d in this study, while waste plastic shavings and polyurethane generally do not need to be replaced due to high wear resistance and strength [15,16,24,25]. During the initial stage of operation, the removal rate of phosphorus fluctuates between 18.5% and 64.3% due to the release of a large amount of phosphorus from the seed sludge. With the reactor reaching the stable operation stage, the TP concentrations in the final effluent showed a significant decrease from 8.43 to 0.09 mg/L, with an average removal rate of 98.5%. It was speculated that the excellent removal rate of TP was related to the addition of waste iron shavings, which was confirmed by previous studies [17,18].
Pollutants Removal
Moreover, the effluent NH 4 + -N concentrations (6.2 ± 0.8 mg/L) were relatively stable during the entire experimental period, with a wonderful removal rate of 93.8%, which showed that the filling fillers have an excellent removal performance on real wastewater. Despite the high TN concentrations (121.9 ± 26.6 mg/L) in the influent, the removal rate still reached 74.7%, which is related to the operation conditions, such as influent flow distribution, the addition of loofah, and the reflux of nitrifying liquid. In addition, the fillers used can effectively avoid the excessive wash-out of biomass at the initial stage of operation because of the good physical filtration performance. Thus, a stable and high removal efficiency of NH 4 + -N and TN can be maintained during the experimental period. The influent SS fluctuated significantly during operation, with a maximum of 120 mg/L and a minimum of 36 mg/L, and an average of 84.5 mg/L. During the entire operation, the effluent SS remained stable below 15 mg/L, and the average removal rate was 94.0%. The low effluent SS was related to the fillers used in the reactor, particularly waste plastic shavings and polyurethane, which could effectively trap impurities and macromolecules in wastewater due to the microporous structure of the filler and the biological effect of surface attachment. Due to the large variation of pollutant concentration in the influent water, the water quality data was statistically analyzed using the cumulative frequency method with reference to the German ATV-DVWK-A 131E standard [23], and the removal rate of COD, TN, NH 4 + -N, TP, and SS could be stabilized at 84.6%, 65%, 91.5%, 89.5%, and 92.0%, respectively. The above analysis shows that the reactor has a reliable and robust performance in pollutants removal.
Proposed Hypothesis of Pollutants Removal Path
As mentioned above, excellent removal performance was obtained, which is related to the operation mode and the fillers configuration of the reactor. The three-stage nonuniform distribution, and simultaneously two independent reflux pipelines (module 5-1 and module 7-3) can save the carbon sources of raw water, enhancing the removal of NO 3 − . Our previous study [18] has confirmed that the effect of iron shavings micro-electrolysis can produce Fe 2+ and Fe 3+ , which would combine with PO 4 3− to form Fe 3 (PO 4 ) 2 and FePO 4 precipitates, enhancing the removal of phosphorus. In addition, the morphology of fillers after use showed that a large number of microorganisms are attached to the surface of the fillers, which was conducive to the formation of anaerobic-anoxic and aerobic environments and the simultaneous removal of nitrogen and phosphorus. At the same time, the filled fillers also have a physical filtering effect, which can retain iron precipitation and microorganisms, reducing the effluent SS.
Conclusions
In this study, an integrated magic filter filled with waste fillers was proposed and established. The filter consists of an equipment room and functional modules. The equipment room is mainly for controlling the operation of each module, and the functional modules can be filled with different fillers according to the process and needs. In addition, the reliable, and robust removal performance was obtained for wastewater treatment, with average removal rates of COD, TP, NH 4 + -N, TN, and SS being 83.3%, 89.6%, 93.8%, 74.7%, and 94.0%, respectively. The magic filter can be freely combined and installed according to the site conditions and obtain better sewage treatment performance by filling different types of waste fillers. Thus, the magic filter achieves both the recycling of wastes and the treatment of wastewater. | 2022-04-03T15:56:32.415Z | 2022-03-30T00:00:00.000 | {
"year": 2022,
"sha1": "7a6ffe6e3867f2c347cfe91bf3f40b11c066e94d",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/2073-4360/14/7/1410/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "07d378d8629e4a0a35464dbebdf1617f6a694ea6",
"s2fieldsofstudy": [
"Engineering"
],
"extfieldsofstudy": [
"Medicine"
]
} |
242452736 | pes2o/s2orc | v3-fos-license | A Fuzzy Analytic Network Process for Security Risk Assessment of Web based Hospital Management System
,
I. INTRODUCTION
Software has become one of the essential needs of users in all the important fields such as medical, engineering, airlines, etc.Given the integral role of software applications and the increasing reliance of the users' in facilitating their tasks through software, it is imperative to ensure optimum security of the software in use.Developers cannot ignore the risk of having insecure software.Hence, the developers need to find newer techniques and adopt necessary action at the time of security risk assessment.In fact a report by the National Institute of Standards (NIST) in 2002 stated that more than $59.5 billion was consumed on repairs of faulty software annually [1].Such reports are a forewarning for all software developers to research on security risk-management activities and create safeguard mechanisms in both cyber security and software security.Most of the times it is seen that ignorance of security is caused by the cost and time spent on it.The generation and development of a security tool for making software systems risk free at the time of design in SDLC is a highly extensive and costly task for any organization.Security risk analysis is an inseparable part of development process.Also, the assessment of security risk includes seven attributes that include: Human and organization factors, security requirements, insecure design, integrity, availability, resilience and non-repudiation.These attributes are the so called keystones to identify the risk and vulnerabilities in security and to mitigate them [2,[14][15][16].Healthcare organizations and security of its applications is a booming and an expansive field for researchers nowadays.Also, patients' data is considered to be highly sensitive information.The trespassers are always ready to target the applications whether it is that of the system or the user's [3,[17][18][19].Any organization that is developing any application for hospital management should manage the security risks associated with it.Furthermore, security risk assessment is a growing area which includes the development and assessment of new tools and technologies for security risk mitigation.The main aim of security breakers or the intruders is to exploit the vulnerabilities and acquire all the sensitive information of the hospitals and patients through which they can control the running system.Hence to avoid all the possible disruption, security risk assessment provides a mechanism through which security risk can be mitigated before delivering the application to the end stakeholder.An enactment of security risk assessment plays an important role for enhancing the security.Security risk assessment along with the planned process of software development improves the understandability for security and predicts the upcoming vulnerabilities which may occur at any time [2,4,[20][21][22].At the same time, growth of security issues or happenings is a rising concern amongst the investors and the IT industries.The organizations which do not focus on the security risk assessment during the development phase of software deliver applications that can contain dangerous vulnerabilities that may pose enormous risks to the users.Aligning with this contextual framework, the second section of the research paper focuses on the needs and importance of the security risk assessment using MCDM techniques.The third section describes the analytic process in detail.The fourth section demonstrates the assessment of security risk using fuzzy ANP.The fifth and the sixth section discuss the results and conclude the article.
II. NEEDS AND IMPORTANCE
The fundamental aim of this paper is to find the priorities based on weightage and ranking of the security risk attributes by applying multiple criteria choice making process under which the use of analytical network process is shown.The idea is to provide more security and reliability to the application software.
A Fuzzy Analytic Network Process for Security Risk Assessment of Web based Hospital Management System
Nawaf Rasheed Alharbe
Since very few attempts have been made for prioritizing and ranking the attributes of security risk for hospital management system which may affect the success of software security and their trade-offs, the fuzzy ANP techniques can be used for prioritizing the attributes of all the security risk.The hybrid technique of Fuzzy ANP gives more accurate results in comparison to the classical ANP technique because it is helpful in eliminating the problem of uncertainty and vagueness in the decision making process.This process of analysis of prioritization of security risk attributes is a type of Multi-Criteria Decision Problem because of the multiple attribute contribution in it [5,6,[23][24].Multi-Criteria Decision Analysis (MCDA) is helpful for performing various evaluations of conflicting elements like Multi-Attribute Utility Theory and Analytic Network Process [7].Objectives, alternative weights and their ranks are three different parts of MCDA process which are used in ANP technique.Further, it is important to underline in this context that the Analytic Network Process (ANP), introduced by Thomas L. Saaty, is a generalized network form of the AHP or in other words ANP is an increased clarification of AHP.Also, Saaty recommended the use of AHP for solving the independent criteria formed in a hierarchy.Whereas, ANP can be used in dependent criteria formed in a network [8,9].The figure-1 describes the actual differences between AHP and ANP which are given below:
Figure-1. Difference between AHP & ANP
There are some real world problems of decision-making criteria which cannot be systematized as a hierarchy.This is because of the interdependence of criteria on each other for forming a network type of a structure rather than a hierarchical one.Furthermore, ANP is crucial in creating network organization equally in AHP, attributes in the lesser layer may impact on the advanced level of attributes.There is one more concept different from AHP which is called the -super matrix‖.It is present in ANP and not in AHP [11].
III. METHODOLOGY
Several recent applications and research works are available in literature in which ANP method has been employed and it gives better results.For example, Rajeev Kumar et al. uses ANP method for the security of software [5]; Goztepe et al. used fuzzy ANP process towards prioritizing six sigma projects [2].Dagdeviren et al. also used fuzzy ANP model to classify faulty behaviour risk in system [4], etc. Chang et al.
[3] used the fuzzy ANP method to assess the risk level of intra organizational cultures and found that the security risk attribute is related to the enterprise resource planning system.Chinho Lin in 2014 applied Fuzzy ANP for supplier selection at an electronics company.The results showed one of the attributes called triple bottom line as an important attribute among the others [4].The methodology used in ANP is shown below and all the defined stages of ANP method are as follows: In this process, all the attributes, sub-attributes and its alternatives that were taken have been elaborated upon.After these steps, the groups of all the items which are taken for prioritization and ranking are determined.Since, ANP is purely based on network process so the association is formed amongst the groups and within each group also more associations are created.As a result, some associations which are different are formed and these have their own properties [14].Items that show self-interaction is also an effect which provides more flexibility and the last one is a mutual effect in which inter-dependencies among criteria is shown.
Step1: After taking linguistic values from the experts and practitioners, authors convert it into crisp numbers and Triangular Fuzzy Number (TFN).TFN can be defined as (p, q, r), where p, q, and r (p ≤ q ≤ r) are parameters indicating the smallest, the middle value, and the largest value in the TFN, respectively.In addition, a fuzzy number M on F is called TFN, if its membership functions is given in equations (1-2) and shown in Figure 2. ( Experts allocated scores to the attributes affecting the values in a quantitative way conferring to scale which is shown in table 1.Using the equations (3-6) numeric values are transformed intoTFN [14] and signified as (l ij , mi ij , u ij ) where, l ij is said to be lower value, mi ij is middle value and u ij is called as uppermost level values.Moreover, TFN [ɳij] is recognized as the following: …………….(3)…………(5)
Table 1. Linguistic terms and the corresponding TFNs
J ijk in the above equations indicates the values given by expert d and the comparative importance of the values between two criteria also in which i and j signify a pair of criteria given by expert d.Assessment ɳ ij .i.e; TFN is calculated using the geometric mean of expert's opinions for a specificjudgment.Further, equations (7-9) help to aggregate TFN values.Consider two TFNs M1 and M2, M1= (p 1 , q 1 , r 1 ) and M2= (p 2 , q 2 , r 2 ).The rules of operations on them are as: Step 2: Prepare the pair-wise comparison matrix by using the responses received from the decision makers.Calculation of the Consistency Index (CI) using the formula in equation 10 as follows: …….( 10) where CI: Consistency Index and Q : is the number of compared elements.Further, calculation of the Consistency Ratio (CR), using a random index is as following: ………..( 11) where RI = random index (CI of the randomly generated pairwise comparison matrix).Random index is derived from Saaty [9] Step 3: Next step is the construction of the comparison matrix.Also, defuzzified values are calculated to create a numeric value based on the produced TFN values.The defuzzification method used in this article has been taken from [12] as formulated in equation (12)(13)(14) which is said to as the alpha cut method.(12) where 0 ≤α ≤ 1 and 0 ≤ β≤ 1 Such that, (13) (14) α and β in these defuzzification equations are said as preferences of number of experts respectively.αand β values vary between 0 and 1.
Step 4: Next step is the formation of the super-matrix which is the result of the priority vector from the paired comparisons between groups including goal, criteria, sub-criteria, and alternatives.
IV. ASSESSMENT OF SECURITY RISKS
Classical Analytical hierarchy process and Analytical network process techniques do not help in gaining unambiguous and clear results in complex and imprecise situations.Hence the use of fuzzy ANP in situations where decision makers are uncertain regarding the level of weights becomes important to overcome the problem of uncertainty.
This makes FANP an efficient method of decision making in problems having multiple criteria.Security risk has been a thrust area for different fields.Ensuring data integrity of immensely classified information like the patients' record, diseases and medication, etc., are important for hospital management application.Hence, the assessment of security risk and its attributes is important for mitigating the risks at early phases of development.Assessment needs attributes of security risk have to be identified.The elements discussed in the proposed paper for security risk prioritization incorporate the following 7 attributes [3]:- Human and Organization Factor (F1); Security Requirements (F2); Insecure Design (F3); Availability (F4); Integrity (F5); Non-repudiation (F6) and Resilience (F7) Security risk attributes are usually a qualitative measure.So, it becomes a stimulating task to assess the security risk attributes quantitatively.Thus, the weightages and ranks of security risk factors help in gaining high software security.For the development of secure software, prioritization of security risk attributes used a multi-criteria decision-making (MCDM) process [8][9].The authors have used fuzzy ANP which gives a newer approach while considering the interdependence of attributes over each other [12][13].In reality, there is interdependence among the nodes and the alternatives in the real-world scenario.Hence, fuzzy ANP provides a network framework among the various attributes and the many alternatives that could be undertaken when it comes to problems related to decision making.This research is focused on choosing the best and highly prioritized attribute among the number of attributes of security risk by using a hybrid methodology of multi criteria decision methods for security risk attributes.For identification of security risk attributes we have used a questionnaire session with a group of experts specifically for hospital management system, aggregated the attributes in seven groups as shown in table 2. Alpha-cut method for Defuzzification of local priorities of security risk attributes and formation of super matrix from all the local priority vectors are shown in table 3 and table 4, respectively.Further, weighted matrix is obtained from super matrix as shown in table 5.This is followed by limit supermatrix obtained from weighted supermatrix which has been shown in table 6.Finally, in table 7 we obtained global priorities of security risk attributes.In this paper, an application of fuzzy ANP has been proposed on security risk factors for hospital management system.The authors in this work proposed a network structure of multitudes of security risk attributes through which a complete relationship and inter-dependencies among these attributes is realized.The data was collected from different academicians and experts form the area and twenty valid responses were collected for assessment.According to the results achieved from the assessment, it is clear that security requirements (R2) have the highest priority among the seven attributes of security risk.
V. CONCLUSION
The process of assessing security risk contains multiple attributes within it.In this paper, we have identified suitable attributes for hospital management system and we have proposed Fuzzy ANP methodology for assessing security risk.The subjective judgments of experts have been defined using TFN and then converted to crisp values using steps in Fuzzy ANP.We have also studied the interdependencies between different attributes of security risk.This assessment will help the developers to mitigate security risk early in the development life cycle by generating guidelines which could prove to be instrumental for the software developers. | 2019-11-22T00:47:49.029Z | 2019-11-10T00:00:00.000 | {
"year": 2019,
"sha1": "58d426b707b2a436f78e1f6e455d45b21adc3369",
"oa_license": null,
"oa_url": "https://doi.org/10.35940/ijitee.a4593.119119",
"oa_status": "GOLD",
"pdf_src": "ScienceParsePlus",
"pdf_hash": "22eee6e0cf3fe5c8290af7bca0ec4f3393e11249",
"s2fieldsofstudy": [
"Computer Science"
],
"extfieldsofstudy": []
} |
234945038 | pes2o/s2orc | v3-fos-license | First record of naturalization of Scaevola taccada (Gaertn.) Roxb. (Goodeniaceae) in southeastern Mexico
Scaevola taccada (Gaertn.) Roxb. is native of Asia and eastern Africa but has been introduced into the Americas as an ornamental urban plant. This paper reports, for the first time, the presence of Scaevola taccada in natural environments from southeastern Mexico. Several populations of S. taccada were identified during a botanical survey of the coastal dunes of the Cozumel Island Biosphere Reserve (State of Quintana Roo, Mexico) aimed at recording the most common plant species. Scaevola taccada is considered as an invasive species of coastal areas in this region. Evidence of its invasiveness is suggested by the fact that populations consisting of individuals of different size classes are found distributed throughout the island. Furthermore, they appear to belong to different generations since we found individuals of different sizes and with and without flowers and fruits. Our observations suggest that the population of Scaevola taccad a can be in a process of expansion. Further studies are necessary to understand the potential effect on the native flora.
Species from the genus Scaevola L. are commonly found in environments that are either arid or with prolonged dry periods (Grande and Nozawa 2010).Only three native species, S. wrightii M. Gómez, 1890; S. plumieri (L.) Vahl, 1791 andS. rialagartensis Cast.-Campos, 2019 are known from the American continent and the Caribbean, all of which mainly inhabit the beach and coastal dunes (Thieret and Brandenburg 1986;Howarth et al. 2003;Gutiérrez-Báez et al. 2013;Castillo-Campos et al. 2019).Scaevola wrightii is endemic to Cuba (Liogier 1962;Howarth et al. 2003), whereas the other two species are native to Mexico.Scaevola plumieri is found in some states of the Mexican Pacific coast, the Gulf of Mexico and the Yucatan Peninsula (Flores Guido 1983;Téllez Valdés and Cabrera Cano 1987;Avendaño-Reyes 2008;Gutiérrez-Báez et al. 2013;Parra-Tabla et al. 2018), although it is also distributed in other tropical and subtropical regions of the Northern and Southern hemispheres (Howarth et al. 2003).Scaevola rialagartensis is a recently described species, endemic to the northern part of the Yucatán Peninsula (Castillo-Campos et al. 2019).Scaevola taccada (Gaertn.)Roxb., 1814 is a species native to Asia that currently has a pantropical distribution (Emura et al. 2014).In the Americas, the distribution range of S. taccada covers from the USA to Venezuela, crossing the Caribbean basin (Grande and Nozawa 2010;Howarth et al. 2003).In Mexico it is often used as an ornamental plant in hotel and urban gardens in coastal locations (personal observations); however, to date, populations growing in natural conditions have not been recorded.
As part of a botanical survey of the coast of the Cozumel Island Biosphere Reserve, Quintana Roo, Mexico, we observed specimens of the genus Scaevola (Goodeniaceae), which we thought might belong to the recently described S. rialagartensis (Castillo-Campos et al. 2019).Under closer examination, we noticed that the specimens showed characteristics that were different from those of S. rialagartensis.A review of type specimens of Scaevola species available in the JSTOR Global Plants (JSTOR 2019) and Tropicos (Tropicos.org 2019) databases showed that the characteristics of our specimens corresponded to Scaevola taccada.
This paper reports, for the first time, the presence of Scaevola taccada in natural environments in the Cozumel Island Biosphere Reserve, municipality of Cozumel, State of Quintana Roo, Mexico.The species was found naturalizing on sand dunes and beaches where only S. plumieri had previously been recorded (Flores Guido 1983;Téllez Valdés and Cabrera Cano 1987;Collantes-Chávez-Costa et al. 2019).
Materials and methods
A wide literature search for floristic and vegetation studies carried out in the Yucatán Peninsula and/or the Island of Cozumel was conducted by surveying the major academic electronic databases (Journal of Citation Reports [JCR], SCOPUS, Google Scholar), aiming to locate records of Scaevola taccada.
Because the island is a Biosphere Reserve, the collection of specimens for herbaria is not permitted.Thus, in order to make an in-depth comparison of the characteristics of Scaevola plumieri and S. taccada, we examined voucher specimens of the genus Scaevola collected in Mexico and deposited in major Mexican herbaria (CHAPA, CICY, ENCB, MEXU, and XAL, see Index Herbariorum NYBG for more information about herbaria).The characteristics of the voucher specimens were compared to those recorded in at least 20 specimens in the field using a hand lens (20x Coddington, Bausch & Lomb, USA), measurements made with a calliper, and detailed photographs.Afterwards, scientific illustrations were drawn based on high-quality photos taken with a Canon SX270 HS 20x zoom lens and the characteristics recorded in the field.Species characteristics recorded in the field and those observed in the herbarium specimens were used to make the taxonomic description presented here.
Spatial distribution
To determine the distribution of this species throughout the island of Cozumel, in August and November 2019, and again in February 2020, we thoroughly explored the coasts of the island using all the available coastal roads and via boat where there were no roads available, on the northeastern part of the island.Transects 1 to 2 km long, covering most of the coastline of the island, were walked to record as many plants as possible.Points were taken for each individual, using a Magellan GPS.The largest and smallest diameters of the individual were measured to estimate the plant's area using the ellipse formula.When it was not possible to distinguish between individuals, the cluster of individuals were recorded as one single plant.Geographical data were used to generate a distribution map, and with the individual plant cover records we built a frequency histogram with the different plants size classes.
Results
A literature search through the major academic electronic databases (JCR 2020; SCOPUS 2020; Google Scholar 2020) showed that Scaevola plumieri is found along the coast of the Mexican Pacific Gulf of Mexico, and Yucatan Peninsula (Flores Guido 1983;Téllez Valdés and Cabrera Cano 1987;Avendaño-Reyes 2008;Gutiérrez-Báez et al. 2013;Parra-Tabla et al. 2018), whereas S. rialagartensis is endemic to the Yucatan Peninsula (Castillo-Campos et al. 2019).No previous scientific records of the presence of S. taccada in Mexico were found.
Plant communities and spatial distribution
Scaevola taccada is usually found in the shrub stratum of coastal dune scrub vegetation, growing on white sandy soils, at elevations from 0 to 7 m above sea level, mainly on beaches in the northeast part of Cozumel Island (Figure 3).The prevailing climate type at Cozumel island is Am(f)(i), Köppen's classification, i.e., warm, humid with summer rains, mean temperature of 25.5 °C and mean annual rainfall of 1,570 mm (García 1988).The species commonly associated with S. taccada include: Bidens pilosa L. ( Asteraceae We recorded 1,565 plants and plant clumps throughout the Cozumel Island coastal area (Figure 3).Despite the fact that S. taccada is used as an ornamental plant in urban and hotel gardens in the central western and south-western coastal areas, respectively, most individuals of this species are located in the areas specifically destined for the protection of terrestrial and maritime flora and fauna in the north and northeastern side of the island (DOF 2012; SEMARNAT 2016).We also observed different sizes of S. taccada, ranging from very small with only 0.10 m in diameter (0.008 m 2 of coverage), with the largest clumps reaching more than 15 m in diameter (> 200 m 2 of coverage) (Figure 4).Most individuals were found to be between 1 and 5 m in diameter (0.40 m 2 and < 20 m 2 ), which are mature and reproductive plants.This suggests that the S. taccada population has been expanding in the coastal zone, the smaller plants being of a later generation from the larger, more established plants.
Discussion
Scaevola taccada was introduced to the Island of Cozumel as an ornamental plant for urban and hotel gardens.It blooms almost all year round; however, flowering peaks occur between May and November and produces fruits from July to December.The plant escaped from these sites and has spread widely to the entire coastal area of the island.The fact that we observed populations consisting of individuals of various size classes distributed throughout the island is evidence of this species' invasiveness.Furthermore, the species seems to have become naturalized along the Mexican coasts, but mostly concentrated along the Mexican Caribbean, as indicated by the occurrence information found in the database from the Global Biodiversity Information Facility (GBIF.org2020).Here 26 records are reported along the Mexican Caribbean, with 4 additional occurrences on the Island of Cozumel, and three more in northern Yucatan Peninsula (Figure 5).As mentioned earlier, to our knowledge, these findings have not been verified with field observations and are not reported in the published scientific literature.Thus, this is the first report on the invasion by S. taccada along the coasts of the Island of Cozumel, with field corroboration.Apparently, the effective invasiveness of S. taccada is associated with the dispersal attributes of its fruits, rather than by the occurrence of a longlasting seedbank.Although local studies in Mexico do not exist, studies elsewhere can help understand the dispersal modes of the species.In this sense, Emura et al. (2014) observed that fruits collected from the Okinawan islands in Japan had two morphs: one floats on seawater and thus can be dispersed by ocean currents, and the other is eaten and dispersed by birds.On the Island of Cozumel we observed seeds of S. taccada in small vertebrate droppings in the northeastern part of the island.It is also possible that birds are eating the fruits.However, we do not have information on which species might be acting as dispersers in Cozumel.Ocean currents are also likely to have promoted the local dispersal of the species, especially because sea water facilitates seed germination by breaking the hard-coat dormancy (Lesko and Walker 1969).
In contrast with the above, there is no evidence that a permanent longlasting seed bank of S. taccada seeds is inducing its invasiveness, since fruits and seeds are predated in the soil and do not accumulate (Louda and Zedler 1985;O'Dowd and Lake 1991).
The broad expansion of S. taccada along the coasts of the Island of Cozumel is probably affecting the natural ecosystems of this Biosphere Reserve, especially those from the beach and coastal dunes.In fact, on-going studies in Cozumel are showing an impact in community composition and structure of beach and coastal dunes vegetation (Castillo-Campos et al. unpublished).In other tropical locations, such as Florida, Gordon (1998) found ecosystem effects of S. taccada, but did not observe a competitive advantage of this species over the native flora.
The potential impact of this broadly expanding invasive calls for more detailed studies on the local reproductive biology and dispersal forms of the species, as well as the effect that the expansion has had on the local flora.This information is highly relevant to help local authorities to implement control programs and tighter restrictions on using S. taccada as an ornamental plant.
Figure 2 .
Figure 2. Scaevola taccada (Gaertn.)Roxb.a, branch with flowers and fruits; b, detail of the indument; c, flower; d, detail of the flower with the corolla, two sepals and two stamens removed; e, stamen; f, stigma; g, detail of the infructescence.Illustration by Edmundo Saavedra.
Figure 4 .
Figure 4. Frequency of Scaevola taccada (Gaertn.)Roxb.individuals in to the different size categories according to their coverage (m 2 ) registered on the Island of Cozumel, Quintana Roo, Mexico.
Figure 5 .
Figure 5. Occurrence records of Scaevola taccada found in the Global Biodiversity Information Facility (GBIF) along the coasts of the Yucatan Peninsula (map from GBIF.org). | 2021-05-22T00:03:00.796Z | 2021-01-01T00:00:00.000 | {
"year": 2021,
"sha1": "82412783438314854f6898d1782ac8645d59d8c8",
"oa_license": "CCBY",
"oa_url": "https://doi.org/10.3391/bir.2021.10.2.21",
"oa_status": "GOLD",
"pdf_src": "ScienceParsePlus",
"pdf_hash": "e6a8e337b3d041562248b503595a5124e8f4ecdc",
"s2fieldsofstudy": [
"Environmental Science"
],
"extfieldsofstudy": [
"Geography"
]
} |
8412233 | pes2o/s2orc | v3-fos-license | A Genome-Wide Association Analysis Identified a Novel Susceptible Locus for Pathological Myopia at 11q24.1
Myopia is one of the most common ocular disorders worldwide. Pathological myopia, also called high myopia, comprises 1% to 5% of the general population and is one of the leading causes of legal blindness in developed countries. To identify genetic determinants associated with pathological myopia in Japanese, we conducted a genome-wide association study, analyzing 411,777 SNPs with 830 cases and 1,911 general population controls in a two-stage design (297 cases and 934 controls in the first stage and 533 cases and 977 controls in the second stage). We selected 22 SNPs that showed P-values smaller than 10−4 in the first stage and tested them for association in the second stage. The meta-analysis combining the first and second stages identified an SNP, rs577948, at chromosome 11q24.1, which was associated with the disease (P = 2.22×10−7 and OR of 1.37 with 95% confidence interval: 1.21–1.54). Two genes, BLID and LOC399959, were identified within a 200-kb DNA encompassing rs577948. RT–PCR analysis demonstrated that both genes were expressed in human retinal tissue. Our results strongly suggest that the region at 11q24.1 is a novel susceptibility locus for pathological myopia in Japanese.
Introduction
Myopia is a refractive error (http://en.wikipedia.org/wiki/ Refractive_error) of the eye in which parallel rays of light focus in a plane anterior to the retina resulting in blurred vision. Myopia is one of the most common ocular disorders worldwide, and is in much higher prevalence in Asians than in Caucasians. Recent population-based surveys in the elderly reported that the prevalence of myopia was approximately 25% in the Caucasian populations [1] and 40% in the East Asian (Chinese and Japanese) populations [2,3].
Myopia is divided into two distinct subsets, namely, common and pathological myopia. Pathological myopia, also called high myopia, is distinguished from common myopia, also called low/ moderate myopia, by excessive increase in axial length of the eyeball, which is the most important contributor to the myopic refraction [4,5]. The axial length of the eyeball in adults is approximately 24 mm, and its elongation by 1 mm without other compensatory changes results in a myopic shift of 22.5 to 23.0 diopters (D). It has been shown that distribution of the axial lengths of the adult myopic population is bimodal [6], and the subgroup with elongated axial length in the bimodal distribution corresponds to pathological myopia. This group comprises 1% to 5% of the population [3,7], and is commonly defined by axial length greater than 26.0 mm which is equivalent to refractive errors greater than 26 D [8].
The excessive elongation of the eyeball causes mechanical strain with subsequent degenerative changes of the retina, choroid, and sclera. The degenerative changes at the posterior pole of the eye such as chorioretinal atrophy or posterior staphyloma are clinically important and unique to pathological myopia [9]. These unique degenerative changes at the posterior pole result in uncorrectable visual impairment due to decreased central vision and make pathological myopia one of the leading causes of legal blindness in developed countries [10][11][12][13].
The genome-wide association (GWA) study using single nucleotide polymorphisms (SNPs) as markers is an alternative approach to identify genetic risk factors of common diseases. This approach has been successfully applied to identify genetic risk factors for multigenetic diseases including ophthalmic diseases such as age-related macular degeneration [33,34] and exfoliation syndrome [35]. To identify the genetic risk factors of pathological myopia, we conducted a two-stage GWA-based case/control association analysis using 411,777 markers with 830 Japanese patients and 1,911 Japanese controls (297 cases and 934 controls in the first stage, and 533 cases and 977 controls in the second stage).
Characterization of the patients with pathological myopia
A total of 839 pathological myopic patients with axial length greater than 26.0 mm in both eyes were enrolled in the current study. In order to maximize the detection power, patients with axial length greater than 28.0 mm in both eyes were enrolled in the first stage of genome scan. No other clinical features were accounted for the assignment of patients to either stage. 824 out of 839 patients (98.2%) had degenerative changes specific to pathological myopia. Other features of cases and controls who passed quality control procedures of genotyping results (see Materials and Methods) were summarized in Table 1.
Genome-wide association analysis
For the first stage, we scanned the genome of 302 cases using the Illumina HumanHap550 BeadChip, which launches 561,466 relatively frequent SNPs (minor allele frequency .0.05) distributed across the human genome at an average interval of 6.5 kilobases (kb). Five cases and 149,689 SNPs were excluded due to quality control criteria (see details in Materials and Methods) and genotyping results of 411,777 SNPs in autosomes for 297 cases were used for the statistical analysis. They were compared with 934 controls from the JSNP database [36] for association with phenotype using x 2 test for trend. Genomic Control (GC) method [37] revealed only a slight inflation of the test statistics (GC parameter l = 1.068). We identified 29 SNPs in 22 chromosomal regions with P-value adjusted by GC being smaller than 10 24 ( Figure 1 and Table S1). Among them, seven SNPs at chromosome 8p12 were in strong linkage disequilibrium (LD) and likewise two SNPs at chromosome 10q22.2 (pair-wise D9.0.95 and r 2 .0.9). Thus, we selected one representing SNP from each region and tested 22 SNPs in the second stage.
Author Summary
Myopia is one of the most common ocular disorders with elongation of axis of the eyeball. Pathological myopia or high myopia, a subset of myopia which is characterized with excessive axial elongation and degenerative changes of the eye, is a leading cause of visual impairment. Since genetic factors play significant roles in its development, identification of genetic determinants is an urgent and important issue. Although family-based linkage analyses have isolated at least 16 susceptible chromosomal loci for pathological or common myopia, no gene responsible for the disease has been identified. We conducted the first genome-wide case/control association study of pathological myopia in a two-stage design using 411,777 markers with 830 Japanese patients and 1,911 Japanese controls. We identified a region strongly suggestive for the disease susceptibility at chromosome 11q24.1 containing BLID and LOC399959. Their expression was confirmed in human retina with RT-PCR. BLID encodes an inducer of apoptotic cell death, and apoptosis is known to play an important functional role in pathological myopia. We believe that our study contributes to further dissect the molecular events underlying the development and progression of pathological myopia. For the second stage analysis, 537 cases and 980 population controls were genotyped by Taqman method. Among them, four cases and three controls were excluded due to low call rates (,90%). Genotyping success rates of the 22 SNP markers in the remaining 1,510 samples were greater than 96.8%. The genotype counts of the first and second stages were combined for metaanalysis. One SNP, rs577948, showed a strongly suggestive association (P = 2.22610 27 ) ( Table 2) in the meta-analysis whereas the remaining 21 SNPs were not significant (P .10 25 ) ( Table S1).
Evaluation of the region with rs577948
The SNP rs577948 which showed P = 2.22610 27 by metaanalysis with OR of 1.37 (95% confidence interval (CI): 1.21-1.54) for the risk allele (nominal P = 2.80610 25 and P = 1.42610 23 in the first and second stages, respectively) ( Table 2) was located at chromosome 11q24.1 (Figure 2A). Using the results of the first stage, an LD block which extended a 55-kb region containing rs577948 was generated. Six additional SNP markers within the block were included in the genome scan chip ( Figure 2B). Among them, we selected three markers with adjusted P-value smaller than 0.01 in the first stage for further genotyping by Taqman method with DNAs used for the second stage. Weaker associations than that of rs577948 were obtained for these three markers by meta-analysis ( Table 2). As shown in Figure 2B, two genes were located in a 200-kb region containing rs577948. BLID is a cell death inducer containing BH3-like motif [38], which is located approximately 44-kb upstream of rs577948. The other gene, LOC399959, is a hypothetical non-coding RNA [39] which encompassed 114-kb DNA in the region, and rs577948 is located in its second intron.
Expression of the BLID and LOC399959
BLID is known as a cell-death inducer expressed in cytoplasm, in mitochondria at lower abundance, and in various human cancer cells from different tissues [38]. LOC399959 was reported as a hypothetical non-coding RNA with a relatively ubiquitous expression pattern. We assessed the expression of the genes by RT-PCR using cDNAs of human retina and brain and those of HeLa cells as positive control. Expressions of both genes were detected in human retinal tissue as well as in human brain and HeLa cells ( Figure 3).
Discussion
Myopic refraction and axial length are reported to be a complex trait under polygenic control in which contribution of each gene is relatively small [40]. In the current study, two-stage GWA analysis identified a region at chromosome 11q24.1, in which rs577948 showed strongly suggestive P = 2.22610 27 with OR of 1.37 (95% CI: 1.21-1.54) for the allele G.
Our GWA study identified only one strongly-suggestive locus. This may principally be due to the sample size of our study not being adequate. Recent genetic studies of complex traits with higher prevalence enroll much larger number of samples. In contrast, recruitment of patients with pathological myopia is difficult due to its lower prevalence, particularly those with degenerative changes (namely degenerative myopia). In order to improve insufficient detection power, we assigned pathological myopia patients with longer axis (greater than 28.0 mm) to the first stage. This strategy might be the reason we were successful in identifying the candidate region with relatively small number of cases.
Insufficiency of detection power due to a limitation in sample number may be a reason for difference between the findings of preceding linkage studies and ours. OMIM database lists 10 MYP regions (MYP1-5, 11-13, 15 and 16) for pathological myopia [18][19][20][21][22][23][24][25][26][27] and 6 MYP regions (MYP6-10 and 14) for common myopia [28][29][30]. None of these 16 MYPs are on chromosome 11q. Stambolian and colleagues reported heterogeneity LOD score of 1.24 at 11q23 in their linkage study for common myopia in Ashkenazi Jewish descent, which is the closest locus to our region reported to date [29]. Because the linkage signal was not strong and the band 11q23 (chr11, position 110,000 kb to 120,700 kb in the NCBI database) is more than 800 kilobases apart from our LD block in 11q24.1 (chr11, position 121,535 kb to 121,590 kb), whether or not they overlap each other is inconclusive. On the other hand, our study did not identify the associated SNPs in any of MYPs.
Although the insufficiency of detection power may be a reason for difference between our study and the linkage studies, there are other possible reasons. In general, any difference in the study designs could cause heterogeneous results. Firstly, there are two definitions of pathological myopia based on two distinct criteria, namely, the axial length and refractive error. In the current study, we enrolled pathological myopic patients based on the axial length (greater than 26.0 mm in both eyes), and not on the refractive error commonly used in the previous studies (refractive errors greater than 26 D). We focused on patients with visionthreatening degenerative changes [9] and the axial length fits better than refractive error for our purpose. The mean refraction in our myopic patients was 213.1464.57 D (eyes that had undergone cataract surgery or corneal refractive surgery were excluded from this calculation) which indeed correspond to pathological myopic group in the previous linkage studies. On the other hand, it is not clear whether the patients enrolled in the linkage studies fulfill our criteria because the distribution of axial length and degenerative phenotypes in the cases are unknown. The difference in definition of pathological myopia may result in different susceptibility loci between studies.
Secondly, the methodology used is different between studies, namely, linkage analysis and association analysis using linkage disequilibrium mapping. The results of linkage and association studies of complex genetic traits are often different. Family-based linkage analysis is much more suitable for identifying rare genetic variants with large effects whereas SNP-based GWA analysis is more powerful in detection of relatively common variants with smaller effects in complex diseases [41].
Finally, the difference can also be due to the ethnicities of the samples enrolled. In the current study, all cases and controls were Japanese. Only one genome-wide linkage study has previously been published for pathological myopia in Japanese [42] and the others were for non-Japanese populations.
It would be interesting and important to examine the association of our locus in other ethnicities. Ethnic variations in disease susceptibility genes have been reported in various genetic traits including ophthalmological disorders. One such example is an SNP in the complement factor H gene (rs1061170) which has a large effect size with age-related macular degeneration in Caucasians [33,43,44] but much smaller in East Asian populations due to a remarkably lower risk allele frequency (,35% in Caucasians and ,5% in East Asians) [45]. Another example is exfoliation syndrome and LOXL1 where the risk allele of rs1048661 is inverted between Icelandic (allele G) and Japanese (allele T) populations [35,46]. Because of a large variation in prevalence of myopia among ethnic groups, a future trans-ethnic investigation of myopia risk genes will be important to dissect genetic backgrounds underlying the etiology of myopia.
Although the susceptibility locus contains BLID and LOC399959, it seems premature to discuss the involvement of LOC399959 in myopia since it is a hypothetical non-coding gene. BLID plays a proapoptotic role involving the BH3-like domain by inducing a caspase-dependent mitochondrial cell death pathway [38]. Indeed, several animal and pathological studies suggested the functional role of apoptosis in pathological myopia [47,48]. Moreover, a recent genome-wide linkage study followed by a fine-scale association mapping identified a myopia susceptibility gene locus containing the PARL gene which inhibits the mitochondrial pathway of apoptosis by interaction with OPA1 [49]. In this context, BLID seems functionally relevant with the pathogenesis of pathological myopia. However, the true functional origin of association in this region has yet to be determined by further detailed investigation along with replication studies to validate our findings.
Study subjects
All procedures used in this study conformed to the tenets of the Declaration of Helsinki. The Institutional Review Board and the Ethics Committee of each institution approved the protocols used. All the participants were fully informed of the purpose and procedures, and a written consent was obtained from each.
Japanese pathological myopic cases were recruited at the Center for Macular Diseases of Kyoto University Hospital, the High Myopia Clinic of Tokyo Medical and Dental University, and Fukushima Medical University Hospital. All subjects underwent comprehensive ophthalmologic examinations, including dilated indirect and contact lens slit-lamp biomicroscopy, automatic objective refraction evaluation, and measurement of the axial length by applanation A-scan ultrasonography (UD-6000, Tomey, Nagoya, Japan) or partial coherence interferometry (IOLMaster, Carl Zeiss Meditec, Dublin, CA).
As a general population control of the first stage, genotype count data of 934 healthy Japanese subjects were obtained from the JSNP database [36]. For the second stage, 980 healthy Japanese individuals were recruited at Aichi Cancer Center Research Institute. Genomic DNAs were extracted from peripheral blood leukocytes with QuickGene-610L DNA extraction kit (FUJIFILM Co., Tokyo, Japan).
Genome-wide association analysis
We designed to scan the genome in two stages. A total of 839 patients and 1,914 controls were separated into two groups; 302 cases and 934 controls for the first stage, and 537 cases and 980 controls for the second stage. In order to increase the detection power, patients with longer axis of the eyeball (greater than 28.0 mm) were principally assigned to the first stage.
For the first stage analysis, 561,466 SNPs were genotyped in 302 patients of pathological myopia using Illumina Human-Hap550 chips (Illumina Inc., San Diego, CA). This chip covers approximately 87% of the common genetic variations in the Asian population [50]. Cluster definition for each SNP was performed using Illumina BeadStudio Genotyping Module. A systematic quality control procedure of the genome scan results was applied as follows. Samples were evaluated for data quality first and markers were subsequently excluded. Genetic proximity of sample pairs was evaluated with pi-hat in PLINK [51] and four samples with indication of kinship or sample duplication were excluded. Genotypes in X chromosome were used for checking the precision of the phenotype record, and only one sample was removed due to mismatch in gender. The final sample size of pathological myopia was 297. As a population-based control, genotype count data by the genome scanning of 934 healthy Japanese subjects using the same chip were obtained from the JSNP database [36]. The chip contained 515,154 markers in autosomes that are common in the cases and controls. We excluded 78 SNPs due to low successful call rate (,95%) in the cases, 1,760 SNPs due to the distortion of Hardy-Weinberg Equilibrium (HWE) in the controls (P ,10 23 by HWE exact test) and 46,722 monomorphic SNPs. 54,817 SNPs with minor allele frequency less than 0.05 in both cases and controls were also excluded. After these quality control procedures, a total of 411,777 SNPs were used for the statistical analysis. The genotyping call rate was greater than 97.43% (median call rate 99.99%) for DNA sample and 98.21% (median call rate 100%) for SNP marker.
Association between genotypic distribution of each SNP and the disease was examined using a x 2 test for trend. The OR and the 95% CI were estimated using Woolf's method [52]. Inflation in the test statistics was assessed using the genomic-control method [37]. Haploview [53] software was used to infer the LD in the targeted regions. SNPs with P-value adjusted by genomic control being smaller than 10 24 were selected as candidates for second stage. Among the candidate SNPs, LD indices (D9 and r 2 ) were calculated with Haploview and when multiple SNPs were in strong LD (D9.0.95 and r 2 .0.9), one representative SNP was chosen to be genotyped in the second stage.
In the second stage, 537 cases and 980 controls were genotyped with the Taqman SNP assay using the ABI PRISM 7700 system (Applied Biosystems, Foster City, CA). The 302 pathological myopic cases in the first stage were also genotyped to validate the concordance between Illumina Infinium assay and Taqman assay. Samples with low successful call rate (,90%) were excluded from the study. Subsequently four cases and three controls were excluded and data of 533 cases and 977 controls were used for the analysis. The concordance rate ranged between 98.68% and 100% for the 22 SNPs. The genotype counts of the first and second stages were combined for meta-analysis using the Mantel-Haenzel method [54] as a fixed-effect model. The OR heterogeneity between the first stage and the second stage was evaluated using Cochran's Q-statistic P-value. The data from the second stage were also evaluated for association independently from the first stage.
Screening for BLID and LOC399959 expression Human retina cDNAs were obtained from Takara Bio Inc. (Kyoto, Japan). Total RNA of HeLa cells and human whole brain were also obtained from the same manufacturer and cDNAs were synthesized using the First-Strand cDNA Synthesis Kit (GE Healthcare Life Sciences, Piscataway, NJ). Two pairs of oligonucleotides were synthesized for RT-PCR; 59-TTGGGTTCCAA-CAAAGAACC-39 and 59-CTTTTACAGGGCCTCAGCAG-39 for BLID, and 59-GGCGACATCAGACAGACAGA-39 and 59-AGGACCAGCTGAAAGGAACA-39 for LOC399959. Expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was tested for cDNA quantification using 59-GACAACAGCCTCAAGAT-CATCA-39 and 59-GGTCCACCACTGACACGTTG-39. PCR reactions were performed under the following condition: initial denaturation at 96uC or 2 minutes, followed by 35 cycles (for BLID and LOC399959) or 18 cycles (for GAPDH) at 96uC for 20 seconds, 60uC for 40 seconds, and polymerization at 72uC for 40 seconds.
Supporting Information
Table S1 Summary results for the 29 SNPs significant at the P ,10 24 level after population stratification adjustment in the first stage of the genome-wide association analysis. Found at: doi:10.1371/journal.pgen.1000660.s001 (0.06 MB XLS) | 2016-05-12T22:15:10.714Z | 2009-09-01T00:00:00.000 | {
"year": 2009,
"sha1": "63dbe487bdf907945d64f1c0d9cca6b6fe957311",
"oa_license": "CCBY",
"oa_url": "https://journals.plos.org/plosgenetics/article/file?id=10.1371/journal.pgen.1000660&type=printable",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "b35cc881d77e82b655bba0a2fa4e2be4fecca219",
"s2fieldsofstudy": [
"Biology",
"Medicine"
],
"extfieldsofstudy": [
"Biology",
"Medicine"
]
} |
119330459 | pes2o/s2orc | v3-fos-license | A method to compute Segre classes of subschemes of projective space
We present a method to compute the degrees of the Segre classes of a subscheme of complex projective space. The method is based on generic residuation and intersection theory. It has been implemented using the software system Macaulay2.
Introduction
Segre classes are generalizations of characteristic classes of vector bundles and they occur frequently in intersection theory. Many problems in enumerative geometry may be solved by computing the Segre classes of an algebraic scheme. Given an n-dimensional subscheme Z of complex projective space there are n+1 Segre classes of Z. The i th Segre class is a rational equivalence class of codimension i cycles on Z. Thus a Segre class may be represented as a weighted sum of irreducible subvarieties V 1 , . . . , V m of Z. The degree of a Segre class is the corresponding weighted sum of the degrees of the projective varieties V 1 , . . . , V m .
In this paper we present a method to compute the degrees of the Segre classes of Z, given an ideal defining Z. The procedure is based on the intersection theory of Fulton and MacPherson. More specifically, we prove a Bézout like theorem that involves the Segre classes of Z and a residual scheme to Z. The degree of the residual may be computed providing intersection-theoretic information on the Segre classes. This enables us to compute the degrees of these classes.
If Z is smooth, the degrees of the Segre classes of Z carry the same information as the degrees of the Chern classes of the tangent bundle of Z. For instance, when Z is smooth, the degree of the top Chern class, which is equal to the topological Euler characteristic of Z, can be computed from the degrees of the Segre classes of Z. Hence one can compute the topological Euler characteristic using our procedure (provided that the input ideal defines a smooth scheme). The relationship between the degrees of the Segre classes of Z and the Chern classes of the tangent bundle of Z may be extended to the non-smooth case via the so-called Chern-Fulton classes of Z. In case Z is smooth, the Chern-Fulton classes coincide with the Chern classes of the tangent bundle.
We would like to mention two interesting features of our method. One feature is that the algorithm is completely elementary in that it requires no knowledge of intersection theory to understand the steps in the procedure and only very basic background in algebraic geometry. The method therefore provides a way of understanding the computation of Segre classes from an elementary point of view. In addition, the algorithm is easy to implement. Another feature is the fact that our method is implementable in a numerical setting via numerical homotopy methods, see [11] for an overview of this area. This allows the method to be applied in settings that can be time consuming (and even out of reach) of current symbolic methods. One such setting is when the generating set, for an ideal determining Z, has complicated coefficients. An additional setting where current numerical methods can sometimes obtain useful information about the degrees of Segre classes, beyond the reach of current symbolic methods, is when Z is a reduced scheme of high codimension. These gains come through numerical approximation and parallelization (but at the expense of exactness).
The procedure presented in this paper has been implemented in the symbolic setting using the software system Macaulay2 [7] and is available at http://www.math.su.se/∼jost/segreimplementation.htm. A numerical version, utilizing the software package Bertini [3], will be made available pending the release of Bertini Version 1.3. Initial experiments, involving subvarieties of relatively large dimension and codimension, show a great deal of promise for the algorithm in the numerical setting.
In the paper [1] Aluffi formulates an algorithm that also computes the degrees of the Segre classes of a subscheme Z of projective space. In addition he shows how to reduce the computation of the so-called Chern-Schwartz-MacPherson classes of a subscheme of projective space to the computation of the degrees of certain Segre classes. In the present paper we present an alternative method to Aluffi's. Though the two are closely related, they have a rather different computational behavior and seem to complement each other well (see Section 6 for some examples). Apart from the difference in computing speed in various cases one may ask what is the need for another method with the same output as an existing method. One answer is that our approach is different and therefore sheds new light on the problem of computing Segre classes. But more importantly we would answer by repeating the two features mentioned above, namely that our method is elementary and that it is readily amenable to numerical computation.
The paper is organized as follows. In Section 2 we give the basic definitions and state a theorem from intersection theory. In Section 3 we derive a recursive formula for Segre classes which is the basis of our method. The procedure to compute Segre classes is presented in Section 4. Some examples are given in Section 5 and in Section 6 we give a list of run times on examples comparing our method to other algorithms.
The results of this paper are generalizations and variants of the results in [2,4] to the setting of subschemes of projective space.
2.1. Notation. Let Y be an algebraic scheme over C of dimension n. By a subscheme of Y we will mean a closed subscheme. We will denote by C p (Y ) the free Abelian group on irreducible p-dimensional subvarieties of Y . The p th Chow group of Y is the quotient of C p (Y ) by the cycles rationally equivalent to 0, and it is denoted . . , X t are the irreducible components of X and m 1 , . . . , m t their geometric multiplicities in X. In particular, [∅] = 0. If α ∈ A * (X), we will at times consider α to be an element of A * (Y ), omitting in the notation the push-forward under the inclusion map.
For a rank ρ vector bundle E on Y we have the Chern class operations c i (E) : . The operation of a product of Chern classes on the Chow group is defined as the composition of the individual Chern class operations. The map c 0 (E) is the identity homomorphism. If Y is smooth, the Chern classes are well defined elements of A * (Y ) and the operations c i (E) ∩ α correspond to the intersection product. Let X ⊆ Y be a closed subscheme with inclusion i : X → Y and let α ∈ A * (X). By the notational convention mentioned above we will sometimes write c(E) ∩ α to mean c(E) ∩ i * (α) ∈ A * (Y ). By the projection formula, see [5] For a Cartier divisor D on Y , the corresponding line bundle on Y is denoted by O Y (D).
In this paper, varieties are by definition irreducible and reduced. Finally, we use the convention dim(∅) = −1.
Regular embeddings.
Let Y be an algebraic scheme over C. A closed embedding X → Y of a subscheme X of Y is called a regular embedding of codimension d if the following holds. Every point of X has an affine open neighborhood U in Y such that the ideal defining X ∩ U is generated by a regular sequence of length d in the coordinate ring of U .
Proof. Let φ : Y → P r be the map given by the complete linear system corresponding to L and put e = dim(im(φ)). If µ > e then X(σ 1 , . . . , σ µ ) = ∅ for general sections σ 1 , . . . , σ µ and hence we may assume that µ ≤ e. That X(σ 1 ) → Y is a regular embedding for a general section has no zero divisors and any σ 1 = 0 will give a regular sequence in O Y (U ). If µ = 1 we are done. If dim(Y ) ≤ 1 then e ≤ 1 and therefore µ = 1 in this case. Assume that 1 < µ, in particular 1 < dim(Y ). Then e ≥ 2 and it follows from Bertini type theorems that X(σ 1 ) is a variety for a general section σ 1 . In fact, X(σ 1 ) is reduced by [6] Corollary 3.4.9 and X(σ 1 ) is irreducible by [6] 3.4.10. Replacing Y by X(σ 1 ) and restricting L to X(σ 1 ) we have reduced to the case of a lower dimensional ambient variety since the composition of two regular embeddings is regular (see [5] Appendix B.7.4).
2.3. Segre classes and intersection products. Let Y be a complex variety and let X be an n-dimensional subscheme of Y . Suppose that X = Y and let Y be the blow-up of Y along X. Let π : Y → Y be the projection, let X = π −1 (X) be the exceptional divisor, and let η = π| X . The total Segre class s(X, Y ) of X in Y is an element of A * (X) which may be characterized as follows (see [5] Corollary 4.2.2): Remark 2.2. The Cartier divisor X on Y corresponds to a line bundle on Y whose restriction to X is the normal bundle of X in Y . The dual bundle to the normal bundle is denoted O(1). The exceptional divisor X is naturally identified with the so-called projective normal cone P (C X Y ) of X in Y . The total Segre class is given by This definition of Segre classes generalizes beyond normal cones of subschemes to arbitrary cones, see [5] Chapter 4.
We will now recall the Fulton-MacPherson approach to intersection products, see [5] Chapter 6. Let Y and V be complex varieties and put k = dim(V ). Let X be a closed subscheme of Y such that there is a regular embedding i : X → Y of codimension d ≤ k. Let f : V → Y be a morphism and put W = f −1 (X). Then we get the following fibre product diagram which maps x i to the image of f i in J/J 2 . In the general case, C W V may be constructed by covering V with open affine subsets and gluing the normal cones of the affine patches together. In case the embedding W → V is regular, C W V is a vector bundle, namely the normal bundle N W V . In this case the total Segre class of W in V is the inverse of the total Chern class of N W V in the sense that , and let p : N → W be the projection. It is shown in [5] Chapter 6, that C embeds in N and therefore it determines a class [C] ∈ A k (N ). Now, by [5] Theorem 3.3 (a), the flat-pullback An important connection to Segre classes is given by [5] Proposition 6.1 (a): We will now state the Residual Intersection Formula from [5] which is the main result underpinning our method to compute Segre classes. Let Y , X, W , V , N , k and d be as in the above definition of the intersection product X · V . Let Z ⊆ W be a closed subscheme and suppose that Z = V . Let π : V → V be the blow-up of V along Z and put W = π −1 (W ) and Z = π −1 (Z). Let R be the residual scheme to Z in W with respect to V , see [5] Definition 9.2.1. This is a scheme such that, if I ( Z), I ( W ) and I ( R) are the ideal sheaves in O V defining the respective schemes, then The following proposition is Corollary 9.2.3 of [5].
Proposition 2.3. With notation as above,
Computing Segre classes of projective schemes
Let Z be a proper n-dimensional subscheme of complex projective space P k . This paper is about a method for computing the push-forward of s(Z, P k ) to P k given an ideal defining Z. In this section we explain how to derive information about the push-forward given sufficiently general elements from the ideal.
Let s 0 , . . . , s n be the homogeneous components of s(Z, where H ∈ A k−1 (P k ) is the hyperplane class and the product is the intersection product on P k . The numbers {deg(s i )} i is the output of our procedure, they carry the same information as the push-forward γ * (s(Z, P k )). The degree of any α ∈ A p (P k ) is defined similarly by deg(α · H p ).
Let I ⊆ C[x 0 , . . . , x k ] be a homogeneous ideal. For a positive integer m, we use I(m) to denote the m th graded piece of I. Given a homogeneous ideal J, the ideal quotient J : I is given by and the saturation of J with respect to I is Note that (J : I p ) : I = J : I p+1 for p ≥ 1, that the ascending sequence of ideal quotients J : I ⊆ J : I 2 ⊆ J : I 3 ⊆ . . . stabilizes and that J : I p = J : I ∞ for large enough p. In fact, if J = i Q i is a primary decomposition (so each Q i is homogeneous and primary) then To see this, note that J : The following theorem is a Bézout like equality which gives rise to a recursive formula for the degrees of the Segre classes of Z in P k . Using the statement of the theorem, we may express the degree of a Segre class s p in terms of deg(s i ) for i < p and the degree of a certain residual scheme R. Computing the degree of the residual R is the main computational step in the method.
Proof. The proof is divided in steps 0 through 4.
Step 0: setup. Let I ′ be the ideal generated by I(m) and let m = (x 0 , . . . , x k ). Then I : m ∞ = I ′ : m ∞ , and therefore I and I ′ define the same scheme Z ⊆ P k . We may thus assume that g 0 , . . . , g r all have degree m. Let π : P k → P k be the blow-up of P k along Z and put Z = π −1 (Z). The map P k \ Z → P r defined by g 0 , . . . , g r extends to a map φ : P k → P r , see [5] 4.4. In fact, P k embeds in P k × P r in such a way that ( P k \ Z) is the graph of the map P k \ Z → P r and φ is the projection. Let W ⊆ P k be the scheme defined by f 1 , . . . , f d and put W = π −1 (W ). Let R be the residual to Z in W with respect to P k .
Step 1: we will show that R → P k is a regular embedding and that R is either empty or of pure dimension k − d and that no irreducible component of R is contained in Z. By [5] 4.4., In concrete terms, P k is defined by a bi-homogeneous ideal such that K contains the elements g i y j − g j y i for 0 ≤ i < j ≤ r. Observe that Z is given by the vanishing of g 0 , . . . , g r . Consider the affine open set U = U αβ = {(x 0 , . . . , x k , y 0 , . . . , y r ) ∈ P k : x α = 0, y β = 0} and let w 0 = y0 y β , . . . , w r = yr y β with w β = 1. Then w 0 , . . . , w β , . . . , w r are coordinates on C r = {y β = 0} ⊂ P r . Note that for all i, f i = r j=0 λ j i g j , for a general vector (λ 0 i , . . . , λ r i ) ∈ C r+1 . With an abuse of notation, we use g i , f i and w i to denote the corresponding elements of the coordinate ring of U . Then, g j = w j g β for all j. Hence Z ∩U is defined by g β . Also, f i = ( r j=0 λ j i w j )g β . It follows that R∩U is defined by the ideal ( r j=0 λ j 1 w j , . . . , r j=0 λ j d w j ). We conclude that R = φ −1 (L) for a general linear subspace L ⊆ P r of codimension d (if r < d, then R = ∅). Hence R is either empty or of pure dimension k − d and R ∩ Z is either empty or of pure dimension k − d − 1. It follows that no irreducible component of R is contained in Z. Since P k is a variety (see [8] Proposition II.7.16 or [5] Appendix B.6.4), it follows by Lemma 2.1 that the embedding of R in P k is regular.
Step 2: applying Proposition 2.3. Let X 1 , . . . , X d be defined by X ν = {f ν = 0}. Then W = d ν=1 X ν . Let X = X 1 × · · · × X d and Y = P k × · · · × P k (d factors). Let j : W → P k be the inclusion and let f : P k → Y and g : W → X be the diagonal morphisms. The morphism i : X → Y induced by the inclusions X 1 , . . . , X d ⊂ P k is a regular embedding of codimension d. Put N = g * (N X Y ). Letting V = P k , we apply Proposition 2.3 to the diagram Here s(Z, P k ) is regarded as a rational equivalence class on W .
Step 3: we will show that where p = d − (k − n). We shall first see that N = j * (E) where E = d ν=1 O P k (m). Let p ν : Y → P k be the ν th projection and let X ν be the divisor p −1 ν (X ν ) on Y .
Then, by [5] B.7.4, for all ν. Hence We now push both sides of (2) forward to P k by j and then take degrees. By Bézout's theorem, deg(j * (X · P k )) = m d (see [5] Example 6.2.6). By the projection formula, j * (c(N ) ∩ s(Z, P k )) = c(E) ∩ j * (s(Z, P k )). We get that The degree of the latter expression is Step 4: it remains to see that deg(j * (R)) = deg(R) where R ⊆ P k is the scheme defined by J : I ∞ . In fact, we shall see that j * (R) = [R] in A * (P k ). Let η : W → W be the restriction of π to W . Since R → P k is a regular embedding we have that s( R, P k ) = c(N R P k ) −1 ∩ [ R]. Since R is either empty or has pure dimension k − d, Hence R = η * ([ R]). Let R 1 , . . . , R t be the irreducible components of R and let m 1 , . . . , m t denote their geometric multiplicities. Since none of the components R 1 , . . . , R t is contained in Z and π : Remark 3.3. With notation as in the proof of Theorem 3.2, note that the group GL(C k+1 ) × GL(C r+1 ) acts transitively on P k × P r and that P k is regular outside Z. It follows by Kleiman's transversality theorem [10] that for a general linear subspace L ⊆ P r of codimension d, R = P k ∩ (P k × L) is regular outside Z and the multiplicities m 1 , . . . , m t of the components of R are all equal to 1. Moreover, it follows that the scheme R defined by J : I ∞ is regular outside Z. This could be of interest in connection with the computation of the degree of R, which is the main computational ingredient in our method to compute Segre classes.
The method
Theorem 3.2 states that certain conditions hold for a general choice of elements of a given ideal. By choosing these elements randomly we turn this into a probabilistic algorithm. Applying Theorem 3.2 to solve for the Segre classes recursively, we obtain the following procedure to compute the degrees of the Segre classes of a subscheme of projective space. The input is an ideal defining the subscheme. Let J = (f 1 , . . . , f d ).
6:
Compute deg(R), where R ⊆ P k is the scheme defined by J : I ∞ . Remark 4.1. We will use the notation of Procedure 1 and the proof of Theorem 3.2.
7:
In particular π : P k → P k denotes the blow-up of P k along Z. Instead of saturating with respect to the whole ideal I, as is done in Procedure 1, one could saturate with respect to one element of I. Let h ∈ I, h = 0, let R ′ be the scheme defined by J : (h) ∞ and let H ⊆ P k be the hypersurface defined by h. The claim is that we could replace R by R ′ in Procedure 1. Tracing back the conditions on R used in the proof of Theorem 3.2 we see that we only need to show that no irreducible component of the residual R ⊆ P k is contained in π −1 (H). This follows exactly as in step 1 of the proof of Theorem 3.2 where it is shown that R has no irreducible component inside the exceptional divisor Z.
where T P k is the tangent bundle of P k . This definition is independent of the embedding of Z in P k in the sense that if Z admits two embeddings into smooth varieties M and P , then c(T M | Z ) ∩ s(Z, M ) = c(T P | Z ) ∩ s(Z, P ), see [5] Example 4.2.6. Let . If Z is smooth, then the Chern-Fulton classes coincide with the Chern classes of the tangent bundle. Since c(T P k ) = (1 + H) k+1 where H ∈ A k−1 (P k ) is the hyperplane class, the degrees of the Segre classes and those of the Chern-Fulton classes are related by Recall that in the smooth case, the degree of the top Chern class of the tangent bundle is equal to the topological Euler characteristic. Thus, in case Z is smooth, deg(c ′ n ) is the topological Euler characteristic of Z and Procedure 1 provides a way of computing this topological invariant.
Examples
In this section we illustrate Procedure 1 with some examples. In these examples, for an n-dimensional subscheme Z ⊆ P k , we use the notation σ(Z) = (deg(s 0 ), . . . , deg(s n )), where s(Z, P k ) = n i=0 s i and s i ∈ A n−i (Z).
Example 5.1. Let I ⊂ C[x, y, z] be the ideal I = (x 2 , y 2 , xy) and let p ∈ P 2 be the degree three zero-scheme defined by I. Then s(p, P 2 ) = s 0 ∈ A 0 (p), and A 0 (p) ∼ = Z via the degree map. Now let f 1 , f 2 ∈ I be general elements of degree 2 and put J = (f 1 , f 2 ). Then J : I = (x, y) and (J : I) : I = (1). Hence J : I ∞ = (1) and the residual R is empty. Therefore Note that s 0 is not equal to the degree of p (which is equal to 3).
Example 5.2. Consider a plane curve C ⊆ P 2 defined by one element g ∈ C[x, y, z] of degree m. Then it is immediate from Procedure 1 that σ(C) = (m, −m 2 ). In case g = xy, we get σ(C) = (2, −4). Now consider the scheme D ⊆ P 2 defined by I = (x 2 y, xy 2 ). The support of D is the union of the lines L 1 = {x = 0} and L 2 = {y = 0}, but D has an embedded point at p = L 1 ∩ L 2 . The normal cone C D P 2 has three irreducible components, all of dimension 2, and the supports of these components are L 1 , L 2 and p, respectively. The supports are the so-called distinguished varieties of the intersection defined by {x 2 y, xy 2 }. A general element f 1 ∈ I(3) may be written f 1 = xy(ax + by), for general a, b ∈ C. Then (f 1 ) : I ∞ = (f 1 ) : I = (ax + by). Hence the residual R is the line {ax + by = 0} and For a general f 2 ∈ I(3) we have that (f 1 , f 2 ) = I and it follows that In summary, σ(D) = (2, −3). Observe that the Segre classes detect the embedded point p.
The degrees of the Segre classes s 0 , s 1 , s 2 of Z in P k are computed as follows:
Implementation and benchmarks
There is an implementation of Procedure 1 in the symbolic setting using the software system Macaulay2 [7] and it is available at http://www.math.su.se/∼jost/. It uses the improvement of Remark 4.1. The user may choose to be given the degrees of the Chern-Fulton classes as output.
As an alternative to the Gröbner basis computations carried out in Macaulay2, one can use the regenerative cascade algorithm [9] implemented in the software package Bertini [3]. The regenerative cascade algorithm uses numerical homotopy methods to collect data about solution sets of polynomial equations and this data includes the degrees of the residuals that are used in Procedure 1 to compute the degrees of the Segre classes. Table 1 shows run times on some examples, comparing our implementation "segreClass" to two other algorithms. One is the April 2009 version of "CSM" which implements Aluffi's algorithm to compute Segre classes, see [1]. The other is the routine "euler" from Macaulay2 which computes the topological Euler characteristic of a smooth projective variety. Observe that the input to "euler" is a projective variety, not an ideal. Following Table 1, we provide a few details about each example. Additional details on how to generate the equations for each example may be found at http://www.math.su.se/∼jost/. Table 1. Comparison of run times. On an AMD Athlon 64 Processor, 2.2 GHz, and with 1 GB RAM. The computations marked with "-" were terminated after 3 hours.
Input
segreClass CSM euler Rational normal curve in P 6 0.5s 180s 4s Rational normal curve in P 10 --512s Grassmannian G(1, 5) ⊆ P 14 -2s -Smooth surface in P 8 defined by minors 89s --Abelian surface in P 4 175s --Segre embedding of P 2 × P 3 in P 11 -8s - The defining equations of the rational normal curves in Table 1 are given as (2 × 2)-minors of a matrix with variables as entries. The Grassmann manifold is embedded with the Plücker embedding. The surface in P 8 is defined by the (2 × 2)minors of (4 × 3)-matrix of random linear forms. The ideal of the Abelian surface is generated in degrees 5 and 6. The Grassmannian and the Segre product were run over Q and the other examples were run over the finite field with 32749 elements.
Remark 6.1. In connection with the comparison made in Table 1 it should be noted that the routine "euler" computes the topological Euler characteristic by first computing the Hodge numbers of the variety and then taking an alternating sum of them. Thus, "euler" computes interesting information that is not attainable from the Segre classes in any obvious way.
Conclusions
This paper presents an elementary algorithm, based on residual intersection, to compute the degrees of Segre classes of a subscheme of projective space. The algorithm has been implemented in Macaulay2 [7] and can be implemented in a numerical setting using numerical homotopy methods. It is particularly well suited to the regenerative cascade algorithm [9] implemented in the software package Bertini [3]. The table of example run times illustrate the complementary nature of the symbolic implementation of the algorithm to previous symbolic algorithms, in particular to the algorithm of Aluffi [1] in the general case and to the algorithm "euler" found in Macaulay2 when run on smooth projective varieties. The possibility of a numeric implementation shows promise for extending the range of problems to which the algorithm can be applied. | 2012-12-04T12:36:11.000Z | 2011-09-27T00:00:00.000 | {
"year": 2011,
"sha1": "007da9ff7e3f639a38bf5824534716bed88c2a66",
"oa_license": null,
"oa_url": "http://msp.org/jsag/2015/7-1/jsag-v7-n1-p04-s.pdf",
"oa_status": "BRONZE",
"pdf_src": "Arxiv",
"pdf_hash": "007da9ff7e3f639a38bf5824534716bed88c2a66",
"s2fieldsofstudy": [
"Mathematics"
],
"extfieldsofstudy": [
"Mathematics"
]
} |
56200531 | pes2o/s2orc | v3-fos-license | COMPETITIVENESS OPPORTUNITIES FOR TOURISM IN LATGALE REGION
In this research study, opportunities for improvement competitiveness in the context of tourism in Latgale region are analysed and disclosed. Building tourism competitiveness by means of efficiency, quality, and innovation is underlined in this research study. Measuring and providing the efficiency of tourism products in the public sector is a weak point in the development of tourism competitiveness in Latgale region. Findings of this research allow expressing proposals for increasing efficiency and tourism competitiveness overall. The competitive advantage development model for tourism products in Latgale is offered in this research study. Recommendations are related to human resources, which are engaged in regional tourism development, and to strengthening the capacity of the municipalities, that will allow developing the competitiveness of tourism in the region and tourism regional policy in general.
Introduction
In accordance with the findings of tourism analysis in the European Union (EU) (Tūrisms, 2016) international tourism in the world, in 2014, amounted to 1.113 billion travellers (+4.3%), including 582 million persons that have entered Europe, covering 51% of the market (+3%).The EU has the world's favourite tourist destination status.In addition, a study on long-term prospects provides some moderation in the growth of tourism in Europe by 2030, which, in accordance with the estimates, will be 744 million tourists (+1.8%), or 41.1% of the market.In the EU, the tourism industry in its immediate sense (traditional travel and tourism service providers) encompasses 1.8 million businesses, primarily small and medium-sized enterprises (SMEs).Tourism contributes 5% to gross domestic product (GDP) and employs 5.2% of the economically active population (that corresponds to approximately 9.7 million people).Taking into account the close links with other sectors of the economy, this contribution is even higher (more than 10% of GDP and almost 12% of total employment, which corresponds to 13 million workers).
In Latvia, tourism impact on the national GDP approaches 4%.In 2015, the export in the tourism industry in Latvia increased by 7.5% in comparison with 2014.Expenses of foreign tourists increased by 16.6% (Tūrisma un viesnīcu...,2016).In addition, the multiplicative effect of the tourism industry has stimulated the increase of demand in many service industries.Tourism becomes more and more important in the employment and policy of the economic growth.Tourism has very strong environmental and sustainability aspect.
Latvia is divided into five planning regions.14.5% of the population of Latvia live in Latgale region.The territory of Latgale region occupies 22.5% of the territory of Latvia (Valdība atbalsta ...,2016).Two cities and 19 counties are located in Latgale region.Each of them operates in accordance with its development strategy.Tourism as one of the specializations is defined in all of these development strategies.
Latgale region might get a competitive advantage as an international tourism destination, taking into account its geographical location, its rich natural and cultural resources, and opportunities of international support in the development of tourism products.The efficiency of the tourism products, developed within the frame of international projects, is not measured.After implementation of these projects, sustainability of developed tourism products is not properly controlled and managed.It means that the competitive advantage opportunities are not used and developed to the full extent.In this aspect, a vital problem for this region is the competitiveness in tourism.
The aim of this research is to reveal opportunities for the development of tourism competitiveness in Latgale region.
To achieve this aim, several tasks were defined: to interpret the measurement of tourism competitiveness; to carry out an analysis of the factors influencing the competitive advantage of tourism products; to work out proposals on the opportunities for the development of tourism competitiveness in Latgale region.The research subject: the factors influencing the competitive advantage of tourism in Latgale region.
The hypothesis of the research: it is possible to increase the competitiveness of tourism in Latgale region by improving the capacity of human resources and by increasing the efficiency of tourism products in the public sector.
Novelty: the improvement of competitiveness in the context of tourism development in Latgale are analysed and disclosed; these opportunities are mainly related to the development of human resources, which are the major factor in the competitive capacity development in the municipalities.The competitive advantage development model for tourism products in Latgale is offered in the research study.The recommendations developed for municipalities' councils in terms of efficiency of tourism products might be used for increasing competitiveness in tourism.
To accomplish the objectives of the research study, the following methods are used: content analysis of regional/municipality development documents, statistical data analysis; comparative analysis, synthesis, abstract and logical construction methods for studying competitiveness opportunities in the tourism area and human resources development opportunities.
Research results
There are many definitions referring to tourism competitiveness in the various sources.In this research study the author has used the following definition of competitiveness: 'Tourism competitiveness for a destination is the ability of the place to optimise its attractiveness for residents and non-residents, to deliver quality, innovative, and attractive (e.g.providing good value for money) tourism services to consumers and to gain market shares on the domestic and global market places, while ensuring that the available resources supporting tourism are used efficiently and in a sustainable way' (Dupeyras et al., 2013).A. Dupeyras and N. MacCallum have emphasized the absence of a good system of indicators that governments could use for measuring success and competitiveness in tourism: 'Competitiveness in tourism is not currently measured and monitored adequately by governments.One of the reasons for this is the difficulty in identifying a few core indicators to effectively measure and monitor what remains a very broad concept'.The key initiatives for measuring competitiveness in tourism mostly are found at the tourism industry level.Some of the studies look at cities and regions, others are concerned mainly with the hotel industry.At international level, the major undertaking remains the WEF Travel and Tourism Competitiveness Index.Several countries also use the Nation Brand Index which measures the image and reputation of the world's nations.There are some researchers, who have explored the competitiveness of tourism in Latvia.For instance, the Competitiveness Model for Latvian Tourism was offered by Ganijeva and Magidenko (Ganijeva, et al., 2011) on the basis of investigated factors of the tourism competitiveness.Especially, they have stressed the importance of marketing in tourism development.The attractiveness of different tourism destination places in Latvia has been investigated by Kleperis (Kleperis, 2012).The attractiveness of Latgale region for tourists has been investigated by many researchers: Dembovska I., Silicka I., Ežmale S. etc.The author's findings of the recent studies (Silineviča, 2009(Silineviča, , 2010(Silineviča, , 2012(Silineviča, , 2013(Silineviča, , 2014) ) have demonstrated that the attractiveness factors of Latgale region for tourists are as follows: cultural and historical heritage (traditions, handicraft, ceramics, Latgalian language, folklore), sacral traditions, natural resources, restful landscapes, hospitality of the local population (see Figure 1).The number of tourists in Latgale region increases annually.It means that Latgale region becomes more attractive to tourists.The data (Table 1) are approximate because some municipalities in Latgale region have not provided information for this period (the references attached to the statistics prove this).
The competitive advantage development model for tourism products
The attractiveness of tourism products has close links with their competitive advantage.The author offers the competitive advantage development model for tourism products in Latgale region (see Figure 2).The base of competitive advantage of tourism products are tourism resources and capabilities, which are transformed into distinctive competencies.Distinctive competencies allow achieving competitive advantage by using efficiency, quality, innovations, and customers' responsiveness.Quality, innovations, and customers' responsiveness are measured by feedback from tourists.
Latgale is rich in many unique tourism resources.Natural resources in Latgale are unique.Many beautiful hills, covered by forests, are crossed by twisting rivers.Many lakes, embraced by picturesque bays, are rich in many islands.The islands are covered by different broad-leaved species.Two biggest lakes in Latvia (the lake of Lubans and the lake of Razna) are situated in Latgale region.The newest National park in Latvia -Razna National Park -is located in Latgale.This preserved area comprises many rare biotopes and attracts many eco tourists.The history of Latgale has many historical events related to the numerous wars, which have crossed the territory of the region.During centuries, different tribes have had settled in this land.Each of them have influenced Latgalian culture.In this way, Latgalian culture has multicultural character with own uniqueness.
The multicultural character relates to the history of religions in Latgale as well.During centuries, this territory was inhabited by Christians (Catholics, Old believers, Orthodox, Lutherans, Baptists) and by Judaists.Each of the faiths has left their own cultural heritage: the unique sacral buildings and traditions.The Latgalian language is a unique regional language with its own grammar, literature, rich folklore, and traditions.Many Latgalian songs are very popular not just within the region but beyond it.This uniqueness gives great opportunities for the development of unique tourism products.Output indicators (number of tourists, revenues from tourists).The easiest way to measure the efficiency of tourism products in a certain area is dividing the outputs by the inputs: In accordance with the research (Cracolici et al., 2014) findings: 'when a tourist site is not able to produce the maximum possible output, given the inputs, the tourist site is inefficient and will attract relatively fewer tourists compared with competing areas.The reason may be that too many inputs are used, an imbalance between inputs and outputs exists, and/or the input combination is not optimal... that a tourist area should be able to manage its input efficiently; in other words, the territory's physical and human resources constitute the input of a (virtual) tourist 'production process', and the output is then formed by arrivals, bed-nights, value added, employment, customer satisfaction, etc.' The sustainability of the tourist destination product depends largely on the ability to efficiently combine and manage input resources to produce the maximum possible output.
The efficiency of the tourism products, which are produced in private sectors, is measured, because each business must be profitable.In relation to the public sector, the situation is different in Latgale.For example, the Cross-Border Cooperation Programme 2007-2013: Support to Tourism Development and Culture Heritage Conservation in Latgale.This programme has promoted the development of tourism in Latgale through separate projects.The water tourism development project "Water Joy" has been one of the projects, supported by this programme.The project's costs were 1.34 MEUR (13).In terms of efficiency 1.34 MEUR is the input.But what about measuring of the output?It is necessary to have a list of offered water tourism products, which are developed on the basis of this project.
It is necessary to have information about prices of these products, the number of tourists, who used these products, and the expenses of tourists during their travel.Only this information would allow calculating the outputs.If this information is absent, it makes impossible calculation of the efficiency of this project.The efficiency of those tourism products, which were developed within the frame of international projects, is not measured.After finishing these projects, the sustainability of developed tourism products are not fully controlled and managed.It means that the competitive advantage opportunities are not fully used and developed.
The information about the efficiency of tourism products is very important for analysing and planning a tourism marketing strategy at regional/ municipal level.If the efficiency of tourism products was not measured in some regions, tourism development management would be problematic in these regions.
Tourism marketing strategy, which a region/municipality adopts, can have a major impact on efficiency.Latgale region comprises nineteen counties and two cities.Tourism marketing strategies in Latgale region operate at three levels: Tourism is a services industry.Employees are the major input into the production of tourism products.Labour productivity mainly influences costs for labour.Labour productivity increases owing to learning in the most efficient way.Production costs decline because of increasing labour productivity and management efficiency.
Building competitive advantage by quality Tourist destination performance might be evaluated by the measurement of competitiveness in terms of quality.Skilled employees are one of the key factors of quality in the tourism industry.They are the major input into tourism services processes.Highly skilled employees can perform tasks faster and more accurately.Positive attitude to the tourists and environment are provided by the employees.The implementation of innovative processes is provided by the employees.The employees should have attributes that match the strategic objectives of the company.The improvement of labour skills increases the competitive advantage by quality.Cooperation with universities, which deliver tourism management study programmes, with the aim to develop training programmes for the improvement of labour skills in the tourism services is very important.
Tourism quality has a strong connection to the infrastructure.The quality of infrastructure is a significant determinant of the tourism inflows into a destination and, at the same time, an important contributor to development of the other sectors creating important indirect benefits.For example, the construction of new regional roads to improve access to an emerging tourism destination can create economic opportunities for individuals and businesses located in that region.Mobility and accessibility are essential issues for the tourists.The transport infrastructure not only plays a significant role in attracting tourists, but also in improving the quality of life of the local population.Furthermore, it promotes the distribution of socio-economic benefits related to the tourists' expenses (e.g.incomes, jobs) and interactions with the local residents (e.g.cultural heritage diffusion and maintenance) (Albalate et al., 2010).
In the case of Rezekne city, the streets' system, improved and developed in 2013, has a positive impact on the attractiveness of the city and number of visitors, and has also helped to raise the positive impact on the local hotels, restaurants, and other retailers.Municipalities of Latgale region actively use the opportunities of the EU funds for the development of infrastructure projects.These facts are reflected in the annual public reports of Latgale counties and cities, which are published in the counties'/cities' home pages.The municipalities of Latgale region actively use the opportunities of the Cross -border Cooperation Programme as well.
Building competitive advantage by innovation.Tourist destination performance might be evaluated through the measurement of competitiveness in terms of innovation.Unique tourism products could be developed by using the local unique cultural heritage, the Latgalian language, and nature.The main factor in the development of competitive advantage of these products is human resources, which are engaged in these processes.Innovation requires highly skilled employees.Highly skilled employees can better perform tasks, more likely to learn the complex tasks associated with many modern services and methods than individuals with lower skills.
Cooperation with universities, which deliver tourism management study programmes, is very important for municipalities in many aspects: • development of training programmes for the improvement of creative labour skills in tourism services; • development of new tourism products by implementing theoretical knowledge and using the creative approach of students, who perform practical studies in a municipality; • development of training programmes for people, who are engaged in tourism businesses, for the improvement of their administrative capacity.
Conclusions and suggestions
Latgale region is able to get competitive advantages as an international tourism destination, taking into account its geographical location, its rich natural and cultural resources, and opportunities of international support in the development of tourism products.The efficiency of those tourism products, which are developed by using international projects, is not measured.After the implementation of these projects, their sustainability is not fully controlled and managed.In that way, competitive opportunities are not fully used.
The councils of municipalities must provide and control the sustainability of the all international projects after their end.It is necessary to develop a management plan for project sustainability and efficiency.It allows increasing competitiveness in regional tourism.The sustainability of the tourist destination product depends largely on the ability to efficiently combine and manage input resources to produce the maximum possible output.
Efficiency is a very important indicator in building the competitive advantage of tourism products.The competitive advantage development model for tourism products in Latgale is offered in this research study.It is necessary to develop a tourism marketing strategy in each municipality.It will allow achieving a higher efficiency of the tourism industry in each municipality.
Skilled employees are one of the key factors of quality in the tourism industry.They are the major input into tourism services processes.Highly skilled employees can perform tasks faster and more accurately, and are more likely to learn the complex tasks associated with many modern services and methods than individuals with lower skills.
Cooperation with the universities, which deliver tourism management study programmes, is very important in the development of training programmes for the improvement of creative labour skills in tourism services, in the development of new tourism products by implementing theoretical knowledge and using the creative approach of students, who perform practical studies in a municipality, and in the development of training programmes for people, who are engaged in tourism businesses, for the improvement of their administrative capacity.
A. Dupeyras and N. MacCallum (Dupeyras et al., 2013) have suggested to organise the indicators of competitiveness in tourism around the following four categories: 1) indicators measuring the tourism performance and impacts; 2) indicators monitoring the ability of a destination to deliver quality and competitive tourism services; 3) indicators monitoring the attractiveness of a destination; 4) indicators describing support policies and economic opportunities.The findings of the World Economic Forum (WEF) annual Tourism Competitiveness Index research have disclosed that, among 133 countries, Latvia IS the 48-th most attractive tourist destination in the world (Travel & Tourism Competitiveness Report, 2009).In 2015 Latvia ranked the 53rd (out of 141) in the Travel & Tourism Competitiveness Index.In 2015 natural resources of Latvia ranked the 93-rd (out of 141) and the cultural resources and business travel ranked the 106-th (out of 141) in the Travel & Tourism Competitiveness Index (WEF Global Travel..., 2015).
Figure 1 .
Figure 1.The attractiveness of Latgale region for tourists (Source: the author's research)
Figure 2 .
Figure 2. The competitive advantage development model for tourism products in Latgale region (Source: the author's construction based on (Charles et al., 2004))
Table 1
Changes in the tourists' flow in Latgale region in 2010 -2015 Capabilities are very important tools for the development of distinctive competencies.Those are necessary for building the competitive advantage of tourism products.Findings of this research show, that in Latgale region these capabilities are strengthened by the following activities: development of cooperation with the regional universities providing tourism study programmes (in case of Dagda countycooperation with the Rezekne Academy of Technologies).Building competitive advantage by efficiency.Tourism destination performance might be evaluated through the measurement of competitiveness in terms of efficiency.Efficiency is a very important indicator in the building of competitive advantage.Efficiency of tourism products would be measured by using the input and output indicators: Tourism Development Plan of Razna National Park (Rāznas nacionālā..., 2010).Taking into account the fact, that Latgale region is specialized in tourism, it is necessary to develop a tourism marketing strategy for each municipality.It would allow achieving higher efficiency in the development of tourism in each municipality. | 2018-12-18T04:00:06.173Z | 2016-10-24T00:00:00.000 | {
"year": 2016,
"sha1": "fcdd1a5a694689a44e3dcf6e5467e29e3ca7fbfb",
"oa_license": "CCBY",
"oa_url": "https://doi.org/10.17770/lner2016vol1.8.1534",
"oa_status": "HYBRID",
"pdf_src": "ScienceParseMerged",
"pdf_hash": "fcdd1a5a694689a44e3dcf6e5467e29e3ca7fbfb",
"s2fieldsofstudy": [
"Economics",
"Business"
],
"extfieldsofstudy": [
"Business"
]
} |
221108389 | pes2o/s2orc | v3-fos-license | New Formulation towards Healthier Meat Products: Juniperus communis L. Essential Oil as Alternative for Sodium Nitrite in Dry Fermented Sausages
The effect of Juniperus communis L. essential oil (JEO) addition at concentrations of 0.01, 0.05 and 0.10 µL/g on pH, instrumental parameters of color, lipid oxidation (2-Thiobarbituric acid reactive substances (TBARS)), microbial growth, texture and sensory attributes of dry fermented sausages produced with different levels of fat (15 and 25%) and sodium nitrite (0, 75 and 150 mg/kg) was assessed. Reduced level of sodium nitrite (75 mg/kg) in combination with all three concentrations of JEO (0.01–0.10 µL/g) resulted in satisfying physico-chemical (color and texture) properties and improved oxidative stability (TBARS < 0.3 mg MDA/kg) of dry fermented sausages produced with 25% of fat. However, sausages produced with 0.10 µL/g of JEO had untypical flavor. No foodborne pathogens (Escherichia coli, Listeria monocytogenes, Salmonella spp. and sulfite-reducing clostridia) were detected in any sample throughout the storage period (225 days). The results of this study revealed significant antioxidative activity of JEO and consequently its high potential as effective partial replacement for sodium nitrite in dry fermented sausages.
Introduction
Fermented sausages have been manufactured in many countries worldwide. Currently, customers are becoming progressively aware of these meat products for their unique sensory characteristics and important health benefits [1]. Dry fermented sausages are produced using fresh or frozen meat (70-80%) and back fat (20-30%), salt, starter cultures, spices and food additives [2,3]. Owing to the relatively high level of fat and distinctive processing technology (e.g., using diverse raw materials, absence of thermal treatment), fermented sausages are highly susceptible to quality deterioration, including lipid oxidation and bacterial growth [2,3].
Lipid oxidation is one of the chief non-microbial factors in quality deterioration in meat and meat-derived products [4]. It is well known that meat products become very susceptible to oxidative established for its in vitro antiradical and antioxidant activities which are mostly dependent on its chemical shape [25,26].
Due to its strong antioxidant, antibacterial, antifungal, and anti-inflammatory properties, Juniperus communis L. and its essential oil are widely used in food processing, and in the pharmaceutical and cosmetic industries. Terpenoids (e.g., α-pinene, limonene and myrcene) determine the strong and distinctive aroma of juniper essential oil [27].
Recently, the application of Juniperus communis L. essential oil as natural additive was investigated in several studies [28][29][30]. Selim et al. [29] found that Juniperus communis L. essential oil added at concentrations of 0.1, 0.5, and 1% possesses a weak inhibitory effect towards Enterococci and Escherichia coli O157:H7 that were inoculated in ground beef meat, stored at a temperature of 7 • C for 14 days. However, in an earlier study, Schelz et al. [28] determined the strong antimicrobial potential of Juniperus communis L. essential oil against Saccharomyces cerevisiae. In our previous research, we found that Juniperus communis L. essential oil efficiently suppressed lipid oxidation and microbial growth and enhanced the color of cooked pork sausages [30].
A literature review has exposed only a few published research papers that discuss the application of essential oil as natural additive in dry fermented sausage processing. There is also a lack of data regarding the application of essential oils as sodium nitrite replacements in this type of dry cured meat product. Regarding its strong antioxidant and antimicrobial potential, we hypothesized that Juniperus communis L. essential oil could be used as an alternative for sodium nitrite in meat processing. Thus, the aim of this study was to assess the effect of Juniperus communis L. essential oil as an alternative for sodium nitrite in dry fermented sausages. For these purposes, several physicochemical (pH, color and texture), microbiological (total plate count, lactic acid bacteria) and sensory (color, odor and flavor) parameters of dry fermented sausages were determined.
GC-MS Profile of Terpenoid Compounds
Juniperus communis L. essential oil (JEO) was purchased from the manufacturer Herba doo (Belgrade, Serbia). JEO was kept in dark glass bottles at 4 • C prior to the experiments.
For identification of volatile terpenoids from JEO, GC-MS analysis was used according to the method described by Pavlić et al. [31]. Agilent GC890N system coupled to mass spectrometer Agilent MS 5759, with HP-5MS column (0.25 mm inner diameter and 0.25 µm film thickness, 30 m length), was applied for the characterization of terpenoid profile. Flow rate of helium was 2 mL/min. JEO was dissolved in dichloromethane (approx. 1 mg/mL) and 5 µL of solution was injected in the device with split ratio 30:1. Temperature conditions were: injector temperature 250 • C, detector temperature 300 • C, initial 60 • C with linear increase of 4 • C/min up to 150 • C. The NIST 05 and Wiley 7n data base were used for compound identification. Retention equations, which describe dependence of peak area on different concentration (R 2 > 0.99), were obtained using standard compounds dissolved in dichloromethane at different concentrations (1-500 µg/mL). Results were expressed as relative percentage (%).
Samples
Dry fermented sausages were created with two levels (15 and 25%) of pork back fat (FC). In both obtained batters, sodium nitrite (NC) was added at three concentrations (0, 75 and 150 mg/kg). Next, each batter was divided into four parts, and into each part the corresponding concentrations of JEO (0.00, 0.01, 0.05 and 0.10 µL/g) were added. The total number of batches (B) was: FC (2) × NC (3) × JEO (4) = 24 ( Figure 1). Samples were collected at different storage periods (SD) involving three randomly selected dry fermented sausages from each batch at the end of drying (0) and after 75, 150 and 225 days of storage. The total number of samples was: B (24) × SD (4) × 3 = 288. randomly selected dry fermented sausages from each batch at the end of drying (0) and after 75, 150 and 225 days of storage. The total number of samples was: B (24) × SD (4) × 3 = 288.
Preparation of Dry Fermented Pork Sausages
Dry fermented sausages were produced in a local industrial plant (A.D. Dim-Dim, Laktaši, Bosnia and Herzegovina). Batters were produced using lean pork shoulder and pork back fat in the ratio 75:25 and 85:15%. The amounts of other ingredients were calculated in relation to raw material weight, and were as follows: NaCl (2.50%), gluconic delta-lactone (0.70%), spice mix (0.50%), dextrose (0.10%), sodium nitrite (0, 75 and 150 mg/kg) and JEO (0.00, 0.01, 0.05 and 0.10 µL/g.) The meat and back fat were minced using a cutter (Krämer & Grebe, Germany), and then the other ingredients were added and mixed with them until the required temperature (1 °C) was achieved. The sausages were
Preparation of Dry Fermented Pork Sausages
Dry fermented sausages were produced in a local industrial plant (A.D. Dim-Dim, Laktaši, Bosnia and Herzegovina). Batters were produced using lean pork shoulder and pork back fat in the ratio 75:25 and 85:15%. The amounts of other ingredients were calculated in relation to raw material weight, and were as follows: NaCl (2.50%), gluconic delta-lactone (0.70%), spice mix (0.50%), dextrose (0.10%), sodium nitrite (0, 75 and 150 mg/kg) and JEO (0.00, 0.01, 0.05 and 0.10 µL/g.) The meat and back fat were minced using a cutter (Krämer & Grebe, Germany), and then the other ingredients were added and mixed with them until the required temperature (1 • C) was achieved. The sausages were stuffed in 37 mm diameter collagen casings and were placed in a climate chamber (Frigovent, Serbia) for 21 days. The processes of fermentation, smoking, drying and ripening were performed at a temperature of 14-16 • C and a relative humidity of 80-95%. Produced sausages were vacuum packed (Multivac C500, Wolfertschwenden, Germany) and stored at 15 ± 1 • C for 225 days. The proximate chemical compositions of the sausages produced with 15 and 25% of back fat at the end of drying process are presented in Table S1 (Supplementary material).
The pH was evaluated using a digital pH meter Testo 205 (Testo AG, USA). Before measurement it was calibrated using standard buffers (pH = 4.00 ± 0.05 and pH = 7.00 ± 0.01 at 20 ± 2 • C). pH values were determined for three samples, from each group of dry fermented sausages, in duplicate.
Color (CIE-LAB values: L*-lightness; a*-redness; b*-yellowness) of each sample of the dry fermented sausages was measured on fresh cross cut immediately after slicing. The L*, a* and b* color coordinates were determined using a MINOLTA Chroma Meter CR-400 (Minolta Co., Ltd., Osaka, Japan) using D-65 lighting, a 2 • standard observer angle and an 8-mm aperture in the measuring head [16]. Prior to measurement it was calibrated using a Minolta calibration plate (No. 11333090; Y = 92.9, x = 0.3159; y = 0.3322). Color was measured for three samples (2 cm thick) from each group of dry fermented sausages in triplicate.
The TPA (Texture profile analysis) test was conducted at room temperature using TA.XT2 Texture Analyzer (Texture Technologies Corp., Scarsdale, NY/Stable MicroSystems, Godalming, UK) equipped with a standard ∅ 75 mm cylindrical plate. TPA parameters hardness (g), springiness, cohesiveness, and chewiness (g) were determined as described by Ikonić et al. [36]. The cylindrical shape samples (2.54 cm in diameter, 2 cm thick) were taken from the central part of the sausage, and were analyzed in two cycle compressions to 50% of their original thickness at a constant test speed of 1 mm/s. Peak force during the first compression cycle was marked as hardness. The rate at which a deformed sample goes back to its undeformed condition after the deforming force is removed was defined as springiness. The ratio of the area under the second and first curve was defined as cohesiveness. Lastly, by multiplying hardness, cohesiveness and springiness, chewiness was obtained. TPA was performed for three samples from each group of dry fermented sausages in duplicate.
Lipid oxidation of dry fermented sausages was assessed using the 2-Thiobarbituric acid reactive substances (TBARS) test according to the method of Botsoglou et al. [37], with some modifications. The final step of the extraction procedure was carried out with total volume (10 mL) of TCA (trichloroacetic acid) in ultrasonic bath XUB 12 (Grant Instruments, Cambridge, UK). Spectrophotometer Jenway 6300 (Jenway, Felsted, UK) was used for absorbance measurement at 532 nm. The results of the TBARS test were expressed as milligrams of malondialdehyde per kilogram of sample (mg MDA/kg). TBARS was determined on three samples from each group of dry fermented sausages in duplicate.
Microbiological Analysis
Microbiological analyses were performed on three samples from each group of dry fermented sausages in duplicate. Samples (20 g) were homogenized in 180 mL 1 g/L buffered peptone water (Merk, Darmstadt) for 10 min at 200 rpm (Unimax 1010, Heidolph, Germany) and the serial of decimal dilutions were prepared (up to 7-10). From each dilution 1 mL was placed in a sterile Petri plate and poured with appropriate media depending on the type of tested microorganisms. The following microorganisms were determined: total plate count (TPC), lactic acid bacteria (LAB), Escherichia coli, Salmonella spp., Listeria monocytogenes and sulfite-reducing clostridia count [38][39][40][41][42][43]. TPC was enumerated in Plate Count Agar (PCA) (Merk, Darmstadt, Germany) and incubated at 30 • C for 72 h; LAB was enumerated in de Man, Rogosa and Sharpe (MRS) Agar (Merk, Darmstadt, Germany) and incubated at 30 • C for 72 h; Escherichia coli was determined on Tryptone Bile Glucuronic Agar (TBX agar) (Merk, Darmstadt, Germany) after an incubation at 44 • C for 24 h; Salmonella spp. was determined on Xylose Lysine Deoxycholate (XLD) agar (Merk, Darmstadt, Germany) after an incubation at 37 • C for 24 h; Listeria monocytogenes was determined on Listeria agar acc. Ottaviani and Agosti (ALOA) (Merk, Darmstadt, Germany) after an incubation at 37 • C for 24 h; sulfite-reducing clostridia count was determined on Tryptone Sulfite Cycloserine (TSC) Agar (Merk, Darmstadt, Germany) after an incubation at 37 • C for 24-48 h under anaerobic conditions. After incubation, microscopic observation of cell morphology and biochemical tests were used for typical and atypical grown colonies identification. Results were expressed as a log number of colony forming units per gram (log CFU/g).
Sensory Analyses
Sensory analysis was carried out by a trained panel consisting of ten members, aged 25 to 50 years, per two sessions. All panelists work at the Faculty of Technology Novi Sad, Serbia, and have wide expertise in the sensory evaluation of foods. Panelists were trained according to methods described in ISO 8586 [44], in a sensory laboratory equipped according to ISO 8589 [45]. Evaluation of sensory attributes (color, odor and flavor) was performed using the difference-from-control test [46]. Prior to analyses, sausages were equilibrated to room temperature for about 15 min. and marked with a three-digit sample number. The sausages were sliced into 2 mm thick pieces and placed on a white porcelain plate. Consumers were firstly questioned to evaluate the control sample (without JEO and with the corresponding contents of fat and nitrite) and afterward to determine how different the coded samples were from the control one. The difference was rated on a scale from 0 to 6, where 0 = no difference; 1 = very slight difference; 2 = slight/moderate difference; 3 = moderate difference; 4 = moderate/large difference; 5 = large difference; and 6 = very large difference.
Statistical Analysis
The statistical program STATISTICA 13.0 (TIBCO Software Inc., Palo Alto, CA, USA) was used for data analyses. The main effects (fat content, nitrite content, JEO content and storage day) were compared. All data were expressed as mean value with their standard deviation (Stdev). The two-way, three-way and four-way interactions between these effects were also tested. Differences among treatment means were compared according to t-test and Duncan's multiple range test (p < 0.05).
pH and Instrumental Parameters of Color of Dry Fermented Sausages
The pH values of dry fermented sausages are presented in Table 2. The fat content and storage time had a significant (p < 0.05) effect on the pH values. The samples produced with 15% of fat had a higher pH value. Regarding storage time, it can be observed that pH values inconsistently increased throughout storage, probably as the result of formation of amino-compounds during the proteolysis in fermented sausages [52,53]. The two-way (SD × JC), three-way (FC × NC × SD, FC × SD × JC) and four-way (FC × NC × SD × JC) interactions had a significant (p < 0.05-0.001) effect on the pH values (Table S2-Supplementary material). Values of pH ranged from 5.11 (FC = 25%, NC = 0 mg/kg, SD = 0, JC = 0.05 µL/g) to 5.63 (FC = 15%, NC = 150 mg/kg, SD = 75, JC = 0.10 µL/g). Similar results were observed by Kurćubić et al. [52] and Ozaki et al. [54] in fermented meat products.
Color is one of the key quality parameters for meat and meat products [16]. The instrumental parameters of color (L*, a* and b*) are displayed in Table 2. The contents of fat, nitrite and JEO, as well as storage time, had a significant (p < 0.05) effect on L* values. As expected, the samples produced with 15% of fat had lower L* values. Moreover, storage time had the effect of decreasing L* values, according with the findings of Pateiro et al. [55]. Finally, the addition of JEO decreased the L* value, probably as the result of interactions among bioactive compounds of JEO (phenolics, terpenes) and myoglobin [30]. The two-way (FC × SD) and four-way (FC × NC × SD × JC) interactions were also significant (p < 0.05) for L* values (Table S2). The L* values ranged across a wide interval from 43.92 (FC = 15%, NC = 150 mg/kg, SD = 225, JC = 0.10 µL/g) to 56.64 (FC = 25%, NC = 0 mg/kg, SD = 0, JC = 0.00 µL/g).
Fat content and storage time had a significant (p < 0.05) effect on the a* values. As expected, the samples produced with 15% of fat had higher a* values. Concerning storage time, the increasing of a* values after the 150th day of storage can be noticed. This is in accordance with the findings of Pateiro et al. [55]. The increase of a* values could be related to the growth of the Staphylococcus species [56]. Faustman and Cassens [56] reported that enzymes (NADH-cytochrome b5 reductase systems, metmyoglobin reductase and nitrate reductase) of S. carnosus or S. xylosus can alter metmyoglobin to form red myoglobin derivatives and enhance the color of meat products. Two-way (FC × NC) and three-way interactions (FC × NC × SD and FC × SD × JC) suggested a significant (p < 0.05-0.01) effect of using both sodium nitrite (150 mg/kg) and JEO (0.10 µL/g) for enhancing the redness of low-fat (15%) dry fermented sausages (Table S2). The lowest (9.38) and the highest (15.95) a* values were determined in the samples: FC = 25%, NC = 0 mg/kg, JC = 0.00 µL/g, SD = 0; FC= 15%, NC = 150 mg/kg, JC = 0.10 µL/g SD = 225. No significant (p > 0.05) four factor interaction was detected for the a* value.
TBARS Values of Dry Fermented Sausages
Lipid oxidation is one the most important parameters of quality for meat and meat products [30]. TBARS values of dry fermented sausages are presented in Table 2. The contents of nitrite and JEO and storage time had a significant (p < 0.05) effect on TBARS values. The inclusion of sodium nitrite (75 and 150 mg/kg) decreased TBARS values. This was probably the result of the antioxidant activity of sodium nitrite [11]. Furthermore, Honikel [11] reported that antioxidant activity of nitrites is associated with the ability of NO to fix and stabilize heme iron (Fe) of meat myoglobin, making it unavailable to catalyze reactions of oxidation. Also, Karwowska et al. [58] reported that the reduction of nitrites, from 150 to 50 mg/kg, increased TBARS values in cooked meat products. Moreover, samples produced with the addition of JEO (0.05 and 0.10 µL/g) had lower TBARS values compared to samples produced without JEO. This is the consequence of the strong antioxidant potential of JEO. Höferl et al. [25] reported that juniper berry oil significantly prevented the formation of lipid peroxidation by-products caused by TBA. Certain compounds, such as α-terpinene, γ-terpinene and α-terpinolene exhibit strong antioxidant activity in prevention of lipid oxidation which could be compared with α-tocopherol [59]. On the other hand, certain compounds from JEO (pinene, sabinene and limonene) have a rather weak effect. Similar findings of the antioxidant effects of JEO in meat products were observed in our previous study [30]. As expected, storage time had a significant (p < 0.05) effect on increasing TBARS values, as the result of lipid oxidation [2]. The two-way interactions (FC × SD and NC × SD) were significant (p < 0.05) for TBARS values. Moreover, three-way (FC × NC × SD and FC × SD × JC) and four-way interactions had a significant (p < 0.05-0.001) effect on TBARS values (Table S2) Melton [60], the TBARS value of 0.3 mg MDA/kg is marked as the threshold for rancidity of meat products. The obtained results suggested that interaction between sodium nitrite (75 mg/kg) and JEO (0.01 and 0.05 µL/g) efficiently reduced the lipid oxidation in high-fat (25%) dry fermented sausages. Regarding the strong lipo-solubility of terpenoid compounds (e.g., β-myrcene, sabinene, β-pinene, limonene) JEO possessed a higher antioxidant potential in dry fermented sausages produced with higher fat content (25%).
Microbiological Analysis of Dry Fermented Sausages
Total plate count (TPC) and lactic acid bacteria (LAB) of dry fermented sausages are presented in Table 3. The contents of fat and nitrite did not exhibit a significant (p > 0.05) effect on TPC and LAB. In the case of JEO, the addition of this essential oil (≥0.05 µL/g) had a tendency to reduce the TPC (for 0.14 log cfu/g), but differences among the samples were not significant (p > 0.05). Hence, further optimization with a higher concentration is necessary. Moreover, in our previous study [30] we found that JEO addition (≥0.10 µL/g) efficiently reduced TPC in cooked pork sausages. On the contrary, Selim et al. [29] showed that JEO had no effect on the reduction of microbial growth in fresh beef meat. The antimicrobial potential of essential oil depends of its chemical shape. Generally, monoterpenes from the JEO (α-pinene, β-pinene, sabinene, γ-terpinene, β-myrcene, and limonene) are not efficient antimicrobials when applied singly [61]. However, a mixture of these compounds with the presence of other JEO constituents present as a minor content could result in additive or synergistic antimicrobial effects [62]. As mentioned, the utilization of a novel extraction technique (e.g., SFE) could be a good solution in order to improve the chemical profile of JEO [50]. Orav et al. [50] found that JEO obtained using SFE contained less monoterpenes (5.1%) and more sesquiterpenes and oxygenated sesquiterpenes (69.8%) with a higher antimicrobial potential. As expected, storage time significantly (p < 0.05) affected TPC and LAB. TPC decreased during the first 75 days of storage, then increased until the 150th day of storage and again decreased until the end of storage. This trend could be related to the reduction of LAB during storage, especially after the 150th day of storage. As mentioned, the population of LAB decreased throughout storage, probably as the consequences of low storage temperature (15 • C) and the exhaustion of sugar [63]. No significant (p > 0.05) two, three or four factor interactions were detected for both TPC and LAB (Table S2). It can also be noticed that foodborne pathogens (Escherichia coli, Listeria monocytogenes, Salmonella spp. and sulfite-reducing clostridia) were not detected in any sample throughout the storage. The obtained results suggested that all treatments provided a satisfactory microbiological quality according to EU regulation [64].
Texture Analysis of Dry Fermented Sausages
Results of instrumental determination of texture characteristics are shown in Table 4. As expected, fat content significantly changed (p < 0.05) the texture parameters (hardness, springiness, cohesiveness and chewiness). Samples with lower fat content showed a higher value of hardness and chewiness. An increase in hardness and chewiness as fat content decreases in dry fermented sausages was also reported by other authors [65,66], probably as a result of a more pronounced moisture loss in sausages with higher proportions of lean meat [66]. During the storage, hardness and chewiness values showed significant (p < 0.05) increase until the 150th day of storage, followed by decrease till day 225. Springiness value showed constant increase, while cohesiveness showed constant decrease until the 150th day of storage. Rubio et al. [67] reported increase of hardness, springiness, cohesiveness and chewiness of dry fermented sausage over the whole storage period (till 210 days), while Severini et al. [68] reported decrease in firmness, due to proteolysis. In the case of nitrites, it can be observed that the addition of sodium nitrite significantly (p < 0.05) affected the cohesiveness of dry fermented sausages. A similar finding was observed by Dong et al. [69] in cooked pork sausages. Moreover, Villaverdre et al. [70] found that the sodium nitrite addition at the levels of 75 and 150 mg/kg increased the hardness of fermented sausages. This could be related to the ability of sodium nitrite to promote protein oxidation and Schiff base formation [70]. Regarding JEO, it has been noticed that JEO addition had no impact on the texture parameters (hardness, cohesiveness and chewiness) of dry fermented sausages. Similarly, Viuda-Martos et al. [71] reported that rosemary essential oil has no effect on texture parameters of cooked sausages. Two-way (FC × SD) and four-way interactions were also significant (p < 0.05-0.001) for all texture parameters (Table S2). The main texture parameters (hardness and chewiness) ranged in interval from 3539 (FC = 25%, NC = 75 mg/kg, SD = 0, JC = 0.00 µL/g) to 10,990 g (FC = 15%, NC = 0 mg/kg, SD = 150, JC = 0.05 µL/g) and from 926 (FC = 25%, NC = 0 mg/kg, SD = 225, JC = 0.05 µL/g) to 2759 g (FC = 15%, NC = 150 mg/kg, SD = 225, JC = 0.05 µL/g), respectively. Similar results for hardness of different dry fermented sausages were observed by Triki et al. [65] and Rubio et al. [67].
Sensory Analysis of Dry Fermented Sausages
Results of sensory analysis are shown in Table 5. The fat content, JEO content and storage time had a significant (p < 0.05) effect on the sensory attribute of color. Two-way (FC × NC, FC × SD, NC × SD, FC × JC, SD × JC), three-way (FC × NC × SD, FC × NC × JC, FC × SD × JC, NC × SD × JC) and four-way interactions were also significant (p < 0.05-0.001) for this sensory attribute (Table S2). Moreover, nitrite content, JEO content and storage had a significant (p < 0.05) effect on sensory attribute of odor. Furthermore, two-way (FC × NC, NC × JC, SD × JC) and three-way (FC × NC × JC, NC × SD × JC) interactions had a significant (p < 0.05-40.001) effect on odor (Table S2). It should also be noticed that the numerical data for the sensory attributes of color and odor did not exceed the values of 2.0 (slight differences, less than 1.67 (color) and 1.78 (odor)), in any samples. Hence, the obtained results suggested that the fluctuations of fat, sodium nitrite and JEO, as well as storage time had no negative impact on these sensory attributes. Nitrite content, JEO addition and storage time had a significant (p < 0.05) effect on sensory attribute of flavor. Two-way (FC × JC, NC × JC, SD × JC), three-way (FC × NC × JC, FC × SD × JC, NC × SD × JC) and four-way interactions had also a significant (p < 0.05-0.001) effect on the flavor (Table S2). The highest differences (>3, higher than moderate) of typical flavor were observed in the samples: FC = 25%, NC = 150 mg/kg, SD = 225, JC = 0.10 µL/g; FC = 15%, NC = 150 mg/kg, SD = 225, JC = 0.10 µL/g. Regarding JEO content of 0.05 µL/g, the highest difference (1.33) was observed in samples: FC = 15%, NC = 150 mg/kg, SD = 225, JC = 0.05 µL/g; FC = 15%, NC = 75 mg/kg, SD = 150, JC = 0.05 µL/g). This difference could be the result of interaction among the sodium nitrite and terpenoid-compounds of JEO. In our previous study we also determined that a high percentage of JEO had a significant effect on the strong aroma of cooked pork sausages [30]. Using novel extraction techniques (e.g., supercritical fluid extraction) at optimum conditions results in extracts which possess a strong antioxidant and antimicrobial potential, as well as mild flavor, which enables their application at lower concentration in meat processing [17,23].
Conclusions
Monoterpene hydrocarbon β-myrcene (14.12%) was the most abundant compound identified in JEO. The sausages produced with a lower fat content were significantly darker and redder (p < 0.05). Moreover, the values of hardness and chewiness were significantly (p < 0.05) higher in the samples produced with a lower fat content. The variations in the contents of nitrite and JEO had no negative impact on the color and texture parameters of dry fermented sausages. No foodborne pathogens were detected in any samples. The highest concentration of JEO (0.10 µL/g) had negative impact on flavor. The addition of JEO (0.01 and 0.05 µL/g) combined with reduced concentration of sodium nitrite (75 mg/kg) efficiently retarded the lipid oxidation of high-fat (25%) dry fermented sausages during 225 days of storage. Hence, JEO with evident antioxidant potential could be used as a partial replacement for sodium nitrite in fermented sausages processing. In order to enhance the antimicrobial potential of JEO, the usage of novel extraction technique (e.g., SFE) could be an effective solution. Further investigations are needed to analyze the synergistic effects of different natural extracts, isolated from various plant sources, on improving the quality and shelf-life of meat products.
Supplementary Materials: The following are available online at http://www.mdpi.com/2304-8158/9/8/1066/s1. Table S1. Proximate chemical composition of dry ferment sausages; Table S2. The effect of two-way, three-way and four-way interactions among processing parameters on the quality of dry fermented sausages expressed as p-value.
Conflicts of Interest:
The authors declare no conflict of interest. | 2020-08-13T10:09:59.775Z | 2020-08-01T00:00:00.000 | {
"year": 2020,
"sha1": "e047d086216b429e9d25eb93d90622dfd2c995e8",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/2304-8158/9/8/1066/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "51cd5c5e28bb6e2529a6cd5ffc63516c601121b3",
"s2fieldsofstudy": [
"Agricultural and Food Sciences"
],
"extfieldsofstudy": [
"Medicine",
"Chemistry"
]
} |
52914176 | pes2o/s2orc | v3-fos-license | Cell Cycle Changes after Glioblastoma Stem Cell Irradiation: The Major Role of RAD51
“Glioma Stem Cells” (GSCs) are known to play a role in glioblastoma (GBM) recurrence. Homologous recombination (HR) defects and cell cycle checkpoint abnormalities can contribute concurrently to the radioresistance of GSCs. DNA repair protein RAD51 homolog 1 (RAD51) is a crucial protein for HR and its inhibition has been shown to sensitize GSCs to irradiation. The aim of this study was to examine the consequences of ionizing radiation (IR) for cell cycle progression in GSCs. In addition, we intended to assess the potential effect of RAD51 inhibition on cell cycle progression. Five radiosensitive GSC lines and five GSC lines that were previously characterized as radioresistant were exposed to 4Gy IR, and cell cycle analysis was done by fluorescence-activated cell sorting (FACS) at 24, 48, 72, and 96 h with or without RAD51 inhibitor. Following 4Gy IR, all GSC lines presented a significant increase in G2 phase at 24 h, which was maintained over 72 h. In the presence of RAD51 inhibitor, radioresistant GSCs showed delayed G2 arrest post-irradiation for up to 48 h. This study demonstrates that all GSCs can promote G2 arrest in response to radiation-induced DNA damage. However, following RAD51 inhibition, the cell cycle checkpoint response differed. This study contributes to the characterization of the radioresistance mechanisms of GSCs, thereby supporting the rationale of targeting RAD51-dependent repair pathways in view of radiosensitizing GSCs.
Introduction
Glioblastoma (GBM) is the most malignant type of primary brain tumor [1]. After initial diagnosis of GBM, standard treatment consists of maximal surgical resection, radiotherapy, and concomitant adjuvant chemotherapy with temozolomide (TMZ) [2]. However, the median time to recurrence is 6.9 months, and the median survival time from diagnosis is 14.6 months [3]. Many GBMs initially respond to treatment, but residual radioresistant cells often survive after radiotherapy and give rise to tumor recurrences. Attempts to improve patient outcomes by increasing radiation doses have been unsuccessful due to necrosis in the surrounding brain, and diffuse cerebral atrophy, as a consequence of exposing healthy cells to levels exceeding their dose tolerance limit [4]. A more efficient strategy would be to suppress or overcome the mechanisms underlying radioresistance in conjunction with radiotherapy.
Several studies have highlighted the presence in the tumor of a subpopulation of cells with characteristics similar to the neural progenitor cells called "Glioma Stem Cells" (GSCs) [5,6]. A pool of these cells can survive exogenous DNA damage, such as radiation-induced double-strand breaks, and repopulate the tumor following treatment, thereby contributing to radioresistance and tumor recurrence [7]. It is now acknowledged that GSCs are highly resistant to chemotherapeutic drugs, including TMZ, and promote radioresistance by preferential activation of a DNA damage response [8]. Initial characterization of the radioresistant properties of GSCs has shown that these cells may resist radiotherapy through preferential activation of a DNA damage checkpoint response and increased DNA repair capacity [8]. Other studies have reported cell cycle defects in the G1/S and S phase checkpoints following irradiation, suggesting that homologous recombination (HR) and cell cycle checkpoint abnormalities may contribute concurrently to the radioresistance of GSCs [9,10]. Existing models established in different cell types have shown that radio-induced double-strand breaks (DSB) undergo faster repair in the G2 phase during the first hours following irradiation [11]. Ropolo et al. indicated that GSC lines display a significantly elongated cell cycle compared to non-stem cells in response to enhanced activities of the DNA damage checkpoint kinases Chk1 and Chk2 [12]. Several other studies have reported that the sensitivity of cells to IR varied widely depending on the phase of the cell cycle at the time of irradiation, and have also shown that cells in the G2 and M phases were approximately 3 times more sensitive than cells in the S phase [13][14][15].
Overexpression of RAD51 in cancer has been widely documented, particularly in glioblastoma [16]. RAD51 protein is crucial for HR, and its inhibition has been shown to sensitize GSCs to irradiation or alkylating drugs [17,18]. RAD51 expression is closely related to the state of cell proliferation and maximally transcribed in the late S and G2 phases [19,20]. In a previous paper, we analyzed the DNA damage response after ionizing radiation (IR) in a panel of 10 GSC lines derived from patients with clearly defined molecular characteristics [21]. While our results underscored efficient DNA repair in all GSC lines, due to variations of intrinsic radioresistance the responses were heterogeneous. Indeed, GSCs could be divided into two groups according to their DNA repair kinetics at 4Gy: a radiosensitive group including GSC-1, -3, -5, -10, and -11, and a radioresistant group including GSC-2, -6, -9, -13, and -14. This study highlights the importance of RAD51 expression in GSCs and, more specifically, of the HR pathway involved in radioresistance. It should consequently be of interest to characterize cell cycle progression and cell cycle changes following RAD51 inhibition in GSCs. The aim of this study was to examine the consequences of IR for cell cycle progression in GSCs. In addition, we intended to assess the potential effect of RAD51 inhibition on cell cycle progression in GSCs.
Analysis of Radioresponse and Cell Cycle Progression of GSCs
Ten GSC lines were subjected to 4Gy IR and we evaluated cell cycle progression through fluorescence-activated cell sorting (FACS) analysis. Variations in cell cycle phases were compared at 24, 48, and 72 h. A significant increase of cell populations in the G2 phase was observed among all GSCs concomitantly with a decrease of G1 and S phase fractions ( Figures 1A,B, S1 and S2). Analysis of later time points indicates that all GSCs were able to sustain G2 arrest for 72 h. In this study, we compared G2 populations between the two groups in the absence of IR (T0) and found no statistical difference ( Figure 1C). We then compared mean variations in G2 phase following IR (∆G2), and once again no significant difference was observed between the two groups ( Figure 1D).
Synchronization and Cell Cycle Progression of GSCs
In unsynchronized cultured GSCs, populations at different stages of the cell cycle can co-occur. In this respect, FACS analysis has demonstrated that, in untreated GSCs, the G2 phase population ranged from 4.8 to 27.2% ( Figure S2), and that these differences may markedly modify their radiosensitivity. To circumvent this problem and to confirm that the previously observed G2 arrest was not biased, we synchronized GSC cultures before IR exposure. Following a double thymidine block, GSC-1 and GSC-14 were efficiently synchronized at the G1/S border with an almost equal distribution of cells in the G1 and S phases ( Figure 2A). After removal of thymidine (T0), cells re-entered into the cycle over the next 24 h, thereby underscoring the reversibility of the G1/S block ( Figure 2A). As was previously observed with unsynchronized cells, IR induced an increase in the G2 phase population of synchronized GSCs ( Figure 2B).
Analysis of Radioresponse of GCSs according to the Verhaak Classification
We first classified GSCs according to the Verhaak signature using an Agilent SurePrint G3 Human GE (8 × 60K) chip. Centroid-based correlation analysis indicated that GSCs could be divided into three subtypes (proneural, classical, and mesenchymal) (Table S1). Both radiosensitive and radioresistant groups of GSCs presented an equal distribution of proneural, classical, and mesenchymal Verhaak subtypes ( Figure S2). We then evaluated G2 populations of GSC lines before IR and found no significant differences in G2 phase distribution between the subgroups ( Figure 3A). Following IR, proneural subtype presented reduced G2 arrest compared to other Verhaak subtypes, but the difference was not statistically significant ( Figure 3B).
The Consequences of RAD51 Inhibition for GSC Radioresponse and Cell Cycle Progression
Targeting RAD51-dependent repair is known to enhance tumor cell radiosensitivity [21][22][23][24]. In the present study, we assessed the effect of RAD51 inhibition on cell cycle profiles by using RI-1 inhibitor in radioresistant (GSC-6 and GSC-14) and radiosensitive cell lines (GSC-1 and GSC-11), with or without IR exposure. In radiosensitive cell lines, RI-1 did not modify G2 phase progression whereas radioresistant GSCs exhibited a more substantial increase of G2 population, aside from at the time point 48 h ( Figure 4A,B). FACS analysis of radiosensitive GSC lines (GSC-1 and -11) showed major G2 arrest after 4Gy IR that was moderately inhibited by RI-1 inhibitor ( Figure 4A). A differential effect of RI-1 inhibitor was noted on radioresistant cell lines (GSC-6 and -14) as it increased the G1 fraction at 24 h post-irradiation (T24-INH = 40% versus T24-INH-IR = 49%) and delayed G2 arrest post-irradiation by up to 48 h ( Figure 4B). Finally, we subjected radioresistant GSC-14 cells to a major radiation dose of 16Gy and, similarly to radiosensitive cell lines, GSC-14 presented a non-delayed G2 checkpoint after a high dose of IR, independently of RI-1 inhibitor ( Figure S3).
GSC Differences and Gene Expression
In the previous [21] and current study, we showed that the two groups differed according to their RAD51 protein expression, their repair kinetics, and their cell cycle profiles post-IR +/− RAD51 inhibition (Table 1). We then performed microarray analysis to investigate the differential expression of genes between the two groups in order to better understand the differences observed in cell cycle responses to IR. We performed gene set enrichment analysis between GSCs 1-3-5-10-11 and GSCs 2-9-6-13-14 according to the C5.bp.v6.1 gene set comprising 4436 gene ontology (GO) terms and according to a target gene set involved in DNA damage checkpoints, the cell cycle, or DNA repair. We found significant enrichment of DNA damage checkpoint and DNA repair process signatures (Table S2). Only the four most illustrative signatures with a Normalized Enrichment Score (NES) ≥1.66 and an False Discovery Rate (FDR) < 0.1 with the list of enriched genes are shown in Figure 5. Interestingly, genes from the Fanconi Anemia pathway FANCA, FANCI, FANCD1 (also known as BRCA2) were enriched in radiosensitive GSCs compared to radioresistant GSCs (Table S2). We then performed a Taqman Low density array (TLDA) assay to compare the expression of 46 target genes involved in DNA damage checkpoints, the cell cycle, or DNA repair between the radiosensitive and the radioresistant groups. Without IR exposure, significantly higher FANCD2 expression was observed in the radiosensitive group (BALBOUS signature, Table S2).
Discussion
HR and cell cycle checkpoint abnormalities can contribute to the radioresistance of GSCs and the targeting of both may represent a potential alternative treatment therapy for GBM patients. However, controversial issues have arisen about resistance phenotypes of GSCs to DNA-damaging agents and IR [8,25,26]. As a consequence of genetic heterogeneity in cancer, GSCs isolated from different patients may differ in their responses to DNA damage and, particularly, in cell cycle checkpoint occurrence [27]. To our knowledge, this study is the first attempt to characterize cell cycle progression and cell cycle changes following RAD51 inhibition in GSCs and to evaluate its potential as a new therapeutic target.
We first assessed cell cycle changes in 10 GSC lines derived from patients with clearly defined molecular characteristics [21,28,29]. Following 4Gy IR, all GSCs displayed a significant increase of cell populations in G2 phase at 48 h, which was persistent for up to 72 h. This finding was consistent with results from Wang et al. showing prominent S-and G2-phase checkpoints in response to IR in embryonic carcinomas [30]. In glioma-initiating cells, Lim et al. described a defect in cell-cycle arrest at the G 1 /S-phase and low dependency on the Non-Homologous End Joining (NHEJ) repair pathway following DNA damage [10].
DSB repair pathways have overlapping roles in DNA repair and are highly dependent on cell cycle phase [31]. Differential cellular sensitivities to IR have been reported depending on the phase of the cell cycle at the time of irradiation [13]. Hence, cells are known to be more radioresistant in the early G1 and late S phases and show increased radio sensitivity in the early S phase [14,15]. Synchronization of GSCs confirmed that a G2 checkpoint is preferentially used in these cells in response to radiation-induced DNA damage and that this outcome is independent of initial cell cycle phase.
It has been acknowledged that HR is active only in the postreplicative stages of the cell cycle, S and G2, during which time the sister chromatide is available [32]. In addition, rather than NHEJ, HR is preferentially activated in GSCs throughout the S/G2 checkpoint in response to DNA damage [33,34]. The G2 arrest observed in all GSC lines in response to IR may then be concomitant with HR activation.
RAD51 is a major component of HR-mediated DNA repair and is overexpressed in many cancers [35][36][37], including glioblastoma [16]. RAD51 activity is partially cell-cycle-dependent, with higher expression in the G2 phase in various cell types [20]. We previously identified two groups of GSCs based on their comet assay profiles with a radioresistant group that showed a significant increase in RAD51 protein expression after IR [21] (Table 1). In this current study, all GSC lines tested displayed a G2 checkpoint response after 4Gy IR, thereby underscoring the importance of G2 arrest in GSC response to DNA damage and supporting the major role of RAD51 in the repair mechanism, at least in the radioresistant group.
The radioresistant set of cell lines also presented a significant decrease in DNA repair after RAD51 inhibition, leading to increased cell death [21] (Table 1). Several attempts to target RAD51-dependent repair have increased the sensitivity of tumor cells to radiotherapy, albeit to a variable extent [22][23][24]. A recent publication confirmed the importance of RAD51 in radioresistance mechanisms, highlighting the potential of RAD51 as a means of targeting specifically radiosensitized GSCs [18]. However, so far, there have been no published reports evaluating the cell cycle profile in GSCs and RAD51 inhibition.
After 4Gy IR, the radioresistant group displayed different G2 profiles compared to radiosensitive cells. Addition of RI-1 inhibitor had only a small influence on the cell cycle of radiosensitive cells, with a slight decrease in G2 populations.
This observation may reflect other repair pathways acting in the radiosensitive group requiring G2 arrest and, to a lesser extent, RAD51. In contrast, radioresistant cells manifested an increased G1 fraction and delayed G2 arrest by up to 48 h in the presence of RI-1. Recently, Chen et al. highlighted a significant time lapse in the cell cycle transition at the G1/S phase using the same RI-1 inhibitor in cervical cancer cells [38]. Thus, the G2 delay observed in radioresistant GSCs after IR and RI-1 treatment might be caused, at least to some extent, by attenuation of the cell cycle transition from G0/G1 to S phase. One of the limitations of this study was the slow doubling time of GSCs (mean doubling time: 7 days), and as the experiment was consequently focused on four independent cell lines representative of the radioresistant and radiosensitive groups, statistical analysis was not possible.
In the presence of RI-1, it is worth noting that radioresistant cells exhibit higher apoptosis and necrotic indexes that, in addition to our cell cycle results, are consistent with the central role of RAD51 in DNA damage response and radioresistance, thereby corroborating previous studies [39][40][41] (Table 1).
This study highlighted significant variation in G2 populations at T0 ( Figure S2 versus Figure 4). In this regard, plasticity of glioma stem cells is a well-known phenomenon that has evolved overtime as a consequence of genetic heterogeneity, generating subclones with differential responses to environmental changes [42]. As basal levels of G2 populations can vary widely in unsynchronized glioma stem cells, variations were taken into consideration when designing our experiments to monitor the cell cycle changes under exposure to IR and other chemicals. In this study, changes in cell cycle population were evaluated after adjusting to the T0 time point. Furthermore, additional synchronization experiments showed that G2 arrest was independent of the initial cell cycle state.
It is recognized that GSCs isolated from different cell lines or patients might present genetic heterogeneity, which could explain differences in the radiosensitive thresholds triggering DNA damage response and repair, as well as differences in cell cycle profiles [27,43]. We studied differences between molecular subtypes according to the Verhaak classification, one of the most widely acknowledged molecular glioblastoma classifications, and nevertheless found the reduced G2 arrest after IR in the proneural subtype not to be significant. GSCs differed according to their intrinsic radioresistance, with higher expression of RAD51 in radioresistant cells after IR exposure. Gene Set Enrichment Analysis (GSEA) revealed significant enrichment of Fanconi anemia pathway genes in the radiosensitive group in comparison with the radioresistant group, and, interestingly, no RAD51 enrichment was pointed out. A TLDAassay was used to compare the expression of 46 target genes involved in DNA repair and confirmed FANCD2 enrichment in the radiosensitive group. The FANCI-FANCD2 complex is known to stabilize RAD51-DNA interaction [44], and direct binding of FANCD1 to RAD51 through the BRC repeat domain is well-established [45]. Therefore, this observation could be indicative of impaired activity of Fanconi anemia genes in radioresistant as opposed to radiosensitive GSCs.
Indeed, Schlacher et al. reported that overexpression of RAD51 in FANCD2-deficient cells can partially rescue replication fork instability and secure fork protection [46,47]. These results are consistent with the increased RAD51 expression observed in radioresistant GSCs after IR [21] and support the assumption that radioresistant cells rely on RAD51 for DNA damage response, whereas radiosensitive GSCs may tend to favor other repair mechanisms. Cell cycle checkpoints and mechanisms involved in DNA repair are clearly interdependent, with the choice of repair mechanism consistently adjusted throughout the cell cycle [34]. Herein, the significant difference observed in repair gene expression between the two groups may to a certain extent explain the differences observed in cell cycle profiles.
To conclude, these data should enhance our understanding of the mechanisms governing radioresistance in GSCs and support the rationale for targeting RAD51 repair pathways in view of radiosensitizing GSCs.
GSC Lines and Cell Culture
Glioblastoma stem cell cultures were derived from freshly resected tumors after written informed consent obtained from each patient enrolled in the study. This study was approved the 30/06/2004 by the ethics committee of Poitiers University Hospital (DHOS/OPRC/FCnotif-tumoro-jun04: 04056), in accordance with the Declaration of Helsinki. The cells were grown as previously described in [18][19][20]. Briefly, GSCs were cultured at 37 • C as proliferative non-adherent spheres in Neurobasal medium (NBE, Life Technologies, Carlsbad, CA, USA) supplemented with 20 ng/mL of basic fibroblast growth factor (bFGF, Life Technologies), 20 ng/mL of epidermal growth factor (EGF, Life Technologies), and culture supplements N2 (100×, Life Technologies) and B27 (50×, Life Technologies). Culture medium was replaced twice a week and, when the spheres became large, they were enzymatically dissociated with accutase (Merck-Millipore, Billerica, MO, USA). GSCs have previously been characterized as regards self-renewal, differentiation, and in vitroclonogenicity by limiting dilution assays [28,29]. Tumorigenicity was evaluated by xenograft experiments in nude mice. The molecular traits of the GSCs were detailed in Balbous et al. [21].
Cell Irradiation
Gamma irradiation was performed at the Department of Radiotherapy (University Hospital of Poitiers, Poitiers, France) with an Elekta Synergy Beam Modulator (dose rate, 4.56Gy/min). GSCs were irradiated at room temperature in tissue culture flasks and cultured at 37 • C. Control cells were subjected to the same experimental conditions.
G1/S Cell Synchronisation with a Double Thymidine Block
As regards GSC synchronization at the G1/S border, the double thymidine procedure was chosen due to the fact that thymidine is an easily reversible block [48]. Cells were grown in a medium containing 4mM thymidine (Sigma-Aldrich, Saint-Louis, MO, USA) for 72 h (first thymidine block), then washed with prewarmed phosphate-buffered saline (PBS) and released for 24 h by culture into a fresh medium. A medium containing 4mM thymidine was added to the cultures for a second 72-h interval (second thymidine block). Finally, the cells were washed and cultured into a fresh medium (T0).
Microarray Analysis
Microarray analysis was performed on an Agilent SurePrint G3 Human GE (8 × 60K) chip (Agilent, Santa Clara, CA, USA). Data were subject to a threshold (values below 1.0 were set to 1.0) and Log2-transformed. Normalization was performed using the per-array and per-gene median scaling method. Genes with very low signals or below noise level were excluded from analysis. All 10 GSCs were analyzed using the 840-gene signature reported by Verhaak et al. [50] and assorted into proneural, mesenchymal, or classical subtypes (Supplementary Table S1). Gene expression of GSCs 1-3-5-10-11 and GSCs 2-9-6-13-14 was compared according to the C5.bp.v6.1 gene set comprising 4436 GO terms and according to a target gene set involved in DNA damage checkpoints, the cell cycle, or DNA repair using Gene Set Enrichment Analysis (GSEA, software.broadinstitute.org/gsea/msigdb/). Lists of two significant enriched signatures are detailed in Table S2.
TaqMan Low Density Array
TaqMan Low density array (Life Technologies) was performed as previously described in [21]. The experiment was conducted in triplicate. A list of 46 genes called "BALBOUS signature" was evaluated (Table S2). Gene expression of GSCs 1-3-5-10-11 and GSCs 2-9-6-13-14 was analyzed according to the 2 −∆∆Ct method using GAPDH expression as an endogenous control. Descriptive statistics of the results were calculated with GraphPad Prism 6 (La Jolla, San Diego, CA, USA).
Statistical Analysis
Statistical analyses of the results were performed with GraphPad Prism 6 (La Jolla). The results are presented as mean ± standard error of the mean (SEM) and statistical significance was evaluated by nonparametric multiple comparison using the Kruskal-Wallis and Mann-Whitney tests. ∆ T-T0 stands for mean variation of cell cycle phases at indicated post-irradiation time points. Statistical group comparison of the TLDA results was evaluated by the nonparametric Mann-Whitney test. p values less than 0.05 were considered statistically significant. | 2018-10-21T21:47:41.039Z | 2018-10-01T00:00:00.000 | {
"year": 2018,
"sha1": "9ab7f754607ea99756c4a67b1eb4e9ffa4c50e78",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/1422-0067/19/10/3018/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "9ab7f754607ea99756c4a67b1eb4e9ffa4c50e78",
"s2fieldsofstudy": [
"Biology"
],
"extfieldsofstudy": [
"Medicine",
"Biology"
]
} |
4690932 | pes2o/s2orc | v3-fos-license | Epidemiologic Profile of an Otolaryngologic Emergency Service
Introduction According to current research, the number of patients seen in the emergency room is progressively increasing. There are few studies on the characteristics of ear, nose, and throat diseases treated in the emergency room. Objectives (1) To establish the epidemiologic profile of patients with these complaints treated at a referral emergency hospital in locoregional city Curitiba, and (2) to evaluate the calls that truly required emergency care. Methods This is a contemporary cross-study of urgent and emergency referrals to a hospital with otolaryngologic services during the year 2012. Data were collected and epidemiologic characteristics analyzed. Results We analyzed 1,067 patients: 312 presented in spring, 255 in summer, 253 in autumn, and 247 in winter. We found 17 diseases that were common during the year, with 244 (23.99%) upper respiratory tract infections being the most frequent disease. There was no statistically significant difference in the incidence of diseases, except that acute otitis media was most common during the summer (p = 0.02); distribution between the sexes was balanced. The predominant age group was adults. We found 9.27% cases were true emergencies. Conclusion Patients were 20 to 40 years, with upper respiratory tract infection the most incident disease; 9.27% of cases were emergencies.
Introduction
The epidemiologic profile of a particular disease can be expressed in absolute numbers (cases in a population at any given time) or characteristics over time. Knowledge of the incidence is important to learn if the disease occurs more or less in a given period. This is a determining factor in the guidance and direction of public, private, educational activities and research. Knowing the incidence aids in configuring and enabling global health care services, especially with respect to highly frequent diseases such as those seen in all bureaucratic factors, and the patient's health and medical situations. 2 It is an internationally accepted fact that measurement of progress and scientific rigor of an area relates to the production of articles published in journals indexed and recognized as prestigious. There are few studies on the characteristics of ear, nose, and throat (ENT) diseases treated in emergency rooms, especially in relation to the severity of treated cases and the adequacy of assistance by the institutions providing care. 4 Timsit et al reported that only 10% of cases presenting to an emergency room were true emergencies. 5 Pino Rivero et al, in work developed from data on ENT emergencies, affirm that it is essential to define frameworks for what constitutes an actual emergency so as to enable adequate planning and care. 6 They also consider that fewer than one third of the visits could be considered real emergencies.
A literature review of studies of the incidence of ENT diseases included searches in VHL (Virtual Health Library), which uses as a database Lilacs; IBEC; Medline; BBO; Cid-Saúde; DISASTERS; MedCarib; REPIDISCA; BDENF HOMEO-INDEX; PAHO; and WHOLIS Cochrane Library. Of the 97 studies found, only three are Brazilian, and of these, two are limited to pediatrics and the third to geriatrics. 5 The objectives of this study are: 1. Establish the annual epidemiologic profile of patients with these complaints treated in a hospital in the city of Curitiba. 2. Evaluate the cases that really would require emergency care.
Methods
This study design was chosen because of its ability to identify possible risk factors associated with ENT diseases, mainly to factors that remain unchanged over time and that may provide valid evidence of statistical associations, without defining the etiologic character. Moreover, it was low cost, with a high description potential and analytical simplicity. We statistically analyzed appropriate factors according to the nature of the data. Inferential statistics were compared for the year 2012 on the following variables: sex, age, upper respiratory tract infections (URTIs), acute rhinosinusitis, tonsillitis, acute rhinitis, cerumen, acute otitis media, acute otitis externa, epistaxis, vertigo, acute laryngitis, secretory otitis media, peritonsillar abscess, temporomandibular joint dysfunction, tinnitus, nasal fracture, nasal foreign body, and foreign body in the ear. For this, we used the standard analysis (Χ 2 ) test with significance level α < 0.05. Graphs of annual frequency of high and low incidence diseases were developed.
Results
Data from 1,067 patients were collected from January 1, 2012, to December 21, 2012. During the spring, a total of 312 patients were seen. Of them, 179 (57.37%) were female, and 30.98% were in the age group 20 to 40 years. During the summer, a total of 255 patients were seen. There was no significant difference between females (n ¼ 132; 51.76%) and males (n ¼ 123; 48.24%) in the emergency service. In summer, 38.32% of patients were in the age group 20 to 40. During the fall, a total of 253 patients were seen; 143 (56.2%) patients were female and 110 patients (43.48%) were male. In the fall, 36.41% of patients were in the age group 20 to 40 years. During the winter, a total of 247 patients were seen; 139 (56.28%) patients were female and 108 patients (43.72%) were male. In winter, 42.34% of patients were in the age group 20 to 40 years.
►Fig. 1 shows the distribution of patients by gender throughout the year. We found a higher incidence of females (55.57%). We evaluated the correlation between sex and seasons to demand emergency service through standard analysis (Χ 2 ) testing. There was no statistically significant difference between them. ►Fig. 2 shows the attendance by age throughout the year. Overall, 37.01% of patients were in the age group 20 to 40 years.
We apply the standard analysis (Χ 2 ) test among the most common diseases in relation to the seasons and found an increased incidence of AOM in the summer compared with other seasons (p ¼ 0.02). With other diseases, there was no significant difference in frequency in relation to the seasons of the year.
Among the most common diseases, epistaxis, peritonsillar abscess, nasal fracture, nasal foreign body, and foreign body in the ear were considered true emergencies. In all, 99 (9.27%) true emergencies were noted (►Fig. 4).
Discussion
The flow of ENT patients in Curitiba is intense during all seasons of the year; with the epidemiologic profile of emergency room cases, we can develop strategies to better serve our patients. In addition, studies on the epidemiologic profile in otolaryngology are scarce, which makes this type of study necessary for our medical specialty.
We found that the demand for ENT care is higher in female patients; however, there was no statistically significant dif-ference between the sexes. The most common age group in all seasons was 20 to 40 years. This demonstrates that our patients are mostly adults. When considering that this age group is the most economically active, removals and medical certificates must be made judiciously. Among the most common diseases, URTI and RSA were the most frequent throughout the year.
We also conclude that our service is primarily composed of urgency care, and only 99 (9.27%) cases were considered real emergencies. Timsit et al 5 found in their series that 10% of cases were considered emergencies, close to the value found in our city.
In future studies, we will continue to evaluate the incidence of disease in each age group as well as the most commonly prescribed drugs in our service.
Conclusion
1. The patients profile was successfully made: 20 to 40 years old, cases of urgency, with upper respiratory tract infection the most common disease. 2. We found an incidence of 9.27% of true emergency cases in emergency care of otolaryngology in Curitiba. | 2016-05-15T02:21:31.035Z | 2014-06-17T00:00:00.000 | {
"year": 2014,
"sha1": "344a98429aa3b099bad13ca35574a3c900a49a80",
"oa_license": "CCBYNCND",
"oa_url": "http://www.thieme-connect.de/products/ejournals/pdf/10.1055/s-0034-1382099.pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "344a98429aa3b099bad13ca35574a3c900a49a80",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
25179819 | pes2o/s2orc | v3-fos-license | Pretreatment prostate specific antigen doubling time as prognostic factor in prostate cancer patients
Despite the prostate-specific antigen (PSA) serum level commonly uses as tumor marker in diagnosis of prostate cancer, it seems that PSA doubling time (PSADT) could be more useful indicator of tumor behavior and of prognosis for patients. The results of hormone and radiation therapy were evaluated for 912 prostate cancer having at least 2 PSA tests before the treatment was started. Clustering procedure (selection of homogenous group) was performed by using PSADT as the classification marker. The rate of PSADT was estimated for different dissemination rate, age, Gleasons's score and education level. PSADT index inversely correlated with the rate of prostate cancer dissemination, Gleason's score and the level of education were directly correlated with the age of patients. Survival time was longer and PSADT index was higher in “slow” tumor growing subgroups in local, local-advanced and metastatic prostate cancer patients than these in “fast” subgroups. The study confirmed the prognostic value of pretreatment PSADT in prostate cancer patients independently of cancer progression. No significant relationship exists between the authors and the companies/organizations whose products or services may be referenced in this article.
INTRODUCTION
Prostate-specific antigen (PSA) -a glycoprotein which is produced and thereby secreted from the epithelium of prostate gland is responsible for liquefaction and ejaculation [1]. Furthermore serum PSA level served as a tumor marker in diagnosis of prostate cancer as well as monitoring of its development. The dynamics of increase in its concentration can be described mathematically in different ways. Among them PSA doubling time (PSADT) is potentially seems most useful. PSADT being sensitive to exponential tumor growth therefore requires a logarithmic analysis [2,3] Distinctly from PSA index, PSADT can help us learn more about biological behavior of cancer, i.e. to tumor progression [4]. At present time, the estimation of PSADT is recommended for [5]: diagnostic of biochemical relapses after radical treatment with the goal to predict tumor-specific survival rate [6,7], to select candidates for active observation group [8,9], as well as in determining the necessity to start treatment of patients who have select active observation.
PSA level without any other clinical data is not considered as a factor in prognosis. Nevertheless it is well known that the pretreatment (primary) PSA level closely correlates with the prostate cancer progression: serum PSA level increase after radical treatment is the main hallmark of tumor recurrence; the dynamics of PSA concentration in response to treatment reflects the effectiveness of therapy [1]. Nevertheless, the rate of PSA kinetics is not considered as the hallmark of tumor growth.
PSADT should be determined before the initiation of treatment in order to obtain information concerning the aggressiveness of prostate cancer, for prediction of treatment results and for monitoring the course of treatment. There are some contradictory data regarding this issue. Some authors proclaim prognostic significance of PSA initial kinetics for disease-free patient survival after beam therapy [10,11] or radical prostatectomy [12]. However, others authors not support this opinion [13,14].
The main aim of this study is to evaluate clinical and prognostic significance of PSADT indices, determined before the starting of treatment in patients with prostate cancer, exposed to combined hormone and radiation therapy.
RESULTS
The PSADT parameters were calculated for 912 prostate cancer patients before starting of their treatment. The differences between the groups were statistically significant (p < 0.00001).
The comparison between PSADT and the primary PSA level also revealed highly significant differences (p < 0.00001). The more PSA level was determined, the less PSADT median value was observed.
The parameters of PSADT in patients of various age and of different education levels are presented in the Table 2. The median PSADT value was 5.5 (IQR 1.4-17.5) months in patients younger than 59 years, 9.0 (IQR 2.5-25.4) months in patients aged from 60 to 69 years; 18.4 (IQR 3.8-52.6) months in patients aged 70-79 months, and 18.6 (IQR 7.6-63.3) months in these older than 80 years.
The PSADT values were also significantly dependent on the patient's education level (p < 0,001) ( Table 2). It was clearly seen that the higher levels of education directly related to the more high parameter of PSADT.
In order to evaluate the PSADT prognostic value, all patients were divided into subgroups designated as "slow" and "fast" PSADT groups. Parameters characteristic for these groups are presented in the Table 3. The long-rank test shows significant difference between slow and fast subgroups in local prostate cancer patients ( Figure 1) (p < 0.01). Similar tendencies were found in patients with generalized (metastatic) prostate cancer. Median PSADT values were 23.4 and 49.1 months in fast and slow subgroups, respectively ( Figure 2). In patients with locally-disseminated prostate cancer survival curve for slow PSADT did not reach median value level ( Figure 3). The difference from the fast PSADT was statistically significant (p < 0.01, log-rank test).
Multivariate analysis revealed that Gleason score (p=0.005), clinical stage (p=0.03) and a pretreatment PSADT (p < 0.001) were independently associated with overall survival. Pretreatment PSA value, hormone therapy timing and duration, and radiation therapy dose were not statistically significant on multivariate or univariate analysis.
DISCUSSION
The prognostic significance of the primary PSA kinetics in prostate cancer patients, subjected to radical prostatectomy or distant radiotherapy has been discussed in several works [11,12]. The clinical significance of PSADT level was not evaluated and not compared with the parameters of prostate cancer in both these studies. Only local prostate cancer patients were under observations in both these works. There was no positive association between the initial PSA dynamics and postoperative pathomorphological alterations, as well as patients' survival rate after radical prostatectomy [7]. However, the number of patients being under observation was rather scarce in this study (n=86) as well as postoperative observation time was 2 years, whereas our work based on a much larger cohort of patients (n=912). For the first time the correlation between PSADT with tumor process characteristics, age and patient's education level was demonstrated. We observed that PSADT decreased with the decrease in prostate cancer differentiation. This parameter also decreased with the increase in its dissemination as well as with the rise in primary PSA level. It was established that the PSADT index inversely correlated with the rate of prostate cancer dissemination, Gleason's score and the level of education were and directly correlated with the age of patients. Survival time was longer and PSADT index was higher in "slow" tumor growing subgroups in local, local-advanced and metastatic prostate cancer patients than these in "fast" subgroups. Some authors believe that with increased education level there is a tendency in aggressiveness decrease in prostate cancer [15,16] which correlates with our data. Our data suggested that the primary PSADT rate could serve as prognostic factor regardless neoplastic dissemination process. The increase in the rate of PSA correlates with reduction of the prostate cancer patients survival. We believe that primary dynamics of PSA level reflected the rate of prostate cancer growth thus allowing to evaluate it as adequate marker for estimation of prostate cancer growth rate [17,18].
It was shown that the higher education level or length is associated with greater use of PSA screening and more likely to have a prostate biopsy than men with low (short) education [19][20][21][22].
PSA testing without clinical manifestation was also associated with higher education in prostate cancer patients [19]. In our study the PSADT index was higher in patients with local prostate cancer than in patients with metastatic cancer. Also, PSADT was much higher in most qualified (educated) patients than those with only secondary education or less. These data are in agreement with observations demonstrated association between high level of education more healthy lifestyle and more survival rate [22][23][24][25].
MATERIALS AND METHODS
The results of hormone and radiation therapy of 912 patients with prostate cancer were evaluated. All patients were treated and monitored at the Russian Scientific Center of Radiology and Surgical Technologies during the period since 1994 to 2012.
Criteria for including the patients in the research were as follow: 1) Morphologically verified diagnosis of prostate adenocarcinoma; 2) Presence of at least 2 PSA measurements, performed with intervals 1-4 months before starting the treatment.
3) Patient compliance to medications which distort actual PSA value (5α-reductase inhibitors and so on); 4) Positive trend of PSADT value (PSA serum concentration rise during re-examination).
In the selected cohort PSADT was determined before the start of treatment. PSA calculation was made accordance to Memorial Sloan-Kettering Cancer Center recommendations using the PSADT calculator [26].
PSADT clinical significance was determined by comparing this criteria to characteristics of the tumor (spread, sum of notes according to Gleason's scale i.e. Gleason's score; PSA primary level -maximal PSA value, based on which biopsy was taken and diagnosis was made), as well as individual patient characteristics (age, education level). The level of education was ranked according to established in Russia system of education as follow: low -secondary school (education length < 11 years;), medium -college for professional education (secondary specialized, 9-12 years), highuniversity (12-16 years). We included into the analysis scholars with academic degrees -candidates of sciences (practically equal to PhD) and doctors of sciences (DSc). It is commonly assumed that academic degrees reflected more high level of education than university level, graduated from university or, and secondary and incomplete secondary. Beam therapy was performed on the linear accelerators of electrons by bremsstrahlung with limit energy 6 to 18 MeV. Patients with localized prostate cancer received local radiation therapy on target organs (prostate and seminal vesicles) of single local dose 3 Gy, with total local dose 54-57 Gy (equivalent dose: 66-70 Gy). Locally disseminated prostate cancer forms were treated in two steps with daily radiation dose. In the first step of treatment, radiation therapy area except the target including regional lymphatic nodes. Based on documented regional lymph nodes involvement, radiation therapy was performed with single 2 Gy doses up to total 40-44 Gy dose.
Step 2 being based on local radiation. Generalized forms of prostate cancer with metastatic pelvic bones involvement were treated with beam therapy starting with segmental radiation step, with single 2 Gy doses up to 20 Gy total, afterwards regional and local radiation was performed according to regimens mentioned above. Patients with generalized metastatic skeletal involvement accompanied with pain syndrome received systemic radiation therapy with 89 Sr-chloride.
The majority of patients (79%) has been underwent hormone therapy with various gonadotropin releasing hormone analogues and antiandrogen medications. Another 21% of patients were underwent bilateral orchiectomy as hormone therapy method.
Patient were under dynamic observations once every 3 months during the first year and subsequently every 6 months. If it was not possible to observe them on a regular basis or in case of patient's absence at certain times regular phone calls were made, follow up letters were sent to their addresses and consultations with the patient's relatives was requested. Accordingly 1st of July 2013 (control point) the comprehensive information concerning every patient in research group was obtained.
To determine the prognostic significance, correlation between PSADT and total survival was evaluated. Clustering procedure (selection of homogenous group) was performed by using PSADT as the classification marker. To measure similarity the Jaccard index was used and as a part of clustering procedure the mean-K method was used. According to tumor dissemination, patients were divided into subgroups based on slow and fast PSADT indices. Kaplan-Mayer's survival curves were plotted for each subgroup using the PC program Statistica v.7 (StatSoft Inc., 2002). Difference between the curves was evaluated using log-rank test. To compare two independent selections Mann-Whitney U-test was determined. ANOVA Kruskal-Wallis test and the median test were used for comparing more than two groups. Pretreatment PSA value, Gleason score, tumor stage, timing and duration of hormone therapy, radiation therapy dose, and PSADT were analyzed for any associations with overall survival by using Cox proportional hazards multivariable analysis. With the p-value < 0.05 the difference was considered statistically significant. | 2018-04-03T01:41:22.368Z | 2017-01-01T00:00:00.000 | {
"year": 2017,
"sha1": "2597dedd80d85f89a25432bfae81bd0d880b6a3c",
"oa_license": "CCBY",
"oa_url": "http://www.impactjournals.com/oncoscience/files/papers/1/337/337.pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "2597dedd80d85f89a25432bfae81bd0d880b6a3c",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
228957851 | pes2o/s2orc | v3-fos-license | Implication of Sensory Evaluation and Quality Assessment in Food Product Development: a Review
The quality of food products is determined by physical properties, chemical composition, the level of contaminants (microbiological and toxic substances) and sensory attributes. Consumers around the world demand consistent supply of quality food products that reflect the value of the price they pay for them. The nature of raw materials and ingredients reflect the quality of food products in the market. Raw materials as well as packaging materials should be purchased based on the quality specifications that suppliers should adhere to. This review aimed at highlighting the importance of using objective assessment tools and consumer/sensory evaluation in determining the quality and acceptability of new food products. Objective tests are used to measure one particular attribute of a food product rather than its overall quality. They are generally rapid, reliable and repeatable. On the other hand sensory methods measure the reaction to stimuli resulting from the consumption of a product. Sensory testing is often used to determine consumer acceptability of a food product and contributes to the design of quality systems hence considered as a technical support for quality assurance during food production. Not only that but also it helps to obtain feedback for making decisions and carrying out proper modification of a particular food product. Therefore objective methods and sensory evaluation are indispensable tools for routine quality control of new food products as well as the existing ones. Current Research in Nutrition and Food Science www.foodandnutritionjournal.org ISSN: 2347-467X, Vol. 08, No. (3) 2020, Pg. 690-702 CONTACT Fabian Dominicus Mihafu fmihafu21@gmail.com Department of Applied Sciences, Mbeya University of Science and Technology, Mbeya, Tanzania. © 2020 The Author(s). Published by Enviro Research Publishers. This is an Open Access article licensed under a Creative Commons license: Attribution 4.0 International (CC-BY). Doi: 10.12944/CRNFSJ.8.3.03 Article History Received: 26 June 2019 Accepted: 11 November 2019
Introduction
Globally consumers demand consistent supply of high quality foods. Food quality may encompass parameters such as organoleptic characteristics, physical, functional properties, nutritional value and consumer protection from adulteration; on the other hand food safety is more associated with the content of various food chemical and microbiological characteristics. 1 Food safety can be measured via the examination of food items with regard to their microbial contamination, chemical contaminants or presence of physical foreign matter. 1 There is general view that food safety is of great concern especially when it comes to human health, and various efforts have been dedicated by several sectors to ensure that safer foods are obtained to the highest degree possible. 1 In many developing countries consumers are substantially conscious about the existence of potential risks in their foods, hence the need for safety and quality standards in all stages along the food supply chain. 3 The impact of this awareness often observed when consumers buy foods that are made by big companies rather than the smaller ones, as they believe that quality products are normally produced by big companies. 4 In developing countries, mid-size and small companies develop new food products and introduce them to the market, however these products are usually made from non-standardized materials and processes resulting into quality variations. 5 Poor quality control is common to small scale food processing enterprises and has been considered as the cause of the enterprises' weakness in terms of price, quality and delivery performance. Identifying internal and external requirements is imperative for a consistent improvement and maintenance of quality hence consumer satisfaction. 6 In any company or organization, employee responsiveness, motivation and satisfaction are needed for ensuring quality, and the prevention of errors and faults should be preferred to detection and correction. 7 Food companies need to build their product specifications on consumer likings and ensure that markets are segmented on the basis of consumer wants and needs. 4 This is why most big companies use objective tests and subjective tests/ sensory evaluation as a daily routine to maintain safety and quality of food products. 8 Objective tests/tools include physical, chemical and microbiological analyses of foods, 8 whereas subjective tools comprise of discriminative tests, descriptive tests and affective tests. They are essential part of both new product development and quality control, they also support marketing and marketing research activities. 9 In many cases quality means different things to different people, for example the appearance of the commodity is the most critical factor in the initial purchase while texture and flavor may determine subsequent purchases 10. The purpose of this review is to highlight the importance of using objective assessment tools and consumer/sensory evaluation in determining the quality and acceptability of new foods or food products.
Quality Control Programs
The major role of quality control programs is to ensure that raw materials and finished products are handled, processed, packaged and stored according to the required standards. Giovannucci and Satin 11 highlighted the fundamental functions of a quality control programs as:
•
Physical and chemical evaluation of raw materials and processed products. • In-process control of i) Raw materials, ingredients and packaging supplies ii) Processing parameters and iii) Finished products. • Microbiological analysis of raw materials and finished products.
•
Control of storage and handling environments. • Sanitation and waste products control. • Assurance that final products are within the established legal and marketing standards.
Type of Objective Evaluation Methods and their Significance In Quality Assessment of Foods or Food Products
Objective methods involve the use of instruments in carrying out evaluation of various attributes in food instead of human sensory organs. They are important in identifying contaminants in foods and uncover faulty processing and adulteration. 8 These tools include chemical tests which are used for the determination of chemical composition as well as chemical and metal contaminants in foods; microbiological methods that are used for detecting microbial contamination, insect excreta and other fragments in foods; and physical methods which are used for measuring characteristics like texture, viscosity, color intensity, turbidity and fill weight among others. 8 These methods focus on the determination of a specific characteristic of a food product than its overall quality. Usually they measure an attribute of the food that has a major effect on quality, so instruments need to be standardized with materials having properties that are comparable to the foods to be tested. To ensure quality and acceptability of new food products, objective tests should go hand in hand with sensory evaluation. 8
Physical Methods
The physical methods deal with attributes like appearance (size and shape), texture, color, viscosity, turbidity, firmness and imperfection, not only that but also process variables such as headspace, fill weight, drained weight and vacuum. 12 These are the rapidest methods which involve the least amount of training. 12,13 Appearance factors are useful in sizing and grading, which ensure uniformity and enhance the process of buying and selling. Therefore appropriate grading improves the quality of end products. Physical properties of food are important for the proper product designing and the prediction of the foods' response to processing, distribution and storage conditions.
Chemical Methods
These are used for quantitative and qualitative evaluations, as well as determination of nutritive values in foods. The composition of a food largely determines its safety, nutrition, physicochemical properties, quality attributes and sensory characteristics. 14 Chemical substances also play an important role in food production and preservation. For instance, coloring agents make food more attractive, flavorings make food tastier and stabilizers prolong the shelf life of food while food supplements are used as sources of nutrition. 14 Chemicals have a variety of toxicological properties, some of which might cause negative effects in humans and animals. 15 The control of food quality and safety is based on the determination of chemical composition as well as chemical and metal contaminants such as heavy metals and other toxic substances which can lead to acute poisoning and other health effects. 16 Usually, chemical substances are not harmful unless we are exposed to levels beyond standards for a long time. In new product development, chemical methods are useful in establishing safe levels which in turn notify decision makers who regulate the use of chemicals in foods or seek to limit their presence in the food chain. 17
Chemical Contaminants in Foods
Chemicals of greatest concern for health are naturally occurring toxins, these include mycotoxins, marine biotoxins, cyanogenic glycosides and toxins occurring in poisonous mushrooms. Staple foods such as cereals (maize, sorghum, wheat) and nuts (peanut) can contain high levels of mycotoxins (aflatoxin and ochratoxin) produced by Aspergillus.
A long-term exposure to these mycotoxins could cause cancer and affect normal development and immune system. 14 Dioxins and polychlorinated biphenyls (PCBs) are example of persistent organic pollutants that can accumulate in the environment and human body. Dioxin is highly toxic substance that can cause reproductive and developmental problems, great human exposure to dioxin is through foods such as meat and dairy products, fish and shellfish. Health effects of dioxins to human include skin lesions and altered liver function. It may also results into compromised immune system, nervous system, endocrine system and reproductive functions. Additionally, chronic exposure of animals to dioxins has been reported to cause several types of cancer. 18, 19, 20
Metal Contaminants in Food
Heavy metals such as lead, mercury, arsenic, cadmium, chromium and nickel are naturally occurring chemical compounds, they can be present at various levels in the soil, water and atmosphere. 21 Metals can also occur as residues in food as a result of human activities like farming, industrial operations and automobile exhausts. 16 Exposure to these metals is through both naturally occurring and manmade chemical compounds present at various levels in the environment including contaminated foods and water. 22 During food product development they must be checked to ensure food quality and safety. 17 Concentrations of these metals are enhanced by human activities and can enter plant, animal and human tissues through inhalation, diet and manual handling which in turn can bind structural proteins, enzymes and nucleic acids hence interfering their normal functioning. 23 Generally, long-term exposure to toxic heavy metals is associated with carcinogenic, circulatory, central and peripheral nervous system effects. 16 Therefore, chemical methods are crucial during food product development as they help safeguarding the quality of new food product and hence consumers' health.
Microbiological Methods
These methods are used in detecting food contamination by microorganisms such as bacteria and fungi (yeast, mold), also used to detect the presence of insect fragments, insect excreta and other foreign materials in food products. 24 For efficient quality control programs, they require significant training of the technical personnel to properly interpret results. Microorganisms can cause food poisoning and spoilage which principally affect food quality, major pathogens for this include Clostridium perfringens, Escherichia coli (O157:H7) and Shigella spp. 25 Microorganisms are all over the place and they are commonly found in faeces, soil and water; rats, mice, insects and pests; domestic, marine and farm animals (e.g. dogs, fish, cows, chickens and pigs); and human body parts (bowel, mouth, nose, intestines, hands, fingernails and skin). 26 Soil-borne microorganisms such as clostridia are common on raw vegetables, while Clostridium botulinum are targeted when designing processing steps to destroy them. Many food borne microbes are present in intestines, hides, feathers of healthy animals which are raised for food. For example, Salmonella serotypes have been scientifically found to infect a hen's ovary. 27 Salmonella spp have been a particular concern with foods of animal origin (e.g. meat, poultry, eggs and dairy products). 28 For Campylobacter spp, the most common foodborne species are Campylobacter jejuni and Campylobacter coli. Members of this genus are susceptible to environmental stresses and cause gastroenteritis associated with headache, diarrhea, fever, abdominal pain and muscle pain. 29 Environmental contamination may result from different sources including ingredients used in processing, either directly or indirectly, worker's hands, shoes, walls and floors. In food processing, food borne microbes can be introduced through humans who handle the food, or by cross contamination from some other raw agricultural materials and/or the establishment environment. 30 Insect damage, physical injury, enzymatic degradation and microbial activity can make food undesirable for human consumption. 27 Toxigenic pathogens create food "poisoning" by producing an enterotoxin in the food, a good example is staphylococcal enterotoxininduced illness. On one hand, food spoilage can be caused by molds, these are the major agents especially in vegetables where bacterial growth is not favored (e.g. low pH). Most molds invade plant tissue through surfaces such as a bruise or crack and spores are frequently deposited at these sites by insects for instance Drosophila melanogaster, the common fruit fly. 30 Yeasts, molds and lactic acid bacteria can as well spoil fermented vegetables such as sauerkraut and pickles, not only that but also other acid foods like salad dressings and mayonnaise. Spoilage in fermented vegetables is often manifested by off odors or changes in the color known as chromogenic colony growth or texture (eg. softening) of the product. 31 Some common routes in which microorganisms enter food products include soil, water, established environment, animal feeds, animal hides, food handlers, food utilities, air, dust, vegetable products and imported and exported products. 27 Most agricultural products and foods of animal's origin carry a wide range of bacteria. During production, processing, packaging, transportation, preparation, storage, and service, any food may be exposed to bacterial contamination and or microbial toxins that may result into mild to severe illnesses. 32 Since it is not possible to produce food with zero pathogens, microbiological methods help to ensure that manufactured food products are within the level of pathogen contamination that provides a scientifically acceptable risk. 14 Generally, objective methods (physical, chemical and microbiological) provide a milestone for quality control. In new food product development these methods should be well implemented and their final results should positively reflect the expected quality of the new food products. Emphasis should be directed to small scale food processing as well as indigenous food production.
Sensory Evaluation Methods
Sensory evaluation measures the reaction to stimuli resulting from the consumption of a food or food product. 33 34 It is concerned with providing answers to questions about product quality and existing competitor products as well as questions that are most often asked by persons concerned with technical, development, research and production; not only that but also factory managers, quality assurance managers and marketing managers. 35 Normally, the personal preference and powers of perception involve the use of sense organs. 8 The right answers to sensory questions are extremely important for the existence of any food product tested as they determine its acceptability and provide effective decisions for future product success in the market. 36 Sensory evaluation helps in eliminating or controlling sources of unwanted error, which can either be through control of the environment, proper experimental design and sampling of human subjects to test the products. 37 When conducting sensory analysis on food quality, it is imperative to know that there is possibility of getting wrong answers, which in turn can risk the competition of a particular food product in the market. Wrong information on food product could lead to miscommunication, improper decisions, wasted consumer research, and much more wasted time, effort and material costs. 8 Panel members form an important tool of sensory analysis. Its value depends on the objectivity, precision and reproducibility of their judgments. Panelists need to be carefully screened, trained, calibrated and validated. 38 Training enhance individual's understanding of sensory attributes and hence enable them provide accurate, valid, consistent, and standardized sensory measurements that can be reproduced. Each panelist must detect, recognize, and agree upon the exact connotation of each descriptive term. 39 The panelists may be guided to ensure consistence in their judgments by using reference standards that demonstrate variation in specific descriptive terms. 9 Panelists are given test samples (food product) and should state their preferences. The word "preferred" can mean most acceptable, tastes best, looks best, would buy and the like. 38 Consumer responses from the product are used to modify the product. The new food product can again be tested and where possible or necessary, modified further. Sensory analysis can therefore play a fundamental role in the management of product quality in the food industry. 35
Types of Sensory Evaluation Methods Discriminative Tests
These tests are designed to determine if a difference exists between food products. 37 Panelists should be knowledgeable about the product in question for easy choice. Each member is required to make a choice among the given food products.
Discriminative tests at some point may be used for different purposes (eg. determining sample differences/ similarities and or quantity of degree of difference/similarities). 34 To perform these tests about 10-50 panelists are recommended. Discriminative tests include triangular test, duo-trio test and paired-comparison test.
Triangular Test
In this test, normally three samples are involved when determining the overall difference between two products. Out of the three samples, two are similar and one is dissimilar. The samples must be coded with individual three-digit numbers. The taster is required to select the sample which is different from others. In these tests the chance of choosing the required sample correctly is greater. It is recommended that no more than six samples be evaluated at one testing session because the method is liable to fatigue of panelists. The tests require fewer tasters, at least 4-8 tasters are considered enough to carry single testing. 37 40
Duo-Trio Tests
This determines whether or not a sensory difference exists between two samples. There is always a reference sample and two test samples; of the two test samples, one sample is identical to the reference, and the other one is the test sample. 41 The panel members are asked to identify the sample that is similar to reference sample. Duo-trio tests are sometimes used instead of triangle tests to compare unknown differences between samples, however they are considered less efficient than triangle tests. At least 7-10 evaluators are recommended. 42 40
Paired Comparison Tests
These are applied when a difference in chemical composition of the sample which requires a sensory assessment is well known. Two differently coded samples are presented to each panelist at the same time and the task is to choose the sample that is perceived higher in the specified sensory attribute. 43 Tasters are asked to judge the samples by comparing them without needing to rate the magnitude of the difference, for example, "are the two samples identical or different?" or "which of the two samples sugary?". 43 Compared to triangular test, paired comparison test is less tedious and frequently used for strongly flavored or complex products. At least 7-10 panelists as for duo-trio are recommended in this test. 40 43
Descriptive Tests
In these tests, sensory attributes of products are characterized in order of their appearances and relative intensities are assigned. 37 They provide more detailed profiles of a product by identifying the different characteristics within the product and quantifying them. Descriptive tests are more comprehensive and sophisticated as compared to discriminative tests. 44
Quantitative Descriptive Analysis
This consists of progressive survey of sensory terms for a product generated by a trained sensory panelists using nontechnical language. Trained judges normally reach a consensus on the relative discrepancies between the samples. 47 QDA and FCP have the same purpose of determining the intensities of all product attributes and also defining the complete sensory profile.
Flavor Profile Analysis
This is useful for identifying sample taste and odor.
It is a technique that provides a written record of noticeable aroma of a product, flavor and aftertaste components. Panelists characterize individual aroma and flavor in the order perceived and assign a constant rating scale. Normally 4-6 panelists are suggested. They independently examine the product and record their impression in terms of aroma, flavor and aftertaste. Finally report is presented to a panel leader in an open discussion. 48 Texture profile analysis: This has been widely applied to test solid and semisolid food products. Usually it involves a panel of 6-9 members; textural attributes and other evaluation procedures are established unanimously by panel members before carrying out the evaluation of the products in question. 49 50 TPA is convenient for rapid evaluation of food texture which is normally measured only by humans. In some experiments, liquid samples that cannot keep their shape but flow under gravity are poured into a cup and subjected to uniaxial compression.
Then the parameters obtained from uniaxial compression are then discussed without considering the physical meaning of these parameters namely hardness, cohesiveness and adhesiveness. 51
Time Intensity Analysis
This is used to estimate the change in intensity of a determined characteristic with time. It has the main role of determining the intensity of any descriptor term in a product with time. TIA and FCP are among the descriptive sensory tests mostly used in scientific studies and by the food companies. 46
Affective/ Consumer Acceptance Tests
Affective methods are also called subjective methods. These are very useful for evaluating food acceptability or preference (which product is liked or preferred). Normally large number of respondents is required (50-150 panelists considered adequate).
Panelists are not trained but selected based on previous use of product, economic social level and geographical area. 33
Preference Ranking
In this technique, three or more samples are rank-ordered with one sample being preferred over the other. This type of test supply information about people's likes and dislikes of a product and determine how various samples differ based on a single distinguishing attribute. In consumer analysis, the panelists are asked to rank the coded samples according to their preference. 52
Hedonic Rating Scale
This is among of the widely used sensory evaluation methods that measure consumers' level of liking of food products. 53 In practice there are 9-point Hedonic scale, 7-point Hedonic scale and 5-point Hedonic scale. The 9-point Hedonic scale range from "like extremely'' to "dislike extremely''. Practically, not fewer than five points are recommended. 34
Some of the Requirements for Sensory Evaluation
There should be clear objective for conducting sensory analysis, appropriate area for preparation of food samples to be tasted, test procedures which should be properly implemented in a way that reduce risks to health of participants, good laboratory, proper method of sample presentation, sensory panel members suitable for evaluation (If training is needed, they must be trained), utensils / glass wares suitable for different foods to be tested, sophisticated sensory booths with controlled temperature (20 o C -22 o C) and relative humidity at 40 ± 5%, suitable lighting in booths, right coding of samples (usually 3 digit coding is preferred) and special software for statistical analysis of sensory data. 35 54 Sensory evaluation is very crucial in the new product development. When properly implemented with controls, careful screening of panelists, effective training as well as proper statistical interpretation, it can provide many benefits and form a basis on which accurate decisions can be made relating to the food product in question. 55 -Free choice profile -Time intensity analysis -Flavor profile analysis -Texture profile analysis What is the acceptability of Affective/Acceptance -Hedonic rating scale (5-Point, a product? Is one product 7-Point and 9-Point Hedonic scale) preferred over another?
-Preference ranking Source: Stone 34 Cruz et al 46 Rosenthal 50 Kemp et al 54
Linkage between Objective Evaluation Methods and Subjective/Sensory Evaluation
Objective tools use equipment for routine quality control, involve physical, chemical and microbiological techniques. They are faster and more efficient in a way that one can perform the test on many samples in a day. 8 On the other hand subjective/sensory evaluation methods involve the use of human subjects by means of their sense organs, can determine sensitivity to changes in ingredients, processing and packaging hence marked crucial for food product development and evaluation of market performance of new foods or food products. 37 38 They are expensive and time consuming (for example, it can take almost a day to perform a complete sensory analysis of two samples only). Sensory evaluation can determine the overall acceptability of food or food products than can objective tools do. 8 Therefore, in order to provide reliable information regarding acceptability and routine quality control of a particular food or food product, objective evaluation methods must be linked with sensory evaluation.
Controlling Food Quality and Safety in Small-Scale Food Production Low compliance to Good Manufacturing and Good
Hygiene Practices by small scale food producers and processors constantly lead to food contamination resulting into poor food quality and safety hence exposing consumers to risks of foodborne diseases. 5 56 Small food producers, processors and distributors are faced with several challenges associated with food supply chain which involves stages like sourcing of raw materials, production, processing and packaging, storage, wholesale distribution and retail redistribution to consumers. Disruption of any of the stages would put the whole supply chain at risk. 57 Small-scale food industry is confronted with practical challenges of developing, implementing and maintaining food safety and quality programs. 11 Major problems associated with small-scale food production especially in developing countries include; lack of right technology, inadequate power for industrial operations, poor water supply for processing and portable use, lack of traceability, failure to control inventory in warehouses and stores, inability to maintain the safety and quality of food or food products, lack of qualified personnel with experience and technical expertise in food safety, lack of training, poor equipment layout, lack of in-house microbial risk evaluation and insufficient funds for maintaining quality standards. 57 58 59 Due to these factors, the hygienic position of facilities in small-scale food production and processing is generally low and pose a great public health risk to consumers. It is therefore necessary to address food quality and safety challenges faced by small-scale businesses by good planning and risk assessment, regular trainings and routine observation of quality standards. 57 56 It has to be remembered that, an efficient food supply chain not only ensure quality of food products but also maintain customers for the particular food or food products.
Global Application of Objective Methods and Sensory Evaluation in Quality Control of Food Products
Globally, physical, chemical, microbiological and sensory evaluation methods are among the analytical procedures used to provide informed features on diverse food products (eg. composition, physicochemical properties and sensory qualities). Such information enhances the ability of most food companies/ industries to economically produce foods that are constantly safe, nutritious and suitable for consumers. 36 However, current findings indicate that these methods are underutilized in a wide range of food industries/ companies particularly under small scale production. For example, in most African countries application of quality assessment and sensory evaluation methods in product development is still a big challenge. This is reflected by a wide range of unsafe foods available in the local markets across the region. Most foods used are subject to hazards and microbial contamination, including mycotoxins which in turn represent serious threats to human health within the continent. 60 61 Problems occur as a result of poor postharvest handling practices. Not only that but also inadequate or absence of facilities such as storage facilities (eg. cold stores), and transport facilities increase microbial contamination and food loss. 59 Poor delivery of quality food products is also contributed by low access to modern equipment and inappropriate packaging materials as well as low quality of raw materials by small-scale food processing enterprises. 5 Therefore authorities responsible for quality control and food safety management at the national level are given an alert that they still have a lot to do in ensuring quality and safety of foods towards safeguarding consumer's health.
Challenges Facing Food Control Authorities In Developing Countries
According to FAO/WHO, 56 some of the challenges that food control authorities still encounter are: • Increasing burden of food-related illness and emerging foodborne hazards. • Changes in technological advancement in food production, processing and marketing. • Problems to food processors as related to timely delivery of raw materials. • Lack of trained analytical staff in most of food control laboratories impedes quality standards. • Need for harmonization of food safety and quality standards globally. • Rapid urbanization and increasing consumer awareness about food safety and quality issues.
These influence changes in lifestyle.
Strategies to be Implemented at The National Level to Ensure Food Safety and Quality Control
Attainment of food control system at the national level requires consideration of current or emerging food safety and quality issues.
• Effective quality control should be focused on the following crucial areas; the factory building; machinery and equipment design and installation; manufacturing system; process and product identification; product design; raw and ingredient materials and packaging materials, and the market through intensive consumer response program which can provide the required feedback. • Appropriate and consistent training of food inspectors should be compulsory. They must be well trained to understand the industrial processes, identification of potential safety and quality problems, and have all the required skills to inspect the premises, collect food samples and carry out general assessment. The food inspectors should also be well trained and equipped with Hazard Analysis and Critical Control Points (HACCPs) and handle HACCP audit responsibilities. • Routine hygienic control of street foods should be observed, these foods are mostly prepared and retailed under unclean environments. • Control measures should mostly focus on training of food producers, shippers, processors, distributors and handlers in hygiene practices to improve safety of particular foods. • Governments should effectively support small food processing units so that they absorb better technology and finally deliver quality food products. • The Food Control Management should lay down the standards for food control laboratories and regularly monitor their performance.
Additionally, food control agencies should prioritize and promote the specific requirements for the training of their food inspectors and laboratory analysts.
Conclusion
Ob je cti ve to ol s (ph ysi cal , ch emi ca l a nd microbiological tests) and sensory /consumer evaluation are very significant in new product development as they ensure food product quality and safety. Product manufacturers and service industries/ companies require persistent and committed efforts for enhancement of products and service quality. This will maintain constant acceptability of the new food products and hence competition in the local and global markets. Consumer acceptance mainly determines the sensory quality of products, therefore sensory evaluation should always go hand in hand with instrumental analysis in laboratories and food manufacturing companies. It is also important to have specialized professionals within food industries/companies, who can help avoiding the use of wrong sensory results which in turn may threaten the success of the particular food or food products during marketing.
Acknowledgement
The Author is grateful to all who made this work comprehensive.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article. | 2020-12-14T21:04:22.169Z | 2020-10-28T00:00:00.000 | {
"year": 2020,
"sha1": "7e780c43eefa6a3e5b8c9ddcbd0a6ab5ce3ca599",
"oa_license": null,
"oa_url": "https://doi.org/10.12944/crnfsj.8.3.03",
"oa_status": "GOLD",
"pdf_src": "MergedPDFExtraction",
"pdf_hash": "ffb97b54944892713d4e1f7535f273e22ebf83f1",
"s2fieldsofstudy": [
"Agricultural and Food Sciences"
],
"extfieldsofstudy": [
"Psychology"
]
} |
251264935 | pes2o/s2orc | v3-fos-license | Using signalling theory to assess the Government of Ghana’s risk communication during the COVID-19 pandemic
Background The proliferation of information through social media and on other communication networks during the corona virus disease 2019 (COVID-19) era altered information transfer in many countries. The content of the messages from government officials, media coverage and alternative narratives, affected the level of compliance in adhering to the various health protocols amongst the public. Aim This article aimed to determine the relationship between the message used, media coverage, alternative narratives, the public’s attitude towards staying at home and their commitment to stay at home during the COVID-19 pandemic campaign period in Ghana. Setting A total of 352 respondents was sampled from the Kumasi metropolis. Methods A survey sample strategy and a convenience sampling technique were used while structural equation modelling with Partial least square (PLS) version 3.0 was used for the analysis. Results The study revealed that the nature of media coverage employed and the alternative narratives had a significant positive effect on the attitude of the respondents, whilst the content of the message had insignificant effects on the attitude of the public. Finally, the attitude of the people had a significant positive influence on their respective commitment to stay home. Conclusion Developing countries in Africa need to fight pandemics using purely subsidised health officials or directorates rather than have government-appointed health experts and officials spearheading activities during a pandemic. Contribution This study contributes to the clear understanding on some COVID-19 misinformation, the distinctive cost it poses to public health delivery in developing economies and the potential strategy of using neutral public health persons to curb the spread of the virus.
Introduction
Ghana recorded its first case on 12 March 2020 after two people tested positive on their return to Ghana from Norway and Turkey. As of 12 May 2020, at 08:31 h GMT, Ghana had recorded 4700 cases, with 494 recoveries and 22 deaths (Worldometer 2020). Given the spread, the Government of Ghana instituted measures and presented frequent updates on the pandemic as part of the campaign strategy employed to curb the spread of the virus. These measures included a 14-day mandatory quarantine imposed on persons who arrived in the country before the closure of the country's borders, periodic televised addresses by the President of the Republic of Ghana, periodic updates on the cases by the Ministry of Information, publicity by the frontline hospital staff or experts from the Ministry of Health, establishing coronavirus hotline telephone numbers, introducing the GH COVID-19 Tracker App and enforcing the restriction of movement in an attempt to control the spread of the virus. In spite of that, many of these preventive measures were flouted by people leading to the spread of the virus.
According to Abdulai et al. (2021), assessing the right health information was a test for many people in Ghana, who did not have the medical knowledge to access and comprehend the information available on many platforms. Studies have shown that factors such as political gimmicks during the COVID-19 campaigns by opposition parties, unstructured media coverage on the fight (Wen et al. 2020), inadequate message structure on COVID-19, unscrupulous business advantage during the COVID-19 period and inadequate health professionals, all deluded people not to stay home to minimise the spread. Overy (2020) observed that some people used the COVID-19 period to carry out vendettas against their opponents as one factor that affected the whole campaign structure in Ghana. It is always challenging when governments mediate in emergencies, such as the coronavirus pandemic. Stories are adjusted by many recipients to frame the organisers of the campaign in ways that fit a political storyline in most emergencies (Bennett & Perez 2020;Terry 1997). Also, excessive blaming of people in emergencies makes it tough to separate actions from discourse and that causes harm to public awareness (Luton 1994;McSwite 2005;Patterson 2001). Likewise, receivers of messages usually attach weights to the signals based on preconceived notions of the signaller (Branzei et al. 2004;Ehrhart & Ziegert 2005).
Studies by Zhang and Wiersema (2009) and Basdeo et al. (2006) investigated the effectiveness of signals sent by firms. Goranova et al. (2007) and Carter (2006) analysed the effect of signals on individuals or managers whilst Chung and Kalnins (2001) investigated on the effect of product signals, with Miller and Triana (2009) focusing on directors' 'message signal-receiver effect'. According to Gammoh, Voss and Chakraborty (2006), the use of the 'signal-receiver' relationship has been common in consumer research and limited in other areas of study. An inference could therefore be drawn on scant research on the use of signals in creating awareness about a pandemic, such as the COVID-19. Commentaries from politicians, religious leaders, celebrities, health experts and self-acclaimed traditionalists all contributed to the possibility of sending mixed messages to the public. People were tired of the blameworthiness, fabrications and political oratory and so disconnected from the reality of the coronavirus spread (Zavattaro & McCandless 2020). Ram (2020) investigated the 'Coronavirus Research Trends' using a '50-Year Bibliometric Assessment' with productive countries, productive institutes, productive authors, productive journals, characteristics of highly cited papers, author keyword analysis and hot research areas that were found from the African continent.
Besides the studies in epidemiology, clinical characteristics, treatment and clinical outcomes of confirmed cases, there are limited studies on the human infection of COVID-19 (Addo et al. 2020;Cortegiani et al. 2020;Li et al. 2020). According to Sibiri, Zankawah and Prah (2020), although developed countries received more attention because of their scientific innovativeness in their quest to contain the spread of the virus, African countries received little attention because their innovations were few although significant. Whilst several studies have been carried out on COVID-19 in Asia, Australia, North America, South America and Europe, there is little research on communication or messages used during the pandemic.
In this study, the authors quantitatively reviewed the content of messages during the height of Ghana's COVID-19 period, the media coverage on the pandemic and how alternative narratives impacted people's commitment to stay at home. We examined the association between messages used and attitudes of people towards staying at home during the pandemic period, investigated the relationship between media coverage of COVID-19 and attitudes of people, established the relationship between alternative narratives on COVID-19 and attitudes of people and analysed the relationship between attitudes of people and their commitment to stay home during the COVID-19 period.
Signalling theory
Signalling theory (ST) propounded by Spence (1973) explains how a signaller transports substantial information about his or her activities as in products or services to a recipient or receiver. The theory is appropriate for defining behaviour when two parties have access to varied types of information (Connelly et al. 2011a;Connelly, Ketchen & Slater 2011b). The signallers are insiders, officials and administrators who obtain information about a person (Spence 1973). According to Kirmani and Rao (2000), signals could be sent on a product as well as on organisations (Ross 1977) that are not accessible to outsiders. Signalling, therefore, takes place within or between organisations (Lester et al. 2006;Rynes, Bretz & Gerhart 1991). As the pandemic period became sensitive, and politicians, celebrities, opinion leaders and reference leaders expressed diverse views on government communication, environmental distortions created the propensity to reduce the observability of the signal. In such a situation, people were likely to doubt the information provided by the signaller. According to Carter (2006), the moment the media reports on press releases in an environment where ideas are diverse, potential distortions are likely to be introduced. Branzei et al. (2004) also observed that external referents could change the information unevenness. Earlier research demonstrated that signals must be noticeable to the recipient and the level at which signalling is obvious depends on how attentive the receiver is in scanning the environment for indications (Bruton et al. 2009;Connelly et al. 2011;Janney & Folta 2006).
Message, media coverage and attitude
A message is a discrete constituent of communication proposed by the source for consumption of the receiver or a group of receivers (Salloum et al. 2018). According to Vasterman, Yzermans and Dirkzwager (2005) and Young et al. (2013), a message shrouded in an unbalanced reporting of healthcare crises has a likelihood to create a split in 'actual versus perceived' risks, leading to under-and overreaction to decisions. A well-crafted message on COVID-19 will influence people's attitude to either accept or reject an idea based on instructions from health professionals. Media coverage refers to all blog articles, video contents or other types of digital content that are produced by individuals or organisations to attract attention, interest, desire and be received by their target audience (Brossard 2013). The nature of media coverage highly influences people's feelings and attitudes whenever there is an outbreak of transmittable diseases (Mairal 2011;Young et al. 2013). As a result, news associated with infectious diseases has a greater likelihood of causing alarm and influencing people's feelings (Tetlock 2007;Wen et al. 2020). In the current study, the signaller is the Government of Ghana providing information through its Ministries of Health and Information to the people of Ghana who are the receivers. Any inconsistency between the signaller and the message being sent can lead to poor signalling, which can distort the information and influence compliance. Also, people's disposition towards the carriers of the message is likely to positively or negatively influence the way message is received.
Attitude is a significant internal influencer, affecting people's decision processes (Abou-Youssef et al. 2020), and 'a learned predisposition to behave in a consistently favorable or unfavorable way with respect to a given object' (Kanuk & Schiffman 2000:253). Attitude has an influence on people's thoughts towards an object (Fazio 1986) and is a motivational factor that drives them towards an action or hinders a certain behaviour . In this study, it was observed that people's attitudes during the COVID-19 period (whether positive or negative) were likely to influence their behaviour to act favourably or otherwise towards the guidelines they were expected to observe.
Health policy campaigns and inadequate health officials
Policies and support procedures were instituted to create awareness within the COVID-19 period. Although introducing health officials to the public in the pandemic period was laudable, the use of frontline workers or health experts who were mostly government appointees in the campaign process created an aversion for people not to comply with most of the directives. As Bennett and Perez (2020) observed, it is challenging when the government mediates in emergencies, such as pandemics. Some members of parliament from the major political parties also politicised the campaign by posting their pictures on veronica buckets and tanks that were provided for hand washing. Other politicians also alleged that their constituents were not supplied with some treats or handouts because they were not members of a particular political party. The back-and-forth of ideological views about the campaign process affected the assimilation of information during the pandemic. That notwithstanding, it would have been expected that people would realise that the virus infected and killed many people irrespective of their political affiliations. Leaders have to be more pre-emptive, sensitive and recognise the effects of their short-term engagement (Gardiner 2020).
The lack of adequate health officials, such as public health directors, public health nurses and community health nurses, who could have salvaged the situation by providing strong case signals about the pandemic also contributed to the attitude of the people in compliance. Some comments culled from responses of some open-ended questions in the questionnaires indicated that: • The campaign should have been carried out by neutral health officials, other than the communication minister.
• The pandemic should have been handled by the health officials in the country.
• Involving public health officers and nurses would have had a greater effect than the government controlling everything • Ghana has become too political, so the involvement of health officials from the various community institutions would have been more appropriate than the government.
Health-related campaigns were more likely to be positive when they were drafted and carried out by the health officials instead of the government. Using health officials in the pandemic campaigns, to a large extent, reduces doubt and misinformation as the external nature of the threat of COVID-19 and its dynamics can make the ultimate impact unknown to government officials or politicians.
Hypothesis statements
The study's hypotheses were presented as follows: H1: The message on the COVID-19 has a significant positive effect on the attitude of people.
H2:
The media coverage on the COVID-19 has a significant positive effect on the attitude of people.
H3:
The alternative narrative on the COVID-19 has a significant positive effect on the attitude of people.
H4:
The attitude of the people has a significant positive effect on people's commitment to stay home within the COVID-19 period.
Methodology Sample and data collection
This study employed a survey strategy in collecting data from respondents during the COVID-19 period. The population comprised the entire people in the Kumasi metropolis; the exact sampling frame could not be determined, whilst the convenience sampling technique was employed in the month of May 2020 for data collection. A total of 364 respondents were sampled from 20 April 2020 (a day after the 3-week lockdown imposed by the Government of Ghana) to 10 May 2020 and 352 participants responded willingly. Data collection was mainly done within the Kumasi metropolis, adhering strictly to the physical or social distance order that was in place. A combination of self-administered questionnaire and interviewer-administered questionnaire was used in the entire data collection process. Convenience sampling technique was used in eliciting the views of the respondents because there was no specific sampling frame. Data collection was carried out until the required samplesubject to related studies -was gathered for the analysis.
The measurement items such as the 'message', 'media coverage', 'alternative narratives', 'attitude' and 'commitment to stay home' on the questionnaire were organised on a fivepoint Likert scale, with 1 representing 'strongly agree' and 5 'strongly disagree'. The questionnaire had two sections: information on participants in Section A, and all the variables (messages, media coverage, alternative narratives, attitude and commitment to stay home) in Section B (see Appendix 1). The items for the variables were developed from academic literature based on previous and current pandemic works.
In minimising the likelihood of probable bias resulting from a response, 'N/A' was included in each question as an alternative to guarantee the soundness of the questionnaire because it was a newly developed questionnaire. Two experts from the Kwame Nkrumah University of Science and Technology (KNUST) School of Business in Ghana and the Department of Marketing at the University of the Witwatersrand, South Africa, evaluated the measurement items by looking at the structure and phraseology of the instrument. Their recommendations were then used to update the questionnaire and tested on 10 people from the Kumasi metropolisspecifically, Oforikrom and Anwomaso communities. The comments from the pilot testing enhanced the adequacy of the content and wording of the instrument. Thus, the measurement item was seen to be dependable. As English is the indorsed language in Ghana, many of the participants could express themselves in the language, therefore making the data collection easier. However, those who could not write also participated as the research assistants were trained to translate it into the 'Akan' language -which many understood and which was also used in the collection process (Brislin 1986).
Analysis
The analysis was carried out in two stages. Firstly, exploratory factor analysis (EFA) was used to validate the number of measurement items. Exploratory factor analysis was conducted using IBM SPSS version 24.0. The hypothesised relationship of the model was carried using structural equation modelling (SEM) via partial least square (PLS) version 3.0.
Test on unidimensionality
A dimensionality assessment was carried out in testing the study's sampling adequacy using the EFA. The Kaiser-Meyer-Olkin (KMO) of sampling adequacy was greater than 0.700 and Bartlett's test of sphericity was significant (p < 0.05), indicating the variables were suitable for factor analysis (Meyer & Collier 2001;Pallant 2010:187). A principal component analysis with varimax rotation was used to extract appropriate factors for the study's analysis. Significant factors were determined using the criterion of an Eigenvalue greater than 1 for factor loadings (Pallant 2010:192). Eigenvalues for the study's constructs were message content on the COVID-19 (2.782), media coverage on the COVID-19 (2.601), alternative narratives on the COVID-19 (2.348), attitude (2.671) and commitment to stay home (2.822). The cumulative variance for the five factors was 71.0% (see Table 1). Table 1 explains the number of items that were retained and rejected. The respective numerators and the denominators indicate the total and the numbers that were used for the study. The study used five items for each of the measurement variables, out of which some were not used because they loaded less than 50%. The retained values are shown in the factor loading section, with the 'message content on COVID-19' scoring 4 of the 5 questions (4/5), 'media coverage on COVID-19', 3 of the 5 questions (3/5), 'alternative narratives on COVID-19', 4 of the 5 questions (4/5), 'attitude towards COVID-19', 3 of the 5 questions (3/5) whilst 'commitment to stay home', 3 of 5 questions (3/5).
Common bias assessment
Data were collected through a survey using questionnaires. Succeeding the recommendations of Podsakoff et al. (2003), https://www.hsag.co.za Open Access several steps were taken to lessen the prospective risk of common method bias from the use of a single respondent. Firstly, the authors and the research assistants ensured informants were normatively committed by reassuring them of confidentiality and anonymity. Secondly, the measurement items were judiciously constructed to avoid any possible uncertainties.
Partial least squares-structural equation modelling
Factors derived from the EFA were subjected to structural equation modelling using Partial least squares (PLS)'s version 3.0 of PLS Graph (Chin 2001
Reliability and validity
Reliability was measured using internal consistency and composite reliability. Internal consistency was measured by Cronbach's alpha values using IBM SPSS and composite reliability with the partial least squares-structural equation modelling (PLS-SEM). Reliability is demonstrated through high Cronbach's alpha values ranging from 0.701 to 0.846 and composite reliability values ranging from 0.813 to 0.906 (Table 2). According to Nunnally (1978), a value of 0.7 as a standard for 'modest' reliability is appropriate in assessing composite reliability usefulness. All tested indicator loadings exceeded the recommended yardstick of 0.7 indicator reliability for all first-order dimensions (Chin 2010). Average variance extracted (AVE) values were used to show convergent validity. As indicated by Zott and Amit (2008), Croteau and Bergeron (2001) and Fornell and Larcker (1981), an AVE value ought to exceed 0.5 to ensure the convergent validity of the items. Also, the application of the exploratory principal component analysis is employed as part of the process of validation (Wilson 2014), demonstrating factor validity.
Discriminant validity
Discriminant validity measures the scope at which concealed factors are distinctive and should not relate so greatly that they seem to measure the same fundamental dimension (Siekpe 2005). Discriminant validity is recognised for assessment if the AVE values of a variable are larger than the squared correlation coefficients between variables (Barclay, Higgins & Thompson 1995;Fornell & Larcker 1981). In the current study, the squared correlations of the corresponding pairs of constructs were lower than the separate AVEs, indicating that the measurement model had good discriminant validity (Hair et al. 2014;Preacher & Hayes 2004). The results in Table 3 show a strong discriminant validity.
Structural model of the relationships
Results for the structural model used to examine the proposed hypotheses are shown in Table 4 and Figure 1. The study examined the impact of the independent variables: message, media coverage and alternative narratives on the mediating variable (attitudes of people), and the relationship between the mediating variable and the outcome variable, commitment to stay home. It was observed that alternative narratives on Note: AN, alternative narratives; AT, attitudes; C, commitment to stay home; MIC, message; MV, media coverage; p < 0.05. ***, p-value is less than 0.001; **, p-value is less than 0.01; *, p-value is less than 0.05.
https://www.hsag.co.za Open Access COVID-19 and media coverage of COVID-19 were significant on the attitude of people. The message on COVID-19 did not significantly influence the attitude of people. The attitude of people had a significant relationship with the commitment level of people to stay home. This implied that H2-H4 were accepted whilst H1 was rejected.
Goodness-of-fit assessment
The goodness-of-fit (GOF) statistics was assessed using a procedure by Tenenhaus et al. (2005), where the averages of the AVE were first multiplied by the averages of the R² value, after which the multiplied value was squared to determine the model fit. The calculated global GOF was 1.10. It can be concluded that the model had a superior fit because this statistic surpassed the threshold of GOF > 0.36 as recommended by Wetzels, Odekerken-Schröder and Van Oppen (2009). Table 5 presents the profile of the study sample. It was observed that 53% of the participants were women and 34% of the participants were between 26 and 35 years old, followed by 35 years and above, which recorded 26%. A majority of the respondents (63%) were married, whilst respondents with 'first degree' as their highest qualification were the most, representing 32%.
Discussion
This study was conducted to determine the relationship between the message, media coverage and alternative narratives on the attitudes of people in the Kumasi Metropolis during the COVID-19 pandemic. It examined the relationship between the attitudes of people and their commitment to stay home during the pandemic period. The analysis showed that alternative narratives were strongly related to attitudes, and media coverage had a weaker coefficient. These findings are consistent with Raghubir (2008), who observed from a high health risk research stream that information rate has a greater likelihood to influence the perception of health risk. It shows that adopting diverse strategies could have had a greater influence on the attitude of people to stay home during the COVID-19 period, as revealed by the respondents through An alternative narrative of a crisis event describes an account of procedures that discards the principal narrative (Nied et al. 2017). The alternative narratives, such as involving public health-related personnel in spearheading the activities during the COVID-19 campaigns, would have affected the attitude of people regarding the information they would have received. Nyarko, Serwornoo and Azanu (2021) revealed how politicians in Ghana used the COVID-19 in diverting attention to their political activities instead of focusing on the fight against the pandemic. The findings also show that media coverage significantly influenced the attitudes of people, but its impact was not very strong, whilst the message on COVID-19 was very weak in terms of its influence on the attitude of people to stay home during the pandemic period. According to Stremersch (2008), people or consumers face a high level of doubt and imperfect information in a context in which wrong decisions have an important impact on their well-being. Farooq, Laato and Islam (2020) observed that the persistent provision of information by different people contributed to the information overload and overconcern amongst people during the pandemic. Using the government communicators and government-appointed health officials in spearheading the campaign weakened the message that was communicated to people. It was observed that people regarded most of the campaign activities as an opportunity for the government to make a good impression on the international community. Political opponents also created countermeasures in refuting the claims of the government in the fight against the virus. This, amongst other factors, affected the content of the actual message. Sarkhel et al. (2020) posited that the increase in the consumption of news led to an increase in people avoiding the news during the COVID period. Relying on the outcome of the study, the findings have shown that non-partisan experts or health professionals from various health sectors in the country ought to have been used during the entire campaign process and the various associations under Ghana Health Service would have been more appropriate in sending the right messages to the people. The overwhelming effect of the COVID-19 was attributed to the information overload and 'infodemic' (Rathore & Farooq 2020).
Finally, it was also observed that the attitude of people influenced their commitment to stay at home. According to Mohammed et al. (2021), valuable information could become a help if it is handled well. It was evident from the study that people were willing to stay home as long as the campaign was free from indirect political inclinations, structured messages and other televised information on affected persons. The findings are consistent with Zavattaro and McCandless (2020) who observed that others respond well to actual solutions.
Implication for theory and practice
This study contributes to the literature on the relevant dimensions of health communication campaigns. Most governments, groups or individuals employed different approaches in minimising the spread of the coronavirus.
Signalling theory provides a way by which signals from independent experts could help convey messages to people. Whilst the traditional operationalisation of the campaign was planned mainly by governments in Africa, this study has found it to be an insignificant component in behaviour modification towards compliance. Involving professional health officials in leading health campaigns has a greater likelihood of changing people's behaviour. While most previous studies on the coronavirus were conducted in the Western world, limited studies have addressed issues within the sub-Saharan African context, and this study with humanities and business research perspectives fills the gap. This research provides empirical evidence from Ghana, a developing economy. The study has some managerial implications for building a resilient health communication strategy that could improve the compliance level of targeted respondents. The need for health authorities to be given the full responsibility by governments to have campaign blueprints and train personnel to convey definite information cannot be overemphasised. Finally, alternative narratives were also found to be the greatest predictor of the attitude of the people. Strong management and linkage with popular celebrities could be used by the health officials to create strong remembrances and images for people to adhere to directives associated with campaigns. Policy makers ought to pursue these aims as part of their campaign-building compliance activities.
Conclusion
In this article, the authors investigated the effect of the content of the message, media coverage and alternative narrative interventions on the attitude and commitment levels of people to stay at home during the COVID-19 pandemic in Ghana. The results showed that the content of the message did not correlate with the attitude of Ghanaians, whilst the alternative narratives and media coverage had a strong and positive correlation with attitude. Also, the attitude of people significantly influenced their commitment to stay home. It was evident that the message by the government during the COVID-19 period insignificantly influenced the attitude of people. The involvement of public health officials was a good signal and would have greatly influenced people's attitude and commitment to stay at home during the pandemic period. The authors, therefore, conclude that developing countries in Africa ought to fight healthrelated pandemics using a non-partisan health directorate and officials rather than government-appointed health experts or government officials in spearheading the activities in such situations.
Limitations of the study and directions for future study
This study had some limitations. First and foremost, the study employed a non-probability sampling technique. Hence, the findings of the study cannot be generalised to represent the population of the study area. Secondly, the study focused on participants within the Kumasi Metropolis.
Future studies should include respondents from other regions of Ghana. Finally, there is the need for an exploratory study to be conducted on two different countries in the subregion -regarding the use of government officials and neutral health professionals in the quest to control probable pandemics like the COVID-19. | 2022-08-03T15:10:19.052Z | 2022-07-27T00:00:00.000 | {
"year": 2022,
"sha1": "b6956f815945b9abfc5ff70c04eff54911b2bc33",
"oa_license": "CCBY",
"oa_url": "https://hsag.co.za/index.php/hsag/article/download/1623/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "9752f7b05697bed88a5a6474f9755de149da7a87",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": []
} |
229399854 | pes2o/s2orc | v3-fos-license | Medical Simulation-Based Learning Outcomes in Pre-Clinical Medical Education
Introduction Medical simulation is widely used in the United States medical curriculum. However, learning outcomes based on simulation have yet to be reported. In this study, we aim to characterize the objective performance of first- and second-year medical students following eight weeks of medical simulation-based learning. Methods First- (n=25) and second-year (n=15) medical students were recruited for this study. We designed and administered a novel pre-experience examination to collect participant demography and assess simulation and non-simulation knowledge. Following 14 high-fidelity simulation scenarios over the course of eight weeks, we administered an identical post-experience examination and compared performance, primarily using a within-subjects analytic design. Results Student performance improved by an average of 18% following the medical simulation experience, and first-year students demonstrated greater benefit (22%) as compared to second-years (12%). Relative to first-years, second-year students showed higher overall performance on both pre- and post-examination. Demographic factors and prior medical experience were not significantly associated with assessment performance and score improvement. Conclusions Our data supported the efficacy of simulation-based learning as evidenced by the significant improvement in objective performance on a standardized examination. That is, both first- and second-year medical students demonstrated test-score improvement following an eight-week medical simulation program. Of note, the first-year students exhibited greater benefit (at the group level). Importantly, these findings were statistically unrelated to participant demographic and background variables. Collectively, this study provides preliminary evidence that medical simulation in the pre-clinical phase of undergraduate medical education is an effective tool for student learning.
Introduction
Medical simulation-based learning is a multidisciplinary approach to medical learning by which subject matter experts produce realistic medical scenarios to facilitate learning in a controlled setting [1]. Contemporary simulation encourages active learning, reinforces didactic material, and presents clinically relevant topics as an authentic assessment of medical knowledge [2][3][4]. Simulation learning also provides an engaging environment in which students can practice high-stress medical decision making without jeopardizing patient safety [5][6]. As such, simulation training provides a pragmatic and affordable way to implement standardized patient-care experiences earlier in a student's medical education [7].
Traditionally, pre-clinical US medical education has been carefully tailored to prepare students for the first round of medical licensing examinations, which are highly correlated to residency placement and largely focus on the fundamentals of biomedical sciences [8][9]. In an unprecedented advance, the United States Medical Licensing Examination (USMLE) Step 1 will transition from a nominal score value to a dichotomous "pass-fail" scoring system beginning in the year 2022. In response to this change, residency program directors report that an increased weight will be placed on the score of USMLE Step 2 exams, which emphasize clinical management [10]. Commensurate with this change, we expect to see an earlier introduction of clinical experiences to the medical school curriculum. However, this is problematic given the competition between interprofessional students requiring preceptorships and advances in medical management have led to decreasing lengths of stay in the hospital. As such, there are fewer opportunities for gaining clinical experience than ever before [11]. Therefore, the exploration of simulation modeling into pre-clinical education may provide a viable solution to introducing clinical experience amidst a shortage of qualified clinical preceptorships.
There is an extensive body of literature discussing the effectiveness of low-and high-fidelity simulation in clinical practice to enhance the ecological validity of medical instruction [12][13][14][15]. Despite this, there is a paucity of studies directly examining the utility of simulation learning in the pre-clinical phase of medical education. More specifically, the current literature lacks a detailed comparison of simulation learning outcomes between first-and second-year medical students. To address this gap in the literature, we investigated the effects of an eight-week medical simulation experience between first-and second-year medical students. We also chose to evaluate the impact of a student's background (ie., demographics, year in medical school, and past medical experiences) on simulation learning. We predicted that first-and secondyear medical students would benefit, albeit differently, from participation in medical simulation, as assessed by score improvement on a standardized examination. We also hypothesized a relation between past student experiences (medical, educational, and simulation) and examination performance such that these modifying factors would produce greater improvement in post-test scores compared to students without past medical experiences or fewer years in medical school.
Participants
A total of 40 medical students at a college of osteopathic medicine located in the southeastern United States, voluntarily participated in an eight-week high-fidelity medical simulation program between September 2019 and November 2019. Participants were limited to first-and second-year medical students. This study was approved by the college's institutional review board (IRB# HS190820EX). Written consent was obtained separately from each participant prior to the administration of pre-and post-simulation examinations.
Assessment
Within one week from the beginning of the simulation learning experience, participants were administered a 31-question multiple-choice pre-test designed to assess the following: 1) six questions assessing demographic variables; 2) eight medically related questions that were not reviewed during the simulation experience (note: these questions served a 'control' function where they were purposely unrelated to the simulation experience to account for/measure artifact score change); 3) 16 medically-related questions whose content was taught during the simulation experience, which served as a measure of simulation performance. Test questions were created by study authors based on the specific cases used during the eightweek simulations. Participants were not given feedback or answers to test questions. Scores were deidentified using a two-digit number, which was only known by the participant. This number was used to match pre-and post-test scores for comparison.
Participants were not informed of the existence of the post-test until just before the post-test was administered. A post-test, composed of identical questions as the pre-test, was administered less than one hour after completing the final training case. One medically related, non-simulation question was inadvertently released to participants during a debriefing session (prior to the post-test); consequently, we excluded this item from all analyses. See Appendix A for copies of the pre-and post-simulation examinations.
Simulation Experience
Participants were self-arranged into eight groups of five participants, each of who met weekly for the simulation experience. Each participant was assigned a specific role for each simulation patient encounter (e.g., team captain, scribe, history taker, physical examiner, and case coordinator). Students were allowed to change roles within groups during subsequent encounters.
In each weekly simulation experience, two groups met at the host institution's simulation complex. Participants were not provided any information before each encounter. One group actively participated in an acute care scenario featuring a high-fidelity human patient simulator programmed using simulation software. The remaining group observed the unfolding scenario via live remote audio and video streaming to the debriefing room. The observation group was provided with the same laboratory values and imaging by a second clinical simulation instructor to keep the observation group engaged during the patient encounter.
During the case, participants were presented with the patient's chief concern, medications, past medical history, family history, and social history to varying degrees of detail. Participants were not given explicit instructions on how to work as a team to obtain relevant history and physical findings. Instead, they were given full autonomy. Participants could obtain the history from the patient or caregiver, perform any physical examinations, order laboratory tests and imaging, perform procedures, administer medication with real-time feedback, and order results pushed to an in-room monitor. Scenarios were programmed to progress through a variety of stages based on ideal management strategies. Each scenario lasted for approximately 15 minutes and was concluded at the discretion of the instructor. Immediately following the scenario, clinical simulation instructors conducted a debriefing session with both participant groups. The debriefing session consisted of an interactive 20-minute discussion summarizing key aspects of the case such as patient presentation, laboratory and imaging findings, and ideal patient management strategies. At the conclusion of the debriefing session, the two participant groups switched roles and experienced a second unrelated encounter. Each team participated in a supplementary voluntary session, which reviewed the capabilities of the high-fidelity simulator, including heart sounds, lung sounds, dysrhythmias, and supplies available in the crash cart; this was an adjunctive experience in which no team practiced an additional vignette.
In total, each group experienced seven encounters via direct participation and seven encounters via observation for a total of 14 experienced simulation cases. A summary of the 14 clinical vignettes is available in Appendix B.
Measures
The demographic variables -sex, class year (e.g., 1st year, 2nd year), the highest educational level achieved, prior medical experience, and prior competitive simulation experience -were collected as part of the preand post-test examination.
Performance variables: Pre-and post-test scores were used as performance metrics. Data were compiled and entered into Qualtrics computer software, which collated and parsed the data into desired outputs. An independent manual tabulation of data was also conducted to confirm the data was collected and recorded correctly.
SIM score: Sixteen medically related multiple-choice questions associated with simulation cases served as a measure of cohort simulation performance. SIM questions are generally related to the following content areas: basic knowledge, cardiology, case management, neurology pathology, pharmacology, respiratory, and trauma.
Non-SIM score: Eight medically related multiple-choice questions not related to cases covered during the simulation served as a control.
Change score: Differences in the post-and pre-test SIM and non-SIM scores were tabulated to assess a participant's relative change in performance. Separate change variables for the SIM and non-SIM scales were computed as follows: Change (Δ) Score = Post-Test Score -Pre-Test Score. Accordingly, positive values on change variables reflect score increase while negative values indicate an interval score decrease.
Statistical analysis
Prior to conducting main analyses, demographic and performance variables were examined for group differences (i.e., between first-year and second-year students) using chi-squared (χ2) tests for categorical variables and independent t-tests for continuous variables. In addition, performance variable distributions were examined for normality (i.e., skewness and kurtosis). Lastly, bivariate correlations were performed to determine whether demographic variables were associated with any of the performance-based outcome variables for the total sample.
For main analyses, paired-samples t-tests were performed on SIM and non-SIM-related questions to evaluate pre-post score differences following the eight-week simulation experience. The analysis was conducted across the total sample as well as within the first-and second-year students (separately) to examine the consistency of the effects. Finally, the degree of test-score change was alternatively assessed using a computed variable of score change for each performance-based variable. Doing so allowed for direct comparison of first-and second-year medical students' score change for SIM and non-SIM-related evaluative questions.
The main measure of effect size for t-tests was Cohen's d, with values of 0.2, 0.5, and 0.8 corresponding to small, medium, and large effect sizes, respectively [16]. Additional effect size indices interpreted were Cramér's V (φc) for χ2 tests and Pearson's r correlation coefficient for bivariate correlations; values were interpreted similarly, where 0.1, 0.3, and 0.5 reflected weak, moderate, and strong associations, respectively [16]. The level of statistical significance (α) was set to 0.05 (two-tailed) for all analyses. Participants with missing or improperly completed examinations were excluded from the analyses. The Statistical Package for the Social Sciences (SPSS) program, version 26 (IBM Corp., Armonk, NY) was used for all statistical analyses. Table 1 displays demographic and performance variable descriptive statistics for the final total sample as well as stratified by medical student year. From the original sample of 40 participants, three (7.5%) were excluded from data analyses due to missing post-test scores (n = 2, 5%) and having to repeat first-year coursework (n=1, 2.5%). The majority of final samples had attained a bachelor's degree as the highest level of education (n=28, 75.7%), were male (n=25, 67.6%), and were first-year medical students (n=22, 59.5%) who possessed some level of prior medical experience (n=28, 75.7%). Overall, prior medical experience consisted mostly of being a prior medical scribe (n=16; 43%), followed by nursing assistant (n=8; 22%), emergency medical technician (n=4; 11%), nurse or nurse practitioner (n=2; 5%), and pharmacist (n=1; 3%). Demographic comparisons between first-year (n=22) and second-year (n=15) students revealed no
Performance variables
Visual inspection of performance variable distributions as well as skewness and kurtosis statistical tests were not concerning for non-normality. Accordingly, parametric statistical analyses were deemed appropriate. The total evaluation scale (i.e., all 24 items summed) demonstrated acceptable reliability, as measured by internal consistency, for both the pre-and post-administrations (Cronbach's α= .72 and .76, respectively). Bivariate correlations were carried out to examine relations between demographic/background variables and performance outcome variables in the total sample. Education, gender, and prior medical experience were not significantly correlated with any of the SIM and non-SIM outcome variables. Next, independent t-tests were conducted to examine group differences in each performance variable. Lastly, we investigated whether the degree of performance change differed as a function of the medical student group by way of computing a distinct score change variable for the SIM and non-SIM-related questions. Independent samples t-tests were then performed to compare SIM and non-SIM-related performance changes between first-year and second-year medical students. The analysis revealed a significant difference in score change for SIM-related questions [t(35)=2.42, p=.02, d=0.82] such that the first-year medical students demonstrated greater interval change in scores (M=3.6, SD=2.3) as compared to second-year students (M=1.9, SD=1.9). Alternatively, there was no significant group difference in non-SIMrelated question score change [t(35)=-1.85, p=.07, d=0.64] though the second-year medical students displayed a mild trend of greater score improvement (M=0.5, SD=1.1) than the first-years (M=-0.4, SD=1.5).
Discussion
Simulation learning in undergraduate medical education remains understudied, despite its potential role in providing low-risk clinical experiences to medical students. In this study, we used objective pre-and postexamination performance scores to compare first-and second-year medical students following an eightweek medical simulation experience. We found that both first-and second-year students had significant improvements in performance. While student scores improved overall, first-year students were found to have greater performance improvements compared to second-year students. Conversely, there were no significant differences attributed to a participant's sex, education level, or past medical experience on performance. The lack of correlation found between the previously mentioned variables and percent change remained unmodified even after factoring in the student's year in school. Medical students, irrespective of past experiences, demonstrated acquisition of knowledge over the interim period, likely reflecting a significant benefit from simulation learning experiences [17].
On average, medical students showed an improvement in performance of 18%. Performance improvement was twice as high in first-year students (22%) compared to second-year students (12%). This suggests highfidelity medical simulation may be an effective learning tool in pre-clinical medical education, especially for first-year medical students. That is, the degree of benefit from the simulation experience may be moderated by medical student year, as first-year students benefited more than their second-year counterparts. Generally, students have a greater foundation of knowledge by the second year of medical school. For instance, at the host institution, first-year students had only experienced three months of basic science courses with minimal clinical instruction at the time of the study. Conversely, second-year students had completed system-based clinical courses in multiple body systems. As expected, second-year test performance was higher in both pre-and post-examinations. In a similar vein, a student's year in medical school was positively associated with higher pre-test performance (r=0.56, p<0.01) such that second-year student status was strongly associated with higher pre-test scores. Meanwhile, a similar correlation was not observed with post-test performance. While first-year students scored 19% lower than second-year students on the pre-test, this deficit was attenuated to 8% on the post-test. Post-test scores for non-SIM items were not associated with the same degree of improvement but rather, there was a slight decline in scores following the eight-week training. Together, this improvement in first-year students may suggest medical simulation may help students assimilate a broader foundation of knowledge, especially in those who have less experience prior to medical school.
In this study, 76% of participants had prior medical experience while 24% had a master's degree or higher. Interestingly, neither of these attributes were associated with better post-test performance. This may be the result of context-dependent factors within the simulation scenarios themselves. For example, acute care settings, such as those simulated in these scenarios, require proficiency in situational awareness, medical decision-making, and overall clinical management to achieve desirable outcomes [18]. It is possible that medical experiences that are not directly related to acute care may not play a functional role in improving simulation-related performance. Similarly, even though a participant's level of education may correlate to higher didactic performance, higher education levels may provide little value without being able to apply acquired knowledge within the confounds of a rapidly evolving medical scenario [19][20]. Since participant characteristics appear to not influence performance outcomes, students of varying backgrounds may find value in simulation learning. This is particularly important in the setting of undergraduate medical education, considering that the diversity of matriculating medical students in the United States has steadily risen over the past 40 years [21].
Several limitations are present in this study. First, this study only examined performance among first-and second-year medical students in the pre-clinical phase of medical education. Therefore, conclusions are not generalizable to third-and fourth-year students. Next, students were given identical pre-and postexaminations eight weeks apart, which introduces possible bias from practice effects. However, threats from this potential bias were remedied by the fact that (a) we analyzed interval change (lack thereof) in nonsimulation based content (which all participants received at pre-and post-test periods), and (b) participants were not informed of the presence of a post-test to avoid recall bias in the form of purposeful memorization of questions and independently sought-out answers. Nevertheless, it remains a possibility that practice effects could have been differentially present in the question sets (e.g., stronger in simulation-related questions). Pertinent to our study, within-subject designs are also susceptible to subject history (i.e., participants having different life experiences across the study period) and maturation effects (i.e., participants differentially maturing over the interval period, which may influence performance. Ordereffects and nonsymmetrical carry-over effects were not a concern by virtue of our study's simulation intervention and identical tasks. Lastly, while an independent control group was not used as a part of the study design, an inherent strength of the within-subjects design is that individual differences are wellcontrolled (as participants act as their own controls).
Conclusions
The results of this study demonstrate that preclinical medical students show significant improvement in objective performance measures following eight weeks of medical simulation. First-year students benefited most from the simulation experience, as evidenced by the highest change in post-test scores. Additionally, a student's sex, the highest level of education, and prior medical experience had no bearing on performance outcomes. We predict that the changing of the USMLE Step 1 from a scored exam to pass-fail will shift medical education towards emphasizing early clinical contact. In this fluid environment, medical simulation serves to reinforce biomedical concepts and acts both as a teaching strategy as well as a tool to assess clinical competency. Together, this information supports the use of simulation learning in pre-clinical undergraduate medical education.
Appendices
Appendix A Figure 1, Figure 2, and Figure 3 comprise the written assessment that was used as the pre-test and post-test.
Additional Information Disclosures
Human subjects: Consent was obtained by all participants in this study. Alabama College of Osteopathic Medicine Institutional Review Board issued approval HS190820EX. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work. | 2020-12-10T09:05:18.914Z | 2020-12-01T00:00:00.000 | {
"year": 2020,
"sha1": "cb648a9ad30e444fded17e130e8e83d43f8d30cc",
"oa_license": "CCBY",
"oa_url": "https://www.cureus.com/articles/39723-medical-simulation-based-learning-outcomes-in-pre-clinical-medical-education.pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "e996e3f667f259868da68ff69cb2b932f4f4e78d",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
237912404 | pes2o/s2orc | v3-fos-license | Fully FPGA-based implementation of a modified control strategy for power electronic transformer in railway traction applications
Due to its several advantages (especially low volume and weight), power electronic transformer has gained notable attraction in railway applications. The main focus of this work is devoted to development a fully filed programmable gate array (FPGA)-based control platform for power electronic transformer in the aforementioned application. Due to the parallel processing of FPGAs that results in a faster control algorithm execution, reliable operation (which is of major importance in traction applications) is guaranteed. For this purpose, an input series output parallel power electronic transformer structure is built, and various considerations for reliable and stable operation of the power electronic transformer in traction applications, such as safe start-up and bidirectional power flow, along with the required control and pulse generation schemes, have been designed and implemented in a Xilinx FPGA control platform. Moreover, a modified control algorithm is proposed for controlling the power electronic transformer in a simple and more reliable manner. This control scheme is developed based on the output voltage of the DC–DC–LLC resonant converter, and can effectively control the sum of rectifier DC-link voltages and track the input sine reference current, with a lower number of required sensors. Finally, the effectiveness of the proposal has been tested from various aspects by experimental tests.
INTRODUCTION
The electric railway system has been highly developed today, due to its several advantages such as being convenient, economical and making possible the massive transportation [1]. The main research efforts in the field of modern electric railways concentrate on improving the speed, efficiency and reducing the system volume and weight. A notable amount of the overall weight of these systems is the line-frequency bulky transformers [2]. In order to solve this issue, the power electronic transformers (PETs) have been proposed, which significantly reduce the weight and volume by employing a high-frequency transformer (HFT) in their configuration [3]. Besides, they can provide several other advantages such as more controllability and four-quadrant operation, compared to the conventional transformer. These advantages could be efficiently achieved through employing a control scheme implemented in a digital control platform [4].
By reviewing the literature, it could be seen that most research efforts on power electronic transformer (PET) have been devoted to different control objectives such as power control [5], voltage balancing [6] and also topology improvement [7][8][9][10]. In order to realise efficient performance of the PET, appropriate voltage-balancing strategies and power-sharing schemes between different cells of the PET should be employed. Moreover, several configurations are suggested for different applications. For instance, PET implementations based on conventional [7] and SiC semi-conductors [8] have been studied. Also, in order to fulfil the requirements of the medium-voltage highpower applications, modular multilevel converters [9] and CHB topology [10] are proposed and developed.
The most popular configuration for the PET in traction application is composed of a single-phase cascaded H-Bridge (CHB) as the active front end (AFE) rectifier, and several parallel DC-DC-LLC resonant converters. This structure is called input series output parallel PET [11]. To balance the output voltage of CHB-AFE and also realise output power sharing of parallel DC-DC converters, She et al. [12] suggest a straightforward solution based on the PET model in dq reference frame. Also, this work proposes a powerbalancing strategy for PET in traction applications, which eliminates the sensors used for measuring the inductor current. In the later strategy, the CHB converter is controlled based on the power component of duty-cycle controller, while the DC-DC stages are controlled based on a feed-forward strategy. However, it should be noted that the efficiency of the DC-DC parallel stages could be low, due to the single-phase shift (SPS) control. Some modified SPS schemes are suggested in [13][14][15] in order to improve the PET efficiency. However, these approaches suffer from relative complexity for hardware implementation.
On the other hand, the real-time hardware implementation of the PET control algorithm could face many challenges in practical applications such as traction railway systems. If a control platform with a single digital signal processor (DSP) chip is employed, some major problems such as low number of PWM units for high-voltage systems, and limited sampling rate value would exist. In [16], a single-phase PET has been realised through employing a digital control platform, which is composed of DSP and PIC controllers. Moreover, the combination of FPGA and DSP is studied in [17][18][19] in grid-connected applications. In these works, since there are two different controller ICs, there may exist a communication delay. Moreover, development and implementation of the control algorithm in these control platforms needs significant prior knowledge.
This paper employs a control platform based on a single FPGA chip in order to implement the control and pulse generation algorithms of PET for railway traction applications. This would solve the issues regarding the multiple-chip control platforms mentioned above. Moreover, the designed control algorithms are more reliable and could be implemented with a very high sampling rate, by taking advantage of the parallelprocessing capability of the FPGAs. Also, a novel control algorithm based on the output voltage value is proposed with a lower number of voltage sensors, which results in controlling the PET in a simple, more reliable and cost-effective manner for traction applications.
PET CONFIGURATION FOR RAILWAY TRACTION APPLICATIONS
The PET configuration for traction applications generally consists of N two-stage cells. A two-cell topology is depicted in Figure 1. Both PET stages, that is, AC-DC and DC-DC converters, provide bidirectional power flow capability. The AC-DC stage of each cell is an AFE H-bridge rectifier and is connected in series with the AC-DC stages of the other cells. This topology is called CHB topology. The inductor used at the input of CHB is responsible for boost operation of the AFE rectifier, improving the quality of the input current and protection against short-circuit current. The DC-DC stage consists of a DC-DC-LLC resonant converter, which is made of two fullbridge converters, an HFT, a high-frequency inductor and a resonance capacitor. The output of DC-DC stages in different cells is connected in parallel. In order to efficiently design the corresponding controllers, it is necessary to develop appropriate models for AC-DC and DC-DC stages. These models are described in the next subsections.
AFE H-bridge operating principles and modelling
The H-bridge topology of a single cell is shown in Figure 2. According to this figure and considering unipolar modulation, In order to obtain the small-signal equations of the system, the average model and also the steady-state equations of the converter should be achieved. The average state matrixes are provided in the Appendix.
Steady-state equations
In order to determine the operating point, the converter should be modelled in the steady-state condition, by omitting the derivative terms. The steady-state model equations could be found in the Appendix.
Small-signal equations
Before determining the small equation, the A, B and C matrixes could be defined as follows: where D is the duty cycle. As the coefficients of the state matrixes are constant for all the operating points; therefore, the small-signal model will be same for different operating points. The deviations of different variables at the operating point are as follows: In the above equations, ⃗ x, x andx, represent the operating point, large signal and deviation variables, respectively. By substituting the above equations in (4)- (6), and omitting the multiplication of two small deviation terms, the final small-signal state equations of the AFE rectifier are as follows: According to the calculated state equations, the transfer functions of the AFE H-bridge are as follows: Finally, it should be noted that all the equations of different models developed above are for a single H-bridge topology. For modelling the overall active front end cascaded H-Bridge (AFE-CHB), the sum of the all capacitors should be considered in these equations.
DC-DC-LLC resonant converter operating principles and modelling
An extremely important part of the PET configuration is the DC-DC-LLC resonant converter, which is depicted in Figure 3. As shown in this figure, two full-bridge structures are employed at each side of the HFT. The HFT is the most important component, which is responsible for matching the high-and lowvoltage DC links with a galvanic isolation, and realising softswitching operation. Moreover, an LLC circuit is employed in the topology of the DC-DC stage. In order to completely form the LLC circuit, an external capacitor should be used besides the HFT. To design and characterise the LLC resonant circuit, the two resonant frequencies that are resulted by the leakage (L r ) and magnetising (L m ) inductances, and the resonant capacitor (C r ) are considered: where the ratio of (L r + L m /L r ) is represented by L n . Figure 4 depicts the LLC resonant circuit gain versus the normalised switching frequency diagram, for different values of Q (the LLC impedance ratio to the equivalent load impedance) and L n = 17. According to this figure, this gain will be independent of the load value in the vicinity of the operating points with a switching frequency equal to the resonant frequency defined in (12). This value is nearly 1. Therefore, a 50% duty cycle at a fixed switching frequency is considered, where the input-tooutput voltage ratio of the DC-DC stage will be determined by the transformer turns ratio. In this condition, one side operates as diode rectifier and the other side experiences switching, based on the power flow direction. The best operating point for the LLC circuit will be in the frequency region of f p < f s < f o . In this region, the primary side switches will operate at zero voltage switching (ZVS) condition (during the forward operation condition), and will turn off of at the maximum value of the magnetising current. Moreover, the secondary side rectifier will operate at zero current switching (ZCS). These will result in the decrease in the losses.
Since this stage operates in the steady-state condition, the transient components could be approximately omitted. Moreover, as the LLC circuit variables have nearly sine waveforms, they could be approximated using their first-order harmonic. On the other hand, the non-linear terms can be approximated using extended describing functions, by their DC or main fundamental harmonic. These will finally result in achieving the reduced large-signal state equations, which could be linearised employing the Taylor expansion and then, the corresponding transfer functions are obtained. Figure 5 depicts the equivalent circuit of the DC-DC-LLC resonant converter. According to this figure, the state equations are as follows: The sgn function is defined as follows:
Approximation of variables
The typical input waveform of the LLC circuit, where its pulse width is determined using the phase-shifted modulation or the duty-cycle change, can be approximated using a sine term that its Fourier series is as follows: If the pulse width modulation is considered: Since u Cr , i Lm and i Lr have nearly sine waveforms, they can be approximated by their first-order sine and cosine components. By this approximation, each state variable is decomposed into two different variables and considering the output voltage, the state equations would include seven variables in overall: By derivation and rearranging the above equations, the approximated equations are achieved as follows, based on replacing the non-linear terms with their DC and first-order harmonic components:
Steady-state equations
The steady-state model could be achieved from the above largesignal equations, by considering all derivatives and the output current i o to be 0. The resulted equations could be found in the Appendix.
Small-signal equations
In order to control the desired output variables, a small-signal model should be developed for identifying the system performance against changes. The small-signal equations could be obtained as in (33)-(40), by similar procedure discussed for the AFE-CHB stage, and based on decomposing the operating point state variables into large signal and deviation variables, and rearranging the equations: . (40) The definitions of various parameters in above equations are provided in the Appendix. Finally, it should be mentioned that the required transfer functions for designing the control system could be achieved by applying the Laplace transform to the above equations.
PET CONTROLLER DESIGN
In this section, the required controllers for the AFE-CHB rectifier and DC-DC-LLC resonant converter stages of the PET are designed, based on the transfer functions determined in the previous sections. Then, a modified controller is proposed based on the output voltage of the PET.
Design of the AFE-CHB controller
The configuration of the AFE-CHB rectifier controller, which consists of the voltage and current control loops, has been depicted in Figure 6. The voltage control loop is responsible for maintaining the sum of AFE-CHB DC link voltages at the desired value. On the other hand, the current control loop provides the input reference sine waveform, based on the desired input power factor. Since the input current has a sine waveform, the Proportional-Resonant (PR) controller is used in our work in order to improve the control performance. The output of the current control loop generates a sine reference waveform, which is then compared with the carrier waveform.
The voltage control loop of the AFE-CHB rectifier, which includes a Proportional-Integral (PI)-type controller, has been shown in Figure 7. The transfer function u c /I s is required for the voltage control loop. By substituting the designed PET parameters, this transfer function will be as follows: The simplified model of the voltage loop transfer function is 1/C dc s, which can be used. Figure 8 shows the current control loop, which consists of a PR controller with the transfer function shown in (42).
In this equation, K p , K r and 0 are, respectively, the proportional and resonance gains, and the resonance frequency. The required transfer function for the current control loop is I s /U i , which is calculated as follows, based on the PET parameters: 500s + 2500 s 2 + 51s + 53, 050 .
The simplified form of the control current transfer function is 1/L s . Finally, the controller coefficients are determined using the tuning tools of the MATLAB/Simulink software.
Considering the calculated coefficients for current and voltage control loops, the closed loop transfer function will be as follows: (44) Figure 9 shows the bode diagram of the current control loop for the open and closed loop cases. As it is obvious in these figures, the PR controller has a notable gain in a small range at the grid frequency. Figure 10 depicts the transfer function of u c /u c *, which corresponds to both the current and voltage control loops. According to the frequency response, the system has a gain margin of 15.3 in gain crossover frequency of 77 Hz, and a phase margin of 54 o in phase crossover frequency of 7.7 Hz. Therefore, the system is stable for a wide range of the loads. Moreover, suitable values are considered for the phase and gain margins, in a way that the system response would be fast and also the oscillations are low. Figure 11 depicts the control block diagram for the DC-DC stage, which consists of the voltage control loop of the parallel cells. The output of the DC-DC stage could be controlled via two parameters: (a) duty cycle; (b) converter switching frequency. The fixed duty-cycle control scheme can be employed due to its superior efficiency performance. Moreover, only one of the H-bridges is active in DC-DC-LLC resonant topology at each instance, based on the forward or backward operation. A controller is used for recognising the operating condition. The operating principles of this controller is such that when the output voltage exceeds the upper band limit, the backward state is activated; and when the output voltage is below the lower limit, the forward state is employed.
Proposed simplified control scheme for the PET
The block diagram of the proposed simplified control scheme is depicted in Figure 12. In the traditional controller, the block diagrams in Figures 6 and 11 are performed together. Therefore, it is required to measure all the output voltages of the AC-DC stages. In the proposed control algorithm, by only measuring the PET output voltage with only one voltage sensor, the control goals could be fulfilled. If the number of input cells is high, the required voltage sensors and the corresponding ADC units are reduced from N to 1 by the suggested method; and hence, the system reliability is increased and the cost of the system is reduced. Moreover, the PET output voltage could be directly controlled by this control scheme. In order to calculate the PI controller coefficients, the dynamics of the DC-DC-LLC resonant converter could be neglected. As a result, the output voltage is simply multiplied by the cell numbers and also the transformer turns ratio, in order to control the input stage. Finally, it should be mentioned that in the proposed control scheme, the DC-DC stage operates with a fixed duty cycle. Moreover, a hysteresis controller is employed for recognising the forward or backwards modes. The operating principles of this deadbeat controller are like what was discussed in Section 3.2.
FPGA IMPLEMENTATION OF THE CONTROL SCHEME
This work employs the control platform that is designed based on the XC6SLX45 chip, manufactured by Xilinx. Since the FPGA chips do not have analogue inputs, four AD9231 chips, which are 20 mega samples per second (MSPS) ADCs, are used. In order to eliminate the high-frequency noises, low-pass filters are employed at the input of these ADCs. In order to implement the designed continuous controllers in s-domain into the FPGA platform, they were firstly discretised and transformed into the z-domain, employing the backward transformation, in which the s operator is replaced with Generator blocks are employed to implement and simulate the control algorithm in MATLAB. Next, a module of codes is generated from this model, which could be employed in ISE. Other components such as ADC, clock generation and virtual inputoutput units could be also considered in ISE. When the controller is implemented in the ISE, all different parts of the developed code can be executed simultaneously and in the parallel form; since the logic sources in the FPGA chip can be configured and operate independently. This fact not only significantly improves the reliability but would nearly eliminate the restriction regarding the sampling rate is DSPs, which is of high importance, especially during fast dynamic changes. Finally, note that ISE automatically applies some optimisation procedures during executing the developed code, by considering a tradeoff between the amount of code and execution speed. In order for capturing the waveforms and debugging purposes, the Chipscope tools of the ISE software are employed. Figure 13 presents different parts of the overall control scheme implemented in the FPGA chip, using the MATLAB System Generator. Figure 13(a) depicts the overall schematic of the AFE-CHB switching and control strategies. As the first stage, the grid voltage and current, and also the PET internal voltages are measured and after passing ADC units, are used as the algorithm inputs. These variables are fed into an low pass filter (LPF) and gain block, in order to attenuate the noises and improve their signal quality. Next, block 2 determines whether the conventional or proposed method is selected. The phaselocked loop (PLL) (block 3), which is of TD type, determines the grid current reference angle, using the grid voltage and reference reactive power (lagging or leading). In order to calculate the reference current magnitude, the main voltage controller (block 4) is employed, which consists of the designed controllers and their corresponding coefficients. After passing through a rate limiter that smoothens its sudden changes (especially during start-up), the reference DC voltage is compared with its real value and the resulted error is fed to this block. In the next stage, the reference magnitude and angle of the input current are combined in block 5 and compared with the real input current. The resulted error is used as the input of current control loop (block 6). This stage generates the appropriate modulation index, considering the calculated controller coefficients in the last section. Finally, the unipolar modulation is responsible to determine the switching states. Figure 13(b) shows the employed TD-type PLL structure, which is based on αβ-dq transform. Since its input voltage is single phase, a 90 o delay of this voltage is used as the second input. Then, the output q component is used for regulating the PLL. The output of the internal PI controller is added to reference ω. By integrating this value, the angle is determined. Figure 13(c) presents the DC-DC stage controller. The output low voltage (LV) voltage is fed to this part as the input. Based on the value of this voltage and also the limits of the hysteresis controller, the forward or regenerative mode is activated. After selection of the operation mode, the switching pulses for this stage are determined with a fixed duty cycle. Figure 13(d) depicts different protections considered for the PET. These include the PLL enable command, overcurrent and overvoltage protections and the user enable command. Finally, Figure 13(e) depicts the digital implementation of the PI controller. Moreover, the antiwind-up function has been considered by the K c gain, in order to compensate the integral component operation.
HARDWARE IMPLEMENTATION RESULTS
In order to evaluate the effectiveness of the developed control platform and also verify the performance of the designed controller different aspects, a laboratory-scale PET ( Figure 14) has been built. Different parameters of the designed PET have been shown in Table 2. These parameters are calculated based on the rated power and operating voltage and frequency of the scaled-down prototype. The semi-conductor switches for the AFE-CHB are IKW30N60T IGBTs, with a maximum voltage and current ratings of 600 V and 30 A. Two 1 mH inductors have been employed in series at the input of the PET. It should be noted that in a real-scale PET, semi-conductor devices with higher voltage and current ratings should be employed at the AC-DC stage, and their switching frequencies are significantly lower. Moreover, a lower value of inductance is utilised at the input. The switches of the DC-DC stage are implemented by 2SK3681 MOSFETs, which their voltage and current ratings are, respectively, 600 V and 40 A. Also, two variable resistors are used as the output loads. Each of the series-connected AC-DC H-bridge cells operates in the boost mode; and provides a higher voltage than its input, at the high voltage (HV) DClink capacitor. The value of the DC-link capacitors has been determined based on the maximum allowable CHB-AFE output voltage ripples, which is 2% in our work. In a real-scale system, the maximum allowable ripple is more (5%, for example), and these capacitor values could become smaller. Moreover, the area product approach has been considered for designing the HFT [20]. The PQ cores have been use for both the HFT and inductors, due to their reduced leakage flux. The sampling rate considered for implementing the controller in the FPGA-based control platform is 10 MHz. Note that if the controller was intended to be implemented on a DSP chip, the sampling rate could not go further than a few tens of kilohertz. On the other hand, the sampling rate for FPGAs is not restricted by the minimum time needed for algorithm execution, due to the parallelprocessing feature. However, some factors such as resolution of
PET start-up
In the start-up condition, due to the small input impedance, charging of high-voltage DC-link capacitors, the inrush current of the high-frequency transformer and charging of the output capacitors, the PET will draw a significant current from the grid. In order to limit this current and protect the devices, a start-up circuit, which includes that a series resistance is required. The value of the employed resistance here is 1.6 Ω. Figure 15(a) depicts the start-up waveform. In this figure, the HV DC-link voltage, grid current and voltage, and also the input voltage of the CHB configuration (v r ) are shown. The maximum current is 82.4 A in this condition, which reaches zero after 150 ms. The HV DC-link capacitors are charged to 160 V. In order to limit sudden changes of the grid current in Figure 15(a) (that is due to voltage change of the high-voltage DC link), a start-up scheme for the PWM rectifier control is also considered and the results are shown in Figure 15(b) and (c). In the start-up instance, the voltage reference value is set to its minimum value (10% of the grid voltage) and after start-up, the voltage value will increase to the nominal value, with the rate of 20 V/s. As it is shown in Figure 15(c), the start-up scheme of the control strategy will be activated at the first zero-crossing point of the grid voltage and this results in the decrease of the current.
Load change
The PET controller should be robust against instantaneous load changes. In order to verify this, a load change from 60 to 1080 W and vice versa has been considered and the results are shown in Figure 16. According to this figure, the output voltage overshoot will reach 10 V for the load change case. It should be mentioned that in a real-scale high-power PET, this overshoot value will be significantly lower, due to the smaller DC-link capacitances. Also, the response time is 200 ms. Moreover, with the increase in load, the output voltage oscillations' magnitude will approximately increase by 2.5 V. Due to the control algorithm performance, the PET is stable and the output average voltage will remain unchanged in this case.
Input voltage change
The proposed controller for PET can effectively maintain the output voltage value at its desired range, when the grid voltage is Figure 17 depicts the experimental results of this condition. According to this figure, the grid maximum voltage value has decreased by 58 V. With the decrease in the grid voltage, its current is increased; since the output load is fixed. However, the value of the output voltage remains completely the same before and after the voltage decrease. This verifies the grid voltage drop compensation by the designed controller for CHB-AFE.
Reactive power control
The designed controller is capable of providing reactive power control for the CHB-AFE. Figure 18(a)-(c) depict the experimental results of the reactive power control with the unity, inductive and capacitive power factors. Figure 18(a) depicts the normal condition of the CHB-AFE with unity power factor, where the root mean square (RMS) value of the input current is 5.5 A. By changing the reference output of the PLL by 60 o (lagging), the results in Figure 18(b) are obtained. In this case, the grid current RMS value increases from 5.5 to 8.37 A, and an active power of 330 W is drawn from the grid. The waveforms of the system with a 60 o reference leading power angle are presented in Figure 18(c). The RMS value of the current is 8 A in this case, which results in injection of 300 Var to the grid.
Regenerative operation
One of the most important advantages of PET, is the regenerative operation capability. The CHB-AFE stage inherently provides the regenerative capability. In order to determine the DC-DC-LLC resonant converter operation mode, the output voltage is used and upper and lower limits are considered for its value. If this voltage exceeds the upper band due to the regenerative current injection, the regenerative mode is activated. As stated before, the primary side H-bridge of DC-DC-LLC resonant converter acts as a diode rectifier, and the secondary side H-bridge will be active in this condition. Figure 19 demonstrates the waveform for the regenerative mode. As it can be seen, there is a phase difference of 180 o between the grid voltage and current.
Current sharing performance
The experimental results of the output currents for both cells are shown in Figure 20. According to these waveforms, it is obvious that the values of these currents are equal (3.9 A); and hence, the output power is equally provided by the both cells. This is achieved since only two input cells are connected in series, and the switching frequency is fixed. Note that if there is a serious problem regarding current sharing performance in very high-power applications with N > 2, simple feedback controllers could be designed for current sharing; and easily implemented in the FPGA-based platform.
Grid current THD
The THD values of grid current waveforms with different magnitudes have been presented in Table 3. These values are less than 5% for a wide range of operating points (the ones above 15% of the nominal current magnitude); which would be acceptable according to well-known standards such as IEEE 519. Note that for a high-power PET, although the switching frequency and input inductance values are smaller, but the THD
PET efficiency
The efficiency of the PET set-up for different loads has been shown in Figure 21. According to these results, the maximum efficiency is 90.43%. Moreover, the efficiency is more than 88% for a wide range of output currents. Note that in a real-scale system, the switching losses of the AC-DC stage will be lower, due to the smaller switching frequency. Regarding the conduction losses, despite the increase of current magnitude, their value won't significantly increase due to the lower values of R on for high-power switches. Therefore, the overall efficiency will not be degraded, and even could be improved.
Computational time analysis
In this subsection, the computational burden analysis is conducted for the developed algorithm. The execution time of an algorithm can be defined as the amount of the time between the instance of the input variables transfer from the ADC blocks to the FGPA unit; and finally obtaining the duty-cycle values at the FPGA unit output. If there was no delay, the whole algorithm could be performed in a single sampling period, that is, 1 10 MHz = 100 ns, due to the parallel-processing capability of FPGA. However, there are several delays due to the existence of components such as input registers of different blocks, multiplier units, PI controllers etc. The total number of delay components between the input and output of our implemented algorithm is 18. Therefore, the total execution time would be18 × 100 ns = 1.8¯s.Moreover, the utilisation percentage of different FPGA chip resources is provided in Table 4.
5.10
LLC circuit currents at different loads
CONCLUSION
This research discusses a fully FPGA-based prototype of a single-phase PET for railway applications. The designed control platform includes a single Xilinx FPGA chip, along with other required components such as ADC blocks. The system generator package of the MATLAB software is employed, in order to easily implement different designed control loops and also additional requirements such a PLL, protection schemes etc. Moreover, some modifications have been made in order to simplify the traditional controller implementation and meeting the requirements of the traction application. According to the experimental results, it could be seen that the PET experiences
FIGURE 22
High-frequency transformer (HFT) waveforms: (a) switching pulses, (b) HFT currents for an 1.5 A output load, (c) HFT currents for a 7.5 A output load and (d) HFT currents for a 13.6 A output load a smooth start-up with a limited inrush current. Moreover, the implemented control scheme is robust against output load changes and input voltage variations. The regenerative operation capability and also the reactive power control are effectively realised by the implemented controller. By analysing the DC-DC stage waveforms, it can be seen that ZVS and ZCS operations are realised at different operating points. Therefore, the experimental results validate the satisfactory performance of the system in terms of efficiency and grid current THD. Finally, for the real-scale high-power traction systems, if the power converter is carefully designed and its parameters are correctly chosen, the above-mentioned benefits could be achieved by the same control platform that is developed and designed in this work; and hence, the proposal could be a good choice to be employed in practical railway applications.
A.1 AVERAGE MODEL EQUATIONS OF AFE H-BRIDGEẋ
where D is the duty cycle and C 1 and C 2 could be defined as in (10). In the above equations, V e , α and β are related to i p , u o and i in with the following equations: | 2021-09-01T15:12:07.472Z | 2021-06-22T00:00:00.000 | {
"year": 2021,
"sha1": "ebf49792b4e4f85a2efc6ebc976d5f82b225f968",
"oa_license": "CCBY",
"oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1049/pel2.12158",
"oa_status": "GOLD",
"pdf_src": "Adhoc",
"pdf_hash": "0d01fb521f9893d756782e26b4f507ba4b5b180c",
"s2fieldsofstudy": [
"Engineering",
"Computer Science"
],
"extfieldsofstudy": [
"Computer Science"
]
} |
85445775 | pes2o/s2orc | v3-fos-license | The transcriptome analysis of Protaetia brevitarsis Lewis larvae
Larvae of the pest Protaetia brevitarsis are used to treat infections in traditional Chinese medicine. However, genomic information about this non-model species is currently lacking. To better understand the fundamental biology of this non-model species, its transcriptome was obtained using next generation sequencing and then analyzed. A total of 7.62 Gb of clean reads were obtained, which were assembled into 169,087 transcripts corresponding to 142,000 annotated unigenes. These unigenes were functionally classified according to Gene Ontology (GO), euKaryotic Ortholog Groups of proteins (KOG), and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotations. A total of 41,921 unigenes were assigned to 56 GO terms, 21,454 unigenes were divided among 26 KOG categories, and 16,368 unigenes were assigned to 32 KEGG pathways. In addition, 19,144 simple sequence repeats (SSRs) were identified. Furthermore, several kinds of natural antimicrobial peptides and proteins, 4 histones with potential antimicrobial activity, and 41 potential antimicrobial peptide sequences were identified. These data are the first reported whole transcriptome sequence of P. brevitarsis larvae, which represents a valuable genomic resource for studying this species, thus promoting the utilization of its medical potential.
Introduction
The beetle species Protaetia brevitarsis Lewis is widely distributed in China, and it is a pest of plants, including vegetable crops. In traditional Chinese medicine, its larvae are used to treat microbial infections. However, very little research has been conducted on this species. Moreover, there is also no genomic data from this non-model organism.
Transcriptome data represent an essential type of genome data and offer an opportunity to explore the genomes of non-model organisms. Owing to the rapid development of next generation sequencing (NGS) technologies, such as the Illumina sequencing platform, the transcriptomes of non-model organisms can be characterized quickly, inexpensively, and accurately [1,2]. Accordingly, the transcriptomes of many non-model organisms have been characterized by NGS, such as the green odorous frog Odorrana margaretae [3], the inland robust scorpion Urodacus yaschenkoi [4], the land snail Koreanohadra kurodana [2], the Chinese red-headed centipede Scolopendra subspinipes mutilans [5], and several luminescent beetle species [6]. PLOS The P. brevitarsis larvae live on the ground and feed on rotten food. This type of habitat and diet lead them to encounter many different kinds of pathogenic microbes, which has imposed strong selective pressures on their immune systems. The insect immune system is composed of two elements: cellular defense responses and humoral defense responses [7]. Insect antimicrobial components secreted by the insect humoral immune system, especially antimicrobial peptides and proteins (AMPs), play an important role in defense against microorganisms, which may be the molecular basis of the anti-infection effects of P. brevitarsis larvae used in medicine. Many insect AMPs have been identified, characterized, and categorized according to familiar sequences and structures, including linear-form, disulfide bridges, and proline-or glycine-rich regions [8]. Additionally, AMPs from insects also show great potential as antiinfection agents [9].
In this study, the transcriptome profile of P. brevitarsis larvae was constructed using the Illumina sequencing platform. After de novo transcriptome assembly, bioinformatic analyses were conducted to annotate the functional direction of the assembled sequences. The analyses focused on P. brevitarsis larvae AMPs. This research provides the first large-scale transcriptome data from P. brevitarsis larvae, which improves the current understanding and medical utilization of this species while enabling the discovery of important functional genes.
Ethics statement
All animal collection and utility protocols were approved by the Henan University of Science and Technology of Biology Animal Use Ethics Committee.
Insect collection
The Protaetia brevitarsis Lewis larvae used in this study were obtained from a breeder in Henan, China. After being washed in 0.9% saline, the samples were immediately frozen in liquid nitrogen and stored at -80˚C until use. cDNA library construction, Illumina sequencing, de novo assemble, and gene annotation Total RNA was extracted using TRIzol reagent (Life Technologies, Carlsbad, CA, USA), and cDNA library construction and Illumina sequencing of the samples using an Illumina HiSeqTM 2000 sequencer (Illumina, San Diego, CA, USA) were performed by Novogene Bioinformatics Technology Co., Ltd. (Beijing, China).
De novo transcriptome assembly was conducted using the program Trinity [1]. Gene annotation was based on searches of various public databases, including NCBI Nr (non-redundant protein sequences), NCBI Nt (non-redundant nucleotide sequences), Pfam (Protein family), KOG (euKaryotic Ortholog Groups of proteins), Swiss-Prot (a manually annotated and reviewed protein sequence database), KO (Kyoto Encyclopedia of Genes and Genomes Ortholog database), and GO (Gene Ontology). SSR markers were isolated from the transcriptome using MISA (http://pgrc.ipk-gatersleben.de/misa/misa.html).
A detailed description of cDNA library construction, Illumina sequencing, the de novo assemble and gene annotation is given in S1 File.
Potential AMP sequences identification
To identify more potential AMPs and peptide sequences with potential antimicrobial activity in the transcriptome of P. brevitarsis larvae, the assembled unigenes were also BLASTed with known AMPs from The Antimicrobial Peptide Database (APD) [10], Collection of Anti-Microbial Peptides (CAMP) [11], and the Linking Antimicrobial Peptides (LAMP) database [12] using BLAST 2.2.31+ with a sequence similarity cutoff of 80%.
Transcriptome sequencing and de novo assembly
The P. brevitarsis larvae cDNA library was sequenced using the Illumina HiSeqTM 2000 sequencer, ultimately obtaining 50,796,336 clean reads with 98.09% Q20 and 42.88% GC content after the removal of adaptor sequences, ambiguous nucleotides, and low-quality sequences. The clean reads were assembled into 169,087 transcripts, which corresponded to 142,000 unigenes. A summary of sequencing and assembly results was presented in Tables A and B in S1 File, and the unigene sequences was presented in S2 File. The clean data had been deposited into the Sequence Read Archive (SRA) database of the National Center for Biotechnology Information (NCBI) under the accession number PRJNA516097.
Functional annotation of unigenes
After public databases were used for sequence similarity searches, 56,937 unigenes (40.09% of all unigenes) were successfully matched to annotations in at least one database (Table C in S1 File), which is similar to rates reported in other beetle species [13][14][15], indicating that a large number of species-specific genes, noncoding gene regions, and non-conserved domains could not be matched to annotated genes or that chimeric sequences had occurred as a result of assembly errors [16]. Among the annotated unigenes, 25.87% (36,743) had significant matches in the Nr database, 9.07% (12,881) in the Nt database, 11.52% (16,368) in the KO database, 20.23% (28,727) in the Swiss-Prot database, 28.52% (40,511) in the PFAM database, 29.52% (41,921) in the GO database, and 15.1% (21,454) in the KOG database. Furthermore, there were 4,618 unigenes (3.25%) annotated in all seven databases.
The E-value distribution and similarity distribution of the annotated unigenes in the Nr database identified 49.2% (Fig 1A) of annotated sequences as having strong homology (Evalue less than 1E-30), with 77.8% (Fig 1B) of the annotated sequences having a similarity index higher than 60%. The top-hit species distribution showed that 27.1% ( Fig 1C) of P. brevitarsis larvae unigene sequences matched with sequences from red flour beetle Tribolium castaneum. This overlap of unigenes was lower than that for other beetle species, such as the pine shoot beetle Tomicus yunnanensis (62.48%), the salt marsh beetle Pogonus chalceus, and the nipa palm hispid Octodonta nipae (72.6%) [17][18][19]. However, this value was similar with respect to the seven-spot ladybird beetle Coccinella septempunctata (28.5%) [15]. In the NCBI protein database, the number of protein sequences from T. castaneum and the mountain pine beetle Dendroctonus ponderosae were 42,319 and 49,666, respectively. But, only 3.4% of P. brevitarsis larvae unigene sequences matched with those of D. ponderosae. It was indicated that Protaetia brevitarsis Lewis had near evolution distance with T. castaneum.
GO categories were used to classify the functions of the unigenes, and the classification of GO categories was conducted using the GOseq R package based on the Wallenius non-central hyper-geometric distribution [20]. According to GO annotation, 41,921 unigenes were classified into three main GO categories: biological process, cellular component, and molecular function (Fig 2). Among the biological process terms, 25 level-2 categories were identified, and cellular process and metabolic process were the most abundant terms. For cellular components, 20 level-2 categories were identified, and cell, cell part, and organelle were the most abundant groups. For molecular function, 11 level-2 categories were identified, and the genes that were associated with binding and catalytic activities were the most represented. Similar gene GO classification distributions had also been reported in transcriptomic studies of other beetles [17,21,22]. To further predict putative protein functions, a KOG analysis was performed. Based on our results, 21,454 unigenes were divided into 26 categories (Fig 3), and the largest category was 'general functional prediction only' (16.02%), followed by 'posttranslational modification, protein turnover, chaperones' (14.38%), 'signal transduction mechanism' (12.40%), and 'translation, ribosomal structure and biogenesis' (12.04%). The KEGG [23] database was used to identify potential biological pathways represented in the P. brevitarsis larvae transcriptome, and the classification of KEGG pathways was performed using KOBAS software [24]. A total of 16,368 unigenes were classified into five categories, including cellular processes, environmental information processing, genetic information processing, metabolism, and organismal systems, and then assigned to 32 sub-terms (Fig 4). Among the pathways, translation (2391 unigenes) and signal transduction (2207 unigenes) were most highly represented. Similar gene KEGG classifications distributions had also been found in coconut leaf beetle Brontispa longissima [25]. These annotations provide a valuable resource for future research on specific processes, functions, and pathways in P. brevitarsis and related species.
SSR detection from the assembled unigenes
As SSR markers are highly polymorphic, co-dominant in their inheritance, highly specific in their amplification, and repeatable in their genotyping, they are widely used in genetic studies to analyze genetic diversity and biological evolution [26]. And transcriptome sequencing using NGS technologies has greatly accelerated the process of discovering molecular markers in non-model organisms [27].
In this study, a minimum of ten repeat numbers for mononucleotide repeats was selected, six for dinucleotide repeats, and five for tri-, tetra-, penta-, and hexa-nucleotide repeats. Thus, 19,144 SSRs were isolated from 13,933 sequences, with 3,136 sequences containing more than one SSR. Additionally, 1,522 SSRs were found in the compound form. Among the identified SSRs, mononucleotide repeats accounted for the highest percentage (13,550 SSRs, 70.77%) followed by trinucleotide repeats (3888, 20.30%), dinucleotide repeats (1575, 8.24%), and tetranucleotide repeats (110, 0.62%). A similar distribution of repeat units were also found in the nipa palm hispid beetle Octodonta nipae [28]. Frequency of SSRs based on the number of repeat units in Protaetia brevitarsis Lewis Larvae was provided in Table D in S1 File. The five most frequent motif types in the transcriptome were A/T (13,195 SSRs, 68.92%), AAT/ATT (1207, 6.30%), AGG/CCT (1001, 5.23%), AT/AT (861, 4.50%), and AAG/CTT (669, 3.49%). These SSR markers will be useful for assessing genetic diversity and population structure in P. brevitarsis and related species.
AMPs
Drug-resistant bacterial pathogens have caused significant health problem worldwide [29,30]. AMPs have become a particularly promising class of candidates for overcoming this problem, as they have unique modes of action that differ from those of classical antibiotics and are difficult for bacteria to develop resistance to [31,32].
Natural AMPs play an important role in immune systems, and they protect multicellular organisms by controlling and/or combating pathogens by killing microbes directly or through functions related to diverse immunomodulatory activities [33]. A greatest diversity of AMPs have been identified in different kinds of insects, including beetles [34,35]. In this study, several kinds of natural AMPs were identified in the P. brevitarsis larval transcriptome.
Lysozymes. Lysozymes are a group of enzymes with antibacterial activity that ubiquitously exist in diverse organisms, including invertebrates and vertebrates. They can catalyze the hydrolysis of β-1,4-glycosidic bonds between N-acetylglucosamine and N-acetylmuramic acid in the peptidoglycan of bacterial cell walls, which causes bacterial cell lysis and thereby prevents bacterial infections [36]. Based on structural and functional features, lysozymes are categorized into three major groups in the animal kingdom: invertebrate-type (i-type) lysozymes, chicken-type (c-type) lysozymes, and goose-type (g-type) lysozymes [37]. The c-and gtype lysozymes are found in various vertebrate and invertebrate species [38]. And the g-type lysozymes have a transglycosylase SLT domain (Bacterial lytic transglycosylases Domain) which contains three conserved catalytic binding sites, three conserved substrate binding sites, and six conserved cysteine residues. In this study, five of these unigenes (c16827_g1, c82138_g1, c85114_g1, c89942_g2, c91216_g2) were annotated as i-type lysozymes, which is similar to the number found in the beetles Harmonia axyridis and Meligethes aeneus [39]. As shown in Fig 5, the unigene c16827_g1 shares 41.77% identity with the i-type lysozyme (XP_022920100.1) from the beetle Onthophagus Taurus (Fig 5A), c82138_g1 shares 59.87% identity with the i-type lysozyme (XP_023027517.1) from the beetle Leptinotarsa decemlineata (Fig 5B), c85114_g1 shares 50.00% identity with the i-type lysozyme (ALM25916.1) from the beetle Harmonia axyridis (Fig 5C), and c89942_g2 and c912216_g2 share 46.71% and 48.50% identity with the i-type lysozyme (ALM25918.1) from the beetle Harmonia axyridis (Fig 5D). Two of these unigenes (c87216_g2, c89288_g1) were annotated as c-type lysozymes, and they shared 51.15% and 31.49% identity with a c-type lysozyme (BBE27867.1) from Locusta migratoria (Fig 5E). One of these unigenes (c124211_g1) was annotated as a g-type lysozyme, and it shares 71.21% identity with a g-type lysozyme (ADV36303.1) from Physella acuta ( Fig 5F), which has shown activity against both E. coli and MRSA [40].
Defensin. Defensins are small, cationic, and cysteine-rich peptides that include 3-4 intramolecular disulfide bonds. Hundreds of insect defensins have been identified since the first defensins were reported in the flesh fly Sarcophaga peregrina [41] and the black blowfly Phormia terraenovae [42]. They are mainly active against Gram-positive bacteria, and some are also active against Gram-negative bacteria as well as fungi [43]. In this study, one of these unigenes (c23254_g1) was annotated as a defensin, and two (c158345_g1, c79927_g1) were annotated as defensin-like proteins. As shown in Fig 6, the unigene c23254_g1 shares 89.16% identity with defensin-3 (AFD01290.1) from Pinus sylvestris (Fig 6A), c158345_g1 shares 56.52% identity with defensin-like 1 protein (AJQ21502.1) from Mytilus galloprovincialis (Fig 6B), and the c79927_g1 sequence was identical to a defensin-like protein (XP_021821153.1) from Prunus avium (Fig 6C).
Peritrophin. Peritrophin was first isolated from insect intestines, where it was inferred to protect insects from microorganismal infection [52]. These proteins have a signal sequence and one of three kinds of peritrophin domains: peritrophin-A domains, peritrophin-B domains, or peritrophin-C domains. Peritrophin-A domains are also referred to as chitinbinding type 2 domains [53]. Today, more and more peritrophin-like proteins have also been found in many organisms [54]. In this study, the unigene c72353_g1 was annotated as a mucin-like peritrophin and shared an identity of 28.21% with a mucin-like peritrophin (ACN62986.1) from Popillia japonica (Fig 10A). The unigene c75073_g1 was annotated as peritrophin and shared an identity of 24.02% with peritrophin (ACN62985.1) from Popillia japonica (Fig 10B). The unigene c77120_g1 was annotated as peritrophin-1-like protein and shared an identity of 35.51% with the peritrophin-1 (XP_003251779.1) protein from Apis mellifera (Fig 10C).
Many other proteins and peptide sequence derived from other protein or peptide have also shown antimicrobial activity [55,56]. To identify more proteins and peptide sequences with potential antimicrobial activity, the transcripts were BLASTed against AMP databases (APD, CAMP, and LAMP) with an identity cut off 80%. Finally, we identified four unigenes that were annotated as histones and shared high identity with histones with antimicrobial activity in the database. As shown in Fig 11, the unigene c13753_g1 shares an identity of 85.40% with the H2A histone (ID in APD: AP02804) from the Pacific white shrimp Litopenaeus vannamei ( Fig 11A), which has anti-Gram positive bacteria activity [57]. The unigenes c117375_g1 and c84048_g2 share identities of 99.03% and 95.15% respectively, with the H4 histone (ID in APD: AP02807) from American cupped oysters Crassostrea virginica (Fig 11B), which has anti-Gram negative bacteria activity [58]. The unigene c92891_g1 shares an identity of 81.75% with the H2B histone (ID in APD: AP02808) from cattle Bos taurus (Fig 11C), which has anti- Gram negative bacteria activity [59], while c86640_g1 and c87143_g1 share identities of 86.76% and 88.24%, respectively, with the H3 histone (ID in APD: AP02809) from cattle Bos taurus (Fig 11D), which has also shown anti-Gram negative bacteria activity [59]. Transcriptome analysis of Protaetia brevitarsis Lewis larvae Furthermore, we also identified 41 peptide sequences (Table E in S1 File) predicted to have potential antimicrobial activity on the basis of sequence similarity.
Conclusion
This study is the first whole transcriptome analysis of larvae from the non-model species P. brevitarsis, which accordingly will contribute to the exploration of the species through the resulting genetic resources and will thus facilitate further comprehensive studies of this species. Moreover, the natural AMPs, proteins with potential antimicrobial activity, and potential antimicrobial peptide sequences identified from transcripts in P. brevitarsis larvae provide templates for further development of new antimicrobial agents and will promote the utilization of the species for its full medical potential. | 2019-03-23T13:02:58.166Z | 2019-03-21T00:00:00.000 | {
"year": 2019,
"sha1": "f8db045a8fbb7cc63bf0b674d861b09d2f30be88",
"oa_license": "CCBY",
"oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0214001&type=printable",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "f8db045a8fbb7cc63bf0b674d861b09d2f30be88",
"s2fieldsofstudy": [
"Biology"
],
"extfieldsofstudy": [
"Biology",
"Medicine"
]
} |
119277224 | pes2o/s2orc | v3-fos-license | X-Ray Multimodal Tomography Using Speckle-Vector Tracking
We demonstrate computerized tomography (CT) reconstructions from absorption, phase and dark-field signals obtained from scans acquired when the x-ray probe light is modulated with speckle. Two different interlaced schemes are proposed to reduce the number of sample exposures. First, the already demonstrated x-ray speckle-vector tracking (XSVT) concept for projection imaging allows the three signal CT reconstructions from multiple images per projection. Second, a modified XSVT approach is shown to provide absorption and phase reconstructions, this time from a single image per angular projection. Reconstructions from data obtained at a synchrotron facility emphasize the potential of the approaches for the imaging of complex samples.
I. INTRODUCTION
X-ray computed tomography [1] has become an invaluable tool for nondestructive testing and 3D rendering of samples in fields as broad as material sciences, cultural heritage, paleontology and even medical science. In the latter domain, the dose necessary for imaging in vivo samples often becomes a critical issue and so must be carefully controlled and maintained below an acceptable hazard threshold [2,3]. Henceforth, phase and dark-field contrast-imaging methods are regarded as sensible approaches to eventually enhance the rendering quality and quantity of information on soft tissues with a dose level equivalent to or lower than that encountered by using conventional absorption contrast radiography.
Nowadays, a few methods and instruments are available to recover the phase shift induced by a sample exposed to an x-ray light wave. Most techniques use the partial transverse coherence property of an x-ray beam to generate interference between optical waves and enable the calculation of the phase shift. This applies, for instance, to propagation-based methods relying on either the contrast transfer function (holotomography) [4] or on the transport-of-intensity equation [5]. In parallel, the demonstration of grating-based phase-contrast imaging using a laboratory x-ray source promotes the attractiveness of this method [6]. Therein, a clever setup enables an extended source to be split into a series of smallersized sources to permit the generation of constructive interference compatible with shearing interferometry [7,8]. Another method sensitive to the differential phase of an x-ray beam is analyzer-based imaging that employs crystals with a narrow Darwin bandwidth to render different contrast for rays having different propagation orientations [9].
Near-field speckle-based methods [10][11][12][13][14] form a more recent class of x-ray imaging techniques and are proven to be applicable both at synchrotrons and at laboratory * berujon@esrf.eu sources [15,16]. Their attractiveness lies in their low requirements on coherence and in the simplicity of the wave-front modulator, reduced to a simple diffuser such as a piece of sandpaper or a biological filter.
Three main speckle methods relying on pattern correlation in the near field are available today. We purposely leave aside the near-field ptychography technique [17] that is based on a different principle (iterative phase recovery) but which often employs a scattering object to improve its efficiency. The first introduced method is xray speckle tracking (XST) [11,12,18]; therein, small subsets of speckle are tracked between two images, one being taken with the sample present in the beam, and the other being used as a reference. The second method requires a slightly more complex arrangement, as the scattering membrane needs to be scanned transversally to the beam with micrometer-or submicrometer-scale steps. This x-ray speckle-scanning (XSS) method [13] provides very high sensitivity at the cost of a number of sample exposures that can quickly rise up to a few dozen. In the last and most recent method [14], the scattering object used to generate the static speckle is moved to various transversal positions with respect to the beam while vectors built from speckle images are tracked across reference vectors. We decide to label this method the x-ray speckle vector tracking (XSVT) technique on account of the aspects it shares with the XST technique and the vectors involved in it.
In the first section of this paper, we briefly review and compare these three speckle-processing approaches. Efforts are made to highlight their differences and to demonstrate theoretically, and later experimentally, that XSVT is particularly well suited to performing 3D multimodal computed tomography. Indeed, without any a priori assumption on the samples, only a few images per projection are required for multimodal tomography when employing the appropriate scheme. In contrast to previous work [19,20], the XSVT approach provides 3D rendering of the dark-field signal, in addition to the absorption and phase signals, whilst the spatial resolution of the imagery is limited only by the detector used.
In the last part, we introduce a mixed approach based on the combination of the XST and XSVT methods. This original scheme permits the extraction of the phase signal with a resolution approaching that of the XSVT technique with a lesser number of exposures than for XSVT alone. Experimentally, the phase-contrast tomography reconstruction obtained from this mixed approach shows that the number of sample exposures necessary is comparable to the case of absorption tomography.
II. SPECKLE-PROCESSING SCHEMES
We give here a short comparative review of the speckleprocessing approaches.
The three speckle schemes available today differ in the kind of data collected, the numerical treatment applied and, for two of them, their intrinsic sensed signals. Nevertheless, the concept shared by all three schemes is the use of cross-correlation operations to track the displacement of a distorted pattern with respect to a reference pattern [21] caused by x-ray refraction through the sample. In all schemes, the speckle pattern used as structured illumination is generated by interference upon the propagation of coherent photons through a scattering object (scatterer) with small random features, often made of sandpaper or of a biological filtering membrane with a known statistical feature size.
To be more precise, the insertion of a sample into an x-ray beam generates a dephasing of the beam waves by the amount φ = −k δ(x, y, z)dz where k is the wave number and δ is the optical refractive index of the sample material. Upon propagation along an e z axis, this dephasing is responsible for the distortion of an input reference modulation pattern, i.e., here, the speckle pattern. The local speckle shift is equal to v = d k ∇φ in the basis (e x , e y ) transverse to the photon-propagation direction. Therein, ∇ is the del operator and d the propagation distance from the sample to the detector. Note that this equation is valid in the x-ray regime due to the small diffraction and refraction angles involved with such light.
This linear relationship between the local displacement v of the modulation features by comparison to a reference position on one side and the phase gradient induced by the sample on the other provides a simple and common way of accessing beam phase information. This core principle is, for instance, applied within the grating interferometer [7] and coded aperture [22] approaches. Here, the use of near-field speckle as a wavefront modulator is particularly beneficial in the hard x-ray regime because such a pattern is shown to be invariable upon propagation over distances that can range from millimeters to meters [10]. Furthermore, this kind of speckle is easily accessible even with beams presenting low transverse coherence properties.
The distinctive aspects of each speckle scheme lie in the approach employed for the recovery of v(x, y). Figure 1 illustrates the data that are collected and used within each method.
Within XST (Fig. 1.(a)) [11,12], the scattering object is fixed and only two images are necessary: one with and one without the sample present in the beam path. The processing consists of taking a small subset of pixels containing some speckle pattern and tracking its shift from the sample image position across the reference speckle image using a cross-correlation algorithm. The operation is repeated for all subsets centered on each pixel. Thus, the vector map v(x, y) is calculated with subpixel accuracy and the two orthogonal differential wave-front gradients are recovered. Generally, the size of the tracked subset can range from 5 × 5 pixels to more than 21 × 21 pixels. Here the resolution is limited by the grain size of the scatterer, as the superimposition of the speckle grains onto sample features of equivalent size renders the tracking of the modified speckle pattern unstable. Despite the existence of algorithms that take into consideration the effect of subset distortion to enhance the tracking robustness [18], the method gets, locally, less robust and/or accurate in the presence of a strong phase shift, e.g. at the sample boundaries.
With XSS [13], whose principle is shown in Fig. 1.(b), the scattering object is mounted on a highly accurate bidirectional translation stage in order to raster scan the scatterer transversally to the x-ray propagation direction. Two sets of images are then collected by performing the same scan, whose step is on the order of 100 nm to a few micrometers, with the sample present in the beam and without it (reference). For each pixel, as marked by an eye symbol in Fig. 1.(b), two data arrays are collected and the analysis is done pixel by pixel. A crosscorrelation-based treatment computes the shift v(x, y) between the recorded patterns with a substep accuracy. Finally, the differential phase gradient is retrieved after linear normalization by a geometrical factor [14]. This scheme can offer a very high spatial resolution when combined with a strongly magnifying setup. Moreover, since the definition of the wave-front gradient is here proportional to the step size employed, it is possible to reach single nanoradian accuracy. On the other hand, such performance is accessible only at the cost of many sample exposures due to the necessity of performing 2D mesh scans, or at least, two wide-range 1D scans [13]).
XSVT is the most recently introduced speckle approach [14]. It requires the mounting of the scattering object on a moving stage capable of high repeatability with respect to the effective pixel size of the detector upon geometrical demagnification brought by the setup. However, unlike XSS, no accurate calibration on the motor step size is required. Figure 1.(c) shows the stacking up of images when the sample is present and when it is removed from the beam. The stage repeatability is needed for the scatterer to be in exactly the same position when recording images with and without a sample. Under this condition, the data-acquisition procedure consists of recording each η th pair of sample and reference images of the two stacks with the scattering object located at every given posi- tion of the scatterer. It is also necessary that a fixed pair of images possesses locally a speckle pattern that is uncorrelated from those of the other images. Data analysis starts by building two speckle vectors for each pixel using the two image stacks. Then, each vector from the sample stack is tracked across those of the reference stack. The location of the peak of maximum correlation between the samples tracked vectors and the reference ones provides a mean of recovering the local displacement vectors v(x, y) and hence the differential phase gradient induced by the sample. This protocol means that a resolution close to that of the detector can be reached and that the sensitivity scales with the detector pixel size and the propagation distance. Table I presents a brief survey of the attributes of each of the three techniques. Note that the figures provided are discussed in more detail elsewhere [11,13,14,18,[23][24][25]. We recognize that, in all three methods, the darkfield signal can be obtained by considering the change in speckle visibility between the references and sample images. However, although this capability is demonstrated with XST [15], only the XSS and XSVT approaches presently prove to be quantitative and accurate enough for 3D tomographic reconstruction. The moderate number of images per projection necessary for XSVT to access the 2D differential phase gradient in addition to the darkfield signal constitutes the main asset of the technique. The data redundancy in the XSVT technique permits the sensitivity to be improved down to 10 nrad. From these considerations and from Table I we are able to conclude that the XSVT method represents the best compromise on the number of necessary images per projection with regard to the sensitivity and spatial resolution. Therefore, although tomography is demonstrated with XST [19,20], XSVT represents our approach of choice to achieve highresolution phase-contrast tomography at a cost-effective number of sample exposures. Nevertheless, one can foresee that XSS will be able, in the future, to perform 3D tomographic imaging of samples presenting very small electronic variations when the dose is not an issue. A sketch of the experimental setup is shown in Fig.2. The experiment is conducted at the ESRF beamline BM05 [26], where a precision tomography station is permanently installed. At BM05, the photons are produced by a 0.85 T bending magnet on the circulating 6.02 GeV electrons of the storage ring. The photon energy is narrowed down to a band of ∆E/E ∼ 10 −4 centered around 17 keV by means of a double-crystal Si(111) monochromator. The sample is placed at 55 m from the source and the detector, a CCD-based FReLoN (Fast Read-out Low Noise) camera coupled to a scintillator, d = 1 m further downstream from the sample. The effective pixel size of the imaging system is h = 5.8 µm. A piece of abrasive paper with a grit designation of P800 is mounted on a linear translation stage allowing displacements in the beam transverse directions and located d p = 400 mm upstream of the sample. At a distance of d = 1 m from the sample, denoting by< > and σ, respectively, the mean value and the standard deviation of the vector scalar components, the visibility of the speckle presenting a nearly Gaussian distribution is of (I max − I min )/2 < I >= 0.4, or σ(I)/ < I >= 0.13. The recorded images have a size of 2048 pixels horizontally by 1850 pixels vertically.
The selected samples presented here are juniper berries and red currant berries. The juniper sample is interesting for the presence of features at very different spatial frequencies and for the significant amount of x-ray scattering generated by structures at the submicron scale. By its very nature the juniper berry is a difficult sample to image for many techniques. In comparison the red currant presents fewer parts that generate strong x-ray scattering but has large homogeneous features that can sometimes become problematic as they render less-accurate quantitative results with propagation-based methods.
IV. XSVT-BASED TOMOGRAPHY WITH AN INTERLACED SCHEME
The method of structured illumination employed in this section has been quickly presented in Sec. II and is detailed for projection imaging Ref. [14]. Here in addition, an interlaced scheme is used to optimize the number of images collected for each projection. The proposed interlaced XSVT method allows us to greatly reduce the overall number of sample exposures whilst conserving the accuracy of the phase and dark-field signals by factoring in the images recorded at consecutive projection angles.
The data-acquisition scheme is sketched at the top of Experimentally, the full data set results from the collection of 20 different scans of 360 projections, the starting scan angle being shifted by 0.1 degree every four scans. Then, for each projection n, the interlaced scheme consists of using the image intensity i(Ω n , x, y, k p ) recorded at the projection {n − 2, n − 1, n, n + 1, n + 2} as a multivariate random variable to build speckle vectors s v (Ω n , x, y) of size 20: contain and then reduce the accuracy of the vector correlation calculation [14]. Another two sets of images are collected below the 0 degree and beyond the 180 degree positions in order to complete the data sets of the first, second, penultimate, and last projections.
A reference image g(p, x, y, k p ), i.e., without the sample inserted into the beam, is taken before each 20 of the scans. Hence, a total of five reference matrices of size 2048 × 1850 × 20 could be built by a circular permutation of the speckle-vector elements and used as reference vectors r v (Ω n , x, y): where ζ is the elementary cyclic permutation over the elements of p with order q = mod (n−1, 5). From these speckle vectors of length 20, the absorption and dark-field signals in addition to the two differential phase gradients could be calculated for each Ω n using the XSVT method sketched in Fig. 1.(c) and detailed in Ref. [14].
In concrete terms, the displacement of the speckle vectors v(Ω n , x, y) induced by the sample refraction is calculated using the Pearson correlation coefficient ρ: with v(x, y) = v x e x + v y e y and where the variable Ω n is dropped for clarity. Note that the vectors v are calculated with subpixel accuracy [21]. Next, the differential phase ∇φ is obtained using the formula already mentioned above and valid for the small angles: where λ is the photon wavelength. We call the wavefront gradient ∇W = v(x, y)/d, which corresponds to the angular deflection of the rays caused by refraction. The transmission signal T for each projection is calculated by taking the ratio of the average speckle-and referencevector intensity elements: Besides, the dark-field signal D f is calculated by using: where σ denotes the standard deviation of the vector elements.
Note that this last formula is based only on statistics. Hence, in the case of large vectors (r v ,s v ), the approach becomes also valid in the case where s v and r v are built from images with different illuminations, i.e. with the membrane located at uncorrelated transverse positions between the reference and sample image stacks [27].
The phase-imaging method employed here is equivalent to x-ray grating interferometry or analyzer-based imaging as the measured signal is the differential phase signal ∇φ [6,8,28]. In these methods, a specific filter is often used in the inverse Radon transform to account for the specificity of the signal and perform a one-dimensional integration in the Fourier space [29,30]. Here, as the XSVT technique provides the two transverse differential phase maps, in contrast to the previous techniques, the phase images can be recovered by 2D integration by matrix inversion via, for instance, the Cholesky decomposition.
A set of raw, directional phase gradient, phase and dark-field images is shown in Fig. 3.(a-h) for the first angular projection of the juniper-berry sample. Figure 3.(c-f) makes us realize that whilst the traditional XST fails largely to track the speckle subsets due the presence in the sample of high spatial frequency features, the XSVT method succeeds perfectly to sense the very turbid wave-front. With the proposed interlaced XSVT technique, both the vertical and horizontal phase gradients are correctly recovered for each pixel, which means that upon 2D integration, the correct phase images are conserved despite the presence of fast-varying structures as seen in the inset of Fig. 3.(g). Similarly, the statistics offered by the speckle vectors permits a clear sensing of the dark-field signal as seen in Fig. 3.(h).
From the three imaging modalities, 3D reconstructions are operated using the ESRF PyHST (Python High Speed Tomography) [31] implementation with the classic filtered back-projection algorithm. Volume reconstructions from the phase and dark-field signals are shown in Fig. 4.(a-b). Transverse reconstruction slices are also displayed for comparison in Fig. 4.(c-e). These three signals do not render the same contrast for numerous sample features. For instance, the scattering features in the outer part of the juniper-berry (noted 1) render a strong contrast in the dark-field signal (slice in Fig. 4.(c)) while a low contrast is observed in the phase slice ( Fig. 4.(d)). Similarly, within the slice of the phase signal, homogeneous features observed within the yellow circle exhibit a large phase shift while presenting very low absorption. The comparison cuts in Fig. 4.(f) provide a different view on these results.
While the total number of images collected (7200) may seem large, it could be readily diminished by reducing the number of projections in the angular tomography scan. This figure could also be compared with holotomography for instance, which also requires a minimum of four images per projection or with grating interferometry, which demands phase-stepping scans of four steps per projection. For this latter technique, although a special scheme could slightly reduce the number of images necessary for reconstruction [32], the presence of an absorption grating is unfavorable when the irradiation dose received by the sample is an issue. The presented scheme can also benefit from a smaller ensemble of images per projection by using images of further projections in order to maintain the length and then the amount of statistics contained in the speckle vectors.
V. A MIXED XST-XSVT APPROACH
We present in this section a second scheme for phase sensing and tomography reconstruction with the aim of reducing the number of acquisitions per projection even further. The principle of this method consists of a mixed scheme between the traditional XST technique [11,19] and the optimized XSVT scheme presented in the pre- vious section. The experimental protocol consists of operating the data acquisition according to the procedure sketched in Fig. 2 (top), while recording only one image per projection angle, i.e. k p = 1. This means that during the scan the light is now modulated with five different speckle patterns, with a cyclical repetition of every five angular projections.
As done earlier, the five interlaced images centered on each considered projection are used to build a vector with sufficient statistics to achieve an accurate tracking. Furthermore, we enlarge the vectors by using not only the values of the centered pixels of each image but also the ones of the eight surrounding pixel values as displayed in Fig. 5. This approach permits to generate, in our case, vectors containing 35 components: , ..., i(Ω n , x + ε x , y + ε y ) (εx,εy) , ..., i(Ω n+2 , x + ε x , y + ε y ) (εx,εy) with here (ε x , ε y ) ∈ {−h, 0, h} 2 , h being the pixel size. Reference stacks are built in the same manner from the reference images: , ..., g(ζ q (C), x + ε x , y + ε y ) (εx,εy) , ..., g(ζ q (E), x + ε x , y + ε y ) (εx,εy) (8) The following step of the processing method is the one of XSVT described by Eq. 3-6 applied to these new vectors.
The calculated differential phase and phase images obtained with the mixed XST-XSVT technique are shown in Fig. 6.(a-d) for the juniper sample. Figure 7 shows the equivalent calculated maps for the red-currant sample. Images (e-f, i-j) are extracted with the interlaced XSVT technique previously presented while images on the right side, labeled (g-h, k-l), are the maps calculated with the present mixed XST-XSVT technique. From these two sample projection images, although the two methods apparently render similar results, a meticulous observation shows that the images obtained with the mixed XST-XSVT method reveal less detail. This difference is better emphasized when comparing the insets of the wave-front gradient maps. In the juniper-berry sample, some features are smeared out with the mixed technique, thus preventing a correct three dimensional reconstruction. Likewise, the scattering image obtained with the mixed method offers less signal-to-noise ratio because the statistics required for its calculation have less independent data.
Conversely, the red-currant sample, which generates much less scattering than the juniper-berry sample, could be easily reconstructed by using the classic filtered-backprojection algorithm from the different signals extracted with this mixed approach. Reconstruction samples are shown in Fig. 8. This example demonstrates the correctness and greater appropriateness of this approach over the traditional XST technique which, in this particular example, failed to accurately 3D reconstruct the smaller features.
More generally, the results presented in Figs. 3, 6, 7 and 8 demonstrate a better fidelity of the signal recovery with the interlaced XSVT scheme than for the other speckle approaches. In comparison to the traditional XST approach [19], the mixed scheme processing is highly beneficial to the robustness of the phase sensing. Where fast and sharp variations of the sample electronic density generate wave-front features with sizes comparable to the speckle, the direct XST of speckle subsets becomes quickly limited and/or the spatial reduction drastically reduced. In this regard, the interlaced mixed method offers much higher efficiency and reliability albeit at the cost of a slight degradation in spatial resolution with respect to the XSVT method.
This second scheme, integrating both multi-image and multipixel values together in a single vector, allows the statistics necessary for accurate correlation and tracking of the speckle displacements to be improved. Nevertheless, such artificially enlarged vectors do not improve the statistics as much as one would expect by a straightforward calculation based on the ratio of integrated pixels (here nine). The fact that the values of the neighboring pixels are partially correlated results in a statistics gain lower than the number of pixels considered. This can be illustrated by comparing the sensitivity of the two techniques when measuring the standard deviation of the wave-front gradient maps in areas of 100 × 100 pixels in the absence of a sample in the x-ray beam. For the XSVT method of Sec. IV, the measured standard deviation is of ∼0.35 µrad, while it is of ∼0.75 µrad for the mixed XSVT-XST method.
As no additional images are required within this processing scheme compared to the XST technique or propagation based methods, this approach presents a high potential for the imaging of samples for which the dose is a sensitive issue.
VI. CONCLUSION
In conclusion, we demonstrate that the XSVT technique is suitable for high-resolution absorption, phase and dark-field contrast tomography. The several images collected for each projection permit us to overcome the limited spatial resolution of the XST technique, whilst the interlaced scheme and the use of a nonabsorbing scattering membrane allows one to keep the dose at a moderate level. Furthermore, the alternative processing scheme presented in the second part is a very promising way of improving the performance of the XST technique in terms of robustness and spatial resolution. Overall, the XSVT method is expected to find many applications at laboratory sources, synchrotron, and also Comptonaccelerator-based x-ray sources, for samples where the dose and the image resolution are a problematic combination. | 2016-04-20T20:46:46.000Z | 2015-07-21T00:00:00.000 | {
"year": 2015,
"sha1": "3852edb612d797cd54020b8899f44f9f17744257",
"oa_license": null,
"oa_url": "http://arxiv.org/pdf/1507.05807",
"oa_status": "GREEN",
"pdf_src": "Arxiv",
"pdf_hash": "3852edb612d797cd54020b8899f44f9f17744257",
"s2fieldsofstudy": [
"Physics",
"Mathematics"
],
"extfieldsofstudy": [
"Physics"
]
} |
265062815 | pes2o/s2orc | v3-fos-license | Recording Data from Metal-Detecting Activities in the Czech Republic: The Portal of Amateur Collaborators and Register of Individual Finds (AMCR-PAS)
This article discusses the importance of the Portal of Amateur Collaborators and the Register of Individual Finds of the Archaeological Map of the Czech Republic (AMCR-PAS) in preserving data from metal-detecting activities in the Czech Republic. The context and legal framework of metal detecting in the country are explained along with the establishment of AMCR-PAS. The collaboration between amateur and professional archaeologists is also discussed. The recorded finds are analysed in terms of chronological, typological and spatial distribution. The AMCR-PAS system is recognised as a crucial tool for the preservation of data from metal detecting, and provides a valuable resource for national and transnational archaeological research, enabled especially by aggregation of AMCR-PAS data in the ARIADNE infrastructure. Overall, this article highlights the significance of the AMCR-PAS system and its potential to contribute to the understanding of the country's rich archaeological heritage. It also examines the shortcomings and challenges that accompany applying the system in practice.
Introduction
Metal detecting in the Czech Republic emerged as an amateur activity and gradually evolved in the 1990s in tandem with the transformation of social conditions following the fall of the communist regime in 1989.This new phenomenon encompassed a range of activities, from looting cultural heritage to attempts to participate in its preservation.From a legal standpoint, privately conducting metal-detector surveys to search for archaeological finds was considered an illegal activity and there were no methodological interpretations of how such activities should be carried out in collaboration with professionals in the context of the law.There was an ongoing debate about the ethics of amateur detecting and the professional approach to it.Archaeologists were uncertain about the legality of collaboration, had no experience with this form of citizen science, and thought it was out of the question to include it in professional fieldwork (for discussions, see Čižmář 2006;Kuna 2006;Smrž 2006;Vencl 2006;Vích 2006).Nevertheless, cultural heritage was rapidly finding its way into private collections or onto the black market, and archaeologists could only helplessly watch this irreplaceable loss (e.g.Navrátil 2015).Despite the controversies surrounding amateur detecting, individual collaborative platforms, tied to specific archaeologists and often regionally limited, began to be formed to address this unsustainable situation.It took a considerable amount of time for the archaeological community to form a consensus on how to approach this issue, and many cultural heritage artefacts had not been officially recorded or had been completely lost in the meantime.For a general overview of metal detecting in the Czech Republic, see Komoróczy 2022 andMařík 2013.For a review of the current state of metal detecting in Europe and beyond, see e.g.Thomas et al. 2022 andDobat et al. 2020.In 2017, a questionnaire conducted to monitor the attitudes and experience of the archaeological and detectorist communities in the Czech Republic revealed that most archaeologists considered collaboration to be necessary, beneficial and acceptable, and that metal-detecting finds were deemed scientifically significant (Komoróczy 2022).The resulting consensus intensified collaboration between archaeological institutions and the metal-detecting community, which led to more data being generated but still not consistently recorded.At the same time, such collaborations did not meet the legal obligations for conducting archaeological fieldwork, e.g. the submission of fieldwork reports to the archives of the Institutes of Archaeology of the Czech Academy of Sciences in Prague and Brno (IAP/IAB).
Both the IAP and IAB had been working for a long time to cultivate and formalise legal amateur archaeological activities.The archives at these institutes had systematically collected, recorded, and made available the results of archaeological fieldwork in the country for more than a century.In 2021, these two endeavours synergistically resulted in the establishment of a tool called the Portal of Amateur Collaborators and Register of Individual Finds (AMCR-PAS).The portal provides a formal framework for meeting the legal requirements for the announcement of archaeological fieldwork projects based on metal-detecting surveys within the amateur-professional collaboration, submitting results, and collecting, preserving, and presenting data from this collaboration.Such an approach meets the criteria of citizen science, as only data that are collected, preserved and published can be used to further the development of research and benefit the transfer of knowledge to society.
The social aspects of metal detecting are in themselves fascinating.Even though it is primarily a hobby activity, it plays a substantial role in the formation of cultural identity, particularly in relation to local communities.From this perspective, detectorists are an invaluable source of information that helps us to evaluate the social dimensions of cultural heritage.Interaction with the metal-detecting community can be used to better understand this large stakeholder group and formulate strategies for collaboration, communication and the presentation of cultural heritage.
Context and Legal Framework
Relevant Czech legislation cannot be considered as a traditional 'Treasure Act' because, since 1941, the definition of an archaeological find is based not on 'the internal content of the monument, nor its external features [authors' note: the material it is made of, its value, etc.], but the method by which the monument is acquired' (Böhm 1941).The legal definition thus corresponds well with the archaeological definition, when it states that 'An archaeological find is a thing or set of things that provide evidence or relics of human life and activities from the early stages of human evolution to the modern era and was usually preserved underground' (Act No. 20/1987 Coll.).One weakness in this definition is the vague chronological scope as there are multiple ways to understand the term 'modern era'.To clarify this issue, the Ministry of Culture stated that finds originating from WWI and WWII, particularly militaria, should also be considered archaeological finds (Statement of the Ministry of Culture regarding the term 'archaeological find' within the provisions of S. 23, Act No. 20/1987 Coll.).
In the Czech Republic, the law generally stipulates full public ownership of archaeological finds.In practice, this is exercised by municipalities, regional governments and state institutions.Individuals, therefore, are basically prohibited from owning archaeological finds.Only organisations with a valid authorisation issued by the Ministry of Culture and an agreement with the Czech Academy of Sciences (referred to as licensed organisations in this text) are allowed to conduct archaeological fieldwork, including metal-detecting surveys.Metal detecting is, therefore, considered a form of archaeological fieldwork.Conditions for conducting fieldwork, i.e. territorial jurisdiction, essential parts of find reports, etc., are specified in the agreements with the Czech Academy of Sciences (S. 21 of Act No. 20/1987 Coll.).
For all types of archaeological fieldwork, an agreement with the landowner or developer is required (S. 22(1) of Act No. 20/1987 Coll.), and this also applies to all types of surveys (fieldwalking surveys, use of metal detectors, etc.).The use of metal detectors and other tools is not explicitly addressed by Czech legislation and, therefore, not prohibited.To comply with the above rules, individuals who are interested in metal detecting must be affiliated with an organisation licensed to conduct archaeological fieldwork and participate in its specific survey project.As part of establishing such a relationship, amateur collaborators undergo methodical training in the technical and professional standards for recording finds and the duration of the collaboration is set.
As of March 2023, there are 100 licensed archaeological organisations, which include 18 institutions specialising in development-led excavations, 72 national, regional and local museums, 7 universities, and the National Heritage Institute.The territorial jurisdictions of these organisations vary greatly, from those with a licence for the entire Czech Republic (typically universities, National Heritage Institute, IAP and IAB, totalling 8 organisations) to those with a licence for several cadastral territories, as in the case of local museums.The distribution of licensed organisations varies greatly across different regions (Figure 1).For more detailed information on these organisations, including maps with polygons of their territorial jurisdictions and the actual distribution of fieldwork events recorded in the last five years, see the Map of Archaeological Organisations (Pajdla 2022).This tool also allows for a spatial search of licensed organisations based on a given point on a map, making it easier for metal detectorists, builders and other stakeholders to find and contact an organisation active in their area of interest.This is particularly important for metal detectorists, as for various reasons not all organisations are willing to collaborate, and without a concise and updated list it can be difficult, if not impossible, to find an appropriate collaborating institution.
Independent searching for archaeological finds outside of authorised archaeological fieldwork is a criminal offence and can result in fines of up to CZK 4,000,000 (approximately EUR 150,000).However, in practice the fines are much lower and rarely issued, although in the case of metal detecting the penalty usually includes confiscation of the metal detector or other tools used.
AMCR-PAS
The AMCR-PAS schema and module was established as an integral part of the Archaeological Information System of the Czech Republic (AIS CR) infrastructure to address the need to record and preserve data from citizen science collaboration, particularly metal detecting.AIS CR is a large research infrastructure that integrates digital tools, services and resources for Czech archaeology and is jointly operated by IAP and IAB.
The Archaeological Map of the Czech Republic (AMCR, Kuna et al. 2015) is a key component of the AIS CR infrastructure, serving as a comprehensive information system for managing archaeological fieldwork, including reporting on construction activities, fieldwork management and submission of final reports.As of 2021, it also provides an interface for collaborators of licensed archaeological organisations to record individual finds in the context of survey projects.The overall objective of the AIS CR is to ensure the long-term preservation of recorded data and documents, as well as to increase their FAIRness.The AMCR data is published in several ways, first and foremost via the AMCR Digital Archive.Secondly, an OAI-PMH API allows users to access metadata about the finds in the AMCR database in a machinereadable format.
Access to the AMCR data is controlled by various user access groups, each with different levels of permissions.The first group includes anonymous users (anyone on the internet), who have limited access to the system.This group can browse the metadata records and view a selection of documents.The second group is registered users ('researchers'), who have access to view and download most of the documents.The third group is composed of archaeologists from licensed organisations who are eligible to create and edit fieldwork projects, collaborate with amateurs and accept individual finds.The final group is AMCR archivists who have full control over the system.Overall, the AMCR has a strict access control system in place to ensure that sensitive archaeological data is only accessible to authorised individuals.
AMCR-PAS workflows
To record finds under the AMCR-PAS scheme, a user account in the AMCR system is required, which requires identification of the user by their name, email address, and organisation (if available).For metal detectorists, the 'amateur collaborator' option is usually selected as their organisation.All registered amateur collaborators are part of the 'researchers' access rights group, enabling them to establish collaborations with archaeologists, record finds, and view most of the published documents, photographs, etc., in the AMCR Digital Archive.Establishing a collaboration is the first step, which occurs both within and outside AMCR-PAS.It can take time to establish trust between the collaborator and archaeologist and may even result in the collaborator being employed part-time by the licensed organisation or forming another form of association.The collaboration is confirmed or rejected in AMCR-PAS by the archaeologist upon the request of the collaborator.An active collaboration can be terminated by either party at any time, and a collaborator may establish multiple collaborations with different archaeologists simultaneously.This allows detectorists who are active over large territories to submit finds to organisations operating in the relevant area.
Finds in AMCR-PAS are recorded under individual survey projects.A 'Project' is an ontological concept and the resulting record represents a coherent unit, determined by the licensed organisation conducting the fieldwork, stating the purpose of the fieldwork and the approximate area under study.In rescue excavation projects, the incentive is usually a planned building activity reported directly by the investor, while survey projects are created by archaeologists to record finds under the AMCR-PAS scheme.Archaeologists have discretion in defining survey projects, with three main scenarios observed in practice: In general, projects within AMCR-PAS have an annual scope, and at the end of each project, a brief report is submitted detailing the project's progress, the interpretation and importance of the finds, and the modes of cooperation with metal detectorists, among other summarising information.
Once the archaeologist confirms the cooperation and communicates the project identifier to the metal detectorist, finds can be recorded in the system.The process of recording a find involves several stages, outlined in Figure 2. The interface of AMCR-PAS is a web-based application with responsive design (see Figure 3), which allows for directly recording finds on the find spot using a mobile device (mobile phone, tablet, etc.) or later using a computer or a combination of both approaches.A unique identifier is generated for each entry once a project is chosen under which it will be recorded.This identifier can be used to label the find bag or any associated field note entries.A metal detectorist usually fills in only the basic information about the find, while uploading one or more photographs and geolocating the find either by entering coordinates, clicking on the map, or using the built-in location service of the device, typically the GPS of a mobile phone.
The 'Finding Context' section (see Figure 3) of the recording sheet includes the project identifier choice field, the name of the cadastral area, which is automatically detected based on the geolocation, and a field for details on the location.Any local names or plot numbers from the land registry are mentioned if necessary.Furthermore, there are fields for the date, context (surface find, in topsoil, etc.) and depth of the find.
The 'Find' section of the recording sheet (see Figure 3) contains metadata about the discovered artefact, including a controlled vocabulary field for the period (Kuna and Novák 2019), a field for exact dating or additional notes on the period, and controlled vocabulary fields for the type and material of the find.Additionally, there are fields for recording the number of finds, allowing multiple potsherds or similar cases to be recorded as one entry, and a general note field for any other relevant information.While some fields are compulsory for the metal detectorist (period, type, and material), the information is verified and corrected by the archaeologist upon submission of the finds.The recording process of a find is also intended as an educational tool.Metal detectorists can conduct their research to fill in some of these fields correctly (e.g.chronological and functional determination of the finds), which can gradually increase their awareness of the material culture they may encounter during their practice.
Once the metal detectorist finishes recording details of the find, the record can be submitted to the archaeologist for review.We recommend that archaeologists meet with their collaborators at least once a year to review the submitted records and receive the finds.As mentioned earlier, archaeologists can edit or correct most of the fields in the 'Finding Context' and 'Find' sections of the find sheet.For instance, if a collaborator prefers to remain anonymous, the archaeologist can remove them from the name field.
The 'Deposition' section (see Figure 3) is filled in exclusively by the archaeologist.Here, the archaeologist's organisation is recorded as the authority to which the finds were transferred, and where they are deposited.Additionally, an evidence number is recorded to allow for identification of the physical find in the collection with the database record in AMCR-PAS.Finally, the accessibility of the geolocation information is selected, including the name of the cadastral area, notes on the location, geographical coordinates and the position on the map.The accessibility is based on the AMCR user access groups as described above.The majority of finds are released with access to spatial data limited only to archaeologists (approximately 70%), while around 30% are disclosed only to AMCR archivists.After the archaeologist finishes editing the record, it is submitted for archiving to AMCR archivists at IAP and IAB (see Figure 2).The archivist performs a formal check, whether all the recorded information is according to good practice, the photographs are of sufficient quality (etc.), and archives the find.After the record is archived, it is published in the AMCR Digital Archive (see Figure 4).By publishing the finds, the contribution of metal detectorists to the preservation of cultural heritage is acknowledged.The accessibility of the data in the Digital Archive is controlled according to the AMCR user access groups and ensures that sensitive information, especially the geolocation of finds, is only available to authorised users.
AMCR-PAS data in ARIADNE
Archived finds are also periodically aggregated in the ARIADNEplus infrastructure (Richards an Niccolucci 2019) and available through the ARIADNE Portal (see Figure 5).The process of data integration into ARIADNE is based on harvesting data from AMCR through the OAI-PMH API.Data in native XML format is mapped to the unified ARIADNE Catalog Ontology (Felicetti et al. 2023) derived from the CIDOC-CRM standard using the 3M Mapping Memory Manager.All controlled vocabularies are also mapped to the authoritative Getty Art and Architechture Thesaurus based on the SKOS standard using the ARIADNE Vocabulary Matching Tool.This step ensures harmonisation of vocabularies across different datasets in ARIADNE.Similarly, all the chronology data is then mapped to the standard period definitions in the PeriodO portal, which again allows archaeological data of different origin to be presented on a single time scale.All these steps are set up to be repeatedly applied within an automated workflow and the data updated at selected cycles.
Given the ongoing re-implementation of some components of the AMCR, we anticipate that in 2024 there will be further optimisation and improved integration so that the ARIADNE Portal offers the AMCR-PAS data to users in the most understandable, complete and reusable form.It is experience with the ARIADNE aggregation process that has helped to significantly optimise and standardise some of the functions of the current API and some of the data models, which are going to further improve the interoperability of the entire AMCR system, including the AMCR-PAS module, and not only with respect to the ARIADNE services.
First Look at the Data
In this section, we provide a preliminary analysis of data on finds resulting from metal-detecting activities and collaborations between archaeologists and amateurs, as recorded in AMCR-PAS.As the system has been in operation for only a short period, from its introduction in 2021 until mid-March 2023, the insights into spatial, temporal and other aspects are limited but are included to showcase the AMCR-PAS dataset as an emerging source of archaeological data on the Czech Republic.
Collaborators and archaeologists
At the time of writing (mid-March 2023), there are more than 3000 recorded finds in the AMCR-PAS system, with over 50% of these finds already archived, i.e. in the final stage of the AMCR-PAS cycle, and approximately 40% submitted to archaeologists that are waiting for further processing or transfer to licensed organisations.Since the AMCR-PAS launch in 2021, around 800 finds have been recorded annually (see Figure 6) and close to 1200 finds were recorded retrospectively (columns in grey in Figure 6).This highlights the demand for a central solution for preserving and presenting data resulting from collaborations with citizen science contributors from the community.Currently, there are 343 active collaborators in the system and 142 of these have recorded at least one find, i.e. at least 40% from active users.The number of registered collaborators is expected to rise as more organisations adopt the AMCR-PAS system.The majority of active collaborators (90%) work with one archaeologist at most, with 42 archaeologists from 35 organisations having active collaborators and 25 archaeologists having at least one collaborator with one or more recorded finds.
The relationships between collaborators and archaeologists are visualised on a network in Figure 7.The network is divided into different clusters, which mostly have a regional meaning (see Figure 10 with corresponding labels).Clusters A, B (Figure 7) and several unlabelled ones show how the long-term joint support of the IAB and the South Moravian Region regional government influences how archaeologists cooperate with metal detectorists in terms of the number of collaborations and recorded finds.Both the clusters are star-shaped networks with a highly centralised structure around an archaeologist from a museum (Figure 7B) and from the IAB (Figure 7A).Cluster C is an example of two archaeologists from the same organisation (IAP) sharing most of the active collaborators in the Kutná Hora District.Cluster D shows an interconnected network in the mid-Central Bohemian Region with metal detectorists cooperating with multiple archaeologists from different institutions at the same time.The structure of other unlabelled clusters is mostly similar to one of the cases described above, with a prevalence of star-shaped networks centred around a single archaeologist.We presume that the network will grow in size, i.e. the number of nodes, as new organisations and archaeologists join AMCR-PAS, and in the number of connections between different clusters created by metal detectorists collaborating with multiple archaeologists across organisations.The majority of collaborators have recorded between 1 and 20 finds, with an average of around 12 finds per collaborator (see Figure 8).Some highly active metal detectorists have recorded over 100 or even 300 finds, often specialising in longterm surveys of endangered archaeological sites or having a particular interest in chosen regions.The collaborators who recorded a low number of finds were mostly occasional contributors who recognised the importance of submitting both the data and the finds to responsible institutions.The majority of finds were discovered during spring and autumn periods when the fields were bare and allowed for walking surveys with metal detectors (see Figure 9).
Nature of finds
While the previous section focused on the nature of collaborations, in this section we explore the nature of the discovered finds.The spatial distribution of recorded finds follows the activity of archaeologists in different regions (see Figure 10 for a distribution map and Figure 7 for the network of collaboration).The largest concentration of recorded finds is observed in the South Moravian Region, including the Boskovice area where long-term support for collaboration between metal detectorists and licensed organisations is provided by the region's governing body (see Figure 10A and B).Another notable concentration of finds is found in the Kutná Hora District (Figure 10C), as well as in the mid-Central Bohemian Region and Prague (Figure 10D).Despite some emergent concentrations of finds in other regions across the Czech Republic, there are still many areas where no metaldetecting finds have been recorded in the system.Although the focus has been on the collaboration of metal detectorists, it is important to note that not all the recorded finds are made of metal.Non-metallic artefacts, including pottery, lithics, polished stone tools, glass (etc.), make up to 10% of the recorded finds (see Figure 11).Among the metallic artefacts, the majority are made of bronze (65%).This is in accordance with the typological distribution (see Figure 12), where the most commonly found category of dress components is dominated by bronze artefacts.Specifically, fibulae make up 70% of this category.The rest of the distribution of artefact types recorded is relatively balanced between coins, personal adornments, building elements, weapons, tools, etc.The temporal distribution of finds shows a notable predominance of Roman-period artefacts, with another peak in the Late Medieval era and a significant amount of Bronze Age finds (see Figure 13).The disproportionate focus on Roman-period finds is partly the result of the involvement of the Research Centre for the Roman and Migration Periods of the IAB in collaboration with metal detectorists (represented by cluster A in Figure 7).The decline in finds between the Bronze Age and the La Tène period, as well as after the Roman period during the Early Medieval period, is currently challenging to explain.Inevitably, the adoption of the AMCR-PAS system by more organisations in more regions will result in increased discoveries from these time periods.The aoristic sum in Figure 13 represents the modelled number of finds in a given 100-year timeframe, obtained by distributing the probability of a discovery uniformly across each temporal fraction of the considered period.Currently, both the spatial and temporal distribution of finds are not sufficiently representative to be able to draw any large-scale conclusions.
Discussion and Conclusions
Finds obtained through metal detecting over the years since the first individual collaborative platforms were formed have helped to fundamentally change the perception of archaeologists on numerous aspects of past societies.For example, they have significantly enriched our understanding of the extent to which coins were used by the communities of the La Tène period (450-30 BC; Goláňová et al. 2020), or they have brought knowledge of the previously unsuspected variety and quantity of Roman products (e.g.brooches and other fashion accessories or metal vessels) that occur in the Czech Republic, i.e. beyond the borders of the Roman Empire, in the first four centuries of our era (Komoróczy et al. 2017).Although the archaeological community has recognised the benefits of incorporating amateur field activities into a formal framework, the practical implementation of this approach is not without challenges.While amateur collaborators, estimated to be between 15,000 to 30,000 individuals in the Czech Republic, can be of significant help in fieldwork surveys, the limited time capacities of archaeologists often make it difficult to establish a comprehensive collaboration that includes communication, administration, training, regular contact and education.In Czechia, there are approximately 500 professional archaeologists who conduct fieldwork, curate museum collections, educate and present archaeology to the public, and conduct research.Their scarce time is frequently the main reason limiting cooperation with metal detectorists (Komoróczy 2022) and will become a problem on the data curation level as well with the rising numbers of AMCR-PAS users.Most archaeologists engage in collaboration with metal detectorists outside their regular work activities and, therefore, the impact of formalising the processes and providing tools to record finds within AMCR-PAS is significant.Although the tool cannot solve the staffing capacity issue on its own, it provides a framework for collaboration to be carried out.Additionally, the system provides valuable data on which it is possible to demonstrate how cultural heritage is being saved (and endangered).This in turn is important for dealing with official bodies that can address staffing capacity issues.The AMCR Digital Archive where finds are published as the last step in the AMCR-PAS workflow is essential for the personal presentation and satisfaction of the finders by giving them appropriate credit, sharing a cultural heritage with the public and reinforcing a responsible approach to amateur detecting.
Active collaboration between archaeological organisations and amateur collaborators also faces the challenge of inadequate capacity for finds conservation, as the number of finds often exceeds laboratory capacities.With the data available, it is possible to negotiate with competent authorities to find practical solutions.Although the limited capacities of licensed organisations are the largest issue, AMCR-PAS implementation is not without problems either.One of the challenges frequently mentioned by metal detectorists is the requirement to manually enter information about their finds into a web-based database form.They also note that there is currently no dedicated mobile application available to simplify and streamline the recording process.Currently, it is challenging for us to change this owing to the technical background of a project, but there is a possibility for future modifications if/as the number of AMCR-PAS users continues to increase.Furthermore, the refusal of amateurs to cooperate is not an exception, especially with reference to personal freedom and an unwillingness to submit to formal rules.Some of the archaeological organisations still opt not to use AMCR-PAS, since they can submit standard fieldwork reports instead.
While AMCR-PAS cannot solve the structural problems associated with metal detecting, it formalises the collaboration between amateurs and professionals and creates an official platform for recording its results with clear principles that correspond to all national legislative requirements.This comprehensive approach helps to make collaboration transparent and professional, making metal detecting at least a partly beneficial public activity, with results available both to the research community and the general public, rather than a completely illicit and unethical hobby.
In conclusion, this article has highlighted the significance of the Portal of Amateur Collaborators and the Register of Individual Finds of the Archaeological Map of the Czech Republic as an important tool for the preservation of data from metaldetecting activities in the Czech Republic.We have discussed the legal and contextual framework of metal detecting in the country and the role that AMCR-PAS plays in the collaboration between amateurs and professionals after its establishment in 2021 as a part of the national archaeological infrastructure AIS CR.
Although it has been in operation for two years and only some institutions and detectorists have been involved in its use so far, we can clearly assess the benefits and weaknesses that need to be addressed in the future.Through the analysis of recorded finds and collaborations, we have demonstrated how AMCR-PAS is being adopted by the community and presented the quality and coverage of data in this emerging collection.The Czech case of the AMCR-PAS recording scheme paves the way for initiatives in countries with similarly restrictive legal frameworks and could help to address issues associated with metal-detecting activities and promote collaboration between amateurs and professionals in archaeology.Aggregation of AMCR-PAS data in the ARIADNE infrastructure enhances data visibility and promotes transnational cooperation both in preservation and analysis of such data and further promotes transparent metal-detecting workflows and practice.Despite the challenges involved in this collaborative approach, the transparency and coverage of data provided by AMCR-PAS will prove invaluable to both the scientific community and the public for years to come.
Note on Data and Software
Data analysed in the article is deposited in the AMCR repository.A portion of the data that is archived is accessible through the AMCR Digital Archive and the OAI-PMH API interfaces.As most of the exact geographical coordinates of the find spots are protected and accessible only to selected user groups, a full dataset is not published here.Most of the visualisations in the article were created in the R ecosystem for statistical analysis and graphics (R Core Team 2023) using the tidyverse family of packages (Wickham et al. 2019), especially the ggplot2 package (Wickham 2016) for plotting and dplyr (Wickham et al. 2023) for data manipulation.The network was created using the igraph package (Csardi and Nepusz 2006), spatial data were analysed with the sf package (Pebesma and Bivand 2023), treemap charts were created with the treemapify package (Wilkins 2021), and the aoristic analysis was computed using the kairos package (Frerebeau 2022).Spatial layers for the Czech Republic were obtained via the RCzechia package (Lacko 2023).
Figure 1 :
Figure 1: Map of the Czech Republic with the number of archaeological organisations licensed in a given area summarised on a hexagonal grid.Out of the 100 organisations, there are 8 with a licence for the whole territory of the Czech Republic
Figure 2 :
Figure 2: The workflow of recording a find under the AMCR-PAS scheme
Figure 3 :
Figure 3: Screenshot of the AMCR-PAS finds recording interface.The exact location of the find on the map was changed to a random spot
Figure 4 :
Figure 4: Screenshot of finds published in the AMCR Digital Archive
Figure 5 :
Figure 5: Screenshot of a record published in the ARIADNE Portal
Figure 6 :
Figure 6: An overview of finds recovered per year.The years corresponding to finds recorded retrospectively, i.e. found before the AMCR-PAS launch, are in grey
Figure 7 :
Figure 7: A network of active collaborations.The blue squares represent archaeologists from licensed organisations and the yellow circles amateur collaborators.The size of the circle indicates the number of finds recorded by the given collaborator.Chosen regional clusters are labelled the same as on the map in Figure 10: A -South Moravian Region, B -Boskovice area, C -Kutná Hora District, Dmid-Central Bohemian Region
Figure 8 :
Figure 8: A histogram of the number of finds reported by individual collaborators.The strongly right-skewed distribution shows that most of the collaborators recorded between 1 and 20 finds although in some cases, collaborators recorded more than 100.The figure includes finds from all processing states
Figure 9 :
Figure 9: Number of finds recovered in different months showing seasonality in the metal detecting practice
Figure 10 :
Figure 10: Map of the Czech Republic with a spatial distribution of recorded finds (with logscaled colour range).The exact locations of the finds are generalised on a hexagonal grid with a cell size of 25km².The figure includes finds from all processing states.Chosen concentrations of finds are labelled the same as in the network in Figure 7: A -South Moravian Region, B -Boskovice area, C -Kutná Hora District, Dmid-Central Bohemian Region
Figure 12 :
Figure 12: A treemap chart of artefact types recorded in AMCR-PAS
Figure 13 :
Figure 13: Aoristic sum of finds in time with visible peaks.The Bronze Age (yellow), Roman (green) and Late Medieval (blue) periods are highlighted | 2023-11-09T16:09:41.469Z | 2023-11-01T00:00:00.000 | {
"year": 2023,
"sha1": "1537009894aebc87da8c891f69b84618c944c9cf",
"oa_license": "CCBY",
"oa_url": "https://intarch.ac.uk/journal/issue64/13/ia.64.13.pdf",
"oa_status": "HYBRID",
"pdf_src": "Anansi",
"pdf_hash": "78d95953ae8929f48c75ac70b3d4a499d6b1b281",
"s2fieldsofstudy": [
"History"
],
"extfieldsofstudy": []
} |
227037247 | pes2o/s2orc | v3-fos-license | Conditioning Factors of Linearized Wood’s Function Lactation Curve Shape Parameters, Milk Yield, Fat and Protein Content in Murciano-Granadina Primiparous Does
Simple Summary Strategies aiming to improve milk yield and quality are essential to maximizing profitability and dairy goat chain efficiency, which in turn optimizes the commercialization of the products derived. A total of 137,927 official controls were traced and recorded from 22,932 Murciano-Granadina primiparous goats between 1996 and 2016. In this regard, the effects of conditioning factors related to kidding (such as type, year and season) and farm were evaluated to detect the phenotypic sources of variation of milk yield, fat content, protein content and curve shape parameters in Murciano-Granadina primiparous goats to provide useful information for dairy goat early selection. Abstract A total of 137,927 controls of 22,932 Murciano-Granadina first lactation goats (measured between 1996–2016) were evaluated to determine the influence of the number of kids, season, year and farm on total milk yield, daily milk yield, lactation length, total production of fat and protein and percentages of fat and protein. All factors analyzed had a significant effect on the variables studied, except for the influence of the number of kids on the percentages of fat and protein, where the variation was very small. Goats with two offspring produced nearly 15% more milk, fat and protein per lactation compared to goats with simple kids. Kiddings occurring in summer–autumn resulted in average milk, fat and protein yields nearly 14, 19 and 23% higher when compared to winter–spring kiddings. Lactation curves were evaluated to determine the effects of the number of kids and season, using the linearized version of the model of Wood in random regression analyses. Peak Yield increased by about 0.3 kg per additional offspring at kidding, but persistence was higher in goats with single offspring. The kidding season significantly influenced the lactation curve shape. Hence summer-kidding goats were more productive, and peak occurred earlier, while a higher persistence was observed in goats kidding during autumn.
Fixed Effects (Factors) Number of Levels Frequency
1580
Variables and Factors
The variables studied were total milk production (kg), average daily production (kg/day), lactation length (days), protein and fat total production (kg) and protein and fat percentages (%). The fixed effects (factors) studied were farms (n = 184 farms), parity year (n = 20, from 1996 to 2016), number of kids (1, 2, ≥3) and parity season (winter, spring, summer and autumn).
Total Production Calculation
Total milk production, fat and protein calculation were performed using the method by Fleischmann [9]. The estimate of total milk production (MP) is obtained by interpolation in the lactation curve, as expressed below: In which P 1 , P 2 , . . . , Pn are the kilograms of milk (or amount of fat or protein) produced in 24 h of the control day and d 1 , d 2 -d 1 , . . . d n -d n−1 are the intervals, in days, between the delivery and the first control (d1) and between successive controls (d 1 and following), respectively.
Fleischmann method computes total production by adding the cumulative production in different sections, defined by the controls carried out. Thus, the days from birth to the first control are multiplied by the production in the first control to obtain the cumulative production in the first section. In the following controls, the average of successive controls is obtained [for example, (P 1 + P 2 )/2] and multiplied by the interval between controls (d 2 -d 1 ) to calculate the cumulative production between controls 1 and 2. Afterward, this is repeated for the remaining controls. At the end of lactation, production is computed after multiplying the production in the last control by the interval from the last control to drying.
Milk Fat and Protein Content Determination
To calculate the total production of fat and protein in lactation, fat and protein percentages in each control are multiplied by the amount of milk in the same control and standardized using the Fleischmann methodology.
Factor Effect Quantification
The total amount of milk, fat and protein produced per goat and lactation, as well as the average daily milk production, lactation length and fat and protein percentages, were subjected to an analysis of variance (ANOVA) with the PROC GLM of SAS to detect differences in the least-square means across farms, parity years, number of kids and parity seasons.
Curve Shape Parameter Evaluation
The validated data for milk production, fat and protein contents derived from the different production controls were analyzed with the PROC GLIMMIX of SAS to build lactation curves. In this analysis, the parameters of the curve were considered as regression coefficients with a random and normal distribution, without any particular structure (type = un). The curves were estimated individually for each animal (subject = animal). The fixed effects of the number of kids (1, 2, ≥3) and parity season (winter, spring, summer, autumn) were considered. To fit the lactation curve, the (1967) was used. The incomplete gamma function of Wood can be expressed using the following formula: where MP (t) is the production of milk on day t of lactation; a is the parameter related to the level of production at the beginning of lactation; b and c are the growth and decrease rates of milk production, and e is the basis of natural or Napierian logarithms.
To solve the previous equation, the PROC GLIMMIX routine of SAS was transformed into its logarithmic form as follows: The aforementioned estimated parameters were used to calculate the production at lactation peak (PP) and the day on which lactation peak (DP) took place: Additionally, persistence percentage (P%) was estimated as the production in a given month, compared to previous months and calculated for the time intervals (t) = 50-100, 100-150 and 150-200. Then, persistence was expressed as P% 100; P% 150 and P% 200, respectively. Table 3 shows the level of significance, the least-square means, the coefficient of determination (R 2 ) and the residual standard deviation (RSD) of each of the different variables studied across the levels considered in the study. The results for average milk production were 301.03 ± 125.74 kg, with an average milk daily production of 1.44 ± 0.46 kg/day and a lactation length of 206.1 ± 24.7 days. The average fat and protein production percentages were 3.55 ± 0.3% and 5.08 ± 0.7% for protein and fat, respectively. This translated into a total of 10.94 ± 4.45 kg of protein and 15.57 ± 6.4 kg of fat per lactation. The different variables studied closely resemble the normal distribution (p < 0.05); hence a parametric approach was considered. Table 4 shows the results for the least-square means for the different variables studied across the number of kids possibilities (single, twins and triplets). Average total milk production difference of 40 kg was found between singleton and twin-pregnant goats and of 38 kg between twin and triplet-pregnant goats, respectively. Average daily production was 1.34, 1.51 and 1.62 kg/day, and average lactation length was 195.9, 199.8 and 206.5 days for singleton, twin and triplet-pregnant goats, respectively. Fat percentage was 5.17% for singleton and twin-pregnant goats and 5.19% for goats pregnant with triplets or more kids, which corresponded to a total fat production of 13.64 kg, 15.79 kg and 17.96 kg, respectively. Protein percentage was 3.56%, 3.55% and 3.53% for singleton, twin and triplet-pregnant goats, respectively. These values corresponded with a total protein production of 9.43 kg, 10.85 kg and 12.20 kg, respectively. The reduced variation in the percentages of fat and protein across the number of kids resulted in a lack of significant differences in the least-square means across the number of kids possibilities, as shown in Table 3 (p > 0.01). a-c The same superindex letter is indicative of the lack of significant differences in the least-square means for the levels sharing the same initial. Table 5 reports the least-square means values for the variables tested in the study across parity seasons (winter, spring, summer and autumn). Total average productions per lactation of 287.2 kg, 278.5 kg, 329.9 kg and 318 kg were reported in winter, spring, summer and autumn. The maximum difference of 51.4 kg was found between summer and spring. Summer was the season for which maximum productions were reached, while the lowest total milk productions were found in spring. Average daily productions of 1.48, 1.45, 1.50 and 1.52 kg/day were reported for winter, spring, summer and autumn, respectively. Average lactation lengths were 192.4, 188.7, 214.4 and 207.2 days during winter, spring, summer and autumn, respectively. found for spring, summer and autumn, respectively. On the other hand, total protein contents and percentages of 10.04 kg and 3.45%, 9.74 kg and 3.45%, 12.08 kg and 3.66% and 11.45 kg and 3.62% for winter, spring, summer and autumn were reported, respectively.
Effect of Parity Year
Milk production varied across the years, following cyclic oscillations of about 80 kg between the least-square means of the years for which minimum and maximum production was reported ( Figure 1). Maximum average productions ranging between 300 and 350 kg were found for the years 1996,2001,2006,2011,2015. The percentage of fat described an upward trend from 1999 (4.83% on average), reporting increasing percentages until 2011 (average of 5.33%), which stabilized thereafter. Protein percentage described oscillations through the years. The highest percentages of around 3.68% were found during the last years, during which the study took place. Total fat and protein production and average daily milk production reported similar oscillating patterns to those reported for milk production. On one hand, total fat production and fat percentage for winter were 14.38 kg and 4.97%, respectively. The values of 14.08 kg and 4.98%, 17.96 kg and 5.41% and 16.875 kg and 5.35% were found for spring, summer and autumn, respectively. On the other hand, total protein contents and percentages of 10.04 kg and 3.45%, 9.74 kg and 3.45%, 12.08 kg and 3.66% and 11.45 kg and 3.62% for winter, spring, summer and autumn were reported, respectively.
Effect of Parity Year
Milk production varied across the years, following cyclic oscillations of about 80 kg between the least-square means of the years for which minimum and maximum production was reported ( Figure 1). Maximum average productions ranging between 300 and 350 kg were found for the years 1996, 2001, 2006, 2011, 2015. The percentage of fat described an upward trend from 1999 (4.83% on average), reporting increasing percentages until 2011 (average of 5.33%), which stabilized thereafter. Protein percentage described oscillations through the years. The highest percentages of around 3.68% were found during the last years, during which the study took place. Total fat and protein production and average daily milk production reported similar oscillating patterns to those reported for milk production.
Effect of Farms
Average milk production ranged from 220 to 440 kg, while fat and protein percentages ranged from 4.5 to 6.5% and from 3.3 to 4.1%, respectively. As a result, average fat and protein production per lactating goat varied between approximately 9 and 30 kg and 6 and 21 kg, respectively. Frequency histograms for the variables studied across farms are reported in Figure 2.
Effect of Farms
Average milk production ranged from 220 to 440 kg, while fat and protein percentages ranged from 4.5 to 6.5% and from 3.3 to 4.1%, respectively. As a result, average fat and protein production per lactating goat varied between approximately 9 and 30 kg and 6 and 21 kg, respectively. Frequency histograms for the variables studied across farms are reported in Figure 2.
Lactation Curve Shape Parameters
3.6.1. Effect of Type of Kidding Table 6 reports a summary of the average curve shape parameters across the number of kids. The graphical representation of the lactation curves for each number of kids possibility is shown in Figure 3. The linearized Wood function was fitted to the milk records with the parameters associated with initial daily yield (a), rate of increase prior to the peak yield (b) and rate of decrease after it (c). Figure 3 shows the lowest initial production is reached from goats pregnant with singletons (a = 0.65 kg), with this being slightly surpassed by twin-pregnant goats (a = 0.75 kg). Higher initial production was reported for goats pregnant with triplets or more kids (a = 1.1 kg).
Regarding the production in the peak (PP) and the day when the peak is reached (DP), it is observed that in goats with three or more kids, the highest productions are obtained in the peak (PP = 1.86 kg), and the peak is reached earlier (DP = 45), if we compare it with goats with one kid (1.35 kg at 57 d) and with goats with two kids (1.59 kg at 58 d), indicating that the average productions in the peak increase near 0.25 kg for each additional kid.
Lactation Curve Shape Parameters
3.6.1. Effect of Type of Kidding Table 6 reports a summary of the average curve shape parameters across the number of kids. The graphical representation of the lactation curves for each number of kids possibility is shown in Figure 3. The linearized Wood function was fitted to the milk records with the parameters associated with initial daily yield (a), rate of increase prior to the peak yield (b) and rate of decrease after it (c). Figure 3 shows the lowest initial production is reached from goats pregnant with singletons (a = 0.65 kg), with this being slightly surpassed by twin-pregnant goats (a = 0.75 kg). Higher initial production was reported for goats pregnant with triplets or more kids (a = 1.1 kg). Table 6. Average values across the number of kids levels for the parameters of the lactation curve (a, b and c); peak day (PD); peak production (PP); persistence percentage at 100 (P%100), 150 (P%150) and 200 days (P%200). Regarding the production in the peak (PP) and the day when the peak is reached (DP), it is observed that in goats with three or more kids, the highest productions are obtained in the peak (PP = 1.86 kg), and the peak is reached earlier (DP = 45), if we compare it with goats with one kid (1.35 kg at 57 d) and with goats with two kids (1.59 kg at 58 d), indicating that the average productions in the peak increase near 0.25 kg for each additional kid.
Number of Kids
On the other hand, if we go to study the descending phase of the curve, we observe that the slope is less marked and with greater persistence in goats with one kid and two kids, there being few differences between both, in such a way that the persistence at 100, 150 and 200 days, respectively were about 96, 89 and 87% in goats of 1 and 2 kids and of 92, 88 and 86% in goats of 3 or more kids (Table 6).
with initial daily yield (a), rate of increase prior to the peak yield (b) and rate of decrease after it (c). Figure 3 shows the lowest initial production is reached from goats pregnant with singletons (a = 0.65 kg), with this being slightly surpassed by twin-pregnant goats (a = 0.75 kg). Higher initial production was reported for goats pregnant with triplets or more kids (a = 1.1 kg).
Regarding the production in the peak (PP) and the day when the peak is reached (DP), it is observed that in goats with three or more kids, the highest productions are obtained in the peak (PP = 1.86 kg), and the peak is reached earlier (DP = 45), if we compare it with goats with one kid (1.35 kg at 57 d) and with goats with two kids (1.59 kg at 58 d), indicating that the average productions in the peak increase near 0.25 kg for each additional kid. Table 7 shows the data referring to the lactation curve according to the season of kidding. These results are represented in Figure 4. It was observed as the season when the average initial productions are lower is in winter (a = 0.48 kg), followed by autumn (a = 0.73 kg), spring (a = 0.86 kg) and finally the summer (a = 1.36 kg), where the highest initial productions were reached, being a difference between winter and summer of 0.88 kg of milk. The milk production peak of 1.51 kg was reached on day 23 in summer, much earlier than in other seasons. It was followed by spring, for which the highest production peak of 1.47 kg occurred on day 40. Milk production peaks of 1.44 and 1.41 kg were reached on day 68 for autumn and on the 61st day of lactation for winter. Table 7. Average values across kidding season levels for the parameters of the lactation curve (a, b and c); peak day (PD); peak production (PP); persistence percentage at 100 (P%100), 150 (P%150) and 200 days (P%200). Regarding persistence values, parities taking place in autumn presented higher persistence values (99, 93 and 91% at 100, 150 and 200 days of lactation) while parities taking place in spring reported the shortest persistence values (90, 85 and 83% for the same aforementioned stages of lactation). Curves from goats kidding in summer started with average higher yields and reached a production peak shortly afterward. Then, the descending phase is characterized by a smooth slope, which translated into a constantly maintained relatively high production.
Discussion
Average production values in first lactation goats in the present study were lower than the values reported in other studies. For instance, León et al. [10] reported productions of 372.58 kg for the first lactation of goats of the same breed. However, the findings in this study were similar to those by other authors. In these regards, Martínez Navalón and Peris Ribera [11] conducted a study between 1999 and 2002 and reported average production of 309 kg for first-parity goats with lactations of 216 days on average. Similar values were also reported by Pérez [12] Differences between Murciano-Granadina goats milk production and that from other Spanish native breeds have been reported in the literature. For instance, Florida breed primiparous goats milk productions of 412.83 kg in 225 days lactations have been reported [14]. However, García et al. [15] reported higher average yields of 441.07 kg and longer average lactation lengths (276.74 days) for the Florida breed. Malagueña primiparous goats presented comparatively higher average milk productions of 380.9 kg of lactations with average lengths of 256 days [16]. Contrastingly, lower milk productions of 286.85 kg with lactations of 210 days have been reported for the Palmera dairy goat [17]. Among all the Spanish breeds, Payoya production values were the closest ones to those in our study, with values of 314.2 kg and average lactation lengths of 234 days [18].
The values for fat and protein percentages reached the upper margin of the results found in the bibliography for Murciano-Granadina goats. Hence, similar results were found for Murciano-Granadina goats officially controlled in Castilla León. Fat and protein percentages were 4.98 and 3.56% in 2017, 5.33 and 3.55% in 2016 and 4.94 and 3.66% in 2015. Pérez [12] obtained similar results as well for their 10-year-long study involving farms of CAPRIGRAN Y ACRIMUR, results which were also similar to those reported by Martínez Navalón and Peris Ribera [11] who reported a protein percentage of 3.55% and a fat percentage of 4.76% in the same breed.
Compared to first birth goats from other native breeds such as Florida, Malagueña or Palmera, fairly similar fat and protein percentages were found, with small variations occurring in regards to fat percentage. For instance, in Florida, goat fat percentages of 4.9% and protein percentages of 3.54% were found [14]. In Malagueña goats, lower fat percentages with 4.27% and higher protein percentages of 3.64% were reported [16], while the primiparous Palmera breed goats presented
Discussion
Average production values in first lactation goats in the present study were lower than the values reported in other studies. For instance, León et al. [10] reported productions of 372.58 kg for the first lactation of goats of the same breed. However, the findings in this study were similar to those by other authors. In these regards, Martínez Navalón and Peris Ribera [11] conducted a study between 1999 and 2002 and reported average production of 309 kg for first-parity goats with lactations of 216 days on average. Similar values were also reported by Pérez [12] Differences between Murciano-Granadina goats milk production and that from other Spanish native breeds have been reported in the literature. For instance, Florida breed primiparous goats milk productions of 412.83 kg in 225 days lactations have been reported [14]. However, García et al. [15] reported higher average yields of 441.07 kg and longer average lactation lengths (276.74 days) for the Florida breed. Malagueña primiparous goats presented comparatively higher average milk productions of 380.9 kg of lactations with average lengths of 256 days [16]. Contrastingly, lower milk productions of 286.85 kg with lactations of 210 days have been reported for the Palmera dairy goat [17]. Among all the Spanish breeds, Payoya production values were the closest ones to those in our study, with values of 314.2 kg and average lactation lengths of 234 days [18].
The values for fat and protein percentages reached the upper margin of the results found in the bibliography for Murciano-Granadina goats. Hence, similar results were found for Murciano-Granadina goats officially controlled in Castilla León. Fat and protein percentages were 4.98 and 3.56% in 2017, 5.33 and 3.55% in 2016 and 4.94 and 3.66% in 2015. Pérez [12] obtained similar results as well for their 10-year-long study involving farms of CAPRIGRAN Y ACRIMUR, results which were also similar to those reported by Martínez Navalón and Peris Ribera [11] who reported a protein percentage of 3.55% and a fat percentage of 4.76% in the same breed.
Compared to first birth goats from other native breeds such as Florida, Malagueña or Palmera, fairly similar fat and protein percentages were found, with small variations occurring in regards to fat percentage. For instance, in Florida, goat fat percentages of 4.9% and protein percentages of 3.54% were found [14]. In Malagueña goats, lower fat percentages with 4.27% and higher protein percentages of 3.64% were reported [16], while the primiparous Palmera breed goats presented similar fat percentages (5.04%) to those in the present study and higher than 4.44% protein percentages for standardized lactations at 210 days [17].
Compared to the productions reported in the literature for other highly productive European breeds, great differences can be found. For example, in controls carried out in the Alpine breed caprine breed in 2016 in France, much higher milk productions were found for primiparous goats, with 856 kg of milk in lactations with 308 days on average. The milk from these goats also presented lower fat and protein percentages (3.89 and 3.35%, respectively). In the Saanen breed in France, first-kidding goats reached productions of 1000 kg of milk, in lactations with an average length of 337 days and with fat and protein percentages of 3.69 and 3.23%, respectively [19].
Prolificity has been reported to be responsible for remarkable alterations in milk yield and composition. In fact, the tendency to present higher numbers of kids may be the result of selection with adequate reproductive management. In the present paper, higher dairy productions for multiparous goats were found, which was also found in publications by other authors such as Hayden et al. [20]; Salvador and Martínez [21]; Martí Vicent [22]; Pérez [12], which suggests litter size has a positive correlation with milk production. This may be due to a higher volume of the placenta in animals with more than one kid, which determines a higher production of placental lactogen, which is the primarily responsible hormone for the development of the udder in mammals. This influence of the number of goats on production is independent of other factors such as the number of lactations, lactation length, age, body weight, among others.
Additionally, Byatt et al. [23] studied the positive effect of placental lactogen by stimulating milk production in cattle using increasing doses of recombinant placental lactogen. In turn, authors such as Delouis et al. [24], Salvador and Martínez [21], or Martí Vicent [22] identified the pivotal role of lactogen but also suggested it may not be the only hormone responsible for the development of the mammary gland during pregnancy. In these regards, other hormones such as prolactin and fetus-placental and ovarian steroids and other hormones involved in general metabolism may be involved as well.
The number of kids did not affect fat and protein percentages in Murciano-Granadina goats' milk in our study, which agrees with the results by other authors such as Olechnowicz and Sobek [25], Ibnelbachyr et al. [26] and others [21,27]. Oppositely, some authors such as Gómez et al. [28], Ciappesoni et al. [29] and Bagnicka et al. [30] have reported significant differences across the different possibilities within the number of kids factor. For instance, fat and protein levels have been reported to be lower in goats with a higher number of kids as an indirect effect of the increase in milk production. Still, the results of different studies are influenced by different factors, for example, feeding, body condition; hence, it is not easy to distinguish which influences may indeed have a genetic origin from those which may not.
Although component percentages were not significantly different as the number of kids increased, significantly higher means were found for total protein and fat production. These trends were also described for multiple births, Florida goats [31], when a predominant number of female kids was present in litters. As a result, females kidding multiple female kids presented not only higher milk yields, which was also supported by our results, but also higher fat, protein and lactose production. Parity season is one of the most commonly reported conditioning factors to influence milk production and composition. This may be due to the different climatic factors which can affect the animals during each season. The effect of the different factors may become accentuated in animals, which present a marked seasonality, for which temperature, photoperiod, humidity may have a greater impact. Many of these variations are related to pasture curve and food availability during each season, which will influence not only milk production but also lactation length and the shapes which lactation curve describes [32].
Pizarro et al. [33] found similar results to those in the present study on Murciano-Granadina goats and ascribed the lack of significance in literature found for protein and fat contents to studies only recording the information from animals with single and twin-parities, hence the influence of higher numbers of kids may be disregarded and responsible for our findings.
Summer lactations were more productive than autumn lactations, and these, in turn, were better than those of winter, with the worst lactations occurring after the births in spring. This is because late summer births often result in high and longer-lasting initial productions. Furthermore, as seen in Figure 4, even if this occurs, lactation peaks do not reach higher levels above the initial productions and are reached within a few days after the lactation started.
Likewise, the slope of the curve, and therefore the decline of production, is not so sharp. On the contrary, births in spring are followed by lactations characterized by higher peaks, which is due to the positive effect of the abundance of food at this time, which coincides with the higher demands of the animals. Nevertheless, the descending phase of the curve is sharper since it is simultaneous to the summer. During summer months, the quality and quantity of pastures are the lowest of the year, which will also determine these lactations may present the shortest average lactation lengths. Lactations after the births occurring in winter and autumn have intermediate characteristics. These results are very similar to those obtained by Deroide et al. [34]; Pérez [12]; León et al. [10]; Fernández et al. [35]; Verdejo et al. [36] and Carrizosa et al. [37] in Murciano-Granadina goats and Sánchez et al. [38] in Florida goats.
In terms of protein, the highest percentage was obtained in summer, followed by autumn, with the lowest percentages being reported for winter and spring. These results are similar to those obtained by other authors, such as those published by Carrizosa et al. [37], Gómez et al. [28], Fernández et al. [35], Pérez [12] and Deroide et al. [34] with Murciano-Granadina goats. For these authors, the lowest percentages were found for parities occurring during the spring months, with the highest percentages being reported after the parities taking place around the month of September.
The highest fat percentages are reached in summer and autumn, with the lowest ones being reported for winter and spring. Similar results were found in the studies by Gómez et al. [28] and Deroide et al. [34] in which the highest fat percentages were found for births taking place between August and December, while the lowest were reported for the months between January and July.
Such fluctuations are in line with the results by other authors such as Martínez Navalón and Peris Ribera [11] who reported similar fluctuations in average milk productions during the period 1995-2001, also reported by CAPRIGRAN and ACRIMUR associations between 2002 and 2012. Similar results, such as those obtained by Lôbo et al. [39], also reported oscillations in average milk productions over the years that their study took place.
In this context, milk production and average daily production describe fluctuations. In recent years, milk production peaks close to 350 kg have been reached, and then declined. This has translated into maximum production differences of around 80 kg. On the other hand, the fat percentage described an upward trend from 1999 on, reporting an average of 5.35%. For protein percentage, average values of around 3.68% were found in recent years.
Protein percentages also presented an upward trend. However, oscillations were reported for the period of the study. Other authors, such as Byatt et al. [23], found significant differences per year of lactation for fat and protein percentage in milk. Avilés et al. [40] and Martínez Navalón and Peris Ribera [11] also reported an upward trend for fat and protein percentages, both in primiparous and multiparous goats. Likewise, Lôbo et al. [39] reported fluctuations in fat and protein percentages, describing upward trends and reporting the highest percentages found in the literature for the last years comprised in the study. This finding may be ascribed to the effects of the selection of individuals by breeders based on dams' fat and protein indices.
The significant effect of livestock on the variables considered in the present study may be a sign of the underlying variability. Variability found may be ascribed to the large difference of conditions across populations, among other factors, such as operating system, reproductive, food and/or health management or infrastructure. Even microclimatic conditions to which each population is exposed [21] may be responsible for a certain degree of the variability that can be observed. Among the studies in literature, the best representation of interfarm variability was reported by Pérez [12], who concluded that 20% of interfarm variations among the herds comprising CAPRIGRAN and ACRIMUR associations were due to the effect of livestock. Authors such as Palma [41] found variability may depend on the conditions to which each livestock unit is exposed or to the ones which are present where herds are located. In this context, food management may be one of the pivotal factors to explain the greatest interherd variability. Simultaneously, according to Kučević et al. [42] and Sandrucci et al. [43], these significant differences may also occur in milk composition when different herds are compared.
Conclusions
The compositional quality of the milk of Murciano-Granadina goats stands out among the Spanish and International dairy goat panorama. Contextually, even if lower milk yields are reported, higher total contents of protein and fat are present, which may translate into an increased technological value of this milk for the production of cheese and other milk derivatives. Highly prolific goats, responsible for this increased component production, may result from the application of adequate reproductive management seeking the maximization of selection practices. The effect of the different factors may become accentuated in animals, which present a marked seasonality, for which temperature, photoperiod, humidity may have a greater impact. Analyzing the variability of the different parameters studied over the 20 years of study, we found significant differences in both milk production and its composition. This effect may be related to climatic differences, which, as aforementioned, has often been studied through the assessment of the effect of seasons on animals. It may also be due to the progressive incorporation of farms with a lower level of production into the official dairy control. | 2020-11-19T09:17:54.491Z | 2020-11-01T00:00:00.000 | {
"year": 2020,
"sha1": "dde3946587e36347389704762a0265b3cbb23c54",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/2076-2615/10/11/2115/pdf",
"oa_status": "GOLD",
"pdf_src": "Adhoc",
"pdf_hash": "47e67cc319acb1d343be89c0e5f242a066d129de",
"s2fieldsofstudy": [
"Agricultural and Food Sciences"
],
"extfieldsofstudy": [
"Medicine"
]
} |
118937333 | pes2o/s2orc | v3-fos-license | Direct observations of different sunspot waves influenced by umbral flashes
We report the simultaneous presence of chromospheric umbral flashes and associated umbral waves, and propagating coronal disturbances, in a sunspot and related active region. We have analyzed time-distance maps obtained using the observations from Atmospheric Imaging Assembly (AIA) on-board Solar Dynamics Observatory (SDO). These maps show the simultaneous occurrence of different sunspot oscillation and waves such as umbral flashes, umbral waves, and coronal waves. Analysis of the original light curves, i.e., without implementing any Fourier filtering on them, show that the amplitudes of different sunspot waves observed at different atmospheric layers change in synchronization with the light curves obtained from the umbral flash region, thus demonstrating that these oscillations are modulated by umbral flashes. This study provides the first observational evidence of the influence of sunspot oscillations within the umbra on other sunspot waves extending up to the corona. The properties of these waves and oscillations can be utilized to study the inherent magnetic coupling among different layers of the solar atmosphere above sunspots.
INTRODUCTION
Waves play an important role in the heating of upper atmosphere of the Sun. Different features observed over sunspots at different atmospheric heights host a variety of waves, such as the 5-min photospheric oscillations, the 3-min chromospheric oscillations, umbral flashes and waves, running penumbral waves, and propagating coronal waves (see for e.g., reviews by Bogdan & Judge 2006;De Moortel & Nakariakov 2012;Sych 2016). Although these oscillations and waves have been studied for decades, we are still far from understanding the physics behind their origin and the possible coupling among them. It has further been suggested that sunspot waves and oscillations may play an important role in the initiation of solar flares and coronal mass ejections (CMEs), as well as solar wind acceleration (see, e.g., Sych 2016). Recent studies show that sunspot waves may also play an important role in the triggering of coronal jets (Chandra et al. 2015). Jets were triggered during the growing amplitude phase of the waves, however, the cause of such an amplitude increase is still unknown.
Umbral flashes are observed as sudden strong brightenings occurring at random locations in the sunspot umbrae with a period of around 3-min in chromospheric lines and are considered as the first observations of sunspot oscillations (Beckers & Tallant 1969). These are strongly nonlinear oscillations with asymmetric light curves, where the increase in the amplitude is steeper than the decrease, giving it a saw-tooth shape. Such light curves are interpreted as signatures of upward propagating magneto-acoustic shock waves (e.g. Rouppe van der Voort et al. 2003;Centeno et al. 2006). The shock wave nature of sunspot oscillations has also been recently reported in the transition region lines (Tian et al. 2014). The running penumbral waves (RPW) are outward propagating intensity waves with a period of about 5-min and are observed in chromospheric penumbrae girjesh@iucaa.in of sunspots (Zirin & Stein 1972). These oscillations are interpreted as upward propagating magneto-acoustic waves guided by the magnetic field and originate in the lower atmosphere (Bloomfield et al. 2007;Jess et al. 2013).
The relationship between 3-min umbral waves and 5-min running penumbral waves are still not fully understood. While some studies have advocated that they are the different manifestations of a common phenomenon (Christopoulou et al. 2001;Tziotziou et al. 2006;Thomas & Weiss 2008), other studies suggest an unclear relationship between them (Christopoulou et al. 2000;Kobanov & Makarchik 2004;Kobanov et al. 2006). Recently Madsen et al. (2015) have claimed that both umbral flashes and running waves originate from photospheric p-mode oscillations, where umbral flashes were preceding the running waves in both the spatial and temporal domains.
Propagating intensity disturbances along various coronal structures with the period between 3-20 min are ubiquitous in the solar corona (De Moortel 2009;De Moortel & Nakariakov 2012). The loop like structures, which are often rooted in the umbra show outward propagating intensity disturbances with periods around 3-min, whereas those rooted in non-sunspot regions show periods around 5-min (e.g. De Moortel et al. 2002). Furthermore, open plume and interplume structures in the polar region also show outward propagating intensity disturbances with periods around 10-30 min (e.g. Gupta et al. 2010;Krishna Prasad et al. 2011). These propagating disturbances are found to have wave-like properties and are often interpreted in terms of propagating slow magneto-acoustic waves (e.g. Kiddie et al. 2012;Gupta et al. 2012). Although these coronal wave disturbances are ubiquitous in the different structures, observational evidence of their source region is still missing.
Recently, Jess et al. (2012) found 3-min magneto-acoustic waves in the coronal fanloops which were rooted into the photosphere at locations where large-amplitude 3-min umbral dot Figure 1. Analyzed sunspot region observed in different AIA and HMI filters. Inner black contour on the top of HMI continuum image shows the boundary between umbra and penumbra, whereas outer one shows penumbra outer boundary obtained from HMI continuum.The umbra-penumbra boundary is also shown by a white contour in AIA 171 Å for reference. Over-plotted blue contours on HMI continuum show the locations of fanloops observed in AIA 171 Å passband. oscillations were observed. Krishna Prasad et al. (2015) compared the period of amplitude modulation on Fourier-filtered light curves obtained in different atmospheric layers above the sunspot and associated the presence of slow magnetoacoustic waves in coronal loops with the photospheric pmode. Zhao et al. (2016) traced p-mode waves from the photosphere to the corona in active regions using a time-distance helioseismology analysis technique. However, direct observation of any connection or influence among different sunspot waves and oscillations at different atmospheric layers is still missing.
For direct and unambiguous detection, it is mandatory to have excellent wave signal at different atmospheric layers, which is not always the case. Here, we present an observation where sunspot oscillations were strong enough to show the influence of the perturbation caused by one of the waves on the other waves. Previous such analyses have utilized light curves at individual locations at different atmospheric heights and performed co-spatial analysis. However, here we present multi-wavelength analysis on various locations obtained from the time-distance plots. It has helped us to establish a connec-tion between waves in different layers of the solar atmosphere using observations recorded by the Atmospheric Imaging Assembly (AIA, Lemen et al. 2012) on-board the Solar Dynamics Observatory (SDO, Pesnell et al. 2012). We show that umbral flashes influence the propagation of umbral and coronal waves and investigate the characteristics of the different waves with respect to each other. We present the details of the observations in § 2, data analysis and results in § 3, and finally summarize our results and conclude in § 4.
OBSERVATIONS
We have analyzed the multi-wavelength observations of an active region (AR) NOAA AR 11133 observed by SDO on December 11, 2010 between 09:30:00 to 10:15:00 UT. We have used AIA/SDO observations in two of its UV channels (1700 Å, and 1600 Å) and all of its EUV channels (304 Å, 131 Å, 171 Å, 193 Å, 211 Å, 335 Å, and 94 Å). The datasets for UV have a cadence of 24 s, while those of the EUV channels have a cadence of 12 s. We have also used data from Helioseismic and Magnetic Imager (HMI) on-board SDO to provide context. The cadence of HMI data is 45 s. The spatial Figure 2). White contour on the top of AIA 171 Å marks the location of approximate umbra-penumbra boundary obtained from HMI continuum. Overplotted yellow arrows are directed to the two fanloop systems rooted inside sunspot umbra. resolution of both AIA and HMI images are 0.6 ′′ per pixel. The AIA and HMI observations are processed using standard processing software provided in the solar software (SSW) distribution. All the images are co-aligned and de-rotated with respect to the AIA 171 Å image taken at 9:30:00 UT.
DATA ANALYSIS AND RESULTS
The observed AR mainly consists of a sunspot with fanloops emanating from its upper half. Figure 1 displays the AR in different AIA and HMI passbands. The top left panel shows the analyzed active region in HMI continuum. Black contours obtained from the HMI continuum show the approximate locations of umbra-penumbra (inner contour) and penumbra outer (outer contour) boundaries. Over-plotted blue contours show the fanloop configuration as observed in the AIA 171 Å passband.
Umbral Flashes, Umbral Waves, and Coronal Waves
We spotted five bright umbral flashes between 09:46:41 UT and 09:57:05 UT in AIA 1700 Å and 1600 Å passbands, as shown in Figure 2. The over-plotted white-box encloses the region within which the different umbral flashes occur. In Figure 3, we show the location of umbral flashes on AIA 1600 Å, 171 Å, and 211 Å images. We overplot the approximate umbra-penumbra boundary (white contour) obtained from HMI continuum on the top of AIA 171 Å image (middle panel of Figure 3). We find that there are two fanloop systems with their coronal footpoints located at different lo-cations of the sunspot umbra (as marked by yellow arrows in Figure 3). To study the effect of perturbation caused by umbral flashes on the surrounding sunspot waves, we adopt a time-distance analysis technique. We show the location of the artificial slit to be used for time-distance technique in Figure 3. We choose the artificial slit in such a way that it passes through the umbral flashes and also traces a fanloop to observe any influence of flashes on the fanloop. In Figure 4, we show the time-distance maps obtained along this slit in different AIA passbands covering the chromosphere and corona above the sunspot. Maps were obtained by subtracting the background trend of ≈ 8-min running average from each spatial pixel along the time. We tried several ranges of running average windows, and found that 8-min running averages represent the background/trend signal very well. The time-distance maps clearly show the presence of propagating disturbances in the different layers of the sunspot atmosphere. Five umbral flashes at chromospheric height can be seen in the upper panels of AIA 1600 Å and 1700 Å. The white arrow in AIA 1700 Å panel locates the umbral flashes. In the top panels of Figure 4, the yellow dashed lines pass through the approximate location of the umbral flashes, whereas, the white dashed lines pass through the umbral waves. The blue dashed lines show the umbra-penumbra boundary. The timedistance maps clearly reveal the presence of umbral waves emanating from the location of umbral flashes and moving radially outward. Umbral waves are found to be confined to the region between the location of umbral flashes and the umbra- Figure 3. AIA 1600 Å is plotted from 0 ′′ to 8 ′′ whereas AIA 171 Å is plotted from 8 ′′ to 16 ′′ . Intensities are normalized by time averaged variation along the slit length. penumbra boundary i.e., the region between blue and yellow dashed lines in Figure 4. The blue arrow in the AIA 1700 Å panel shows the propagation of umbral waves originating from the location of umbral flashes. We drew several lines on these propagating features and obtained the average slope and standard deviation which provided the wave propagation speed and associated errors. The umbral wave speeds are found to be quite similar (within errors) in different passbands with around 66.1 ± 8.7 km s −1 for 1700 Å, 49.0 ± 7.1 km s −1 for 1600 Å, and 56.7 ± 5.1 km s −1 for 304 Å passbands.
Propagating coronal waves are omnipresent along the fanloop in all the AIA coronal passbands for the observed time duration except in 94 Å, where the signal is too poor to make any conclusive statement. Coronal waves are also detectable for the other fanloops of umbral and penumbral origin (i.e., coronal footpoints co-spatial to umbra and penumbra of the sunspot) as visible in the coronal images of Figure 1. In the bottom panels of Figure 4, we show the presence of coronal waves for AIA 171, and AIA 211 Å passbands propagating along the analyzed fanloop rooted in the umbra. The white dashed lines in the bottom panels of Figure 4 pass through the coronal waves. The coronal wave speeds are found to be around 50.9 ± 4.9 km s −1 for 171 Å, 46.2 ± 5.3 km s −1 for 193 Å, 46.9 ± 3.6 km s −1 for 211 Å, 62.4 ± 9.2 km s −1 for 335 Å, and 44.8 ± 6.2 km s −1 for 131 Å passbands.
The time-distance maps reveal a peculiar noticeable characteristic for the different sunspot waves. We find an enhancement in the amplitude of the umbral and the coronal waves for the duration of occurrence of the five bright umbral flashes. Enhancements in the amplitude of coronal waves, which resulted in the triggering of coronal jets were also observed by Chandra et al. (2015). In order to have a clear picture of the simultaneous amplitude enhancement between different sunspot oscillation and wave modes, we show a combined time-distance map of chromospheric AIA 1600 Å and coronal AIA 171 Å passbands in Figure 5. Cadence of AIA 1600 Å images is 24 s whereas that of AIA 171 Å images is 12 s. Therefore, we interpolated the AIA 1600 Å images to 12 s cadence to create the combined time-distance map. In this map, we plot AIA 1600 Å from 0 ′′ to 8 ′′ and AIA 171 Å from 8 ′′ to 16 ′′ . The resulting map clearly shows an amplitude Figure 4). We choose the locations on the basis of signal strength. The umbral flash location is averaged over 3 ′′ , while the umbral and coronal wave locations are averaged over 1.2 ′′ and 1.8 ′′ respectively. The detailed analysis performed on these light curves are described in the following subsections.
Wavelet Analysis
We obtain temporal intensity variations of the umbral flash region, umbral waves, and coronal waves for locations marked in Figure 4. The time evolution of intensities obtained from various AIA passbands for different sunspot waves are plotted in the top panels of Figures 6, 7, and 8. All these intensity light curves show prominent growth in the amplitude of oscillations for the similar time as that of the occurrence of umbral flashes. In all figures, time runs from 9:30 UT to 10:15 UT.
To obtain the period of these oscillations, we performed wavelet analysis (Torrence & Compo 1998) on all the light wavelet plots which are obtained after considering the white noise in the data. We also obtained first two power peaks from the global wavelet which are printed at the right top corner of wavelet plots. Global wavelet plots for umbral and coronal waves show very similar nature of power distribution near the peak period of ≈ 2.8-min. Results from wavelet analysis reveal the clear presence of ≈ 2.8-min period oscillations for all the three sunspot oscillation and waves over the whole observed duration. However, we also noticed that wavelet powers for this period are not constant and change with time.
In the time range between ≈ 15 − 30 min, wavelet power increases with time for all the three sunspot oscillation and waves, and later decreases. This almost co-temporal increase in wavelet power with time in different waves is suggestive of coupling among them which was also visualized from the time-distance maps in Figure 4. We further refine our findings by obtaining oscillation amplitudes of different wave types shown in top panels of Figures 6, 7, and 8 and plotted in Figure 9. Oscillation amplitudes are obtained with respect to the background signals, which were obtained from 8-min running average of original light curves as previously. Figure 9 clearly reveals a similar pattern of growth in all the oscillation amplitudes. The amplitude of oscillations grew by more than 20% for umbral flashes observed in AIA 1600 Å, whereas that for umbral and coronal waves grew up to ≈ 10% and 5% respectively. We also see a saw-tooth pattern where the amplitude first increases sharply, and later decreases slowly for umbral flash oscillations. This pattern is also visible in umbral and coronal wave amplitudes, however, to a lesser extent. The appearance of the saw-tooth pattern may indicate the propagation of shock waves as suggested by Tian et al. (2014) in the transition region lines. Similarity in the growing amplitude of oscillation, and almost cotemporal appearance of umbral flashes with those of umbral and coronal waves is a strong indication that these waves are influenced by umbral flashes.
To quantify the amplitude growth of these 2.8-min oscillations, we look at the oscillatory power of these waves with time. Since the wavelet transform provides a temporally variable oscillatory power, we obtain the oscillatory power of these waves with time using the wavelet transforms shown in Figures 6, 7, and 8. Henceforth, we obtained the wavelet oscillatory power at around 2.8-min period averaged over the range of 2.3-3.3 min. In Figure 10, we show oscillatory wavelet power for different oscillation and waves. The upper two panels are shown for coronal waves in AIA 171 and 211 Å, middle panels for umbral waves in AIA 1600 and 304 Å, and bottom panels for umbral flashes in AIA 1600 and 1700 Å. On each panel, we over-plot green curves to show the errors associated with these oscillatory power curves. These error-bars are obtained by carrying out the same wavelet analysis on Monte Carlo bootstrapped light curves. In this method, we generate new light curves from the observed one, including point-wise error estimates on the intensities. This is obtained by adding the normalized random distribution of errors to the original light curves. For the purpose, we generated 100 such new light curves. Then we performed the same wavelet analysis to get a measure of the fuzziness in the results due to statistical fluctuations. Respective errorbars on AIA light curves were obtained using routine aia bp estimate error (Boerner et al. 2012). The plots show almost similar power characteristics for all the waves. Given the range of errorbars, we conclude that consistent growth observed in wavelet powers (in the period range 2.3-3.3 min) between 09:44:00 to 10:00:00 UT is real. Thus, findings of almost co-temporal increase of oscillatory power in around 2.8-min period further strengthens our claim of association between umbral flashes and waves, and coronal waves.
Time-Delay Analysis
To further strengthen and understand the probable coupling among different waves and oscillations, we performed a crosscorrelation analysis of these waves for the duration 09:43:00 UT to 10:00:00 UT. The time is chosen such that it covers the time of occurrence of the umbral flashes. This enables us to observe the time lags associated with the maximum correlation co-efficients, and hence, to determine the time delays between different waves. We choose the light curve of umbral flashes obtained using 1700 Å images to perform the crosscorrelation with light curves of umbral flashes observed in 1600 Å and 304 Å, umbral waves observed in 1600 Å, and 304 Å, and coronal waves observed in 171 Å, and 211 Å. Figure 11 displays the results of cross-correlation analysis in terms of correlation coefficient obtained for different time lags. The analysis is performed using standard IDL routine c correlate that finds the correlations amongst the amplitude of oscillations of different sunspot waves and oscillations. Plots reveal around 70% correlation for all the waves with respect to AIA 1700 Å umbral flash oscillations. We observe an increase in time delay corresponding to the peak correlation coefficient as we go from chromospheric umbral flashes and umbral waves to coronal waves. The time delay increases because distance at which light curves were obtained increases for umbral waves and coronal waves with respect to umbral flash location (see Figure 4). However, time delays obtained from AIA 304 Å passband are relatively larger for umbral flash and wave as compared to AIA 1600 Å passband. This may indicate that AIA 304 Å forms at higher atmospheric height compared to AIA 1700, and 1600 Å passbands. Furthermore, we do not find any significant time delays among the coronal passbands. This could be attributed to the fact that emissions in different AIA passbands are coming from the lower temperature components as fanloops are typically of 1 MK temperature (e.g., Ghosh et al. 2017). The significantly correlated light curves observed in chromospheric umbral flashes with umbral waves, and coronal waves, confirm the influence of umbral flashes on umbral waves and coronal waves.
SUMMARY AND CONCLUSIONS
In this paper, we have focused on different types of sunspot oscillations and waves observed at solar chromospheric and coronal heights. We explored the sunspot with AIA 1700 Å, 1600 Å and 304 Å passbands and the fanloop region over it with AIA 131 Å, 171 Å, 193 Å, 211 Å, and 335 Å passbands. We list our findings below: 1. Five bright umbral flashes were identified from AIA 1700 Å, 1600 Å and 304 Å images (shown in Figures 2, and 3). Their locations were found in close proximity to the footpoint of one of the fanloops that were rooted in the umbra (shown in Figure 3).
2. Emergence of umbral waves moving radially outward was observed in AIA 304 Å, 1600 Å, and 1700 Å passbands from the locations of umbral flashes (shown in Figure 4). The amplitude of umbral waves increased during the umbral flashes.
3. Almost all the AIA coronal passbands showed signatures of propagating magneto-acoustic waves along the different fanloop structures of umbral and penumbral origin. However, the fanloop systems that was rooted inside the sunspot umbra showed oscillations with modulations in amplitude (shown in Figure 4). Combined time-distance plot of chromospheric AIA 1600 Å, and coronal AIA 171 Å showed a simultaneous amplitude increase in coronal waves that could be associated with the umbral flashes, and thus, with umbral waves (shown in Figure 5). Hence, the increasing amplitude of the coronal waves could be influenced by the occurrence of umbral flashes. Moreover, the umbral flash light curves, and sometimes (to a lesser extent) umbral waves and coronal waves light curves reveal a clear saw-tooth pattern of oscillations (shown in Figure 9), which can be attributed to chromospheric response to the magneto-acoustic shock due to propagating photospheric p-mode oscillations (e.g., Centeno et al. 2006;Tian et al. 2014).
4. Using wavelet analysis, we obtained periods of oscillation of the different sunspot waves. For all the waves, i.e., umbral flash, umbral waves, and coronal waves, the dominant period was ≈ 2.8-min (shown in Figure 6, 7, and 8). The co-temporal growth of 2.8-min oscillations for all the sunspot waves and oscillations were also suggested by the temporal variation of wavelet power (shown in Figure 10, which shows simultaneous growth in wavelet power for all the sunspot waves and oscillations).
5. The significant correlations among chromospheric umbral flash, umbral waves, and coronal waves with some time delays is an indication of propagation of sunspot oscillation and waves from the lower atmosphere to the upper atmosphere (shown in Figure 11).
The results obtained here provide the first direct observational evidence of the influence of chromospheric umbral flashes on umbral waves and coronal waves. These results are supported by the time-distance maps and simultaneous growth in oscillation amplitudes obtained from the original light curves. Though our results are based on the analysis of original, unfiltered light curves, we also performed the same analysis using the Fourier filtered light curves obtained within the frequency range 5-7 mHz (≈ 2.3 − 3.3 min). The Fourier filtered light curves also yielded similar co-temporal pattern for different sunspot oscillation and waves in the different AIA passbands. Our results point towards the occurrence of a few strong umbral flashes which influence the propagation of all sunspot waves and oscillations observed at different solar atmospheric layers. Hence, we show the effect of chromospheric umbral flashes in the corona. The analysis presented here also provides important findings to understand trigger mechanism of coronal jets. Chandra et al. (2015) suggested that jets were triggered due to increase in the amplitude of waves. This analysis provides the reason for the increase and therefore, important results for initiations of jets.
To further confirm and establish these findings, coordinated observations of sunspots waves and oscillations using simultaneous ground and space-based facilities are essential. The Solar Ultraviolet Imaging Telescope (SUIT; Ghosh et al. 2016) on board Aditya-L1 will provide excellent coverage of photosphere and chromosphere to study the coupling of these waves in more details.
AS thanks IUCAA for providing local hospitality and support during her stay. GRG is supported through the INSPIRE Faculty Award of the Department of Science and Technology (DST), India. DT acknowledges the support from the Max-Planck Partner Group on Coupling and Dynamics of the Solar Atmosphere at IUCAA. VK was supported through NASA Contract NAS8-03060 to the Chandra X-ray Centre. Part of this work was done during collaborative visits that were part of the ClassACT -an Indo-US Centre for Astronomical Object and Feature Characterization and Classification, sponsored by the Indo-US Science and Technology Forum (IUSSTF). AP acknowledges visiting associateship of IUCAA. Authors thank Drs. Dipankar banerjee, David Jess, and Krishna Prasad for helpful discussions. AIA and HMI data are courtesy of SDO (NASA). Facilities: SDO (AIA, HMI). | 2017-12-12T16:33:39.000Z | 2017-10-23T00:00:00.000 | {
"year": 2017,
"sha1": "ce77aa529c278f411de53dcc8b02efdcb330ecf0",
"oa_license": null,
"oa_url": "http://arxiv.org/pdf/1710.08438",
"oa_status": "GREEN",
"pdf_src": "Arxiv",
"pdf_hash": "ce77aa529c278f411de53dcc8b02efdcb330ecf0",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Physics"
]
} |
5786586 | pes2o/s2orc | v3-fos-license | A Novel Artificial Fish Swarm Algorithm for Recalibration of Fiber Optic Gyroscope Error Parameters
The artificial fish swarm algorithm (AFSA) is one of the state-of-the-art swarm intelligent techniques, which is widely utilized for optimization purposes. Fiber optic gyroscope (FOG) error parameters such as scale factors, biases and misalignment errors are relatively unstable, especially with the environmental disturbances and the aging of fiber coils. These uncalibrated error parameters are the main reasons that the precision of FOG-based strapdown inertial navigation system (SINS) degraded. This research is mainly on the application of a novel artificial fish swarm algorithm (NAFSA) on FOG error coefficients recalibration/identification. First, the NAFSA avoided the demerits (e.g., lack of using artificial fishes’ pervious experiences, lack of existing balance between exploration and exploitation, and high computational cost) of the standard AFSA during the optimization process. To solve these weak points, functional behaviors and the overall procedures of AFSA have been improved with some parameters eliminated and several supplementary parameters added. Second, a hybrid FOG error coefficients recalibration algorithm has been proposed based on NAFSA and Monte Carlo simulation (MCS) approaches. This combination leads to maximum utilization of the involved approaches for FOG error coefficients recalibration. After that, the NAFSA is verified with simulation and experiments and its priorities are compared with that of the conventional calibration method and optimal AFSA. Results demonstrate high efficiency of the NAFSA on FOG error coefficients recalibration.
information is provided from external equipment so that the calibration precision is dependent on the accuracy of the external equipment [26]. The high workload and costs of conventional calibration method are also not affordable for low costs applications. Therefore, the focus on FOG error parameters' recalibration to eliminate these drawbacks is always a hot research point.
The OAFSA was utilized for FOG random drift modeling in the navigation applications of AFSA in 2012 by Wang Tingjun [27]. Meanwhile, the OAFSA also used for the real-time ring laser gyroscope bias temperature error compensation in 2014 by Yu Xudong [28]. Moreover, Gao Yanbin has successfully adopted the OAFSA to calibrate the error parameters of FOG and verified the feasibility of OAFSA on FOG error coefficients recalibration [29,30]. However, OAFSA only balanced the exploration and exploitation abilities during the optimization process by the modification on AFs' Visual and Step parameters. Additionally, the secondary initialization method after certain times of OAFSA optimization manually increased the non-autonomous property of the OAFSA. But the structural and computational complexities of OAFSA remain and the AFs' previous experiences are not used for improving the convergence rate. Therefore, solving these issues and letting the NAFSA recalibrate the FOG error coefficients are of great value to improve the overall navigation precision of FOG-based SINS.
The Monte Carlo simulation (MCS) method is a broad class of computational algorithms that relies on repeated random sampling to obtain numerical results [31]. In this research, it is adopted to simulate the NAFSA process for increasing the credibility of FOG error coefficients recalibration results; hence, the computational results are closer to real conditions. Furthermore, it has the priority of reducing workload and costs over conventional expensive and high-precision turntable calibration methods. So the overall advantages of the MCS-NAFSA for FOG error parameters identification are (1) that the algorithm's structural and computational complexities are reduced to release the high computational cost; (2) that the algorithm's convergence rate is improved by adopting AFs' previous experiences during AFs optimization process; (3) that no external reference information is introduced into the identification process; (4) that the high workload and costs in conventional calibration method are decreased greatly; and (5) that the non-autonomous characteristic of OAFSA on FOG error parameters recalibration is avoided. Therefore, the hybrid MCS-NAFSA technique that utilized on FOG error parameters recalibration is the main contribution of this research.
The rest of this paper is organized as follows. In Section 2, the SAFSA and its disadvantages on FOG error parameters recalibration are first presented. Then, the OAFSA and the corresponding secondary initialization method on FOG error parameters recalibration are briefly dedicated. Finally, the NAFSA and its advantages on FOG error parameters recalibration are described with details. Section 3 indicates the FOG error parameters MCS-NAFSA implementation procedures. After that, the MCS-NAFSA FOG error parameters simulation is conducted, and the results are discussed in Section 4. Next, Section 5 demonstrates the FOG-based SINS navigation experiments and discussion with FOG error parameters recalibrated by NAFSA. Section 6 concludes this article.
SAFSA and Its Demerits on FOG Error Coefficients Recalibration
Generally, fish move to the areas that have more food by their individual or swarm search. The AFs model is depicted by prey, swarm, free moving, and following behaviors [1][2][3]. The AFs food consistency degree in specific areas is the AFSA objective function as well as the AFs approach to the maximum food density point. The state of AF i is denoted as vector X = (x1, x2, …, xn), and xi(i = 1, 2, …, n) are the optimization variables. The current food consistency degree of AF i in position X can be expressed as objective function Y = f(Xi). Visual is the sight field of AFs and Step represents the maximum length of each movement. The distance between two AFs in Xi and Xj positions is shown by Euclidean Distance Disi,j = |Xi − Xj|. Moreover, the best AFs position is loaded in bulletin and crowd factor δ(0 < δ < 1) represents the AFs crowd degree within its Visual range.
According to the characteristics of SAFSA, there are some demerits for its application on the recalibration of FOG error parameters [29,30]. The first one is the AFs lack of the application of previous experiences, which would lead the AFs falling into local extremes during the optimization process. For FOG error parameters identification, this demerit would lead the FOG error parameters to non-optimum values. The second one is that AFs lack the balance between exploration and exploitation during the optimization process, which deteriorates the convergence rate and accuracy of AFs optimization. So the FOG error parameters optimization process would require more time to implement the optimum results. The last demerit is that the structural and computational complexities of SAFSA are high, which will cost more memory loads during optimization process. So it is unsuitable for FOG-based SINS with high real-time computational requirements. Therefore, these demerits should be eliminated before the SAFSA is applied to FOG error coefficients identification.
OAFSA and Its Shortcomings on FOG Error Coefficients Recalibration
Usually, the varied Visual and Step parameters are used to improve the algorithm's precision and convergence rate [5,7,18]. Furthermore, the secondary initialization method is also utilized for higher precision FOG error parameters recalibration [29,30]. When the initialization value of parameters Visual and Step are relatively large, the exploration ability of AFSA is enhanced while the exploitation ability is weakened. Conversely, if the Visual and Step parameters are relatively small, AFSA's exploration ability is weakened and the exploitation ability is enhanced. Therefore, a varied Visual and Step parameters are adopted as [27]: Step Step Step where, G and Gmax denote the current iteration times and the preset maximum iteration times, λ denotes the attenuation function, which could balance the exploration and exploitation abilities during the overall optimization process.
Moreover, after dozens of iterations, the indicator function may present a divergence tendency because the AFs fall into the local extreme by unsuccessful prey behavior [29,30]. The former AFs parameters and the optimized FOG error parameters have reached their limits to implement higher precision. In this case, the secondary initialization method is utilized to the AFs swarm and the related FOG error parameters recalibration procedures. Firstly, the variation tendency of indicator function is observed until it presents divergence tendency, and the AFs parameters and FOG error coefficients are stored in the lowest indicator function point. Secondly, the changed AFs parameters are reloaded and the former saved FOG error coefficients are reloaded manually, which is obtained from the former optimization process. Finally, the AFSA optimization process is executed again to reach higher optimization precision.
However, when OAFSA is used for FOG error parameters identification, only the second drawback of SAFSA is eliminated, but the other two drawbacks are not avoided during the optimization process. Moreover, by inducing the secondary initialization method, the lowest indicator function point selection at first stage is artificially aided. The reload process of AFs parameters and FOG error coefficients are completed manually. This means the method is non-autonomous during the optimization process. Therefore, there are also some shortcomings when OAFSA is applied to FOG error parameters identification.
NAFSA and Its Advantages for FOG Error Parameters Recalibration
NAFSA was first proposed and used in data clustering by Yazdani in 2013 [19]. It solved these mentioned disadvantages by improving AFSA's functional behaviors and overall procedures with some AFs parameters eliminated and several supplementary parameters added. More details on the improvements of AFSA are shown in the following NAFSA parameters and behaviors introduction parts.
Parameters of NAFSA
Suppose there are N AFs in D-dimensional space, the position of AF i could be denoted as vector Xi = (xi,1, xi,2, …, xi,D). AFs Visual could be expressed as vector Visual = (v1, v2, …, vD), the Visual dimensions are determined by the inner coverage of searching space. Therefore, NAFSA could use different Visual in various space ranges. The components of vector Visual are divided into many parts that make the AFs have better global optimization ability. So NAFSA has higher precision in global extreme optimal ranges. Moreover, the Contraction Factor (CF) parameter is utilized to substitute Step and crowd factor parameters. It is introduced to NAFSA for choosing different Visual values in different optimal process, and CF is an integer less than 1, whether a constant or a function. Previously, the inertial weight parameter was applied in PSO for balancing the exploration and exploitation abilities during optimization process [32]. The CF in NAFSA has the similar function to inertial weight in PSO. Here, random function is adopted to generate CF in all iteration process: The above equation generates a random CF in [CFmin, CFmax]. Therefore, the ith element of vector Visual in next iteration could be expressed as: Next, the NAFSA behaviors will be discussed.
Individual Behavior
Individual behavior is made up of prey and free moving behaviors. The AF i in position Xi(t) tries several times of movement to better position. In each iteration process, AF i will occupy the position Xj(t) by prey behavior, and then evaluate their food density. If f(Xi) ≥ f(Xj), then the next position is expressed as: Because the position Xj(t) is within the Visual range of AF i, the move distance of AF i would be less or equal to Visual vector in the same dimension. If f(Xi) ≥ f(Xj), the AF i will move to a better position with several iterations by Equation (6) or by prey behavior and Equation (6). However, if f(Xi) < f(Xj), the AF i would not move towards Xj(t) and it will find a better position from its previous position. Therefore, for single individual behavior, the AF could find better position by trying several times. Otherwise, if AF i could not find a better position after all attempts, the AF could move one Step randomly within its Visual range: In NAFSA, each AF moves towards better position by individual behavior. But when an individual fails, it will perform random behavior in its Visual range and may discard its previous position, which may find a worse position in the searching space. Nevertheless, in order to keep the AFs swarm diversity and find better position in later optimal behavior, performing the random behavior is necessary for the AFs swarm. Moreover, the AFs position search through random behavior would not be used as best AFs position, so the best AFs position would not be lost even if AFs could not find a better position. In this case, the best AFs position is what has been searched previously. Therefore, in NAFSA, the current AFs position is the best position, so the bulletin parameter in SAFSA is no longer a necessity.
Group Behavior
Group behavior performs instead of following and swarm behaviors. Keeping all AFs swarm characteristics and making AFs movement in the best position are two main targets in group behavior. The center position of AFs swarm is obtained by swarm behavior. If f(XCenter) > f(Xi), then the next position of AF i is: , AF i could not move towards the center position, while moving towards the best position in the searching space: Therefore, the AFs in a worse position would move towards center position by comparing with the center position. When the position is better than center position, it will move towards the best AFs swarm position. Therefore, all the AFs will reach the best position by performing group behavior. Consequently, in NAFSA, the best position searched by fish swarm would be adopted to accelerate the convergence rate with all AFs movement. So the group behavior is used to maintain the fish swarm characteristics and avoid reducing swarm diversity.
In group behavior, the center AFs position may have better food density (indicator function) than the best AFs position. AFs move towards center position by Equation (8), but a worse position may exist between the current position and center position. The AFs position may then get worse or even lose their best position by executing Equation (8). Therefore, if the indicator function of center position is better than the best AFs position, the best AFs position is determined by the following equation: The above equation executes only when f(XCenter) < f(XBest), while the other AFs movement by Equation (8) helps to maintain the diversity of fish swarm.
Advantages of NAFSA for FOG Error Parameters Recalibration
According to the introduction of NAFSA, there are three main advantages for its applications with FOG error parameters recalibration. The first one is that the parameters reduction and behavioral simplification of NAFSA reduced the structural and computational complexities, which means the NAFSA will cost less memory loads when it is used in FOG error parameters recalibration. The second one is that the choice of CF parameter can balance the exploitation and exploration abilities during AFs optimization process. Additionally, CF parameters could avoid the local extreme during the AFs optimization process, so the NAFSA could neglect the secondary initialization method in OAFSA and implement the autonomous characteristic when it is adopted in FOG error parameters recalibration. The third one is that the CF parameters would reveal the application of AFs previous experiences, which equates to a faster convergence rate during the AFs optimization process. It could implement the FOG error parameters identification with less time and higher precision. Therefore, the NAFSA is more suitable for FOG error parameters recalibration when it is compared with the previous OAFSA.
FOG Error Coefficients Recalibration by NAFSA
In this section, the tri-axial FOG static error model will be provided at first. And then the optimization indicator function derivation process is presented. Finally, the FOG error coefficients identification procedures by NAFSA will be demonstrated specifically.
FOG Static Error Model
The purpose of error coefficients recalibration is to identify the FOG error parameters accurately, quickly and steadily. There are various recalibration methodologies for FOG error parameters [33][34][35]. The static error model of tri-axial FOG is shown [35]: where, Kgi(i = x, y, z) denote the FOG scale factors; Egij(i, j = x, y, z; i ≠ j) denote the FOG misalignment errors during installation; Ngi(i = x, y, z) denote the FOG output data; ωi(i = x, y, z) denote the turntable alignment axis input angular rate; ωio(i = x, y, z) denote the FOG biases. Therefore, there are 12 static error parameters for tri-axial FOG to be identified in total.
Derivation of the Optimization Indicator Function
The NAFSA is terminated in one of three conditions. The first is when the maximum number of iterations is reached. The second condition is when the optimization indicator function is below a pre-defined threshold during the optimization process. The third condition is that when performing the next iteration, the deviation of the current iteration result and the next iteration result is within an acceptable range. The optimization indicator is a key factor for the terminate condition during the NAFSA optimization process. The following part presents an optimization indicator function for FOG error parameters identification based on NAFSA.
Theoretically, when the static tri-axial FOG at arbitrary space position, the FOG measured angular rate information would satisfy the following equation: where, ωi(i = x, y, z) are tri-axial FOG theoretical input angular rates; ωie = 15.0411°/h denotes the Earth rotation angular rate, which is a constant vector along the Earth rotation axis. Actually, because of the errors caused by FOG itself, the calculated angular rates are different from theoretical values. Therefore, the angular rate mode square error (MSE) is adopted to represent the deviation, which is derived from Equation (12) and expressed as: where, ˆ( , , ) i i x y z denote the angular rates calculated from Equation (11) with the stored FOG output data Ngi(i = x, y, z).
The target of identifying the steady FOG error coefficients is to make the angular rate MSE as stable as possible. So the standard deviation function is utilized to evaluate the discrete degree of FOG error coefficients: In Equation (14), M denotes the number of positions during optimization process.
MCS-NAFSA Implementation Procedures
The implementation procedures of FOG error coefficients identification are demonstrated in this section. Two main steps are indicated to illustrate the NAFSA optimization process. At the beginning, the variation characteristics of the 12 total error coefficients in the tri-axial FOG are discussed and a clustering process is described with different parameters. After that, the specific MCS-NAFSA FOG error coefficients identification procedures are presented step by step.
FOG Error Coefficients Clustering
In the identification of 12 tri-axial FOG error coefficients, different error coefficients have different influences on the angular rate MSE, and also the NAFSA requires all the AFs to have similar characteristics during the optimization process. Therefore, FOG error coefficients clustering is a necessity before the FOG error coefficients can be optimized by NAFSA.
Thinking about the different error coefficients' influences on angular rate MSE and based on our previous experiences, the FOG scale factors have the highest impacts on angular rate MSE, followed by the biases, and the last parameters are FOG misalignment errors, so the FOG error coefficients are divided into three categories. They are: three FOG scale factors Kgi(i = x, y, z) as category one, three FOG biases ωio(i = x, y, z) as category two, and six FOG misalignment errors Egij(i, j = x, y, z; i ≠ j) as category three. Therefore, when adopting MCS-NAFSA to identify the FOG error parameters, there are three main steps of optimization process should be conducted to implement the highest precision.
MCS-NAFSA FOG Procedures
Through the analysis in Section 3.3.1, within NAFSA FOG procedures, the FOG error coefficients identified by NAFSA are mutually independent in different categories. Hence, three phases of optimization process pseudo-code is shown in the algorithm FOG NAFSA.
In the first phase, FOG scale factor Kgi(i = x, y, z) identification is optimized by NAFSA. Firstly, the AFs parameters, category two parameters ωio(i = x, y, z) and category three parameters Egij(i, j = x, y, z; i ≠ j) are all initialized. After that, each AF i performs Individual behavior and moves to a better position based on the outcome. Subsequently, each AF i executes Group behavior with respect to their new position. Finally, this process is repeated for N times, and we could calculate the mean value Kgi_m(i = x, y, z) as FOG scale factors.
In the second phase, FOG bias ωio(i = x, y, z) identification is indicated by NAFSA. At the beginning, the AFs parameters, category three parameters Egij(i, j = x, y, z; i ≠ j) and optimized FOG scale factors Kgi_m(i = x, y, z) are all loaded. Next, all the AFs execute Individual behavior and Group behavior respectively. At last, this process is repeated for N times, and we could obtain the mean value ωio_m(i = x, y, z) as FOG biases.
In the third phase, FOG misalignment error Egij(i, j = x, y, z; i ≠ j) identification is demonstrated by NAFSA. At first, the AFs parameters, optimized FOG scale factors Kgi_m(i = x, y, z) and biases ωio_m(i = x, y, z) are all loaded in initialization process. Second, all the AFs execute Individual behavior and Group behavior, respectively. Finally, this process is repeated for N times, and we could acquire the mean value Egij_m(i, j = x, y, z; i ≠ j) as FOG misalignment errors.
Finally, the three stages above are repeated until the FOG error coefficients meet the terminate conditions, when optimization indicator reaches σ < 10 −8 , or the number of iterations reaches a certain preset number.
Simulation and Discussion
In this section, the FOG error coefficients simulation is conducted by MCS-NAFSA. Before the simulation, the AFs parameters and the non-optimized FOG error parameters at each phase should be preset. Subsequently, the simulation on FOG error parameters is shown by MCS-NAFSA.
Simulation Parameters Preset
The Section 2.3.1 described all the AFs parameters during optimization process. All the preset AFs parameters of tri-axial FOG before the FOG error parameters optimized are listed in Table 1. Meanwhile, in Section 3.3.2, when one category of FOG error parameters are identified by NAFSA, the other two categories' parameters also have impacts on angular rate MSE. Therefore, the preset FOG error coefficients are shown in Table 2. It is worth noting that, in each of iteration, the dimensions of vector Visual listed in Table 1 are equal to the FOG parameter number. So the vector Visual dimensions on FOG misalignment errors are different from FOG scale factors and FOG biases. CF is a positive number <1, so that its minimum value and maximum value are preset as 0.000001 and 0.999999, respectively. In Table 2, in order to reduce the other two categories' FOG parameter influences on angular rate MSE during one category of the FOG error parameters optimization process, the preset FOG error parameters are based on the conventional 24-position calibration method [36,37], which is aided by expensive and high-precision turntable in indoor environment.
Simulation Results and Discussion
After the presetting of all the parameters (i.e., the initialization process) is completed, all AFs start to execute the NAFSA optimization procedures. In order to increase the FOG error parameters' degrees of credibility during the NAFSA optimization process, the random factors are introduced by conducting MCS 100 times after the single NAFSA optimization. Figure 1 shows the three FOG scale factor NAFSA identification curves during the 100 times of MCS. Meanwhile, Figure 2 demonstrates the three FOG bias NAFSA identification curves with 100 times of MCS. Figure 3 presents the six FOG misalignment error NAFSA identification curves with 100 times of MCS. From Figures 1 to 3, all the FOG error parameters fluctuate during the MCS process within a relatively small range, and the results revealed that the FOG error parameters are influenced by their usage environments. Moreover, the precision of simulation results is also likely to be deteriorated by random factors, such as algorithmic error and computer error, which are easily being neglected and unable to be eliminated during a single NAFSA optimization. So, in this phase, in order to reduce the effects of random factors, the data smoothing method is adopted to calculate the mean value of the MCS results. After being processed by the data smoothing method, all FOG error parameters simulation results are listed in Table 3. Simulation Times
Egzy(rad) Egzy
Through the comparison between the preset FOG error parameters and the NAFSA identification results in Table 3, we can summarize that the relative error amplitudes of each parameter are substantially small and the credibility of identification results is enough for the FOG-based SINS navigation requirements. FOG scale factors Kgi(i = x, y, z) relative error magnitudes are small enough at 10 −2~1 0 −3 ppm, to meet the high-precision navigation requirements completely. The relative errors of FOG biases ωio(i = x, y, z) and misalignment errors Egij(i, j = x, y, z; i ≠ j) could also reach 10 −3~1 0 −4 in magnitudes which are also completely enough to satisfy the FOG-based SINS precision demands. However, in OAFSA, the FOG scale factors relative error magnitudes are 10 −1~1 0 −2 ppm and the FOG biases and misalignment errors' relative error magnitudes are 10 −2~1 0 −3 [29,30]. Therefore, theoretically, the precision of NAFSA identification results is an order of magnitude higher than OAFSA in FOG error parameters identification.
Moreover, the standard deviations of the estimates are indicators to show the stability of the estimates [38,39]. In Table 3 In Figure 4, the red dotted curve denotes the SAFSA indicator function variation tendency. After 20 iterations, the indicator function has a slight increase in tendency, and it remains stable after 24 iterations, with the indicator function reaching 0.002584(°/h) 2 . However, the OAFSA's indicator function with the blue star curve has a faster convergence rate than SAFSA, but the indicator function begins to diverge after 20 iterations. The secondary initialization method is adopted in 30 iterations to decrease the indicator function and to improve the convergence precision. For comparison, the NAFSA optimization indicator function is shown with the black plus curve in Figure 4. We can conclude that the NAFSA has a better convergence rate than the OAFSA and SAFSA because of the reduction of the algorithm's structural and computational complexities. Moreover, it is evident that the NAFSA indicator function is always convergent during the optimization progress, which is due to the usage of the previous experience of AF. Therefore, the NAFSA has better performance in convergence rates and reliability of the optimized results compared to SAFSA and OAFSA.
Experiments and Discussion
To validate the feasibility and priorities of the NAFSA on FOG error parameters optimization, the static and dynamic navigation experiments were conducted, respectively. Before these two experiments, the FOG error parameters were calibrated by using a turntable with 24-position method. The navigation information output results are also based on the FOG error parameters that are calibrated by this 24-position method. Additionally, for comparison with OAFSA and NAFSA in navigation experiments, the stored experimental data were also used for navigation mechanization with FOG error parameters identified by OAFSA and NAFSA.
For both experiments, the FOG-based SINS was developed by Inertial Navigation and Measurement & Control Technology Institute at Harbin Engineering University. The main performance indicators of FOG are demonstrated in Table 4. Figure 5 shows the FOG and the FOG-based SINS in experiments. Table 4. FOG performance indicators.
Experimental Procedures and Data Processing
In this section, a static navigation experiment is carried out. At the beginning, the FOG-based SINS and the corresponding monitor are installed on the marble benchmark that is used to eliminate external disturbances on system positioning precision. Then, the SINS is started, the turntable calibrated FOG error parameters and the initial navigation information (initial position and velocity) are loaded. After that, both the inertial measurement unit (IMU) output and the navigation information for 24 h are stored after the SINS completes the initial alignment process.
After obtaining the stored 24 h IMU output and navigation information, we first used the FOG data to recalibrate the FOG error parameters by OAFSA and NAFSA, respectively. Second, the navigation mechanization process was conducted again with the FOG error parameters optimized by OAFSA and NAFSA, respectively. Finally, the positioning error curves were plotted and the positioning error numerical results were obtained with the three methods introduced.
The positioning error is calculated by [40,41]: where, long0 and lat0 are the initial longitude and latitude of the SINS, and long and lat are the calculated longitude and latitude. R denotes the radius of Earth. Figure 6 shows a comparison of positioning errors in 24 h static navigation experiment when the FOG error parameters are identified by the conventional calibration method, OAFSA and NAFSA, respectively. The red dotted positioning error curve represents the OAFSA FOG error parameters identification results. Additionally, the blue dotted curve represents the positioning error curve with FOG calibrated by conventional high-precision turntable method. Both curves present positioning precision of 4.5 nautical miles in 24 h static navigation, which shows that the OAFSA could substitute the conventional calibration method without using high-precision turntable [29,30]. Moreover, it is worth noting that the black solid curve in Figure 6 denotes positioning precision of the NAFSA on FOG error parameters identification. The curve's tendency demonstrated that after 5 h of navigation, the positioning error is lower than the other two methods and the precision is about 0.3 nautical miles better than the OAFSA in one day of navigation.
Experimental Results and Discussion
The corresponding numerical results of static positioning errors with the three different methods are shown in Table 5. Both the conventional calibrated and the OAFSA recalibrated FOG-based SINS have about 4.5 nautical miles positioning error in 24 h. Meanwhile, the NAFSA recalibrated FOG-based SINS has 4.255 nautical miles of positioning error. Therefore, the static navigation experiment demonstrates that the NAFSA recalibrated FOG-based SINS is superior to that of the conventional calibrated and the OAFSA recalibrated FOG-based SINS. Figure 6. The comparison of positioning error with three methods.
Experimental Procedures and Data Processing
In order to validate the feasibility and priorities of the NAFSA in real application conditions, a lake navigation experiment was also conducted in Qiandao Lake for a period of time. Firstly, the FOG-based SINS and the reference system, the difference global positioning system (DGPS) receiver, were installed in a ship. Secondly, after the FOG-based SINS finished the mooring alignment process at the starting point, the ship sailed successively with speed change, heading change, manoeuvres, etc. At the same time, the reference DGPS information, IMU data and the self-developed FOG-based SINS navigation information were all collected and saved. Finally, the data was processed the same way as Section 5.1.1. The trajectories of the lake experiment with GPS and the SINS when FOG parameters are identified by three different ways are all demonstrated in Figure 7. Moreover, the numerical results of the system positioning errors in both the North and East directions are calculated and listed in Table 6.
Experimental Results and Discussion
On one hand, Figure 7, shows that the conventional calibration method, the OAFSA identification method and the NAFSA identification method on FOG error parameters all have the ability to implement the FOG error parameters calculation and reach different degrees of positioning precision in the lake experiment. On the other hand, the green curve shows that the NAFSA recalibration method is superior, such as better robustness when speed and heading change, better tracking capability during the whole navigation process, which means higher positioning precision. Moreover, by utilizing the NAFSA on FOG error parameters identification, some lower precision SINS would have better performance for parameters identification after a specific period of navigation.
The lake experiment positioning errors compared with reference solution at the end of the navigation are listed in Table 6. We found that the conventional calibration method and the OAFSA identification method have almost the same positioning errors. The conventional calibration method has a North direction positioning error around 5.1154 m and East direction error about 20.0253 m. The OAFSA identification method has a North direction positioning error around 5.2131 m and an East direction error about 20.3580 m. While the NAFSA identification method has better performance in terms of positioning error, with a North direction positioning error of 5.0134 m, and East direction error of 8.1689 m. By comparing the North direction positioning errors with these three methods, the NAFSA method has only a slightly smaller positioning error. Furthermore, the NAFSA recalibration method could improve the East direction positioning error of the conventional calibration and OAFSA recalibration methods from about 20 m to 8.169 m, which could clearly demonstrate the priorities of the NAFSA recalibration method. Therefore, the NAFSA recalibration method is a more powerful choice in its engineering application for FOG error parameters recalibration.
All in all, in both experiments, the NAFSA recalibration method has advantages in workload and costs compared to the conventional calibration method. However, it presents better performance in long-term navigation precision and is more acceptable for actual engineering applications than previous OAFSA recalibration methods, which is mainly due to the lower structural and computational complexities and faster convergence rate of the NAFSA recalibration method.
Conclusions
After the FOG-based SINS operated for a period of time, the FOG would be vulnerable to the working environmental disturbances, such as gravitational field, magnetic field and thermal field, which cause nonreciprocal phase shifts except for the rotary movement by the vehicle itself. These exterior disturbances could influence the FOG error parameters' stability directly or indirectly. Even though some advanced measures are taken to eliminate these effects, high-precision navigation application is far from enough.
This research work is based on one of the swarm intelligence algorithms, NAFSA, focusing mainly on its combination with MCS and utilization in FOG error parameters identification. The NAFSA has the advantages of lower structural and computational complexities and higher convergence rates than the previous OAFSA recalibration method during the optimization process. It also has lesser workload and costs requirements than the conventional FOG error parameters calibration methods. Furthermore, the non-autonomous property could be avoided when compared with the previous OAFSA recalibration method. Therefore, the NAFSA FOG error parameters recalibration method could implement longer recalibration interval time with higher precision in some harness application environments.
When the FOG-based SINS applied in navigation conditions, NAFSA-identified FOG error parameters could realize the SINS navigation process rapidly and accurately. Moreover, the NAFSA-identified FOG error parameters have better environmental adaptive ability, which means higher positioning accuracy and better tracking performance. Therefore, the NAFSA recalibration method has better ability than the conventional calibration method and the previous OAFSA in FOG error parameters recalibration application.
However, the AFSA on FOG error parameters recalibration is only in an exploratory phase and all the navigation experiments are based on the stored data. Thus, our work for the next stage is to realize the algorithm in real-time navigation. | 2015-09-18T23:22:04.000Z | 2015-05-01T00:00:00.000 | {
"year": 2015,
"sha1": "5d7e3858eecf1963104aa5138e8c0ebd37972141",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/1424-8220/15/5/10547/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "7846b554c6d627b280d8b85cb1e221dee6306f7a",
"s2fieldsofstudy": [
"Computer Science"
],
"extfieldsofstudy": [
"Engineering",
"Computer Science",
"Medicine"
]
} |
16565516 | pes2o/s2orc | v3-fos-license | Fungal endophytes in spotted knapweed influence its competitive interactions
Fungal symbionts are often overlooked in studies of plant invasion. Nevertheless, their role could be essential to the competitive success of the invader. We studied fungal endophytes in the widespread invasive Centaurea stoebe (common knapweed). A preliminary experiment showed that endophytes in roots of C. stoebe significantly reduced the biomass of evolutionarily na\"ive neighbours (Festuca idahoensis), compared to endophyte-free C. stoebe. In the main experiment non-clavicipitaceous endophytes belonging to six phylotypes, were employed as root inoculants. Each of these endophytes again reduced the growth of na\"ive neighbours (F. idahoensis); and remarkably, each also increased the growth of adapted neighbours (F. ovina) that were tested for the first time. Four of the six endophytes caused C. stoebe to gain a competitive advantage over its na\"ive neighbour that was significantly greater than the endophyte-free C. stoebe over that same neighbour. However, endophyte-free C. stoebe had no greater competitive advantage over F. idahoensis than it had over F. ovina. Therefore, plant-plant interactions were dramatically affected by the presence of endophytes in a way that would favor invasion.
Introduction
In plant invasions, a primary challenge is to understand the superior competitive ability of a successful exotic plant. Typically a successful invader is both less competitive and less abundant in its native range; this range-dependent puzzle of invasiveness is central to invasion biology. Although the contribution to plant invasions of release from fungal pathogens is well known (Mitchell & Power, 1991), studies of the contributions of endophytes have been initiated only recently (Addy et al., 2005;Faeth et al., 2004;Omacini et al., 2006;Rudgers et al., 2005;Rudgers & Orr, 2009).
Recently, Rodriguez and co-authors used symbiotic criteria to group fungal endophytes of plants in four classes (Rodriguez et al., 2009). Class 1 endophytes belonging to the Clavicipitaceae are well known to ecologists as grass symbionts (Clay, 1988), and the pioneering investigation showed that Neotyphodium caenophialum promotes plant invasions (Rudgers et al., 2005). However, the other three classes of endophytes are uninvestigated with respect to their roles in plant invasions. We have found considerable diversity among non-clavicipitaceous endophytes in Centaurea stoebe, the European plant invader in North America that is commonly known as "spotted knapweed" (Shipunov et al., 2008). All 92 sequence-based, fungal phylotypes were obtained from cultures of seed isolates of C. stoebe. Since endophytes in classes 3 and 4 are not transmitted vertically through seed, whereas Class 2 endophytes are (Rodriguez et al., 2009), endophytes from C. stoebe seed are presumed to belong to Class 2. These endophytes can colonize and affect biomass of both root and shoot systems of plants, but their effects on plant competitiveness and invasiveness are unknown. To date, we have determined the effects of only a few of the 92 endophytes on the growth of C. stoebe itself (Newcombe et al., 2009). An important question is whether endophytes improve the competitiveness of their hosts versus plants that they encounter in their invaded range.
Our purpose here was to determine whether the most common endophytes of C. stoebe influence its competitive interactions with two species of Festuca, grasses that co-occur with spotted knapweed in both native (F. ovina) and invaded (F. idahoensis) ranges. In order to determine the existence and magnitude of these putative interactions, we designed a set of experiments (preliminary and main) that involved inoculations of seedling roots of C. stoebe with endophytes followed by competition with either of Festuca idahoensis or F. ovina.
A. Selecting the most abundant phylotypes
Seedheads of C. stoebe were sampled in its invaded range (mostly Northwestern U.S.) and its native range (Middle and Eastern Europe, European Russia, North Caucasus and the Urals). In all, 102 sites were sampled (53 from the invaded range and 49 from the native range). In each site or population of C. stoebe, five plants were sampled, and from each plant, 20 seeds (i.e., achenes), for a total of 100 seeds per site and 10,200 seeds in all. Endophytes were isolated onto potato dextrose agar, PDA, from seeds following "Method II" surface-sterilization (Schulz et al., 1993). Each isolate received its own "Cultivation Identification Number" (CID -Table 1), and was assigned on the basis of morphology and ITS and Alt a 1 sequences to a phylotype of a fungal genus. Methods for extraction, amplification and sequencing of the nuclear 5.8S rRNA gene and the two flanking, ITS regions were as previously published (Ganley et al., 2004). As a proxy for recognizing fungal species, ITS sequences may be conservative because biological species may share the same sequence (Lieckfeldt & Seifert, 2000).
Because undescribed species may be common among endophytic isolates (Froehlich & Hyde, 2004;Ganley et al., 2004;Hartnett et al., 1993;Shipunov et al., 2008), a sequence-based approach is increasingly employed in endophyte studies. For those endophytes of C. stoebe that could be assigned on the basis of ITS sequences to Alternaria and related genera, the Alt a 1 gene was also sequenced to provide additional discrimination of phylotypes (Hong et al., 2005). It is important to bear in mind that 5 a single phylotype does not represent a clone; individuals belonging to the same phylotype here may differ genetically at loci that were not sequenced, and even more significantly they may differ biologically. In other words, variation within a phylotype is akin to intraspecific variation, as would be expected for a species proxy. To determine the most abundant phylotypes for experiments, relative abundances of endophytes were calculated on a phylotype basis, and then representative isolates were selected for the inoculations of the main experiment, described below. Sequence data were deposited in GenBank (http://www.ncbi.nlm.nih.gov/).
Preliminary experiment
Endophyte status was determined by germinating field-collected, surface-sterilized seeds of C. stoebe on 1.5% water agar in Petri dishes. E+ (endophyte infected) seedlings were ones from which endophytic fungi that had been in the seeds grew out into the agar; the roots of these seedlings were examined under a dissecting microscope to directly observe tissue darkening associated with infection. E-(endophyte-free) seedlings did not yield endophytes. These seven-day-old seedlings were then transplanted first to trays and then to pots two weeks later. Five, two-week-old seedlings of F. idahoensis were planted around each seedling of C. stoebe. In total, we prepared forty standard 3.78 dm 3 pots (20 per treatment). In this experiment, endophytes represented a random sampling of endophyte diversity in C. stoebe (Shipunov et al., 2008), as they had not yet been assigned to phylotypes.
Main experiment
For the main experiment, we employed 10-day-old cultures of representative isolates of the most abundant phylotypes (see below) to inoculate roots of seedlings germinated from seeds of C. stoebe 6 plants grown in greenhouse. We had previously observed that individual plants of C. stoebe always produced endophyte-free seeds in greenhouse conditions. The experiment was conducted with representatives of the three most common phylotypes from each of the native and invaded ranges of C. stoebe: 1) isolates or CIDs of phylotypes "alt002b", "alt002c" and "alt002f" from the native range; 2) isolates of "alt002b", "cla063", and "epi066" from the invaded range (Table 1). Each of the six isolates was inoculated into roots of seven-day-old seedlings of C. stoebe by placing seedling roots in contact with a live culture of a particular endophyte for 12 hours. Root tissue darkening associated with infection was again checked under a dissecting microscope. Roots of control seedlings were placed in contact with uninoculated culture medium (i.e., PDA) for the same duration. After two weeks in trays, seedlings of C. stoebe were planted in pots with two-week-old neighbours that were either seedlings of evolutionarily naïve F. idahoensis, or adapted Festuca ovina from the exotic and native ranges of C. stoebe, respectively. This experiment comprised 192 pots, given 12 replicates of each combination of treatment (12 by 6 by 2, or 144 pots) and neighbor including E-control pots (12 by 2, 24 pots); plus 12 replicates of each neighbor without C. stoebe (24 pots).
In both experiments, pots were filled with sterilized "Sunshine" mix (Sun Gro Horticulture Inc., Bellevue, WA, USA). Seeds of F. idahoensis were obtained from the Wind River Seed Co., Manderson WY; seeds of F. ovina were obtained from Grasslands West, Clarkston, WA. Greenhouse conditions included a 16h day, with temperatures between 24 and 27 °C. Each experiment was run for 18 weeks, at which point C. stoebe plants had flowered. Aboveground biomass was harvested, oven-dried to constant weight, and then weighed. If endophytes could affect competition, then the "competitive advantage" of knapweed over fescue, was expected to be enhanced by endophytes and therefore biomass of endophyte-infected knapweed could prevalent over the biomass of fescue more than biomass of endophyte-free knapweed. Statistical analyses were performed both with R and with Systat 7 version 12. The K-S Test (Lilliefors) was used to test data distributions and Levene"s Test was used to test for homogeneity of variances.
Re-isolation experiment
To determine whether inoculation resulted in infection, we attempted to re-isolate inoculants of two phylotypes (CID 63 and CID 120) three weeks post inoculations. E-seedlings were treated as in the main experiment (see above), and then left to grow in a sterile environment for 21 days. Then seedlings were surface-sterilized with 50% ethanol (5 min) and distilled water and placed on the PDA medium.
C. Presence of endophytes in roots of Centaurea stoebe in the field
Field-collected roots of C. stoebe were sampled for endophytes. Because initial sequence data revealed multiple fungal species present in root tissues of plants in the field near Potlatch, Idaho, leading to mixed populations of ITS amplicons, PCR products were cloned, and individual sequences obtained from cloned PCR amplification products. One to three microliters of mixed, unpurified, undiluted PCR product were ligated overnight at room temperature into pGEM-T Easy TA cloning vector (Promega) in 10-microliter ligation reactions, following the manufacturer"s protocol. One microliter of the ligation mixture was used to transform competent JM 109 E. coli cells, which were plated in multiple concentrations on LB/ampicillin plates (100 micrograms/mL) containing X-gal and IPTG. Presumptive recombinant colonies containing the cloned PCR product were screened by PCR for presence of appropriate insert; for each candidate colony, a 30-microliter PCR reaction was prepared containing ITS 1 and ITS 4 primers, PCR conditions and concentrations as described elsewhere (Ganley et al., 2004). Sterile micropipette tips were touched briefly to the surface of the candidate colony, and then rinsed in the PCR reaction by pipetting up and down two to three times. Reaction tubes were then placed into a thermal cycler without further treatment, and PCR carried out as usual. Five-microliter aliquots of completed PCR reactions were run on 1% agarose gels to check for amplification. Those containing insert of appropriate size were directly sequenced.
D. Endophytes in Festuca neighbors
To be sure that endophyte effects in the experiments were not due to Festuca endophytes, 300 seeds of F. idahoensis and 100 seeds of F. ovina were checked for Neotyphodium and other endophytes following surface-sterilization, and isolation as described above.
Competition experiments
In the preliminary experiment, the dry biomass of F. idahoensis in E-and E+ pots averaged 3.08 g and 2.20 g, respectively, on an individual plant basis. Endophytes in C. stoebe were thus responsible for significantly reducing the biomass of neighbouring F. idahoensis (p << 0.01, F = 19.67, df = 1). The biomass of inoculated C. stoebe itself was significantly higher than endophyte-free C. stoebe (p = 0.009, F = 7.21, df = 1) as E+ and E-C. stoebe averaged 13.40 g and 9.75 g, respectively. In sum, in the preliminary experiment, endophytes in C. stoebe were exerting negative effects on F. idahoensis.
However, since the preliminary experiment was conducted with uncharacterized endophytes, we wondered whether observed effects were representative of the most common endophytes that we had isolated from C. stoebe.
The main experiment was conducted with representative isolates of the most common endophytic phylotypes found in seeds of C. stoebe (Table 1), after relative abundances had been determined. As in the preliminary experiment, the biomass of F. idahoensis was reduced by endophytes in C. stoebe (Fig. 1). However, this experiment also contrasted evolutionarily naïve and adapted neighbours, F. idahoensis and F. ovina, from the invaded and native ranges of C. stoebe, respectively.
These neighbours were both affected by endophytes in C. stoebe but in opposite ways (Fig. 1).
Whereas endophytes of C. stoebe generally reduced biomass of the naïve neighbour, F. idahoensis, they increased biomass of the adapted neighbour, F. ovina. Three of six endophytes significantly reduced the biomass of neighbouring F. idahoensis when compared to the effect of E-C. stoebe on F. idahoensis: CIDs 120, 63, and 73 (Bonferroni-adjusted, pairwise comparison p values = 0.003, 0.032, and 0.013, respectively). The first CID, 120, was from the Eurasian range of C. stoebe, but 63 and 73 were both isolated in North America. The effect of the Eurasian CID432 on neighbouring F. idahoensis was marginally significant as well (p = 0.062). CIDs 2, Eurasian, and 66, North American, reduced the biomass of F. idahoensis also (Fig. 1), but not significantly.
In striking contrast, four of six endophytes significantly increased the biomass of neighbouring F. ovina when compared to the effect of E-C. stoebe on F. ovina: CIDs 2, 432, 63, and 66 (Bonferroniadjusted, pairwise comparison p values = 0.009, 0.002, 0.05, and 0.000, respectively). The first two of these were isolated in the Eurasian range of C. stoebe, and the last two were both isolated in North America. CIDs 120, Eurasian, and 73, North American, increased the biomass of F. ovina also (Fig. 1), but not significantly. Thus, the only C. stoebe endophyte to both significantly reduce the biomass of F. idahoensis and significantly increase that of F. ovina was CID 63, a Cladosporium isolate from North America.
Four of six endophytes caused C. stoebe to gain a competitive advantage over F. idahoensis, that was significantly greater than the competitive advantage of endophyte-free C. stoebe over F. idahoensis. These four endophytes were CIDs 2 (p = 0.01), 432 (p = 0.004), 63 (p = 0.03), and 73 (p = 0.001). Interestingly, CID 2 significantly increased competitive advantage of C. stoebe even though it had not significantly reduced biomass of F. idahoensis. Conversely, CID 120 did not significantly increase competitive advantage of C. stoebe over F. idahoensis even though it had significantly reduced biomass of F. idahoensis. The endophyte-free controls showed the lowest mean competitive advantage over F. idahoensis at 4.9 g (Table 2). Thus, CID73, the isolate of the "alt002b" phylotype from North America, increased by over four times the competitive advantage of C. stoebe over F.
Since four of six endophytes significantly increased the biomass of neighbouring F. ovina when compared to the effect of E-controls, one would expect an endophyte-mediated reduction in competitive advantage of C. stoebe over F. ovina. However, only CID 63 significantly reduced competitive advantage over F. ovina (p = 0.03) to -3.8 g per pot (Table 2). Even though CIDs 432 and 66 had significantly increased the biomass of F. ovina, each increased, though insignificantly, the competitive advantage of C. stoebe over F. ovina, when compared to the E-control.
Finally, C. stoebe gained a greater competitive advantage over its naïve neighbour, F. idahoensis, than that which it gained over its adapted neighbour, F. ovina, only when inoculated with endophytes: Biomass Endophyte-infected C. stoebe -Biomass F. idahoensis > Biomass Endophyte-infected C. stoebe -Biomass F. ovina . Endophyte-free C. stoebe actually showed comparable competitive advantages over F. idahoensis and F. ovina (4.9 g versus 7.8 g, respectively - Table 2). In contrast, five of the six endophytes significantly increased the competitive advantage of C. stoebe over the naïve neighbour when compared to the advantage over the adapted neighbour ( Table 2). The one exception was CID66, an Epicoccum isolate that did not cause a significant increase in competitive advantage over F. idahoensis, when compared with the endophyte-free controls (7.1 g versus 4.9 g, respectively - Table 2).
Biomasses of C. stoebe and Festuca species were inversely correlated for both F. ovina (Pearson r = -0.40; p < 0.001) and for F. idahoensis (Pearson r = -0.41, p < 0.001), as one might expect for moderate competition within pots. However, it was only when C. stoebe was growing with F. idahoensis, that biomass of C. stoebe was highly correlated with competitive advantage of the former over the latter (Pearson r = 0.82, p < 0.001). In contrast, there was no correlation between biomass of C. stoebe and competitive advantage over F. ovina (Pearson r = 0.07, p = 0.51), largely because only CID 63 significantly affected competitive advantage over F. ovina, as discussed above.
The endophyte factor, with seven levels (i.e., six isolates plus the E-control), by itself explained 31% of the variation in competitive advantage over F. idahoensis (GLM; F = 5.76, p < 0.001).
However, interaction between C. stoebe biomass and the endophyte factor actually explained slightly more variation, 36%, in competitive advantage over F. idahoensis (GLM; F = 7.36, p < 0.001) than endophytes alone. For both F. idahoensis and F. ovina, competitive advantage of C. stoebe was not as well explained by the interaction of endophytes with Festuca (i.e., biomass) as by the interaction of endophytes with their host, C. stoebe (biomass). C. stoebe biomass was itself significantly affected by endophyte treatments (GLM; F = 6.31, p < 0.001), as it had been in the preliminary experiment.
Re-isolation experiment
Inoculants (i.e., CIDs 63 and 120) were commonly re-isolated indicating that infection had taken place.
In several cases, we obtained isolates from plant tissues formed after inoculation, indicating that further colonization occurred after infection.
Presence of endophytes in roots of Centaurea stoebe in the field
Seed endophytes clearly had significant effects when inoculated into roots of C. stoebe plants in greenhouse experiments. But, did seed endophytes occur naturally in roots of C. stoebe in the field? Our sampling was not extensive but following cloning, all colonies with insert were sequenced, revealing four ascomycetous fungi: 1) a fungus with an ITS sequence identical to an "uncultured ascomycete clone", EU003079, in GenBank; 2) a fungus identical to Protoventuria alpina, EU035444 (Crous et al., 2007); 3) a fungus identical to an uncultured, soil fungus from the humic horizon, EF434053 (Taylor et al., 2007); and 4) the "cla063" phylotype that is the third most common seed endophyte of C. stoebe in its invaded range (Shipunov et al., 2008), and the endophyte that significantly reduced and increased biomasses of F. idahoensis and F. ovina, respectively, as reported here. With minimal sampling, "cla063" was additionally found via cloning (i.e., the same approach used for detecting endophytes in roots) in leaves of C. stoebe in the field. This Cladosporium isolate, "cla063", has thus been isolated from roots, leaves and seeds as one would expect for a Class 2 endophyte (Rodriguez et al., 2009).
Endophytes in Festuca neighbours
Surface-sterilized samples of the seed of F. idahoensis and F. ovina employed in the greenhouse experiments did yield some endophytes: four phylotypes from F. idahoensis and three from F. ovina.
Isolation frequencies were thus low and approximately equal for the seed of F. idahoensis and F. ovina (i.e., 1.7% and 3%, respectively). Neotyphodium isolates, which are known to affect growth and interactions of Festuca (Van Hecke et al., 2005), were not obtained. There was no overlap (i.e., no endophytes in common) between the seven phylotypes from Festuca and the five phylotypes from C.
stoebe of Table 1. The implications of these results in combination with results from the competition experiments suggest that influence of Festuca endophytes on experimental outcomes was minimal.
Discussion
We found that competitive interactions between C. stoebe and its Festuca neighbours were affected by the presence of endophytes in C. stoebe. The identity of the neighbour mattered; effects on evolutionarily naïve F. idahoensis were negative, aiding C. stoebe, whereas effects on adapted F. ovina were positive. Our findings indicate that some of the endophytes of C. stoebe may increase its invasiveness, at least as gauged by competition with F. idahoensis. At a more general level, Class 2 endophytes should be considered an additional group of mutualistic agents that can promote plant invasions (Richardson et al., 2000;Rudgers et al., 2005).
The effects of endophytes were not tied to the range of C. stoebe (native or invaded) from which they were isolated; site of isolation does not by itself indicate the native range of an endophytic fungus (Shipunov et al., 2008;Newcombe & Dugan, 2010). But just as the identity of the neighbouring Festuca species influenced competitive outcomes with C. stoebe, the identity of endophyte inoculants also mattered. For example, the Epicoccum isolate of the "epi066" phylotype did not increase the competitive advantage of C. stoebe over the naïve competitor as compared to the adapted competitor (Table 2). Whereas the phylotype for which the evidence of Class 2 endophyte status was strongest (i.e., the Cladosporium isolate of the "cla063" phylotype) did.
In our experiments, the roots of seedlings of C. stoebe were inoculated to mimic what appears likely to be a natural infection process following germination of endophyte-infected seed, and the reisolation experiment showed that inoculation can result in infection.. Roots are more likely to be colonized systemically by endophytes than shoots (Boyle et al., 2001), but we do not yet know whether the effects reported here even depend on persistent root infection. Root turnover can provide a significant substrate for microbes in soil (Leigh et al., 2002), and it is conceivable that endophytes alternate between in planta and soil phases. Endophytes might retard growth of naïve neighbours (Rudgers et al., 2005). Underground chemical compounds can be produced by invasive plants, as has been postulated for C. stoebe itself (Bais et al., 2003;Blair et al., 2005;Blair et al., 2006;Callaway & Aschehoug, 2000;Callaway & Ridenour, 2004;Vivanco et al., 2004), although this hypothesis is still controversial (Lau et al., 2008). Nutrient parasitism can also be mediated by mycorrhizal fungi (Carey et al., 2004), but the ascomycetous root endophytes employed here are not known to set up networks essential to this possible mechanism (Addy et al, 2005;Jumpponen, 2001). Barrier experiments coupled with observations of cleared and stained roots of both C. stoebe and its neighbors are needed.
Neighbour identity has been shown to affect plant interactions mediated by soil fungi . Similarly, root inoculations with fungi have shifted coexistence ratios of Populus and invasive Tamarix in pot experiments (Beauchamp et al., 2005), and the roots appeared to be colonized mostly by dark septate endophytes that are likely ascomycetous as here. But, in the latter experiments also, mechanism remained unknown. Fungi can produce phytohormones (Tudzynski, 1997); in particular, Alternaria species can produce plant growth regulators (Kimura et al., 1992), and four of the six endophyte isolates employed here belonged to this genus. Mycorrhization can increase rates of net photosynthesis (Allen et al., 1981;Dosskey et al., 1990), but the endophytes employed in our main experiment are not known to do so. Alternatively, various rhizosphere microbes are also known to both up-regulate and down-regulate auxin activity in different plants (Ditengou & Lapeyrie, 2000), by acting on auxin-responsive genes such as Pp-C61 (Reddy et al., 2003).Whatever their underlying mechanisms may be, the effects reported here suggest at the very least that endophytes may play important roles in plant community ecology, and their roles in plant invasions merit further study. | 2012-09-07T19:15:30.000Z | 2012-09-07T00:00:00.000 | {
"year": 2012,
"sha1": "0da6b4fffdd84c07f72c4a850b78b296acc5444c",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "0da6b4fffdd84c07f72c4a850b78b296acc5444c",
"s2fieldsofstudy": [
"Environmental Science"
],
"extfieldsofstudy": [
"Biology"
]
} |
15392820 | pes2o/s2orc | v3-fos-license | The Gromov width of complex Grassmannians
We show that the Gromov width of the Grassmannian of complex k-planes in C^n is equal to one when the symplectic form is normalized so that it generates the integral cohomology in degree 2. We deduce the lower bound from more general results. For example, if a compact manifold N with an integral symplectic form omega admits a Hamiltonian circle action with a fixed point p such that all the isotropy weights at p are equal to one, then the Gromov width of (N,omega) is at least one. We use holomorphic techniques to prove the upper bound.
Introduction
Consider the ball of capacity a with the standard symplectic form ω std = dx j ∧ dy j . The Gromov width of a 2N -dimensional symplectic manifold (M, ω) is the supremum of the set of a's such that B(a) can be symplectically embedded in (M, ω). Computations of Gromov width and, more generally, of symplectic ball packings, can be found, for example, in [4,5,6,8,16,19].
Often in symplectic geometry, equivariant techniques give constructions whereas holomorphic techniques give obstructions. We use both.
Our main technical result is a criterion for the existence of symplectic embeddings of open subsets of C n into a symplectic manifold with a Hamiltonian torus action. See Proposition 2.8. Paul Biran has asked whether the Gromov width of a compact symplectic manifold is at least one if the symplectic form is integral. In Proposition 2.11 we answer his question positively whenever the manifold admits a Hamiltonian circle action with a fixed point p such that all the isotropy weights at p are equal to one.
As a corollary, we obtain embeddings of balls into complex Grassmannians. More precisely, we prove the following theorem.
Theorem 1 Let Gr(k, n) be the Grassmannian of k -planes in C n , together with its U(n)-invariant symplectic form ω , normalized so that [ω] generates the integral cohomology H 2 (Gr(k, n); Z). There exists a symplectic embedding of B(a) into Gr(k, n) if and only if a ≤ 1.
Next, we use holomorphic techniques to show that it is impossible to embed the ball B(a) into Gr(k, n) if a > 1. The proof uses two ingredients: a slight adaptation of the proof of Gromov's non-squeezing theorem, and the calculation of certain Gromov-Witten invariants for Gr(k, n), which we quote from [18] (also see [11,3]).
The Gromov width of the complex Grassmannian was independently computed by Guangcun Lu in [14]. Lu obtained the lower bound by an explicit embedding of a ball. Our results are more general in that they give lower bounds for the Gromov width of many more manifolds, such as other generalized flag manifolds, and in that they allow one to embed sets other than balls.
Lower bounds for Gromov width
In this section, we construct symplectic embeddings of open subsets of C n into symplectic manifolds with Hamiltonian torus actions. The key technique is Moser's method. This section is an extension of our work in [10, §13] and is inspired by [7, §1].
Let a torus T ∼ = (S 1 ) dim T with Lie algebra t act effectively on a connected symplectic manifold (M, ω) by symplectic transformations. A moment map is a map Φ : M −→ t * such that where ξ M is the corresponding vector field on M .
Let p ∈ M be a fixed point. There exist η j ∈ t * , called the isotropy weights at p, such that the induced linear symplectic T -action on the tangent space T p M is isomorphic to the action on (C n , ω std ) generated by the moment map The isotropy weights are uniquely determined up to permutation.
By the equivariant Darboux theorem [20], a neighborhood of p in M is equivariantly symplectomorphic to a neighborhood of 0 in C n . The results of this section allow us to control the size of this neighborhood.
For the applications in this paper it is enough to symplectically embed the ball B(1) ⊂ C n into manifolds with S 1 -actions; see Proposition 2.11. However, we will take this opportunity to develop the relevant machinery in the more general case where we embed other subsets of C n , possibly unbounded, into manifolds with torus actions.
Let T ⊂ t * be an open convex set which contains Φ(M ). The quadruple (M, ω, Φ, T ) is a proper Hamiltonian T-manifold if Φ is proper as a map to T , that is, the preimage of every compact subset of T is compact.
For any subgroup K of T , let M K = {m ∈ M | a · m = m ∀a ∈ K} denote its fixed point set.
Example 2.3 A compact symplectic manifold with a non-trivial T -action is never centered, because it has fixed points with different moment map images.
Example 2.4 Let a torus T act linearly on C n with a proper moment map Φ C n such that Φ C n (0) = 0. Let T ⊂ t * is be an open convex subset containing the origin. Then Φ −1 C n (T ) is centered about the origin.
Example 2.5 Let M be a compact symplectic toric manifold with moment map Φ : M −→ t * . Then ∆ := image Φ is a convex polytope. The orbit type strata in M are the moment map pre-images of the relative interiors of the faces of ∆. Hence, for any α ∈ ∆, is the largest subset of M that is centered about α.
Example 2.6 Let (M, ω, Φ, T ) be a proper Hamiltonian T -manifold. Then every point in t * has a neighborhood whose preimage is centered. This is a consequence of the local normal form theorem and the properness of the moment map.
Proof For simplicity, assume that α = 0. Let There exist a convex neighborhood V of 0 in T and a T -equivariant symplec- These exist for the following reasons. The equivariant Darboux theorem gives an equivariant symplectomorphism from a neighborhood of p in M to a neighborhood of 0 in C n . Because Φ −1 ({0}) consists of a single point, the moment map Φ C n : C n −→ t * is proper. (See, for example, Lemma 5.4 in [10].) Hence, the neighborhoods of p ∈ M and 0 ∈ C n contain the preimages of a neighborhood of 0 ∈ t * .
The Euler vector field on a vector space is the generator of the flow x → e t x. Let X be the Euler vector field on t * ; then −X generates the flow x → e −t x.
wherever it is defined. Hence, since Φ is proper and T is convex and contains the origin, Let X be half the Euler vector field on C n . Then − X generates the flow ψ t : C n −→ C n given by ψ t (z) = e −t/2 z.
We will now apply Moser's method. Let λ be a T -invariant one form such that dλ = ω . Let λ t = e t ( ψ t ) * λ.
Then dλ t = ω t . Also, ι(ξ M )λ t = Φ ξ , because both sides take the value 0 at p, and they have the same differential. Hence, for all ξ ∈ t. Since Φ is proper, this implies that Y t integrates to an isotopy, is a symplectomorphism for all t ≥ 0.
Thus, the composition , ω std ) is an equivariant symplectomorphism. If T is bounded, then we can choose t sufficiently large so that T ⊂ e t V ; hence we are done. Now suppose that T is not bounded. We will modify our constructions of the vector fieldX and the one-form λ so that F t = F s and G s = G t on the intersection Φ −1 (e t V ) ∩ Φ −1 (e s V ).
First, we modify our construction of X so that, in addition to satisfying Φ * ( X) = X , after possibly shrinking V F * ( X) = X on Φ −1 (V ). (2.9) To do this, we construct X on M {p} as before, and then patch with the vector field F * ( X) on Φ −1 (V ), using an invariant partition of unity subordinate to the sets M {p} and Φ −1 (V ).
The property (2.9) implies that F Therefore, since the union of the sets e t V over all t ≥ 0 is all of t * , we can define an equivariant diffeomorphism Clearly, F * (ω std ) = ω t on Φ −1 (e t V ).
Since ( ψ t ) * λ std = e −t λ std , the property (2.10) implies that . Therefore β t = 0, and hence Y t = 0, on Φ −1 (e t V ). Consequently, Therefore, since the union of the sets e t V over all t ≥ 0 is all of t * , we can define an equivariant diffeomorphism Clearly, G * (ω t ) = ω on Φ −1 (e t V ).
Hence, F •G : (M, ω) −→ (Φ −1 C n (T ), ω std ) is an equivariant symplectomorphism, as required. Proof We assume, without loss of generality, that N is connected. Let ξ N denote the vector field that generates the circle action. Our convention is that the circle group is S 1 = R/Z, so that ξ N generates a flow of period one. Let Φ : N −→ R be the moment map, so that ι(ξ N )ω = −dΦ. For simplicity, assume that Φ(p) = 0 Because the isotropy weights are positive, p is an isolated local minimum for the moment map. Since the moment map fibers are connected [2,9] By Stokes's theorem, for any fixed point q , the difference Φ(q) − Φ(p) is equal to the integral of ω over the cycle obtained from a curve connecting p to q by "sweeping" the curve by the circle action. Because [ω] is integral, this implies that Φ(q) − Φ(p) is an integer. So p is the only fixed point that is mapped to [0, 1).
Let M = Φ −1 ([0, 1)). Consider a subgroup K ⊂ S 1 and let Y ⊂ M K be a connected component of its fixed point set. Since Y is closed in M and Φ : M −→ [0, 1) is proper, the image of Y is a closed subset of [0, 1), so it has a minimum. Any point in Y which is mapped to this minimum must be a fixed point. Hence p ∈ Y , and so 0 ∈ Φ(Y ). This shows that M is centered about 0.
Proposition 2.11 then follows from Proposition 2.8.
Lower bounds for Grassmannians
In this section we construct an embedding of the ball B(1) into the complex Grassmannian Gr(k, n), thus showing that the Gromov width of the Grassmannian is at least one.
Proposition 3.1 Let Gr(k, n) be the Grassmannian of k -planes in C n , together with its U(n)-invariant symplectic form ω , normalized so that [ω] generates the integral cohomology H 2 (Gr(k, n); Z). There exists a symplectic embedding of B(1) into Gr(k, n).
Take the induced action on Gr(k, n). Then p = C k × {0} is a fixed point for this action. Since T p Gr(k, n) ∼ = Hom(C k , C n−k ), the isotropy action is complex multiplication by a ∈ S 1 . Proposition 3.1 now follows from Proposition 2.11.
Upper bounds for Gromov width
In this section, we give a short review of Gromov-Witten invariants and how they can be used to give upper bounds to Gromov widths. This material appears in detail in many places; our treatment is adapted from [17]. which is defined, for example, in [17], has the following property. Let J denote the space of almost complex structures that are tamed by ω . Let J reg (A) ⊆ J denote the set of regular almost complex structures for the class A (see [17]). Given J ∈ J reg (A), for generic deformations N ′ i of N i and generic points t i ∈ CP 1 , the number of J -holomorphic maps where a > λ. The standard complex structure J std on B(a) ⊂ C n transports through ρ to a complex structure on ρ(B(a)). By a well known technique, after passing to a smaller a such that a > λ there exists a tamed almost complex structure J on M such that ρ intertwines J std with J . (Since Riemannian metrics can be patched together by partitions of unity, this follows from the existence of an Sp(R 2n )-equivariant projection from the space of inner products on R 2n to the subspace of those inner products that are compatible with the symplectic form.) Because J reg (A) is of second category in J , and by the characterization of Gromov-Witten invariants discussed above, there exists a sequence J i of almost complex structures converging to J , and points p i which converge to p := ρ(0), and, for each i, a J i -holomorphic curves CP 1 −→ M in the class A which passes through p i .
Since the class A is indecomposable, by Gromov's compactness theorem [8,17] there exists a subsequence of the J i 's which converges weakly to a Jholomorphic curve C in the class A. In particular, C ω = λ. Since p i converge to p, this limit curve contains p.
The pre-image of C under ρ is a holomorphic curve in the ball B(a) which passes through the center of the ball and which is closed in B(a). But the smallest area of a such a curve is that of a disk through the center, which is a.
Upper bounds for Grassmannians
It remains to prove the following proposition.
Proposition 5.1 Let Gr(k, n) be the Grassmannian of k -planes in C n , together with its natural U(n) invariant symplectic form, normalized so that [ω] generates the integral cohomology H 2 (Gr(k, n); Z). If a > 1, there does not exist a symplectic embedding of B(a) into Gr(k, n).
Proof The real dimension of M = Gr(k, n) is 2k(n − k). Let A be the generator of H 2 (Gr(k, n), Z) ∼ = Z. Clearly, A is irreducible.
The standard complex structure is tamed by ω , and c 1 (T M )(A) = n. It is easy to check that Gr(k, n) is monotone.
Fix a hyperplane W ⊂ C n and let X ⊂ Gr(k, n) be the set of k -planes that are contained in W . Fix a vector y ∈ C n and let Y ⊂ Gr(k, n) be the set of k -planes that contain y . Let p ∈ Gr(k, n) be any point. In [18], it was shown that Φ | 2014-10-01T00:00:00.000Z | 2004-05-20T00:00:00.000 | {
"year": 2004,
"sha1": "275656a110ffcbe8d2999a8d443929c7d8c3024d",
"oa_license": null,
"oa_url": "http://msp.org/agt/2005/5-3/agt-v5-n3-p03-s.pdf",
"oa_status": "BRONZE",
"pdf_src": "Arxiv",
"pdf_hash": "ebabd97fe00e8dbbb045e6a5aeead96eaf9dde54",
"s2fieldsofstudy": [
"Mathematics"
],
"extfieldsofstudy": [
"Mathematics"
]
} |
227210654 | pes2o/s2orc | v3-fos-license | Unsupervised Word Translation Pairing using Refinement based Point Set Registration
Cross-lingual alignment of word embeddings play an important role in knowledge transfer across languages, for improving machine translation and other multi-lingual applications. Current unsupervised approaches rely on similarities in geometric structure of word embedding spaces across languages, to learn structure-preserving linear transformations using adversarial networks and refinement strategies. However, such techniques, in practice, tend to suffer from instability and convergence issues, requiring tedious fine-tuning for precise parameter setting. This paper proposes BioSpere, a novel framework for unsupervised mapping of bi-lingual word embeddings onto a shared vector space, by combining adversarial initialization and refinement procedure with point set registration algorithm used in image processing. We show that our framework alleviates the shortcomings of existing methodologies, and is relatively invariant to variable adversarial learning performance, depicting robustness in terms of parameter choices and training losses. Experimental evaluation on parallel dictionary induction task demonstrates state-of-the-art results for our framework on diverse language pairs.
INTRODUCTION AND BACKGROUND
With the success of distributed word representation, like Word2Vec , GloVe (Pennington et al., 2014) and FastText (Bojanowski et al., 2017), in capturing rich semantic meaning, the use of these embeddings has permeated a wide range of Natural Language Processing (NLP) tasks such as text classification, document clustering, text summarization and question answering (Klementiev et al., 2012) to name a few. Unsupervised learning of such continuous high dimensional vector representation for words rely on the distributional hypothesis (Harris, 1954).
Motivation. As a natural generalization, methods for obtaining multi-lingual word embeddings across diverse languages have recently gained significant attention in the NLP research community (Wang et al., 2020). Learning cross-lingual word embeddings (CLWE) entails mapping the vocabularies of different languages onto a single vector space for capturing syntactic and semantic similarity of words across languages boundaries (Upadhyay et al., 2016). Thus, CLWE provides an effective approach for knowledge transfer across languages for several downstream linguistics tasks such as machine translation (Artetxe et al., 2018a;, POS tagging (Zhang et al., 2016), dependency parsing (Ahmad et al., 2019), named entity recognition (Xie et al., 2018;Chen et al., 2019), entity linking (Tsai & Roth, 2016), language inference and low-resource language understanding (Xiao & Guo, 2014). In fact, word alignment across languages also finds interesting applications in the study of cultural connotations (Kozlowski et al., 2019) and spatio-linguistic commonalities (Zwarts, 2017;Yun & Choi, 2018;Pederson et al., 1998).
Linguistic Correlation. Monolingual representation spaces learnt independently for different languages tend to exhibit similarity in terms of geometric properties and orientations . For example, the vector distribution of numbers and animals in English show a similar geometric constellation formation as their Spanish counterparts. Further, the frequency of words across languages have been shown to follow the Zipf's distribution 1 , with nearly 70% most frequent word overlap (Aldarmaki et al., 2018) and 60% synonym overlap (Dinu et al., 2015) across language pairs. Existing techniques for extracting cross-lingual word correspondences rely on above inter-dependencies to efficiently learn transformations across the monolingual embedding spaces.
State-of-the-art & Challenges. Early approaches for directly obtaining multi-lingual word embeddings relied on the availability of large parallel corpora (Gouws et al., 2015) or document-aligned comparable corpora (Mogadala & Rettinger, 2016;Vulić & Moens, 2016). However, such methods are not scalable as annotations are expensive and large parallel datasets, especially for low-resource languages, are scarce in practice. To address the above challenges, linear transformations between two monolingual embedding space using small manually created bi-lingual dictionaries were proposed Artetxe et al., 2016). Words having similar surface forms across languages were used to induce seed dictionaries and other augmented refinement strategies were explored in the semi-supervised approaches of Artetxe et al. (2017); Zhou et al. (2019); Doval et al. (2018). Subsequently, improvements in orthogonality and optimization constraints were explored for generalization beyond bi-lingual settings for supervised cross-lingual alignment and joint training methods (Joulin et al., 2018;Jawanpuria et al., 2019;Alaux et al., 2019;Wang et al., 2020).
Unsupervised framework for bi-lingual word alignment was first proposed by Barone (2016); Zhang et al. (2017a; using adversarial training. The use of post-mapping refinements were shown to produce high quality results in the MUSE framework across diverse languages, and was used for machine translation system in . Parallel dictionary construction using CSLS ) (adopted in this paper) or inverted softmax (Smith et al., 2017) was shown to tackle the "hubness problem" (Radovanović et al., 2010) caused due to highly dense vector space regions (called hubs), which adversely affects reliable retrieval of bilingual word translation pairs. However, the performance of adversarial learning techniques have been shown to suffer from instability, convergence issues, and dependence of precise parameter settings. Further, Søgaard et al. (2018) found the above unsupervised approaches to fail for morphologically rich languages. Hence, optimization formulations using Gromov-Wasserstein, Sinkhorn distance, and Iterative Closest Point were explored Alvarez-Melis & Jaakkola, 2018;Xu et al., 2018;Hoshen & Wolf, 2018). Recently, adversarial auto-encoders using cyclic loss optimization in latent space supplemented with refinements (Mohiuddin & Joty, 2019; has achieved state-of-the-art results for bi-lingual word embedding alignment on diverse languages. Proposed Approach. In this paper, we propose BioSpere (Bi-Lingual Word Translation via Point Set Registration), a novel framework for fully unsupervised bi-lingual word correspondence induction. Given two independently learnt monolingual word embedding space, BioSpere uses a combination of adversarial training, refinement procedure, and point set registration approach to efficiently extract word translations. Specifically, the input vector spaces are initially aligned using CycleGAN, a Generative Adversarial Network (GAN) trained using cycle-consistency loss optimization criteria, as word translation pairs are symmetric, i.e., if word w x is a translation of w y , then w y is also a translation of w x . The cyclic loss criteria has been shown to be better in capturing bi-directional distributional similarities (Xu et al., 2018) and in training adversarial networks in (Mohiuddin & Joty, 2020) (auto-encoders with a latent space of the embeddings). The word alignments obtained from CycleGAN are then refined via symmetric re-whitening or spherical transformation (Artetxe et al., 2018b) to remove correlations among the different components of the language embeddings. It is interesting to note that extracting word correspondences is akin to point set registration (Zhu et al., 2019) in image processing. To this end, BioSpere finally utilized the Coherent Point Drift (CPD) algorithm (Myronenko & Song, 2010) to compute an affine transformation between the aligned and refined vector spaces. Our choice of CPD hinges on two key insights: (i) CPD inherently works on the concept of Gaussian Mixture Model (GMM), which has been shown to tackle the "hubness" problem (Zhou et al., 2019); and (ii) CPD being an unsupervised approach might reduce error propagation from the adversarial or refinement steps, as opposed to the supervised Procrustes refinement (Schönemann, 1966) (extensively used in the literature) that requires an intermediate synthetic (possibly erroneous) dictionary creation from the adversarial training stage. Extensive empirical results on diverse languages (reported in Section 3) demonstrate that the proposed BioSpere framework outperforms existing approaches in terms of accuracy for parallel dictionary creation. We further show that BioSpere can robustly handle adversarial convergences issues, sub-optimal parameter settings, as well as morphologically rich and low-resource languages.
Contributions. In a nutshell, the key contributions of this paper can be described as follows: • BioSpere, an unsupervised framework for learning bi-lingual word translations from two independent monolingual embedding spaces -thus aligning the vocabularies to a common vector representation for capturing semantic similarities between words across languages; • A novel combination of adversarial training, refinement procedure, and point set registration algorithm -coupling the advantages of cycle-consistence loss and Gaussian Mixture Model -to alleviate the challenges for word embedding space alignment; • Unsupervised stopping criterion incorporating cycle-loss consistency measure, with better correlation with mapping quality, for selection of adversarial training model parameters; • Experimental evaluation on diverse language pairs showcasing enhanced accuracy (nearly at par with supervised approaches) compared to existing techniques, for parallel dictionary construction task, even for small vocabulary sizes; and, • Robustness study of BioSpere framework in efficiently handling hubness problem, dependencies on adversarial learning convergence and precise parameter choice, as well as morphologically rich or low-resourced languages.
FRAMEWORK
We assume the existence of two sets X = {x n } N n=1 and Y = {y m } M m=1 of word embeddings trained independently on monolingual data from a source and a target language, respectively. The aim of our BioSpere framework is to map each word in the source language to its translation in the target language, in a manner that does not require any cross-lingual supervision. Equivalently, we wish to align the two embedding sets in such a way that words that are semantically similar across languages are close to each other.
To achieve this, we hinge on 4 modules, namely Align, Correspond, Transform and Generate (ACTG) 2 A pictorial depiction of the overview of the functionality of the different modules is presented in Figure 1. We now look at each module individually.
ALIGN
Our first module estimates an initial mapping using a domain-adversarial approach (Ganin et al., 2016). Let x ∼ p data (x) and y ∼ p data (y) be the data distributions. We learn two linear mappings F : X → Y and G : Y → X, that we refer to as forward and backward generators, respectively. We then train a model D Y to discriminate between synthetic target embeddings Y syn = F X = {F x n } N n=1 , and real ones Y . Similarly, we train D X to discriminate between synthetic source embeddings X syn = GY = {Gy m } m=1 M and X. Note the notation overloading: we have used F and G to refer both to the parametric linear operators, as well as to the matrices of their parameters. We continue this way for simplicity, unless the context makes the reference ambiguous.
This results in a two-player game, where the discriminators aim to distinguish real and synthetic embeddings, while the generators aim at making their image as close to their codomain as possible, prevent discriminators from making accurate predictions.
We resemble this game in our training objective, which includes two categories of terms. The adversarial loss, formulated for matching the distribution of the synthetic embeddings to the real distribution. For the forward generator F : X → Y , and its corresponding discriminator D Y , our adversarial loss is: We use a similar adversarial loss L adv (G, D X , Y, X) for the backward generator G : Y → X and its corresponding discriminator D X .
The second objective category is in line with the work of Mohiuddin & Joty (2020). Similar to them, we note that an adversarial generator could map the same set of source embeddings to any random permutation of target embeddings, as long as the synthetic distribution matches the target distribution. To account for this possibility, we argue that the learned generators should not contradict each other, but should be cycle-consistent. That is, given a source embedding x, the forward translation cycle should attempt to produce an output that coincides with x, i.e. G(F (x)) ≈ x. Analogously for the backward translation cycle, G(F (y)) ≈ y. We capture this endeavour with the addition of a cyclic loss to our objective: (2) Following , we make sure F and G remain roughly orthogonal during training by alternating model parameter update with F ← (1 + β)F − β(F F T )F , proceeding analogously for G. Intuitively, this preserves the monolingual quality of our embeddings by preserving their dot product and l 2 distances.
The output of this module are the two sets X A = F (X) and Y A = G(Y ) of aligned embeddings (the images of the learned transformations).
CORRESPOND
Our vanilla CPD results, despite better than previous adversarial networks, are not au par with supervised work. To address this, we perform a set of refinement steps. In Correspond, the first refinement module, we perform symmetric re-weighting, successfully applied in previous work for word embedding alignment refinement (Artetxe et al., 2018a;2016;2017;Mohiuddin & Joty, 2020). This requires a seed parallel dictionary. We induce such a dictionary by considering mutual nearest neighbours across the the original and mapped embeddings in both directions. That is, given mappings f : X → Y and g : Y → X, the similarity between x n and y m is σ nm = δ(f (x n ), y m ) + δ(x n , g(y m )), where δ is a metric in both X and Y . Our metric of choice is cross-domain similarity local scaling (CSLS) , shown by to effectively address the hubbness problem, stereotypical especially when working in highdimensional spaces. Using the bidirectional nature of our adversarial network when computing the similarity has not been done in previous, work to our knowledge, and we found it to considerably improve word translation performance. During dictionary induction, we only consider the 25K most frequent words from the source and target languages.
In the first step of this module we length-normalise and mean-center X and Y , then apply a linear transformation with corresponding whitening matrices W x = (X T X) −1/2 and W y = (Y T Y ) −1/2 , i.e. X w = XW x and Y w = Y W y . This makes the embedding dimensions ucorrelated among themselves.
Next, let X d and Y d be two matrices that reflect our seed dictionary, with X d i being the embedding of a source word Y d i being the embedding of its translation. We perform an orthogonal transforma-tion with symmetric re-weighting. Specifically, we compute X o = X w U S 1/2 and Y o = Y w V S 1/2 where U , S, and V come from the singular value decomposition U SV T = (X d w ) T Y d w . This transposes the source and target embeddings into a common vector space.
In a final step, we perform de-whitening, to restore the original covariance in the embedding dimension distributions. That is, this module outputs
TRANSFORM
In this module we perform a further refinement of the transformed embeddings X C and Y C using affine Coherent Point Drift (CPD), a probabilistic framework suggested by Myronenko & Song (2010) to perform point set registration, particularly in computer vision applications. The main idea is to consider the task of aligning the two embedding sets as a density estimation problem, where one set represents Gaussian mixture model (GMM) centroids, and the other the data points. With the two sets aligned, word translations can be obtained using the maximum of the GMM posterior probability, given a source embedding. Specifically, we consider the embeddings in Y C as GMM centroids and the ones in X C as data points, generated by the GMM. The GMM density has has the form: where p(x|m) = 1 (2πσ 2 ) D/2 exp x−ym 2 2σ 2 and x ∈ X C , y m ∈ Y C . We also add a uniform distribution p(x|M + 1) = 1/N to account for outliers, resulting in a Uniform-Gaussian mixture model. Following the authors, we use equal isotropic covariances σ 2 , and equal membership probabilities P (m) = 1/M for all GMM components. We estimate the GMM centroid locations θ by minimising the negative log-likelihood function: We use the Expectation Maximization (EM) algorithm (Dempster et al., 1977). to find the parameters θ and σ 2 . We direct the interested reader to a more detailed description of CPD provided by its original authors (Myronenko & Song, 2010).
We use the affine version of CPD, which provides a tuple (R, t, s), where R is a rotation matrix, t is a translation vector, and s is a scaling constant. The transformed source embedding set is computed as X T = (RX T C * s + t) T . We run CPD twice for each language pair, once in each directions, providing us with X T and Y T .
GENERATE
We iterate between Correspond and Transform modules until an model selection criterion degrades for two consecutive iterations. The criterion is specified in Section 2.5. Equipped with the final X T and Y T , we compute the final estimated parallel dictionary using the same procedure as in Section 2.2. We compare this with ground truth parallel dictionaries to compute word translation accuracy.
UNSUPERVISED MODEL SELECTION
Being in an unsupervised setting, we cannot use a validation set to direct us in choosing the best performing setting of our framework. We follow approaches suggested in previous work to address this issue, that we adapt to our framework. We follow in considering the closeness of the source and target mapped embedding spaces. Specifically, we consider the 25K most frequent source words, use CSLS to generate a translation for each of them, and compute the average cosine similarity between these pairs. In our scenario, we consider similarity in both the source and target spaces, as specified in Section 2.2, criterion that we found to be better linked to word translation accuracy, compared to the unidirectional setting used in previous work Mohiuddin & Joty, 2020).
EMPIRICAL EVALUATION
In this section, we evaluate the performance of the proposed BioSpere framework in mapping the input word embeddings onto a shared vector space, such that semantically similar words across languages are close to each other (in terms of distance) in the common space. We benchmark the accuracy of BioSpere against several existing approaches on the tasks of bi-lingual dictionary induction and sentence translation retrieval across a diverse set of languages.
EXPERIMENTAL SETUP
Dataset. Our experimental setup closely follows that of , extensively used in the literature. As input vocabulary, we use the FastText monolingual vector embeddings (with a dimensionality of 300) (Bojanowski et al., 2017) of the top 200K most frequent words in each language. We consider seven different language pairs including morphologically rich and lowresourced languages. Specifically, we use English (en), German (de), French (fr), Spanish (es), Russian (ru), Hebrew (he), Finnish (fi), and Romanian (ro) -a diverse mix of isolating, fusional and agglutinative language with dependent and mixed marking as reported in Søgaard et al. (2018).
Evaluation. We report the Precision@1 (P@1) accuracy scores based on the CSLS criteria (Conneau et al., 2018a) for our empirical evaluations. In the word translation task, we use the gold dictionary with 1,500 source test words (and full 200K target vocabulary) for different language pairs (obtained from github.com/facebookresearch/MUSE). We also perform the above evaluations with a smaller input vocabulary, to simulate scenarios of limited domain-specific resources.
For completeness, we also report the accuracies achieved by state-of-the-art supervised approaches: (1) RCSLS (Joulin et al., 2018): State-of-the-art supervised method for training a learning architecture based on optimizing the CSLS criteria ; (2) GeoMM (Jawanpuria et al., 2019): Language specific geometric rotations are learnt, and subsequently a network architecture is trained to align the languages; and (3) DeMa-BME (Zhou et al., 2019): Provides a weakly-supervised approach for learning a Gaussian Mixture Model by characterizing the probability density between embeddings spaces.
Despite obtaining state of the art results, we emphasize that achieving the best possible accuracy was not our focus. Rather, we aimed to build a framework robust to adversarial instability and data noise. Most parameters were set to fixed values. As such, following , we only fed the adversarial discriminator with the 50K most frequent words; the discriminator had an input dropout layer with rate 0.1. Production deployments may consider further parameter tuning.
In our experiments, we only tuned the weight assigned to the cyclic loss between 5 and 10, and ran the framework under different random seeds, always picking the best model using the unsupervised criterion.
RESULTS AND DISCUSSION
Word Translation. Similar to machine translation, this task involves the retrieval of the translation of a given source word for a target language (from the target vocabulary). Observe, polysemy of words and hubness in embedding space provide a significant challenge in this setting. We evaluate the approaches using a similar setting and the ground-truth dictionaries from . From Table 1, we observe that our BioSpere framework provides state-of-the-art translation results in nearly all of the language pairs. In fact, for certain language pairs like fr→en, the performance of BioSpere is almost at par with existing supervised methods (83.8 compared to 84.1 by RCSLS).
However, the challenges in word translation are compounded for morphologically rich and lowresources languages due to high vocabulary variation and limited accuracy of word embeddings respectively. To this end, we explore the performance of the competing algorithms on Finnish, Hebrew and Romanian -generally identified as "difficult" languages in the literature (Søgaard et al., 2018). From Table 2 it can be seen that BioSpere significantly outperforms the existing approaches with an accuracy improvement of 1.5% on average across the languages.
Limited Vocabulary. An interesting application for cross-lingual word embedding alignment is translation tasks in domain-specific context. For example, an organization expanding its scale of operations to geographically distributed markets and consumers. This would necessitate the efficient expansion of supporting languages for existing documents like manuals, FAQs, etc. as well as for customer services like Chatbots (Qiu et al., 2017;Massaro et al., 2018). Observe, that in such cases, the domain-specific vocabulary is relatively small, depending on the organization's range of business range and limited training resources. We simulate such application scenario in this setting, and observe the performances of the algorithms in face of with limited vocabulary.
The input mono-lingual word embeddings are limited to the 10K most frequent words (instead of 200K most frequent words) in each of the languages, which can potentially severely impact the training stages of existing techniques. However, we initially study the word coverage of Wiki articles with varying vocabulary sizes. Figure 2 depicts the percentage of word coverage with varying frequent word vocabulary sizes using the plain texts of Wikipedia articles from 2018 (Rosa, 2018). It can be observed, that all the languages depict similar characteristics, with a plateau around 50- 75K vocabulary size. Recall, that our empirical setting is based on training the architectures on the 50K most frequent word embeddings. However, with around 10K vocabulary size, the coverage is in general not overtly bad (around 5-10% lower), but provides valuable insights as to the robustness to domain-specific or niche applications, and is hence used in this low vocabulary setting. From Table 3, we see that BioSpere performs better than existing methods on most of the language pairs.
Ablation Study. Finally, to understand the effects of the different modules in BioSpere on the overall performance, we perform ablation study by incrementally adding and removing the separate components. Table 4 tabulates the obtained results on different language pairs (including morphologically rich and low-resourced). We observe, that the adversarial network, CycleGAN, using the cycle-loss consistency optimization criteria, in general performs better than MUSE GAN, the traditional GAN framework of . In terms of refinement performed in the Correspond module of BioSpere, we compared the performance of symmetric re-whitening (used in this work) with the orthogonal Procrustes strategy. Both the refinement processes are seen to be comparable in performance, however since Procrustes, by definition, is a supervised approach, errors from the adversarial training in the Align module might be propagated, degrading the efficacy of the entire framework. Finally, addition of the Coherent Point Drift point-set registration in the Transform module (i.e., the complete BioSpere pipeline) is seen to further improve the results over the refinement strategy.
One important criticism for the performance adversarial training based alignment techniques is their dependence on precise parameter settings to tackle convergence instability (reported previously in our empirical results). Hence, we study the robustness of BioSpere to such issues, by intentionally selecting a sub-optimal CycleGAN model (from the training epochs) as the final output from the adversarial based Align module, denoted as Bad-GAN in Table 4. We observe BioSpere to robustly handle such situations, and provide a final accuracy score that is comparable to that achieved with a properly trained adversarial model selected based on our cyclic unsupervised criteria. Specifically, for en → de and fi → en languages, the performance of Bad-GAN is around 15% worse than the properly selected CycleGAN model, however, the final accuracy of BioSpere for word translation is seen to differ by only 1% (Table 4) -depicting robustness to noisy training.
In summary, the above empirical evaluations showcase that our framework, BioSpere, provides better unsupervised cross-lingual alignment of embedding spaces, by not only outperforming existing techniques in terms of translation accuracy even on morphologically rich and low-resource languages, but also demonstrating robustness in gracefully handling potential adversarial training loss.
CONCLUSION
We introduced BioSpere, an unsupervised cross-lingual alignment framework for word embedding. We use adversarial training with a cycle-consistency loss to induce a seed bidirectional mapping, that we subsequently refine and generate word correspondences using point set registration method. Extensive experiments on multiple languages for parallel dictionary creation not only demonstrate state-of-the-art results for our framework, but also depict robustness to variable adversarial performance, a considerable limitation of past work. | 2020-11-30T02:00:57.511Z | 2020-09-28T00:00:00.000 | {
"year": 2020,
"sha1": "ed48204925d00f4407471b8ac5599723917bbf00",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "ed48204925d00f4407471b8ac5599723917bbf00",
"s2fieldsofstudy": [
"Computer Science"
],
"extfieldsofstudy": [
"Computer Science"
]
} |
28364715 | pes2o/s2orc | v3-fos-license | Variceal bleeding: update of recommendations from the Brazilian Association of Hepatology
Since the publication of the Brazilian Association of Hepatology recommendations for the prevention and treatment of variceal bleeding in 2010, new evidence-based data were reported in the literature. This has changed our current management for portal hypertension. This review updates the previous recommendations. It takes the new prognostic staging of cirrhosis into account allowing tailored treatment for advanced fibrosis, compensated or decompensated cirrhosis. An organizing panel of five experts reviewed all recommendations according to available data, which were subsequently scrutinized by all members of the Brazilian Association of Hepatology using a web-based approach. The accepted recommendations are presented in this manuscript. HEADINGS – Gastrointestinal hemorrhage. Esophageal and gastric varices. Liver cirrhosis. Infection. Declared conflict of interest of all authors: none Disclosure of funding: no funding received 1 Hospital Português, Salvador, BA, Brasil; 2 Hospital do Coração de São Paulo, SP, Brasil; 3 Departamento de Gastroenterologia da Faculdade de Medicina da Universidade de São Paulo, SP, Brasil; 4 Universidade Federal de Ciências da Saúde de Porto Alegre e Irmandade da Santa Casa de Misericórdia de Porto Alegre, RS, Brasil; 5 Faculdade de Medicina da Universidade Federal de Pernambuco, PE, Brasil. Correspondence: Paulo Lisboa Bittencourt. Rua Prof. Clementino Fraga, 220 / 1901. CEP: 40170-050 – Salvador, BA, Brasil. Email: plbbr@uol.com.br and management of patients with cirrhosis(7,10). This was recently addressed in two evidence-based manuscripts published by the Baveno VI group and the American Association for the Study of Liver Diseases(11,15). Cirrhosis is classified either as compensated (CC) or decompensated (DC). The appearance of clinically relevant events, mainly ascites, VB, hepatic encephalopathy (HE) and infections marks the progression to the decompensated stage(7,10,11,15). Studies on the natural history of cirrhosis have demonstrated different prognoses in patients with CC according to the presence of varices. In patients with DC, prognosis is related to the occurrence of VB combined with one or more clinical decompensating events, such as ascites, infections or HE (Table 1). It is now well known that PH, defined hemodynamically by a hepatic vein pressure gradient AG-2017-79 dx.doi.org/10.1590/S0004-2803.201700000-79 AHEAD OF PRINT TABLE 1. New prognostic grading of patients with advanced liver disease (advanced fibrosis or cirrhosis) according to levels of PH, presence of varices and variceal bleeding and other clinical events Stage Advanced fibrosis or compensated cirrhosis 1 2 3 4 5 HVPG >5 mmHg ≥10 mmHg >12 mmHg 12-20 mmHg >20 mmHg Varices No varices Varices Variceal bleeding Any Clinical events Mild PH CSPH No other events Other non-bleeding events (ascites/HE) 2nd event Mortality (5 year) 1.5% 10% 20% 30% 88% Progression (5 year) 21% to stage 2 1.5% to stage 3 or 4 8% to stage 3 20% to stage 4 45% to stage 5 48% to stage 5 Aims of therapy Prevention of decompensation Prevention of decompensation and variceal bleeding Control of bleeding Prevention of rebleeding and death Prevention of other clinical events and death Assessment for liver transplantation PH: portal hypertension; HVPG: hepatic vein pressure gradient; CSPH: clinical significant PH; HE: hepatic encephalopathy. Adapted from references 7, 10 and 15. Bittencourt PL, Strauss E, Farias AQ, Mattos AA, Lopes EP. Variceal bleeding: update of recommendations from the Brazilian Association of Hepatology Arq Gastroenterol • 2017 [ahead of print] (HVPG) >5 mmHg, is the main cause of hepatic decompensation. Mild PH, characterized by HVPG levels >5 mmHg and <10 mmHg is usually asymptomatic. Clinically significant PH (CSPH), heralded by the presence of HVPG levels >10 mmHg, can be associated with increased risk or occurrence of clinical events leading to DC. VB, either responsive or refractory to standard therapy, is associated with HVPG levels higher than 12 mmHg and 20 mmHg, respectively(7) (Table 1). Drugs such as traditional non-selective betablockers (NSBB), such as propranolol or nadolol, and NSBB with anti-α1 activity, such as carvedilol, were shown to decrease HVPG levels in patients with cirrhosis. Response to drug therapy, defined as a 20% decrease in baseline HVPG values or a decrease in HVPG to levels below 12 mmHg, are usually associated with improved survival. Management of patients with cirrhosis differs according to the magnitude of PH, presence of CC or DC and the size of varices. Several interventions categorized in our first consensus document as pre-primary prophylaxis, primary prophylaxis and secondary prophylaxis of VB are no longer endorsed. They are now better classified as strategies for management of: 1) Patients with CC and mild PH; 2) Patients with CC and CSPH without varices; 3) Patients with CC and either small or medium/large varices; 4) Patients with acute VB; 5) Patients who have recovered from VB(11,15). Major advances were also notable in the noninvasive diagnosis of PH and varices screening. However, very scarce new data are available regarding the management of PHG, GAVE and schistosomal PH. This manuscript updates our previous recommendations for managing VB, taking new evidence-based data and its applicability in Brazil into consideration. An organizing panel of five experts, the same who produced the previous version, reviewed all recommendations according to available data, which were subsequently scrutinized by all members of the Brazilian Association of Hepatology using a web-based approach. The accepted recommendations are presented in this manuscript. Most of these reccomendations are based on new data published since 2010(1-33), which are briefly summarized in Figures 1 to 4. Screening of varices in patients with cirrhotic portal hypertension 2010 Current Rationale Screening of varices using endoscopy Fixed intervals for endoscopic screening Screening of CSPH using non-invasive methods Screening of varices using endoscopy Different intervals for screening according to disease (in)activity Use of non-invasive methods to rule out CSPH to avoid endoscopy(8,11,12) Reversal of fibrosis induced by treatment may halt progression of portal hypertension(11) Management of patients with compensated cirrhosis and no varices 2010 Current Rationale No role for pre-primary prophylaxis According to stage of advanced fibrosis or cirrhosis No role yet for NSBB Statins look promising(1) For prevention of progression it is important to treat the underlying cause of liver disease, and to recommend alcohol abstinence and weight control(11,15) Management of patients with cirrhosis and either small or medium/large varices 2010 Current Rationale According to the size, presence of red signs on varices and Child-Pugh score either traditional NSBB or EVL According to the size, presence of red signs on varices and Child-Pugh score, either traditional NSBB, EVL or carvedilol can be used(27,31) Avoidance of carvedilol in decompensated cirrhosis and caution with the use of NSBB in refractory ascites(3,6,11,15,19,20, 25,26) Statins are promising(1) Hepatic vein pressure gradient-guided therapy useful but not widely available to assess response to NSBB(11,15) Carvedilol promotes greater reduction of portal hypertension when compared to propranolol, and is as efficacious or better than EVL and can be used in patients unresponsive to traditional NSBB(22,27,31) Simvastatin can lower portal hypertension FIGURE 1. Comparison of the 2010 and current recommended strategies for screening and management of patients with portal hypertension. CSPH: clinical significant portal hypertension; EVL: endoscopic variceal ligation; NSBB: non-selective betablockers. Management of patients with acute variceal bleeding Restrictive strategy for fluid and blood product administration. Antibiotics: ceftriaxone vs norfloxacin in all patients. Early intervention with of vasoactive drugs, preferably terlipressin Endoscopy and endoscopic variceal ligation during the first 12 hours after admission. Sengstaken-blakemore tubes (maintained for up to 24 hours in cases of massive bleeding) Restrictive transfusion strategy(32) Antibiotics (ceftriaxone or. Norfloxacin) can be avoided in child-pugh score a patients(29) Early use of vasoactive drugs: terlipressin, somatostatin or octreotide(24,33) Self-expandable esophageal metal stents (for up to 7 days) instead of sengstaken-blakemore tubes for massive variceal bleeding Early tips (in the first 72 hours) for patients with child-pugh score c or b with active bleeding(13,14) Maintenance of hemoglobin levels between 7-9 g/dl improves survival Comparison of vasoactive drugs have demonstrated similar efficacy of terlipressin, somatostatin and octreotide, but doses employed for terlipressin were lower than currently used(24) Fewer side effects of self-expandable esophageal metal stents, when compared to sengstaken-blakemore tubes Early tips strategy reduces rebleeding, treatment failure and improves survival in some(13,14) but not all studies(23) suggesting the need for better patient selection criteria for widespread adoption FIGURE 2. Comparison of the 2010 and current recommended strategies for management of patients with acute variceal bleeding. TIPS: transjugular intrahepatic portosystemic shunt. Bittencourt PL, Strauss E, Farias AQ, Mattos AA, Lopes EP. Variceal bleeding: update of recommendations from the Brazilian Association of Hepatology Arq Gastroenterol • 2017 [ahead of print] Prevention of variceal rebleeding Traditional non-selective betablockers plus endoscopic variceal ligation Traditional non-selective betablockers plus endoscopic variceal ligation Carvedilol not currently recommended(11,15) Propranolol dose must be titrated according to the stage of liver disease(3,19) Simvastatin is promising(2) Non-selective betablockers should be used with caution in patients with refractory a
INTRODUCTION
The Brazilian Association of Hepatology published evidencebased recommendations on the management of variceal bleeding (VB) in the April/June 2010 issue of Archives of Gastroenterology following a consensus meeting held in Salvador on May 6th 2009 (5) .
The first version covered the screening of varices and prevention of the first bleeding episode; treatment of acute VB; management of treatment failure; prevention of recurrent bleeding; portal hypertensive gastropathy (PHG); gastric antral vascular ectasia (GAVE); gastric and ectopic varices and schistosomal portal hypertension (PH). Improved understanding of the natural history and prognosis of PH has led to major changes in definiton, diagnosis, stratification
Variceal bleeding: update of recommendations from the Brazilian Association of Hepatology
and management of patients with cirrhosis (7,10) . This was recently addressed in two evidence-based manuscripts published by the Baveno VI group and the American Association for the Study of Liver Diseases (11,15) . Cirrhosis is classified either as compensated (CC) or decompensated (DC). The appearance of clinically relevant events, mainly ascites, VB, hepatic encephalopathy (HE) and infections marks the progression to the decompensated stage (7,10,11,15) . Studies on the natural history of cirrhosis have demonstrated different prognoses in patients with CC according to the presence of varices. In patients with DC, prognosis is related to the occurrence of VB combined with one or more clinical decompensating events, such as ascites, infections or HE (Table 1). It is now well known that PH, defined hemodynamically by a hepatic vein pressure gradient Aims of therapy Prevention of decompensation (HVPG) >5 mmHg, is the main cause of hepatic decompensation. Mild PH, characterized by HVPG levels >5 mmHg and <10 mmHg is usually asymptomatic. Clinically significant PH (CSPH), heralded by the presence of HVPG levels >10 mmHg, can be associated with increased risk or occurrence of clinical events leading to DC. VB, either responsive or refractory to standard therapy, is associated with HVPG levels higher than 12 mmHg and 20 mmHg, respectively (7) (Table 1). Drugs such as traditional non-selective betablockers (NSBB), such as propranolol or nadolol, and NSBB with anti-α1 activity, such as carvedilol, were shown to decrease HVPG levels in patients with cirrhosis. Response to drug therapy, defined as a 20% decrease in baseline HVPG values or a decrease in HVPG to levels below 12 mmHg, are usually associated with improved survival. Management of patients with cirrhosis differs according to the magnitude of PH, presence of CC or DC and the size of varices. Several interventions categorized in our first consensus document as pre-primary prophylaxis, primary prophylaxis and secondary prophylaxis of VB are no longer endorsed. They are now better classified as strategies for management of: 1) Patients with CC and mild PH; 2) Patients with CC and CSPH without varices; 3) Patients with CC and either small or medium/large varices; 4) Patients with acute VB; 5) Patients who have recovered from VB (11,15).
Major advances were also notable in the noninvasive diagnosis of PH and varices screening. However, very scarce new data are available regarding the management of PHG, GAVE and schistosomal PH.
This manuscript updates our previous recommendations for managing VB, taking new evidence-based data and its applicability in Brazil into consideration. An organizing panel of five experts, the same who produced the previous version, reviewed all recommendations according to available data, which were subsequently scrutinized by all members of the Brazilian Association of Hepatology using a web-based approach. The accepted recommendations are presented in this manuscript. Most of these reccomendations are based on new data published since 2010 Use of non-invasive methods to rule out CSPH to avoid endoscopy (8,11,12) Reversal of fibrosis induced by treatment may halt progression of portal hypertension (11)
Management of patients with compensated cirrhosis and no varices 2010 Current Rationale
No role for pre-primary prophylaxis According to stage of advanced fibrosis or cirrhosis No role yet for NSBB Statins look promising (1) For prevention of progression it is important to treat the underlying cause of liver disease, and to recommend alcohol abstinence and weight control (11,15)
Management of patients with cirrhosis and either small or medium/large varices 2010 Current Rationale
According to the size, presence of red signs on varices and Child-Pugh score either traditional NSBB or EVL According to the size, presence of red signs on varices and Child-Pugh score, either traditional NSBB, EVL or carvedilol can be used (27,31) Avoidance of carvedilol in decompensated cirrhosis and caution with the use of NSBB in refractory ascites (3,6,11,15,19,20,25,26) Statins are promising (1) Hepatic vein pressure gradient-guided therapy useful but not widely available to assess response to NSBB (11,15) Carvedilol promotes greater reduction of portal hypertension when compared to propranolol, and is as efficacious or better than EVL and can be used in patients unresponsive to traditional NSBB (22,27,31) Simvastatin can lower portal hypertension
Management of patients with acute variceal bleeding
Restrictive strategy for fluid and blood product administration.
Antibiotics: ceftriaxone vs norfloxacin in all patients.
Early intervention with of vasoactive drugs, preferably terlipressin Endoscopy and endoscopic variceal ligation during the first 12 hours after admission.
Sengstaken-blakemore tubes (maintained for up to 24 hours in cases of massive bleeding) Restrictive transfusion strategy (32) Antibiotics (ceftriaxone or. Norfloxacin) can be avoided in child-pugh score a patients (29) Early use of vasoactive drugs: terlipressin, somatostatin or octreotide (24,33) Self-expandable esophageal metal stents (for up to 7 days) instead of sengstaken-blakemore tubes for massive variceal bleeding Early tips (in the first 72 hours) for patients with child-pugh score c or b with active bleeding (13,14) Maintenance of hemoglobin levels between 7-9 g/dl improves survival Comparison of vasoactive drugs have demonstrated similar efficacy of terlipressin, somatostatin and octreotide, but doses employed for terlipressin were lower than currently used (24) Fewer side effects of self-expandable esophageal metal stents, when compared to sengstaken-blakemore tubes Early tips strategy reduces rebleeding, treatment failure and improves survival in some (13,14) but not all studies (23) suggesting the need for better patient selection criteria for widespread adoption
Prevention of variceal rebleeding
Traditional non-selective betablockers plus endoscopic variceal ligation
Traditional non-selective betablockers plus endoscopic variceal ligation
Carvedilol not currently recommended (11,15) Propranolol dose must be titrated according to the stage of liver disease (3,19) Simvastatin is promising (2) Non-selective betablockers should be used with caution in patients with refractory ascites and should be withdrawn in the presence of hypotension, hyponatremia or acute kidney injury, because of their detrimental effect on survival Doses should not be increased beyond 160 mg/day in subjects with large or refractory ascites (3,19) Vasoactive drugs Endoscopic therapy with laser or argon plasma coagulation.
TIPS or surgery in refractory cases
Vasoactive drugs Endoscopic therapy with laser or argon plasma coagulation or hemostatic powder (28) TIPS or even surgery for patients refractory to treatment Hemostatic powder can provide a mechanical barrier over the bleeding point, leading to hemostasis. Experience is too limited.
The value of the technique is yet to be proven.
Gastric variceal bleeding 2010 Current Rationale
Endoscopic cyanoacrilate (preferred) TIPS BRTO Endoscopic cyanoacrilate injection (preferred) TIPS BRTO Endoscopic ultrasound-assisted coil deployment for gastric varices with or without concomitant cyanoacrylate glue injection for selected patients (4,16) Coil provides a special matrix for cyanoacrilate adherence and theoretically decreases the risk of glue embolization. The transesophageal approach avoids interference from food or clots on visualization of fundal varices (few data). Postoperative endoscopic SCL promotes faster variceal eradication and decreases bleeding (9) Decreasing portal vein pressure with surgical treatment enhances the results of the endoscopic treatment. Although valuable, this combination has high incidence of asymptomatic portal vein thrombosis and has not been compared with. EVL, which is the standard endoscopic treatment (9) . III. Management of patients with cirrhosis and either small or medium/large varices 1) Patients with small varices and Child-Pugh A or B cirrhosis without red signs on varices may benefit from primary prophylaxis, but there is insufficient evidence to support a recommendation. 2) Patients with small varices with red signs on varices and/ or advanced cirrhosis (Child-Pugh C) have high risk of bleeding and should be submitted to primary prophylaxis with traditional NSBB or carvedilol.
3) Patients with medium or large varices should be submitted
to prophylaxis independently of the presence of advanced liver disease or red signs on varices. Treatment options include traditional NSBB, carvedilol or endoscopic variceal ligation (EVL). Sclerotherapy (SCL) should be proscribed due to its adverse impact on patient survival. 4) Propranolol or nadolol (although no longer commercially available in Brazil) are traditional NSBB that can be used in patients with CC and DC. Propranolol should be used with initial oral doses of 20-40 mg twice a day, titrated up to 160-320 mg/day to maintain heart rate (HR) between 55-60 bpm and systolic blood pressure >90 mmHg. In patients with ascites, the dose of propranolol should be increased gradually to no more than 160 mg/day, and should be discontinued in patients with large or refractory ascites, acute kidney injury, hyponatremia (Na levels <130 mEq/L) or systolic blood pressure <90 mmHg. 5) Carvedilol should be started in oral doses of 3.125 mg twice a day, and titrated to 6.25 mg twice a day. Doses should not be further increased unless there is evidence of arterial hypertension. Conversely, the dose should be decreased or treatment discontinued if systolic blood pressure falls below 90 mmHg. Because Carvedilol lowers systolic blood pressure, caution is required in patients with ascites. It should be avoided in patients with refractory ascites. 6) EVL should be performed every 2-4 weeks until eradication of varices. Endoscopy should be repeated 3 months after eradication and thereafter every 6-12 months. 7) Therapeutic strategy should be tailored according to the patient's characteristics and preferences, contraindications, adverse events, availability of local resources and expertise. Propranolol intolerant patients should be switched either to carvedilol or EVL. 8) There is no rationale for any combination therapy of drugs (propranolol, carvedilol or nitrates), or any of them with EVL, in order to prevent the first VB episode in cirrhosis. 9) There is no need for surveillance endoscopy in patients under primary prophylaxis. 10) HVPG measurement before and after initiation of pharmacological therapy is valuable in assessing hemodynamic response to propranolol or carvedilol, defined as a 20% decrease in HVPG baseline values, or a decrease to lower than 12 mmHg.
IV. Management of patients with acute VB
1) Acute VB should be initially managed in the intensive care unit. 2) Fluid resuscitation should be employed with caution in order to maintain hemodynamic stability. 3) Airway protection is advisable for patients with impaired consciousness and massive hematemesis, and those who require Sengstaken-Blakemore (SB) tubes. 4) Use of SB tubes at admission for a maximum of 24 hours should be reserved for cases of massive hemorrhage with hemodynamic compromise and for patients who are not responsive to intravenous fluids, as a bridging therapy until definitive treatment can be administered. Self-expanding covered esophageal metal stents, wherever available, are a safer alternative to SB tubes. 5) Red blood cell transfusions should aim for hemoglobin levels between 7-9 g/dL, but higher hemoglobin levels may be necessary depending on patient's age, comorbidities and ongoing bleeding. 6) INR should not be used to guide transfusional policy.
There are no data to support recommendations for management of coagulopathy and thrombocytopenia in patients with cirrhosis. 7) Vasoactive drugs should be started as early as possible, even prior to endoscopy, in patients with suspicion of VB. 8) Terlipressin (TL), somatostatin (SMT) or octreotide (OCT) should be used according to patient's characteristics, and taking cost into account. With currently recommended doses, TL, when compared to SMT or OCT, induces a greater and more sustained decrease in portal pressure. Some studies have shown a beneficial effect on survival when TL is used alone or combined with sclerotherapy (SCL), when compared to placebo or SCL alone. However, more recent data has challenged this finding, and do not demonstrate a clear benefit of any one drug over the others. Use of TL is not advisable in patients with coronary heart disease, severe peripheral vascular disease and uncontrolled arterial hypertension, and the use of vasopressin with or without nitrates should be abandoned in the management of VB. 9) TL should be administered as an intravenous bolus dose of 2 mg followed by 1-2 mg (depending on patient's weight) every 4 hours during the initial 48 hours after admission. The dose should be reduced to a maintenance dose of 1 mg every 4 hours. SMT should be administered as an intravenous bolus dose of 250 µg followed by a continuous infusion of 250 µg/h. OCT should be administered as an intravenous bolus dose of 50-100 µg followed by a continuous infusion of 50 µg/h. These drugs should be adminstered for 5 days in order to prevent variceal rebleeding. 10) Upper gastrointestinal endoscopy should ideally be performed during the first 12 hours of bleeding from esophageal varices. 11) Prior to endoscopy, airway protection is recommended in patients with massive bleeding, grades III and IV hepatic encephalopathy or respiratory failure. In the absence of contraindication, 250 mg of intravenous erythromycin over 30-120 minutes should also be administered to improve the visual field of the operator during endoscopy.
12) EVL should be the first choice for the endoscopic treatment, but SCL is an option when EVL is unavailable or technically not feasible. 13) Combined endoscopic and pharmacological treatment with vasoactive drugs is superior to either treatment alone. Combination therapy is recommended for all patients with suspected VB, but pharmacological monotherapy is capable of controlling VB when therapeutic endoscopy is not immediately available. Vasoactive drugs should be administered immediately at admission. If varices are excluded as the source of bleeding, drug therapy must be discontinued. 14) Infections, particularly urinary tract infections, spontaneous bacterial peritonitis and lower respiratory tract infections should be screened for in all patients with VB. 15) Screening for infections should at minimum include blood cultures, ascitic fluid culture and biochemistry tests, urine sediment analysis and chest X ray. 16) Antibiotic prophylaxis is mandatory, preferably before endoscopy, to reduce the risk of infection, variceal rebleeding and mortality. 17) Oral quinolones, preferably norfloxacin twice a day (400 mg), or third-generation cephalosporin, usually intravenous ceftriaxone 1 g per day, is recommended up to 7 days. Patients with advanced cirrhosis and/or hemodynamic instability should ideally receive ceftriaxone. 18) Based on current data, there is no evidence to recommend any prophylaxis for hepatic encephalopathy in patients with VB. 19) Even though hypovolemia is the most common cause of acute kidney injury in patients with VB, the possibility of hepatorenal syndrome should be taken into account, and treated with combination of albumin and either terlipressin or norepinephrine. 20) Due to its positive impact on survival, early (transjugular intrahepatic portosystemic shunt) TIPS placement has been recommended for high-risk patients (Child-Pugh C with 10-13 points or B with active bleeding). However, refinement of these criteria is needed before it can be generally recommended. In this regard, it is important to emphasize that this strategy is unavailable in most Brazilian centers due to its cost and a lack of local operator expertise. 21) In cases of treatment failure, salvage placement of PTFEcovered TIPS should be recommended, but additional endoscopic hemostasis could also be attempted while waiting for TIPS placement, if it is not immediately available. 22) SB tubes or esophageal stents remain options for patients with massive bleeding as a bridging therapy to a definitive treatment. 23) Shunt surgery should be performed only in patients with well-preserved liver function, and if TIPS is unavailable and bleeding is uncontrolled or recurs after the second therapeutic endoscopy.
Prevention of recurrent VB 1) Combination of traditional NSBB and EVL is recommended for prevention of recurrent VB in patients with cirrhosis. Carvedilol cannot be recommended for prevention of rebleeding at present because it has not been compared to the standard treatment, EVL + traditional NSBB. However, the use of carvedilol and statins for secondary prevention shows potential for future applications.
2) Propranolol or nadolol (not commercially available in Brazil) are traditional NSBB that can be used in patients with CC and DC. Propranolol should be administered in initial oral doses of 20-40 mg twice a day, titrated up to 160-320 mg/day to maintain heart rate (HR) between 55-60 bpm and systolic blood pressure >90 mmHg. In patients with ascites, the dose of propranolol should be increased gradually to values of no greater than 160 mg/day and should be discontinued in patients with grade III or refractory ascites in the presence of acute kidney injury, hyponatremia (Na levels <130 mEq/L) and systolic blood pressure <90 mmHg. 3) EVL should be performed every 2-4 weeks until variceal eradication is achieved. Endoscopy should be repeated 3 months after eradication and thereafter every 6-12 months. 4) PTFE-covered TIPS is recommended for patients with recurrent bleeding despite undergoing treatment with EVL + traditional NSBB. 5) Surgical shunts should only be considered when TIPS is unavailable. When surgery is required as rescue therapy, selective splenorenal, mesocaval or 8mm H-graft portacaval shunts are preferable. The procedure requires a high level of surgical expertise, and should be performed in a specialist center in order to attain positive outcomes. Liver transplantation (LT) is a better treatment option than surgical shunts for patients with high MELD scores.
V. Management of PHG and gastric antral vascular ectasia (GAVE)
1) PHG and GAVE are causes of upper gastrointestinal bleeding in patients with cirrhotic or non-cirrhotic PH, They are considered two separate conditions with distinct management options. 2) Due to a lack of data, no recommendations can be drawn for primary prophylaxis of bleeding in PHG. 3) Traditional NSBB and iron supplementation are the frontline treatment for obscure blood losses from PHG. 4) Injection, thermal or mechanical methods of endoscopic therapy may be attempted in patients with PHG or GAVE with treatable lesions identified at endoscopy. The most commonly used method is argon plasma coagulation (APC), but hemostatic powder, clips and band ligation may also be used. 5) In patients with acute bleeding from PHG, vasoactive drugs (TL, SMT or OCT) can be administered, although data are scarce on its efficacy. Traditional NSBB should be introduced once the acute bleeding is controlled. There is no role for vasoactive drugs or NSBB for treatment of GAVE. 6) Preferably TIPS, or shunt surgery can be tried as a last resort for patients with PHG refractory to standard treatment, but there is no role for them in treatment of GAVE.
VI. Management of gastric and ectopic varices
1) There are no data regarding primary prophylaxis of bleeding from gastric varices. As Traditional NSBB reduces portal pressure, they are an acceptable treatment option. 2) The recommendations for management of type 1 GOV are the same as for esophageal varices. 3) Cyanoacrylate endoscopic injection is the preferred treatment of bleeding caused by type 2 GOV and type 1 IGV. Traditional NSBB should be introduced after bleeding is controlled. However, as cyanoacrylate endoscopic injection can induce fatal thromboembolic events, it should be avoided in patients with hepatopulmonary syndrome and intracardiac shunts. 4) TIPS should be considered as a rescue therapy if active or recurrent bleeding cannot be controlled. 5) Balloon occluded retrograde transvenous obliteration (BRTO) and endoscopic ultrasound-guided coil and cyanoacrylate injection may be used in selected patients with gastric varices and active or recurrent bleeding, but these methods have not gained widespread application due to scarcity of data recommending their employment. They also require equipment and expertise that few centers have. 6) There are insufficient data to make a recommendation concerning ectopic varices, including IGV type 2. Depending on variceal size, cyanoacrylate endoscopic injection, band ligation, endoscopic ultrasound-guided coil and cyanoacrylate injection, TIPS and BRTO can be used.
VII. Management of PH in patients with extrahepatic portal vein obstruction 1) For chronic extrahepatic portal vein obstruction (EHPVO), there is insufficient evidence to recommend traditional NSBB or endoscopic treatment for primary prophylaxis of VB. Both treatment options are acceptable for patients at risk of bleeding. 2) In the absence of specific data, management of acute VB should follow the same measures currently employed in management of cirrhosis, including vasoactive drugs and EVL. 3) EVL or traditional NSBB may be used as secondary prophylaxis, because they are safe and efficient. There are no data that support the use EVL and NSBB in combination to prevent recurrent bleeding in EHPVO. 4) In chronic EHPVO, anticoagulation, whenever indicated, should be started after adequate prophylaxis for VB.
VIII. Management of schistosomal PH
1) In PH caused by schistosomiasis, there are few data that suggest the efficacy of traditional NSBB or EVL as primary prophylaxis of VB. However, both strategies are currently used for patients at risk of bleeding. SCL must be avoided in patients that have not already had a bleeding event. 2) While there is a lack of data, the same treatment options used to control acute VB in cirrhosis can be used by analogy on patients with PH caused by schistosomiasis. 3) For secondary prophylaxis of VB, either EVL alone or a combination of EVL and traditional NSBB are acceptable, but the evidence for their use is not strong. SCL may be used if EVL is not available or not feasible. 4) There are no data to recommend surgery over EVL or traditional NSBB for secondary prophylaxis in PH due to schistosomiasis. Due to its efficacy and safety, EVL alone or in combination with traditional NSBB is recommended. Surgery is recommended as rescue therapy if endoscopic or combined treatment fails. 5) Surgical treatment options influence clinical outcomes.
Azygos-portal devacularization with splenectomy is the preferred surgical procedure for PH caused by schistosomiasis. Postoperative endoscopic treatment improves variceal eradication and decreases bleeding recurrence.
Authors' contributions
Bittencourt PL, Strauss E, Farias AQ and Mattos AA reviewed data and updated the recommnedations. Lopes EP made the final critical revision of the manuscript. | 2018-04-03T01:11:30.170Z | 2017-10-02T00:00:00.000 | {
"year": 2017,
"sha1": "3d26ff8239cd1b627bcac387a6c2bc5c11677ff4",
"oa_license": "CCBYNC",
"oa_url": "http://www.scielo.br/pdf/ag/v54n4/1678-4219-ag-s0004280320170000079.pdf",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "b3bf9a0d3f766c61adcea6d5eda74c2be504128c",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
214268493 | pes2o/s2orc | v3-fos-license | A poem wants to change a perspective on the world. That is its ambition: An Interview with Pat Boran
Pat Boran attended the conference ‘Irish Itinerary 2018 (EFACIS): Trauma and Identity in Contemporary Irish Literature and Culture’ at the University of La Rioja, Spain. The following interview took place there on 15 February 2018, and covered Spanish translations of his work, poetry writing, formal innovation, ecocriticism, Imagism, photography, friendship, Irish poetry and broadcasting. Boran offered a number of insightful responses and shared his most honest thoughts on aesthetics and motivations to write poetry. Portlaoise-born poet, writer and broadcaster Pat Boran currently lives in Dublin. He is an elected member of Aosdána, the Irish association which honours distinguished artistic work. He is one of the most widely acclaimed Irish poets of his generation. His work has been translated into several languages and received numerous awards. In 2008, he received the Lawrence O’Shaughnessy Poetry Award of the University of St Thomas, St Paul, Minnesota. He has published more than a dozen books of poetry and prose – among them The Next Life (2012), Waveforms: Bull Island Haiku (2016), and A Man is Only as Good: A Pocket Selected Poems (2017), as well as the humorous memoir The Invisible Prison (2009) and the popular writers’ handbook The Portable Creative Writing Workshop. Besides these published works, Pat is a former presenter of The Poetry Programme and The Enchanted Way on RTÉ Radio 1, and works part-time as a literary editor of Dedalus Press. He has edited several anthologies of prose and poetry, for example, with Gerard Smyth, the anthology If Ever You Go: A Map of Dublin in Poetry and Song, the Dublin: One City, One Book designated title for 2014, and, with Eugene O’Connell, The Deep Heart’s Core (2017).
PB:
We have this idea that with the Romantic poets, what made them so great was that delivery of emotion, but actually what makes them good, and sometimes great, is the presentation, the physical rememberment, if I can put it like that, of that emotion, and the connecting of the things that they can relate to us. When someone asks you to look at something that has drawn her attention, the act of looking has the power to transfer to you the emotion that makes that original perception possible. My job is to see the world as accurately as I can, from where I am, in space and time, and on my emotional see-saw (as we all are), and trust in that. That is the boat in which I put my perception, my being. Otherwise, there is nothing. If there is an analogy between the scientific approach and that of the poet, it is that exact, careful and precise observation is essential to both. The poet can learn a great detail from the scientific method, striving for accuracy and detail, a kind of empirical subjectivity, I might say, in which the unknown and the contradictory is never to be avoided but the vague flourish or neat conclusion is always suspect.
MT: Some other poems of yours use natural elements recurrently. "Waving" is a good example of it. What do you make of this idiosyncratic aspect of your poetry? PB: Poems, for me, often being as visual observations and animate the way a short film might in the mind of a director. If I feel I can see my way in, and then see my way back out of the world of a poem, that is often enough to persuade me it might be worth spending time in its imagined landscape. The world of the poem might have a lot in common with the world of the writer; in the act of composition it might even stand for the real world, but they are still different and distinct places. The world of the poem is a simplification, an editing down and a condensing of elements that may result in amplifications and even distortions. And what is true, or feels true, in a poem may not be so in the real world, and vice versa.
When someone travels and wants to impart some of the experience of that travel, the obvious method is to draw on the senses and to remake aspects of that observed place, in words or pictures, say, in order to set the stage for the emotions, thoughts and revelations encountered there. So too it is with poems. Feelings, thoughts, perceptions must be embodied; the 'where' and the 'when' of an experience do not just limit or define but also enable that experience, and allow for its recollection, its further exploration and re-examination.
In making a poem, in being, as it were, the first person to enter a world and to have an experience, one wants to recreate for the reader (which, of course, includes the poet him/herself) the particulars of that experience. For this reason alone the sequence, colour, tone and intensity of observations are fundamental. The techniques of the storyteller, the journalist, the painter, the musician -all have to be drawn on in order to produce the most affecting possible experience, whether or not the world described be naturalistic or otherwise. The result of that kind of attention to sequence and detail is often a temporal dimension, a sense in the architecture of the poem that movement not only in place but, deriving from same, in time is also possible. Even in a short lyric, the reader's journey takes time, and the discoveries along that timeline (the things that will determine whether or not the poem succeeds and is worth the patience and commitment of a reader) run in parallel to those of the writer. Just as the reader may, the writer must go through some kind of discovery, change, growth. That seems to me to be the purpose of poetry. Not just to record discover, change, growth, but in fact to enable it.
MT: Now, I would like to talk about your latest work, Waveforms: Bull Island Haiku, because it is quite unique 1 . The following haiku is a good example: Two boys with a kite made from twigs and plastic bags.
Wind shrugs: "Oh, all right." You have expressed reservations about what are often seen as 'traditional' haiku in English, where humour, detailed sensual observation and linguistic tension appear unwelcome or are sacrificed in order to present a determinedly 'watercolour' view of the world. However, the haiku I have just read, "Two Boys with a Kite", and many others from Waveforms, are as economic, full of wisdom, personal and scientific as many visual haikus written by some Modernist English authors, those written by Ezra Pound, for example. What's more, your visual work or photography in Waveforms is as meaningful as the haikus, and the visual aspects of your haikus are evocative as the words themselves. Here, all the reader's senses work at once in order to grasp the sense of the image captured in words. Has Imagism influenced your haikus?
PB: Just to be clear, it's not so much that I dislike 'traditional' haiku in English (there are many great and inspiring examples), it's just that so often the essential perception and moment of realization seem missing for me; often it seems that the haiku writer in trying out a new form abandons all of his or her acquired skills and produces a kind of pastiche rather than an actual poem. For me, I wanted to keep the precision and focus of what I think are the best poems, in whatever form, and in some way marry what I was already doing to a tradition that has in many ways evolved independently, and in doing so discover some new perspectives and ideas. Of course, as has been noted by others before, there are also interesting similarities between the haiku tradition and the small nature poems to be found as marginalia or verbal doodles in the Irish monastic tradition. If the form is not exactly the same, the intention is often similar. It struck me too that when Wordsworth went to write his poem about daffodils, he started out with "I wandered lonely as a cloud" etc., etc., the typical approach of the western writer to nature: here I am, and here I go a-wandering, and, oh, look over there … and then he, at last, turns his attention to the daffodils. The haiku poet, not least because the form demands it, cuts straight to the subject: 'Golden daffodils / beside the lake, beneath the trees, / dancing in the breeze' or whatever. And then, once I had formulated that general idea in my head, I wondered about the 3-line form itself, about how the sequence could make all the difference, or at least subtle differences. Thus, I might try out, for instance: 'Beside the lake, beneath the trees, / dancing in the breeze -/ golden daffodils', which creates a different kind of narrative, a different kind of discovery or relevation. After that, I started thinking of the thorny matter of syllable count. And, after that again, I wondered if I could make things harder on my lazy mind by looking to recognize the language's tendency to rhyme. And out of all that juggling and revising, I ended up with a form that over a year delivered to me about 250 haiku, which were whittled down to the selection that make up the book.
MT:
How have the experiences captured in these photos affected your decision to write haikus in particular? I mean, does the visual aspect of this poetry have any relationship with the characteristics of photography as an art form in your poetry? And its pocket-sized format?
PB:
The connection between the photographer's focus (pointing out and away from himself) and that of the haiku poet was one I found very inspiring. And though the photographs were taken digitally in what is known as RAW format (i.e. with colour and a lot of other information till intact), I further simplified the idea by reproducing them, in this little edition at least, in monochrome and on basic bookwove paper rather than on photographic paper. As I might have said elsewhere, for me one of the real liberations of this project was the large number of limitations that it imposed on me. Without them I don't think I would have stayed with it or, certainly, ended up with a book-length volume.
MT: Pat, is there any relationship between Waveforms and the poem "Waving"? If so, could you explain its significance?
PB: It's interesting you ask about a relationship. It had not occurred to me before but, yes, I can see there is one there. The poem 'Waving' approaches its conclusion by saying that 'whole humans -arms, legs, backs and bellies -/ are waving away, flickering on and off / in time and space …' And I suppose that is a direct prompt (although it took me nearly 20 years to respond to it) to the haiku that make up 'Waveforms'. As I say in the afterword to the book, I had recently lost my close friend and had taken to walking every day on Bull Island, where the book is set, often finding myself looking at only a vague horizon through a haze or mist or what passes for a sand storm in Dublin Bay. And that sense of everything scarcely holding together, of being on the point of blowing apart, was very much what I felt in my period of loss. At least, until I realized that that same sand storm (that felt so painful and, as it happens, on one occasion almost destroyed my camera) was the island creating itself under my feet. The chaos of emotion I felt was not the end as one might easily have interpreted it but the on-going creation of something. These tiny grains that were stinging my eyes were the stuff the world was made of. It seemed only obvious and natural to respond to them though haiku, to write a kind of creation myth that begins with nothingness ('First, a mystery, / the absence of things …' then records the arrival of the sand itself / Grain by tumbling grain, the world forms before our eyes, and may fade again and, after that, aspects of the flora and fauna that not only depend on the new island but, literally, hold it together. Light-headed or not -/ hills of swaying marram grass / rooted to the spot. After that it was a case of turning up every day and paying attention.
MT: Quite recently, The Irish Times website published the tributes of a wide range of poets and writers whose lives and work were enriched by Irish poet Philip Casey 2 . You contributed to that online archive with a prose recollection and introduction to his poem 'Machine Buried'. I know you were a very good friend of Casey and you admired the magic and depth of his poetry. In your tribute, you said something that I found particularly evocative: 'We lose our loves and our friends, but something we write as in a dream, or stumble upon by accident in a public library on a rainy afternoon, becomes our farewell message to the world, and someone's lifelong companion'. Would you mind telling us what makes 'Machine Buried' special to you? Which poem would you like to be your farewell message to the world, and why?
PB:
The loss of Philip Casey was, in so many ways, a huge loss to contemporary Irish poetry, though Philip was the last person who would have thought of himself in such a way. Not only a sensitive, thought-provoking poet and novelist, Philip was also a real champion of writers and writing, of the whole idea of a writing culture. For instance, out of his own energy and enthusiasm, and without payment of any kind, for many years he maintained a website presenting biographical and bibliographical information on many hundreds of Irish poets and writers, when none of the official institutions charged with the promotion of Irish writing was able to do the same. He did this because he truly believed that writing is a meritocracy, that good writing will find its way in the world and that all of us in the writing world owe it to each other as well as to ourselves to encourage and support. Because the truth is, none of us knows where the next great work of literature will come from. It might be from a Nobel Laureate or it might be from a schoolchild up the road or a newly arrived immigrant writer putting her thoughts into words for the first time. As I mentioned in that piece about Philip and his poem 'Machine Buried', which is far from his most accomplished piece, sometimes it is hard to say why a poem continues to haunt us or work its particular magic on us. But it is so often true that the poem that someone makes, almost by accident, almost automatically (with little planning or preparation or thought) becomes the poem they struggle to repeat for the rest of their lives, while the poem that is worked on over a long period and indeed almost perfected (if such a thing were possible) often appears overworked and cold and lifeless when returned to years later. The mystery of the process is that good poems (poems that seem as fresh as the day they were first written) almost always involve a lot of luck, a sense that the writer only had to keep up with whatever was happening or being channelled through the air at that time. Of course, there is often a considerable amount of rewriting to be done and, even where the rewriting is only minimal, the job of changing a single word can be much more difficult than rewriting a whole poem: that is the nature of the small machine that is a lyric poem. | 2020-02-13T09:11:09.236Z | 2019-07-17T00:00:00.000 | {
"year": 2019,
"sha1": "f97d764696bde962cfa5ea4aeec27f8818602554",
"oa_license": "CCBY",
"oa_url": "http://revistas.fflch.usp.br/abei/article/download/3266/2828",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "5861f7a2a1396d8c3daf67bb99c414751618b43a",
"s2fieldsofstudy": [
"Art"
],
"extfieldsofstudy": [
"Art"
]
} |
247857250 | pes2o/s2orc | v3-fos-license | Application of B cell immortalization for the isolation of antibodies and B cell clones from vaccine and infection settings
The isolation and characterization of neutralizing antibodies from infection and vaccine settings informs future vaccine design, and methodologies that streamline the isolation of antibodies and the generation of B cell clones are of great interest. Retroviral transduction to express Bcl-6 and Bcl-xL and transform primary B cells has been shown to promote long-term B cell survival and antibody secretion in vitro, and can be used to isolate antibodies from memory B cells. However, application of this methodology to B cell subsets from different tissues and B cells from chronically infected individuals has not been well characterized. Here, we characterize Bcl-6/Bcl-xL B cell immortalization across multiple tissue types and B cell subsets in healthy and HIV-1 infected individuals, as well as individuals recovering from malaria. In healthy individuals, naïve and memory B cell subsets from PBMCs and tonsil tissue transformed with similar efficiencies, and displayed similar characteristics with respect to their longevity and immunoglobulin secretion. In HIV-1-viremic individuals or in individuals with recent malaria infections, the exhausted CD27-CD21- memory B cells transformed with lower efficiency, but the transformed B cells expanded and secreted IgG with similar efficiency. Importantly, we show that this methodology can be used to isolate broadly neutralizing antibodies from HIV-infected individuals. Overall, we demonstrate that Bcl-6/Bcl-xL B cell immortalization can be used to isolate antibodies and generate B cell clones from different B cell populations, albeit with varying efficiencies.
The isolation and characterization of neutralizing antibodies from infection and vaccine settings informs future vaccine design, and methodologies that streamline the isolation of antibodies and the generation of B cell clones are of great interest. Retroviral transduction to express Bcl-6 and Bcl-xL and transform primary B cells has been shown to promote long-term B cell survival and antibody secretion in vitro, and can be used to isolate antibodies from memory B cells. However, application of this methodology to B cell subsets from different tissues and B cells from chronically infected individuals has not been well characterized. Here, we characterize Bcl-6/Bcl-xL B cell immortalization across multiple tissue types and B cell subsets in healthy and HIV-1 infected individuals, as well as individuals recovering from malaria. In healthy individuals, naïve and memory B cell subsets from PBMCs and tonsil tissue transformed with similar efficiencies, and displayed similar characteristics with respect to their longevity and immunoglobulin secretion. In HIV-1-viremic individuals or in individuals with recent malaria infections, the exhausted CD27 -CD21memory B cells transformed with lower efficiency, but the transformed B cells expanded and secreted IgG with similar efficiency. Importantly, we show that this methodology can be used to isolate broadly neutralizing antibodies from HIV-infected individuals. Overall, we demonstrate that Bcl-6/Bcl-xL B cell
Introduction
The isolation and characterization of antibodies from infection settings and vaccine trials in both humans and animal models can inform rational vaccine development. A better understanding of neutralizing antibodies, including their longitudinal maturation and binding specificity can aid in the development of new antigens and vaccine regimens. In addition, several neutralizing antibodies have either been approved for use, or are currently being tested for use as therapeutics to prevent or treat infectious diseases, such as in the cases of Ebola virus, respiratory syncytial virus (RSV), SARS-CoV-2 and Human Immunodeficiency virus (HIV) (1-4). Thus, efficient strategies to isolate antibodies from memory B cells following infection or vaccination are needed to not only inform future vaccine design, but also to identify novel therapeutics.
Multiple strategies exist to isolate antigen-specific B cells (5). One strategy includes using a recombinant fluorophore-labeled antigen probe to isolate antigen-specific B cells by flow cytometry for subsequent sequencing of the VH region (6,7). While this strategy enriches for B cells based on antigen binding, antibody function can only be tested after antibody sequencing, cloning, expression and purification. Another method relies on the culture of B cells in vitro, where after 10-14 days of culture with cytokines and CD40 Ligand, activated B cells secrete immunoglobulin into culture supernatant and differentiate into plasma cells before cell death (8). This strategy enables screening for antibody function using immunoglobulin-containing culture supernatant, but the small-scale volume of culture supernatant available limits the breadth of screening that can be performed.
An alternative method that can combine both antigenb in d in g f o r B c e l l e n r i c hm e n t a n d s c r e e n in g o f immunoglobulin-containing culture supernatant for antibody function has been applied to isolate rare antibodies from IgG + and IgM + memory B cells in PBMCs (9)(10)(11). This method relies on the retroviral expression of the transcription factor Bcl-6 and anti-apoptotic molecule Bcl-xL in primary memory B cells. When cultured in the presence of IL-21 and CD40 ligand, memory B cells transformed by Bcl-6 and Bcl-xL survive longterm in vitro, secrete antibodies into the culture supernatant, and continue to express the B cell receptor (BCR) on the cell surface (9). Thus, Bcl-6/Bcl-xL immortalization of memory B cells provides a flexible tool for antibody isolation where one can enrich for antigen-specific B cells by flow cytometry using a fluorophore-labeled antigen probe, and also screen for antibody function with a renewable source of culture supernatant.
The efficiency and application of Bcl-6 and Bcl-xL expression in specific B cell subsets has not been fully explored, and in select conditions this technology may be limited. Some level of B cell activation and proliferation must occur in vitro to enable retroviral transduction and Bcl-6 and Bcl-xL expression. Hence, B cell subsets that have a limited ability to proliferate and/or readily undergo in vitro cell death may not be as amenable to B cell transformation.
For example, in cases where biopsy samples or fine needle aspirates of draining lymph nodes are available, antigen-specific B cell responses from B cell follicles and germinal centers could be studied. However, germinal center B cells isolated from secondary lymphoid organs (SLO) more readily undergo cell death during in vitro culture and antibody discovery techniques that rely on longterm B cell culture may be limited (12). In chronic infection settings such as HIV-1, malaria or tuberculosis, the memory B cell population characterized by low expression of CD27 and CD21, sometimes referred to as "atypical", "tissue-like" or "exhausted", can comprise up to 40-50% of the memory B cell pool in PBMCs (13-17). In general, CD27 -CD21 -B cells are characterized by the high expression of inhibitory receptors and have an impaired ability to proliferate and produce antibodies when stimulated (13, 18). Nevertheless, HIV-and malaria-specific B cells can be found within CD27 -CD21 -B cell subsets, so isolation and characterization of antibodies from these subsets is of interest (19,20).
In addition to antibody isolation from memory B cell subsets, immortalized B cells can be used for other applications such as the characterization of naïve B cell repertoires, as has been described for HIV-1-specific precursor B cells (21), as well as the generation of B cell clones that can serve as experimental controls. Clonal B cell lines can serve as valuable tools in the development of new reagents and assays. As novel antigens are developed to serve as probes for flow cytometric isolation and characterization of antigen-specific B cells, immortalized clones can function as controls for testing probe sensitivity and specificity, as well as being used to determine the level of detection when sorting rare B cells by flow cytometry.
Here, we explore the efficacy of Bcl-6 and Bcl-xL-mediated B cell immortalization to isolate B cell clones and antibodies from naïve and memory B cell subsets obtained from PBMCs and tonsil tissue. In addition, we compare the efficiency of Bcl-6/Bcl-xL B cell immortalization of CD27 + CD21 + and CD27 -CD21 -B cells from HIV-1-infected individuals and from malaria convalescent individuals, and demonstrate a strategy to isolate HIV-1-specific neutralizing antibodies by combining Bcl-6/Bcl-xL B cell immortalization with the TZM-bl neutralization assay. Overall, we found Bcl-6 and Bcl-xL-mediated B cell immortalization to be a valuable tool to isolate antibodies and B cell clones in both vaccine and chronic infection settings.
Human subjects
Peripheral blood mononuclear cells (PBMC) from healthy subjects were obtained from donors participating in the NIH research apheresis program. Tonsil cells were acquired from discarded anonymized specimens from Children's National Medical Center (CNMC) with the approval from the Basic Science Core of the District of Columbia Developmental Center for AIDS Research and did not constitute 'human subjects research' as determined by the CNMC Institutional Review Board. Samples from the H7N9 influenza vaccine study (VRC 315; ClinicalTrials.gov; NCT02206464), which was a randomized phase I clinical trial in healthy adults designed to study the safety and immunogenicity of prime-boost vaccination regimens, were obtained through the VRC clinic (22-24). Informed consent was obtained from each volunteer and approved by the Institutional Review Board at NIAID, NIH. PBMC samples from three HIV-1-infected donors were also obtained through the Vaccine Research Clinic, while PBMCs from HIV-1-infected individuals with broadly neutralizing sera (patient 44 and donor N90) have been described previously (25,26). All donors provided informed consent and studies were approved by the Institutional Review Board at NIAID, NIH. PBMC samples collected one week after treatment of the first febrile malaria episode were obtained from a Malian cohort as previously described (27), and was approved by the Ethics Committee of the Faculty of Medicine, Pharmacy and Dentistry at the University of Sciences, Technique, and Technology of Bamako, and the Institutional Review Board of NIAID, NIH (ClinicalTrials.gov; NCT0132258).
Human IL-21 production
Human IL-21 was produced at the Vaccine Research Center. To express human interleukin-21 (IL21), (UniProtKB Q9HBE4) with a C-terminal thrombin cleavage site, an 8xHisTag and a TwinStrep tag the gene was synthesized and cloned into a mammalian expression vector pVRC8400. The plasmid was transiently transfected into freeStyle293F cells (Thermo Fisher). Protein was purified from filtered cell culture supernatant by Ni-NTA affinity column first and then by a StrepTactin (IBA) column.
B cell isolation and activation
B cell populations from PBMC or tonsils were sorted as described above and cultured in 12-well tissue culture plates in 1 mL of I10 medium (IMDM (Gibco) with glutamine, 10% FBS, supplemented with Penicillin, Streptomycin, Glutamine). Wells contained 1 x 10 4 to 2 x 10 5 B cells, 2 x 10 5 irradiated 3T3-msCD40L cells and 25 ng/mL human IL-21. Cells were cultured for 36 to 48 hours at 37°C and 5% CO 2 before retroviral transduction.
Retroviral transduction of activated B cells
B cells that had been activated for 36-48 hours were transduced with fresh viral supernatant containing polybrene (4 mg/mL). Briefly, 1 mL I10 media was removed from the well of the 12-well plate and replaced with 1 mL fresh viral supernatant (48hour retroviral supernatant) containing polybrene. The plate was centrifuged at 1200xg for 1 hour at 32°C, rested for 1 hour at 37°C and 5% CO 2 before the viral supernatant was replaced with fresh I10 containing 25 ng/mL IL-21 and cultured at 37°C and 5% CO 2 overnight. The same procedure was repeated the following day using the 72-hour retroviral supernatant. B cells were then rested for~3 days before determining transduction efficiency by flow cytometry and sorting for microculture.
Transformed B cell culture
Transformed B cells cultured in bulk were maintained in 12well plates. Cells were split 1:4 every 3-4 days into wells containing 2 x10 5 3T3-msCD40L cells and 25 ng/mL human IL-21 in 1mL of I10. Cell counts were measured by flow cytometry using counting beads (BD Biosciences) and data were collected on a modified LSRII or a modified FACSymphony. For microculture, transformed B cells were sorted 3 days following transduction. Briefly, transformed B cells were sorted at 1-5 B cells/well into a 384-well plate containing 5 x 10 3 irradiated 3T3-msCD40L, human IL-21 (25 ng/mL) in 50 mL of I10. Every 4 days cells were fed by replacing 25 ml of media with fresh I10 containing irradiated 2.5 x 10 3 3T3-msCD40L cells and 50 ng/mL IL-21 using a Biomek NX P liquid handler (Beckman Coulter). Cells were cultured for 2-3 weeks before harvesting the final supernatant and either freezing the cultures in the 384-well plate (50ul freezing media per well (FBS + 10% DMSO)), or expanding selected wells to 96-well tissue culture plates.
IgG and IgM ELISA
Concentrations of total IgG and IgM in the culture supernatant were determined by ELISA (ThermoFisher) according to the manufacturer's instruction.
Immunoglobulin amplification and sequencing
B cells were dry-sorted into a 96-well PCR plate at 1-2 cells per well and multiplex polymerase chain reaction (PCR) was used to amplify the heavy and/or light chain sequences as previously described (7). PCR products were sequenced by ACGT, Inc., Eurofins or Genewiz and analyzed using Geneious Prime and the IMGT database (32, 33).
In vitro mutation analysis
To characterize the accumulation of mutations during in vitro culture, IgG + B cells from PBMC of uninfected individuals and germinal center IgG + B cells from tonsil tissue were transduced. Four days post-transduction, GFP + B cells were sorted at one cell per well into 384-well microculture plates. Following 8 days of microculture, clonal GFP + B cells from selected wells were harvested and sorted at 1 cell per well into a PCR plate for BCR sequencing (48 total wells for each clone), and the remaining clonal B cells were placed back into culture. The same selected clonal B cells were again harvested on day 24 and sorted into PCR plates at 1 cell per well for BCR sequencing (96 total wells for each clone). The sequences from the VH genes were aligned using Geneious software to generate a consensus sequence and each individual sequence was analyzed for differences from the consensus sequence. Mutations deviating from the consensus sequence were confirmed by analyzing sequence chromatograms and assessing the quality of chromatogram peaks relative to the basecall recorded. Only mutations with clear, single color, high confidence sequence peaks were considered as actual mutations.
Phosphoflow to measure phosphorylation of ERK HA-specific immortalized B cell clones were resuspended in IMDM + 1% FBS and stimulated for 2 minutes at 37°C with HA probes (250nM), anti-IgG F(ab') 2 (5 mg/mL; Southern Biotech) or anti-IgM F(ab') 2 (5 mg/mL; Southern Biotech). Cells were immediately fixed with pre-warmed paraformaldehyde (4% final concentration), incubated at 37°C for 10 minutes and pelleted by centrifugation. Cells were resuspended in cold Perm Buffer III (BD Biosciences), incubated on ice for 30 minutes and washed twice with stain buffer (PBS + 1% FBS). Cells were resuspended in stain buffer and incubated with anti-ERK1/2 (pT202/pY204 Alexa647) (BD Biosciences) for 30 minutes at room temperature. Cells were subsequently washed and data were collected on a modified LSRII (BD Immunocytometry Systems) and analyzed using Flowjo (TreeStar).
BG505-SOSIP DS binding analysis of supernatants
S t a n d a r d M S D 3 8 4 w e l l s t r e p t a v i d i n c o a t e d SECTOR ® Imager 2400 plates were blocked with 35 µL of 5% (W/V) MSD Blocker A and incubated for 1 hour at room temperature (RT) on a Heidolph Titramax 100 vibrational plate shaker at 650 rpm. All incubations in this assay were performed as described above. The plates were washed three times with 0.05%Tween PBS (wash buffer) and were coated with biotinylated BG505 SOSIP DS-FPV1 10ln QQ AVI protein (Biotin Trimer 7070 capture) at an optimized concentration of 1 mg/mL for 1 hour. 1% MSD Blocker A was used as the diluent in the assay. The test samples (B cell supernatants) were diluted 1 to 1 with assay diluent in dilution plates. After 1 hour of incubation, the plates were washed again with the wash buffer and the serial diluted test samples were added to MSD plates. After an hour of incubation with the samples, the plates were washed again, and SULFO-TAG conjugated goat anti-human secondary detection antibody was used for detection at an optimized concentration of 1 mg/mL. After an additional hour of incubation, the unbound secondary detection antibody was washed off the plates and the plates were read using 1X MSD Read Buffer on the MSD Sector Imager 2400.
Antibody purification from B cell culture supernatants B cells from the CD27 + CD21 + and CD27 -CD21populations that displayed greater relative neutralization against pseudoviruses expressing HIV-1 envelopes BG505.T32N and MW965.26 in the TZM-bl microneutralization assay were expanded in serum free media hybridoma media (Gibco) that was supplemented with irradiated feeder cells and IL-21. Supernatants were concentrated and IgG was purified using Ab Spintrap columns (GE Healthcare/Cytiva).
HIV-1 TZM-bl neutralization assays
Screening small volumes of B cell culture supernatants for HIV-1 neutralizing activity was performed using the HIV-1 microneutralization assay as described (34). B cells from Pt.44 and donor N90 that displayed greater relative neutralization against the JRFL.JB pseudovirus in the microneutralization assay were cloned and expressed. Percent neutralization was calculated as: [((RLU virus with no Ab) -(RLU virus with Ab))/(RLU virus with no Ab)] X 100. Neutralization breadth and IC50 and IC80 values were determined for HIV-1-specific antibodies against a panel of Env-pseudoviruses using TZM-bl target cells as previously described (35).
Statistical analysis
GraphPad Prism software was used for statistical analysis and to prepare figures. Statistical analysis was performed using paired t-tests, Wilcoxon matched-pairs signed rank tests or oneway ANOVA with Tukey's multiple comparison test as indicated. Bars depict mean values with standard deviations shown.
Activation and transduction of multiple B cell subsets from PBMC and tonsil
To better understand the application of Bcl-6 and Bcl-xL B cell transformation to different B cell subsets, we compared the efficiency of Bcl-6/Bcl-xL transformation of naïve and memory B cells isolated from human PBMCs, and naïve and memory subsets isolated from human tonsil. For PBMCs, we sorted IgD + (naïve) and IgD -IgM -IgG + (IgG + memory) B cells and activated them in the presence of recombinant human IL-21 (25 ng/ml) and irradiated 3T3-msCD40-ligand (CD40L) feeder cells for 2 days ( Figure 1A). We spinoculated (1200xg, 32°C, 1 hour) activated B cells with fresh viral supernatant supplemented with polybrene (4 mg/mL) on days 2 and 3 of activation (Supplemental Figure 1). On day 4 post-transduction we observed transduction efficiencies between 45-80% for both IgD + and IgD -IgM -IgG + B cell populations as determined by the expression of green fluorescent protein (GFP) (Figures 1C, E). For tonsil tissue, we sorted IgD + (naïve) B cells, IgD -CD20 + CD38 + IgG + (Germinal Center (GC) IgG + ) B cells, and IgD -CD20 + CD38 -IgG + (non-germinal center (Non-GC) IgG + ) B cells and transduced the populations as described above ( Figure 1B). Similar to B cell populations isolated from PBMCs of healthy subjects, we again observed transduction efficiencies between 45-80% for all B cell subsets based on GFP expression ( Figures 1D, F).
In addition to determining the frequency of GFP + B cells, we calculated the expansion of B cell subsets from the initial activation through 4 days post-transduction (6 days total). Overall, while we did observe differences between subjects, we found that the total number of CD19 + B cells expanded between 4-fold to 40-fold, with IgG + B cells from the tonsil expanding less compared to naïve B cells (Figures 1E, F). Following transduction and expression of Bcl-6 and Bcl-xL, all GFP + B cells continue to expand and to express surface immunoglobulin (Supplemental Figures 2A-C).
Microculture of transformed B cells
To evaluate the efficiency of transformed naïve and IgG + memory B cell subsets to expand in microculture, and to determine the amount of immunoglobulin secreted, we compared microcultures of transformed B cells isolated from naïve and memory B cells from both PBMCs and tonsil tissue. Transduced B cells were sorted into 384-well microculture plates at 1, 2 or 5 GFP + cells per well at four days post-transduction. Cultures were supplemented with fresh media, IL-21 and irradiated CD40L-feeder cells every four days. On day 24 of culture, we measured IgM and IgG concentrations in the culture supernatant of transformed naïve and memory B cells, respectively. When sorted at 1 cell per well, between 20 and 33% of the memory wells had detectable levels of IgG, while between 32 and 38% of the naive wells had detectable levels IgM (Figure 2A). Within the positive wells, the median immunoglobulin concentration ranged from 0.39 mg/mL to 1.5 mg/mL for IgG with greater median concentrations for IgM at 7.3 to 8.3 mg/mL for PBMC and tonsil cells, respectively. Sorting 2 and 5 GFP + B cells per well resulted in an increased frequency of positive wells at day 24 with a slight, but not significant increase in the median immunoglobulin concentration (Figure 2A).
Since B cell receptor sequencing of B cells and cloning of antibodies may not occur until after screening culture supernatants on day 24 or later, we explored whether or not B cell clones acquire mutations during long-term culture, as has been described for an immortalized RSV-specific B cell line (9). We sorted single IgG + transformed B cells from both PBMCs and the tonsil germinal centers into a microculture plate and sequenced individual B cells from the expanded clones both early during culture (day 8) and at the end of culture (day 24) ( Figure 2B). Germinal center B cells were specifically chosen for analysis because they express high levels of activation-induced cytidine deaminase (AID) in vivo, which could result in more mutations during in vitro culture following transformation. After aligning the individual B cell sequences from each clone to determine the consensus sequence bioinformatically, we determined whether each individual sequence contained any base pair substitutions, or insertions/deletions compared to the consensus. The overwhelming majority of sequences acquired zero mutations at both day 8 and at day 24 of culture, although we did detect more sequences containing mutations on day 24 (PBMC IgG + with zero mutations: mean 98.4% on day 8 vs 90.3% on day 24, p<0.05; Tonsil GC IgG + with zero mutations: mean 93.7% on day 8 vs 90.1% on day 24, p<0.05) ( Figure 2C; Supplemental Figures 3A, B). Interestingly, with the exception of one clone, germinal center B cells did not acquire significantly more mutations compared to IgG + B cells from PBMCs during the 24 day culture. The few acquired mutations consisted mostly of one or two base pair substitutions ( Figure 2C; Supplemental Figures 3A, B). These results suggest that the frequency of mutations acquired during transformed B cell culture is almost negligible, and that the immunoglobulin sequence of the primary B cell can readily be determined by aligning multiple clonal B cell sequences to determine the consensus sequence.
Generation of antigen-specific naïve and IgG + memory B cell clones
We next asked whether we could generate antigen-specific memory B cell clones from both naïve and memory B cell subsets. To generate naïve B cell clones, we isolated CD27 -IgD + B cells from PBMCs of a healthy donor and transduced the cells with the Bcl-6/Bcl-xL retroviral vector. Following transduction, we stained the cells with eOD-GT8, a probe that binds to precursor B cells of VRC01-class HIV-specific broadly neutralizing antibodies (bNAbs) that target the CD4 binding site (21), and the eOD-GT8 knockout probe (eOD-GT8 KO), which has been mutated within the CD4 binding site and is used to avoid sorting off-target B cells (29) (Figure 3A). From the GFP + eOD-GT8 KOpopulation we sorted the eOD-GT8 + B cells into a microculture plate at 1 cell per well. After allowing the B cells to expand, we sequenced the clonal B cells to determine their BCR sequence. Based on the sequence data we determined that clone IMMO-A1, which is of the VH1-2 family and contains a 5 amino acid long L-CDR3, is a VRC01 precursor B cell line ( Figure 3B). Clone C1, which contains a VH1-2 sequence without a 5 amino acid long L-CDR3, and clone A2, which is not a member of the VH1-2 family, are also shown. After expansion and freeze/thaw, clone IMMO-A1 maintains its ability to bind to eOD-GT8. Thus, this methodology can be used to produce an immortalized cell line of a rare and highly specific B cell clone from the naïve B cell repertoire.
We also applied B cell immortalization to isolate antigenspecific memory B cell clones after influenza vaccination. To do so, we took PBMCs collected two weeks after a second dose of H7N9 MIV (22-24). To generate H7 HA-specific B cell clones we sorted IgG + B cells from PBMCs and following transduction we stained the transduced cells with H7 HA and H3 HA fluorophore-conjugated probes, and subsequently sorted all H7 HA-positive B cells at 1 cell per well for microculture ( Figure 3C). After clonal expansion we sequenced and characterized clones for binding to H7 and H3 HA, of which Clone 1 and Clone 2 are shown ( Figure 3D). We found the HA clones maintained their specificity from the single cell sort through clonal expansion, with Clone 1 binding to both H7 and H3 HA, and Clone 2 binding to the H7 HA but not to the H3 HA ( Figure 3D). In addition, we measured HA-specific B cell activation as detected by the ERK phosphorylation by flow cytometry, similar to what has been described for tetanus-specific B cells (9). Anti-IgG F(ab') 2 and H7 HA promoted phosphorylation of ERK in both clones, whereas the H3 HA induced phosphorylation of ERK in only Clone 1 ( Figure 3D). Thus, immortalized clones can be used to test whether immunogens not only bind to a specific B cell receptors but can also activate signalling through the B cell receptor.
Because the H7 HA-specific B cell repertoire has previously been characterized in these individuals, we compared the H7 HAspecific immortalized B cell repertoire to the previously characterized repertoire (23). We compared the clonal overlap and VH gene usage of the H7 HA-specific B cells from immortalized and ex vivo sorted B cells from 3 donors at the same time point following vaccination. In general, large clonal expansions are detected by both methods with less overlap observed with infrequent clones (Supplemental Figure 4A). And while there are some differences in the VH gene usage amongst the H7 HA-specific clones, there is not a major skewing of the repertoire when comparing the ex vivo probe-sorted repertoire to that of the immortalized probe-sorted repertoire (Supplemental Figure 4B). Thus, this methodology can be used to produce and characterize immortalized B cell clones that are representative of the vaccine-specific memory B cell pool in healthy individuals.
Immortalization of B cells from individuals with chronic infection
In chronic infection settings, the CD27 -CD21 -B cell population, which has a markedly reduced ability to proliferate in vitro, increases in frequency within PBMCs. Since antigenspecific antibodies can be found within this population, we characterized the efficiency of immortalization within this population. We sorted the CD27 -CD21and CD27 + CD21 + populations from 3 healthy individuals, 3 HIV-1-viremic individuals (viral loads: 385, 4503 and 608,544 copies/mL) and 4 individuals recovering from febrile malaria at one week after treatment (convalescence). Within the HIV-1-infected and malaria convalescent donors, we observed that the ex vivo CD27 -CD21population was indeed increased compared to healthy donors while the CD27 + CD21 + population was decreased ( Figures 4A, B). We observed significant differences in the transduction efficiencies between the CD27 + CD21 + and CD27 -CD21population across all donors (p<0.0001) and differences across multiple subsets based on CD27 and CD21 expression ( Figure 4C; Supplemental Figure 5A). We observed a similar trend with respect to the fold increase in total B cells during the transduction process, where the fold increase in B cell numbers of CD27 -CD21population was consistently lower compared to the CD27 + CD21 + population (p<0.005) ( Figure 4D, Supplemental Figure 5B). On day 7 we analyzed the surface phenotype of the transduced (GFP + ) and non-transduced (GFP -) populations and found that the GFP + cells continued to express CD20 and surface IgG, with GFPcells losing expression of these markers (Supplemental Figure 5C). Interestingly, we found that the GFP + cells expressed CD21 but little to no CD27 irrespective of the starting population. To evaluate the efficiency of B cell expansion and secretion in microculture, we sorted 2 GFP + B cells per well into a 384well plate and cultured the cells for 24 days. Although not statistically significant, we found the CD27 -CD21population tended to result in fewer IgG + positive wells compared to the CD27 + CD21 + population, but that IgG + wells contained similar Figures 5D-F). Thus, while the overall B cell expansion and transduction efficiency is lower for the CD27 -CD21population, the transduced cells do maintain surface IgG expression and secrete antibodies.
In the HIV-1-infected individuals, we identified IgGcontaining culture supernatants that bound to a soluble, HIV-1 envelope protein based on the BG505 sequence (BG505 SOSIP) in both the CD27 + CD21 + and CD27 -CD21populations (36) (Supplemental Figure 6A). After expanding B cells from the wells that showed binding to the BG505 SOSIP, and screening supernatants for neutralization against the BG505.T332N or the Tier1A MW965.26 pseudoviruses, we purified antibodies from 2 wells to further test for neutralizing activity (Supplemental Figure 6B). While well 1 failed to neutralize either the BG505.T332N or MW965.26 pseudoviruses upon further inspection, we did detect weak neutralization against the MW965.26 pseudovirus for well 2, which originated from the CD27 -CD21 -B cell subset (Supplemental Figures 6C, D). These data demonstrate that HIV-1-specific antibodies can be isolated from the CD27 -CD21population by B cell immortalizaton.
Isolation of broadly neutralizing antibodies from immortalized B cells
To determine if B cell immortalization can be used to isolate bNAbs from HIV-1-infected individuals, we applied the technology to donors from which bNAbs have previously been isolated. Antibody VRC38.01 was previously isolated from donor N90 while VRC13 was isolated from patient 44 (Pt.44) (25,26). B cells were sorted and transduced from each donor, and on day 4post transduction GFP + cells that bound to the RSC3 probe (Pt.44) and JRFL.DS.SOSIP (N90) were sorted into 384-well plates at 2 cells per well to maximize the number of B cell positive wells and cultured for 23 to 31 days ( Figure 5A). Single point neutralization screening was performed using supernatants from days 23, 27 and 31 of microculture against the JRFL.JB pseudovirus ( Figure 5B). Cell lysates from wells with greater relative neutralization were sequenced and only one heavy chain and one light chain sequence was obtained from each positive well. Analysis of the antibody sequences indicates the antibody isolated from donor N90 (well 5J18) is a relative of VRC38, while the antibody isolated from Pt.44 (well 3E06) is related to VRC13 ( Figure 5C). Of note, the immortalization procedure was repeated for Pt.44 and a second VRC13 relative was isolated (well E3). The antibody isolated from Donor N90 (well 5J18) utilizes the same V and J genes as VRC38.01 for both the heavy and light chains and share the same CDRH3 and CDRL3 sequences. Similarly, the two antibodies isolated from Pt.44 share the same V and J genes for both the heavy and light chains as VRC13, and within the CDR3 regions residues that differ from VRC13 are highlighted in red ( Figure 5C).
Single point neutralization assays were performed against a 31-virus panel for the three cloned antibodies and between 29 and 32% of viruses were neutralized by more that 50% ( Figure 5D). The cloned antibodies, along with VRC13 and VRC38.01 were further tested for neutralization against a panel of 17 viruses that showed sensitivity in the single-point assay. The antibody isolated from well 5J18 showed similar potency and breadth as compared to VRC38.01, whereas the two antibodies isolated from Pt.44 display less potency and breadth as compared to VRC13 ( Figure 5E). Overall, these studies serve as a proof-of-principle that B cell immortalization can be used to isolate bNAbs.
Discussion
While Bcl6/Bcl-xL B cell immortalization was first described in IgG + and IgM + memory B cell populations from PBMCs from healthy donors, and similar B cell immortalization methods have recently been described in cancer settings and applied to naïve B cells, an extensive comparison of the same Bcl6/Bcl-xL B cell immortalization protocol across multiple B cell subsets, including chronic infection settings, has not been done (9,21,37). Here, we characterized and applied Bcl-6/Bcl-xL B cell immortalization to multiple B cell subsets including naïve and memory B cell subsets from both lymphoid tissue and PBMCs in healthy and chronically-infected individuals. In our hands, we found Bcl-6/BclxL B cell immortalization to be applicable across multiple B cell subsets, although with differing efficiencies amongst distinct B cell populations.
The application of B cell immortalization to B cell populations with a reduced capacity to proliferate and survive in vitro, such as CD27 -CD21memory B cells and germinal center B cells, may provide an alternative means for antibody isolation from these populations, in lieu of traditional B cell culture. For example, although we found primary germinal B cells expanded less when cultured with IL-21 and CD40 Ligand when compared to other B cell subsets, we found that posttransduction, the immortalized B cells from this population expanded and secreted antibodies in similar amounts to that of immortalized IgG + B cells from PBMC of healthy donors, suggesting that immortalization of germinal center B cells could serve as an alternative method to isolate rare antibodies from germinal centers of SLOs. In this regard, HIV-1-specific bNAbs were recently isolated from germinal center B cells of lymph nodes using an Epstein-Barr Virus-based immortalization strategy, demonstrating that rare antibodies can be isolated from germinal centers with B cell immortalization strategies (38). In chronic infection settings, we found that B cell immortalization can be applied to the CD27 + CD21 + IgG + population with similar efficiency to that seen in IgG + memory B cells in uninfected individuals, but we observed decreased transduction and expansion of the CD27 -CD21 -IgG + population. Thus, if antibodies of interest are found within the CD27 + CD21 + B cell population, this technique has a greater capacity to isolate the antibodies of interest, although antibodies can still be isolated from the CD27 -CD21population, as the immortalized B cells from this population do secrete antibodies and express surface immunoglobulin.
One of the greatest advantages of this technology is the flexibility of screening for both antibody binding and function with relatively large quantities of immunoglobulin-containing culture supernatant. In a traditional B cell culture method, where primary B cells are activated for 10 to 14 days before screening immunoglobulin-containing culture supernatant, a limited amount of functional screening can be applied. In the HIV-1 TZM-bl microneutralization assay, which screens primary B cell culture supernatants from 384-well culture plates for HIV-1 neutralizing antibodies, one can only test for neutralizing activity against one or two pseudoviruses due to limited supernatant volumes. Because immortalized B cells are continuously expanding, larger volumes of supernatant can be generated and screened against several pseudoviruses to assay for breadth before antibody sequencing and cloning. We were able to isolate a HIV-1-specific antibody from the CD27 -CD21population of a HIV-1infected donor, where we first screened for binding to the HIV-1 envelope SOSIP trimer, with subsequent screening for neutralizing activity after expanding wells of interest. Importantly, we were also able to demonstrate this technology can be applied to isolate bNAbs from HIV-1-infected indiviudals by using a slightly different strategy where we combined B cell immortalization with the TZM-bl microneutralization assay and subsequently cloned the antibodies for further characterization. Both of these examples, along with our recent study in donors after malaria infection where we isolated MSP-specific IgM + antibodies, can serve as a proof-of-concept to show that Bcl6/ Bcl-xL immortalization can be applied to chronic infection settings to isolate antibodies of interest (39). Because we did detect differences between B cell subsets during activation and following transduction, and because individual clones do expand at different rates, the utility of this methodology as a means to characterize the B cell repertoire is unclear. Using influenza vaccination as a model system, we compared the H7 HA-immortalized B cell repertoire to that of a previously characterized repertoire using the same samples and the same probes as bait as previously published (23). In general, both methodologies detected large clonal expansions but there was less overlap with rarer clones, with some clones detected with only ex vivo sorting, and others detected after immortalization and clonal expansion. And while some of these discrepancies can be attributed to the depth of sequencing, they may also be attributed to the variable efficiency of B cell immortalization in different B cell subsets. In general however, we did not detect a major bias in the immortalized H7 HA-specific B cell repertoire compared to the previously published ex vivo repertoire (23). Therefore, not only can B cell immortalization be used to isolate rare antibodies and generate B cell clones, it could also be used to study the B cell repertoire to a particular antigen.
As B cell clones are expanded for increased screening and analysis, one potential concern is whether the clones acquire mutations during the in vitro culture, and whether the final sequence of the desired clone represents the sequence of the primary B cell. In our experimental setting, while we did detect mutations during the microculture, we found these mutations to be minimal. Some of the mutations may be due to PCR error, although some certainly represent actual mutations as we observed more mutations on day 24 of culture compared to day 8 using the same sequencing protocol. Overall, we found that by sequencing several individual B cells from the same clone and aligning the sequences bioinformatically, we could determine the consensus sequence which likely represents the sequence of the original primary B cell. Of course, as clones are further expanded one should take into account the accumulation of additional mutations which may or may not be desired. Interestingly, we did not detect significant differences in the frequency of mutations acquired during in vitro culture for immortalized B cells isolated from PBMCs and immortalized B cells isolated from germinal centers. Its possible that B cell transformation itself, and/or the in vitro culture conditions result in similar AID expression between the immortalized B cell subsets.
In addition to serving as experimental controls to test for specificity and sensitivity of B cell probe binding, immortalized B cell clones can also be used to characterize the capacity of novel immunogens to induce B cell signaling. In combination with a high-throughput B cell activation assay, such as detection of phosphorylation of ERK by flow cytometry, libraries of immortalized B cell clones can be utilized to measure the diversity and specificity of the B cells that respond to the immunogen. The use of clonal B cell lines may also prove to be especially valuable in the HIV-1 vaccine field where novel immunogens are being designed to bind to and activate germline precursor B cells. Here, naïve precursor clonal B cell lines can serve as controls to characterize novel probes aimed to activate precursor B cells and to also test for assay specificity, sensitivity and reproducibility in the analysis of data from clinical trials.
As with any methodology, there are advantages and disadvantages to be considered before applying the technology, including the efficiency of the methodology. Overall, Bcl-6/Bcl-xL B cell immortalization is a valuable tool for antibody isolation and clonal B cell line development, and our results here show that it can be applied to different B cell populations, including populations that may be less amenable to long-term B cell culture.
Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.
Ethics statement
The studies involving human participants were reviewed and approved by Institutional Review Board at NIAID, NIH; and the Ethics Committee of the Faculty of Medicine, Pharmacy and Dentistry at the University of Sciences, Technique, and Technology of Bamako. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin . KB wrote the manuscript and TW, EC, JS, SA, CH, TZ, WS, PC, LG, PK and RK edited the manuscript. SA, CH, AC, JC, TY, TZ, WS, PC, JL, MC, PK, AM and RK provided reagent resources or clinical samples and RK supervised the work. All authors contributed to the article and approved the submitted version.
Funding
This work was supported by the Bill and Melinda Gates Foundation (grant OPP1147555) and by the Intramural Research Program of the Vaccine Research Center, NIAID, NIH. | 2022-04-02T13:16:33.588Z | 2022-03-30T00:00:00.000 | {
"year": 2022,
"sha1": "8e5a7878316c0de7096a1656f3c94bd88691ca33",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Frontier",
"pdf_hash": "b948f76a0f3eb0a935a73bc5d8c9c82a9c0e28f4",
"s2fieldsofstudy": [
"Biology",
"Medicine"
],
"extfieldsofstudy": [
"Medicine",
"Biology"
]
} |
118162887 | pes2o/s2orc | v3-fos-license | Infinitesimal deformations of complements of plumbings of rational curves
We construct infinitesimal deformations on an open domain of a smooth projective surface given by a complement of plumbings of disjoint linear chains of smooth rational curves. We show that the infinitesimal deformations are not small deformations, that is, they change the complex structure away from the boundary of the domain.
A plumbing of a linear chain D i is a union of small tubular neighborhoods of each rational curves C ij (j = 1, . . . , m i ). Assume that the supports of D i 's are disjoint and each divisor D i can be contracted to an isolated rational singular point. Let f : Z → X be the map contracting D i 's.
In this paper we construct infinitesimal deformations on the complement of the union of certain plumbings of D i 's. In Section 1 we prove that, for each i, there is a plumbing, say U i , of D i corresponding to the ample divisor H on Z such that H 2 (Z \ U i , T Z\Ui ) = 0. Furthermore we show that H 2 (Z \ U, T Z\U ) = 0 where U = ∪ k i=1 U i ; Theorem 1.2. Since Z \U is not compact, the vanishing does not imply that an infinitesimal deformation on Z \ U is integrable. We leave the integrability question for future research. In Section 2 we construct infinitesimal deformations on Z \ U by restricting to Z \ U meromorphic 1-forms on Z with poles along C ij of certain orders. The orders of poles are determined by H uniquely up to constant multiple; Theorem 2.2. We finally show that the infinitesimal deformations are not small deformations, that is, the infinitesimal deformations change the complex structure away from the boundary of Z \ U ; Theorem 3.1.
In order to construct such infinitesimal deformations, we apply a similar strategy in Takamura [1], where he constructed infinitesimal deformations on a complement of a certain negative non-rational curve on a smooth surface of general type.
Throughout this paper we work over the field C of complex numbers.
Holomorphic tubular neighborhood
We first show that there is a special holomorphic tubular neighborhood for each divisor D i = mi j=1 C ij in the surface Z with a vanishing cohomology condition; Proposition 1.2.
Proof. We use a similar method in Takamura [1]. Let A ∈ |lH| (l ≫ 0) be a smooth irreducible hyperplane section of Z. Fix a base point p ∈ C ∪ A of the Abel-Jacobi map α : Div(A) → Jac(A). Let V p be a small open neighborhood of p in A. Choose a nonzero section s ∈ H 0 (A, O A (A)). Set R = div(s), which is an effective divisor on A of degree d = A 2 . Note that the fundamental domain of the complex torus Jac(A) is bounded and α(V p ) is an open set (by shrinking V p if necessarily) which contains the origin in Jac(A); hence we have α(Sym nd V p ) = Jac(A) for sufficiently large n because the Abel-Jacobi map α is a homomorphism of abelian groups. Therefore we have α(nR) ∈ α(Sym nd V p ) for sufficiently large n. Since is surjective. Therefore there is an smooth irreducible curve A ′ ∈ |nA| of Z whose restriction on A is R ′ . Then A ∩ A ′ ⊂ V p . We may assume that A ′ intersects C transversely and that p ∈ A ′ ∩ C because A ′ is also a hyperplane section.
Since Ext 1 (N C,Z , T C ) = 0, we have T Z | C = T C ⊕ N C,Z . Therefore there is a holomorphic tubular neighborhood V of C in Z. We may assume the followings: (i) V ⊃ V p by shrinking V p if necessarily, (ii) V ⊃ A ∩ A ′ , and (iii) at least one component of V ∩ A (and also V ∩ A ′ ) is a fiber of V , where V is considered to be a holomorphic disk bundle over C.
We remark the following observation: Let N be a holomorphic line bundle over an open Riemann surface S and let N 0 be obtained by deleting a tubular neighborhood of the zero section. Then N 0 is biholomorphic to S × B where B is a complement of an open disk in C because any holomorphic line bundle over an open Riemann surface is trivial. Since both S and B are Stein, so is N . In particular, the boundary ∂N 0 is pseudoconvex.
In our case, the set is pseudoconvex in the affine variety Z \ A (respectively, Z \ A ′ ) by the above observation. Therefore By the Mayer-Vietoris sequence, we have the exact sequence . Therefore, according to the vanishing of higher cohomology of a Stein space, we have H 2 (Z \ V, T Z\V ) = 0.
By the Mayer-Vietoris sequence and the induction on is not compact, the vanishing does not imply that a infinitesimal deformation on Z \ U is automatically integrable.
Infinitesimal deformations
Let H be an effective ample divisor on Z. We construct infinitesimal deformations of the open surface Y = Z \ U derived from H; Proposition 2.2.
x kj C kj satisfying the following properties: (iii) L descends to an ample divisor on X. Furthermore a tuple of coefficients (x 0 , x 11 , . . . , x km k ) is uniquely determined by H up to a constant multiple.
Proof. For simplicity we first consider the case k = 1. We denote C 1j by C j , b 1j by b j , and m 1 by m. Set a j = HC j > 0 for j = 1, . . . , m.
For m ≥ 3, the condition (i) would be interpreted as the system of linear equations Since B is of full rank, the solution of (2.1) is unique up to constant multiple. Since a i , b i > 0, it is easy to show that there are positive integers x 0 , . . . , x m satisfying (2.1). Therefore there is an effective divisor We now show that L descends to an ample divisor on X. Let E = x 1 C 1 + · · · + x m C m . Consider the exact sequence Since H is ample, H 1 (Z, nx 0 H) = 0 for n ≫ 0; hence ψ is surjective for n ≫ 0.
On the other hand, setting V = f (U ) and p = f (D), (U, D) → (V, p) is the minimal resolution of the rational singularity p of X. We may assume that V ⊂ X is Stein and contractible by shrinking U if necessary. Then H 1 (U, O U ) = H 2 (U, O U ) = 0 and D is a deformation retract of U , in particular the exponential sequence on U gives an isomorphism Pic(U ) = H 2 (U, Z). Therefore a line bundle L on U is trivial if and only if L · C i = 0 for every irreducible component C i of D. Therefore the restriction of L on U is trivial and nL is also trivial on U . In particular, the restriction map H 0 (nE, nL| nE ) → H 0 (D, nL| D ) is surjective.
Since nL| D is trivial, we can choose s 0 ∈ H 0 (Z, nL) such that s 0 | D is a nonzero constant. On the other hand, since H is ample, we may choose { s 1 , . . . , s l } ∈ H 0 (Z, nx 0 H) so that they gives an embedding Z ֒→ P l−1 for n ≫ 0. Set s i = φ(s i ) (i = 1, . . . , l). Since s i | D = 0 for i = 1, . . . , l, the map π : Z → P l defined by π(x) = (s 0 (x), s 1 (x), . . . , s l (x)) contracts D and gives an embedding of Z \ D; hence, the map π is an embedding of X, which implies that L descends to an ample divisor on X.
In case of k ≥ 2, one may consider the following equation instead of (2.1): Setting x 11 x 12 . . .
Then the proof is identical to the proof of the case k = 1.
Our infinitesimal deformations of Y are obtained by restricting to Y meromorphic 1-forms on Z with poles along C ij . In order to construct such infinitesimal deformations we use a similar strategy in Takamura [1].
Proof. The divisor L = x 0 H + E descends to an ample divisor on X by Lemma 2.1. Let L = L|X. Since L is ample, we may choose an irreducible smooth curve C ∈ nL for n ≫ 0 such that C does not pass through the singular point of X. We denote again by C the inverse image of C under the contraction map f . Then C is linearly equivalent to nL and C ∩ Supp(E) = ∅.
Consider the exact sequence Since C − nE = nH is ample, we have Therefore the map α is injective. On the other hand, since C ∩ Supp(E) = ∅, the map α factors through H 1 (Y, T Y ), i.e., Therefore the restriction map H 1 (Z, T Z (nE)) → H 1 (Y, T Y ) is injective for any n ≫ 1. The infinitesimal deformation H 1 (Z, T Z (nE)) is nonempty; Proposition 2.5. We use a similar method in Takamura [2].
Proof. Since the supports of the divisors D i are disjoint we may assume that k = 1. For simplicity we denote C 1j by C j , b 1j by b j , m 1 by m, and x 1j by x j . We divide the proof into two cases.
Case 1: m = 1. Let C = C 1 , b = b 1 , and x = x 1 for brevity. We first claim that induced from the tangent-normal bundle sequence. Since for 0 ≤ l ≤ nx, it follows from the above exact sequence and the Riemann-Roch theorem that h 0 (T Z (nE − lC)| C ) = 0 and h 1 (T Z (nE − lC)| C ) = 2nx − 2lb + b.
Consider the decomposition sequence Therefore Case 2: m ≥ 2. The proof is similar to the case m = 1. For the convenience of the reader, we briefly sketch the proof. We claim that h 0 (T Z (nE − lC)| C ) = 0 . Consider the exact sequence Hence the claim follows.
From the decomposition sequence and the above claim, we have if n ≫ 0 because α ∼ O(n) for n ≫ 0.
Proof. Consider the exact sequence induced from the exact sequence Then we have By Lemma 2.4, the left hand side of the above equation grows quadratically for n ≫ 3, but dim im β bounds for h 2 (Z, T Z ). Hence dim im α must grows quadratically for n ≫ 0, which implies that h 1 (Z, T Z (nE)) grows at least quadratically in n ≫ 0.
Proof. We only prove the lemma in case i = 1, j = 1, and m 1 ≥ 2. The proof of the other cases are similar. For simplicity we denote C 1j by C j , b 1j by b j , m 1 by m, and x 1j by x j .
According to the claim in the proof of Lemma 2.4, we have Note that x 1 ≤ na1x0+2 b1 + 1 for n ≫ 0. Therefore By (2.2) and the induction on l, it follows that H 0 (T Z (nE)| x1C1 ) = 0.
From the exact sequence we have an induced map H 1 (Z, T Z (nE)) → H 1 (Z, T Z ((n + 1)E)).
Proof. Let E i = mj j=1 x ij C ij . Since the supports of D i 's are disjoint we have . Therefore it is enough to show that H 1 (Z, T Z (nE i )) → H 1 (Z, T Z ((n + 1)E i )) is injective for every n ≫ 0. We prove only the case of i = 1 and m i ≥ 2. For simplicity we denote C 1j by C j , b 1j by b j , m 1 by m, and x 1j by x j .
Consider the exact sequence By Lemma 2.6, we have H 0 (T Z (nE)| x1C1 ) = 0; hence, Consider the exact sequence By Lemma 2.6, we have H 0 (T Z (nE)| x2C2 ) = 0. Therefore it follows from (2.3) that Repeating this process, then we have
Properties of the infinitesimal deformations
A small deformation of the domain Y in Z changes the complex structure only near the boundary of Y while keeping unchanged the complex structure away from ∂Y . In this section we will show that the deformations induced from H 1 (Z, T Z (nE)) are not small deformations. In fact, any non-zero α ∈ H 1 (Z, T Z (nE)) induces a non-trivial infinitesimal deformation of a certain curve away from ∂Y ; Theorem 3.1.
Since the divisor L in Lemma 2.1 descends to an ample divisor on X, we may choose an irreducible smooth curve C ∈ |nL| (n ≫ 0) on Z such that C ∩ U = ∅.
Theorem 3.1. Any infinitesimal deformation induced from a non-zero element H 1 (Z, T Z (nE)) preserves C ∩ Y but changes infinitesimally the complex structure of C ∩ Y .
Proof. We use a similar method in Takamura [1]. Let C ′ = C ∩ Y . Consider the exact sequence Since C ∩U = ∅, C ′ is stein; hence, H 1 (C ′ , N C ′ ,Y ) = 0 and the map ψ is surjective. Therefore any infinitesimal deformation of Y induced from a non-zero element in H 1 (Z, T Z (nE)) preserves C ′ .
Let α ∈ H 1 (Z, T Z (nE)) be a non-zero element. We denote again by α the image of α of the injective map H 1 (Z, T Z (nE)) → H 1 (Y, T Y ). Take any α ∈ H 1 (Y, T Y (− log C ′ )) such that ψ( α) = α. Consider the commutative diagram In order to prove that α changes infinitesimally the complex structure of C ′ , it is enough to show that the image of α by φ : H 1 (Y, T Y (− log C ′ )) → H 1 (C ′ , T C ′ ) is non-zero. In the proof of Proposition 2.2, we showed that the restriction α| ′ C ∈ H 1 (C ′ , T Z | C ′ ) is non-zero because is injective for n ≫ 0. Since the above diagram commutes, φ( α) is also non-zero. | 2010-10-10T08:06:24.000Z | 2010-10-10T00:00:00.000 | {
"year": 2010,
"sha1": "73df60f30e0a85780aa9019bb870797054070c45",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "73df60f30e0a85780aa9019bb870797054070c45",
"s2fieldsofstudy": [
"Mathematics"
],
"extfieldsofstudy": [
"Mathematics"
]
} |
261150363 | pes2o/s2orc | v3-fos-license | Structural design of satellite geodetic networks using mobile monitoring stations
. The article considers a method for solving the problem of structural design of modernized networks of non-request measuring instruments of a ground-based control complex for navigation stations of a space complex. The object of research is satellite geodetic networks using mobile monitoring stations. The type of such networks is information and control systems distributed over the territory, dispersed in space. Structurally, they are determined by multifunctional sets of stationary and mobile elements with advanced technical means for receiving, transmitting and processing information. It is shown that in the structural design of satellite geodetic networks, special attention is paid to the cost of the network, since this can significantly reduce the cost of field work associated with the use of mobile monitoring stations. An analysis of various approaches to the formation of the structure of such systems was carried out, which showed that the tasks of formation can be divided into two groups: the formation of the topological structure of the system and the formation of the functional structure of the system. The paper presents the stages of the analytical and simulation procedure, which includes optimization and simulation models and allows solving the problem of structural design of satellite geodetic networks.
Introduction
Structural design issues are relevant for modern tasks of building modernized networks of non-request measuring instruments of the ground-based control complex for navigation stations (NS) of the space complex (SC), which is used as part of geodetic networks [1][2][3]. It is noted in [4] that solving the problems of designing geodetic networks intended for observing the movements and deformations of the earth's crust and large engineering structures is of great practical interest. As a rule, in the structural design of satellite geodetic networks, special attention is paid to the cost of the network, since this can significantly reduce the cost of field work associated with the use of mobile monitoring stations. The authors in [4] formulate the following network structural parameters (attributes) optimized during design, which affect the cost of work: weights of measured values (angles, distances, etc.); position of monitoring points (network configuration); number of network monitoring points. Accounting for the parameter that reflects the number of network monitoring points is due to the fact that the largest contribution to the cost of all field work is made precisely by the laying of geodetic points, the construction of signals, the cost of materials, transportation costs, etc. For mobile monitoring stations, these costs are significantly lower, which allows reducing the overall costs for the development of a geodetic network.
It should be noted that despite the importance of solving the problem of determining the minimum required number of network points and their optimal position, especially if the construction of points is possible only in places strictly fixed on the ground or in an engineering structure, it has not yet found proper coverage, much less a complete solution in geodetic literature. A number of works [5][6][7] are devoted to solving this problem.
Thus, the structural design of satellite geodetic networks using mobile monitoring stations at the moment, for the indicated reasons, is an unresolved and rather difficult task. This problem is formulated as the problem of finding the minimum number of optimally located points necessary for the correct functioning of satellite geodetic networks using mobile monitoring points. In this case, it is assumed that strictly fixed places (coordinates) are known in advance, where both the construction of stationary and the placement of mobile monitoring points is possible.
Materials and methods
Modern materials and methods that exist today allow, in general, to successfully solve the mathematical problems of fixing the general earth kinematic coordinate system only according to measurement data in space geodetic networks. It is noted in [4] that until now, in world practice, such fixation was carried out by reference to any geological and geophysical model of lithospheric plate movements. The authors state that the refusal to use any hypotheses and models of the Earth's evolution will make it possible to more reliably fix the coordinate system in the Earth's body to provide satellite navigation systems such as GPS/GLONASS. This will also make it possible to more reliably solve such problems of geology and geophysics as independent testing of various hypotheses and models of the Earth's evolution, possible changes in its radius, post-glacial uplift of the earth's crust, etc.
At present, computer algorithms have been developed and implemented for optimal design of the most informative measurements in geodynamic GPS networks for monitoring active volcanoes and faults in the earth's crust [8][9][10][11][12][13]. The algorithms can significantly reduce the cost and time spent on field work and are ready for practical use.
An algorithm has been developed for estimating the average annual displacement rates of geodetic points from short series of average daily coordinates of GPS stations, taking into account a priori information about the amplitudes and initial phases of seasonal variations, which can significantly improve the accuracy and reliability of determining modern movements of the earth's crust.
Experimental work has been carried out to combine various global space geodetic networks and determine the current movements of the earth's crust on a global and regional scale. The point displacement rates in global networks are in good agreement with the NNR NUVEL-1A tectonic model and the ITRF2000 international coordinate system. The materials and methods used in these studies have been repeatedly reported at various international symposia, published in [8][9][10][11][12][13] and other publications.
Thus, today the main theoretical methods related to mathematical processing and optimal design of geodetic measurements are well developed. Developing an approach based on the structural design of satellite geodetic networks using mobile monitoring stations, attention should be paid to the method of fixing the coordinate system.
The physical essence of the method for determining the movements and deformations of the earth's crust and large engineering structures consists, as a rule, in calculating the displacements of fixed points in any coordinate system from geodetic measurements. If these displacements are determined relative to the initial geodetic points fixed in the nondeformable zone, the problem is known to be solved without any problems in the coordinate system fixed by these initial points. Otherwise, which is also typical when using mobile monitoring points, all points of the geodetic network experience shifts and the problem is solved ambiguously, since fixing the coordinate system requires, firstly, to select quasi-stable points according to some criteria and, secondly, set in some way the preliminary coordinates of the points of the geodetic network, which, in fact, will fix the coordinate system by applying the theory of mathematical processing of free geodetic networks using the apparatus of generalized inverse matrices. At the same time, in [4], it is proposed to solve the following system of correction equations with restrictions where B is the vector of free members; K -matrix of correction equation coefficients; σY = (σy1, σx 1 , σz 1 , ..., σy q , σx q , σz q ) T is the vector of unknown station velocities in the rectangular coordinate system X, Y, Z; x, y , z -rectangular coordinates of q network points; C is the vector of corrections to the measurement results. The choice of the system of coordinates depends on the choice of the constraint system (2), in which the velocities of the geodetic points will be obtained. In equations (2), one usually chooses the matrix E = G T I e .
(3) In expression (3), the following designation is additionally introduced: Ie is a diagonal matrix containing ones on the diagonal for quasi-stable and zeros for mobile points. It is noted in [14] that despite the simplicity of expressions (1)-(3), a number of incorrect statements and conclusions are allowed when using them. For example, matrices K and G do not have to be orthogonal and should only complement each other. It is also noted there that the scale of the network is usually always set by the results of direct measurements.
Taking into account that for a planned network it is expedient to fix the origin of point shifts relative to the center of gravity of a stable territory or a block of the earth's crust, the origin of coordinates should be placed at the center of gravity of quasi-stable points located there, but not always at the center of the network. It makes sense to center the coordinates over all points only if all points are considered to be quasi-stable. For global space networks covering the entire globe, the reference point should be combined with the center of gravity, but not the center of gravity of quasi-stable points. Then, during the rotation of the network, the vertical displacements of points should remain unchanged, for example, during linear transformations by the iterative weighted method of S-transformations performed according to known formulas [15].
Results and discussion
The problems of choosing the structural composition of the ground segment of the space complex of the system and the locations of its objects are relevant for the medium term. It is during this period that they will be finally determined. In this regard, the proposed formal model is of a particular nature and, accordingly, will require clarification based on the result of a detailed analysis when conducting model experiments, taking into account the results of ongoing development work. It is obvious that from the network control center in real time and in an automated mode, almost all the main navigation tasks of controlling the NS and controlling the navigation field are solved.
Taking into account the analysis of various approaches to the formation of the structure of such systems [16][17][18], it is shown that the problems of formation can be divided into two groups. The first includes tasks related to the formation of the topological structure of the system, which consists in determining the composition, territorial location and type of control nodes at all levels of the system hierarchy and communication channels between them. The second group includes the problems of forming the functional structure of the system, that is, the distribution of control functions between the nodes of the system, including the control object and the distribution of technical means among the nodes of the system.
From a mathematical point of view, the considered problems of structural design of satellite geodetic networks using mobile monitoring stations belong to the class of mathematical programming problems in which a number of constraints are specified not explicitly, but algorithmically [19].
The following algorithm is proposed for solving the problem of structural design of satellite geodetic networks using mobile monitoring stations. Denote by the set of variants of structures that are admissible under restrictions on attributes specified in the analytical form, and by " -a set of variants of structures that are admissible according to restrictions on attributes specified algorithmically, i.e., the fulfillment of these restrictions can only be verified during simulation. Then the space of admissible variants of structures " ' . Models of formation problems, depending on the method of specifying the objective function and the space of admissible variants of the structure , can be divided into classes, where the objective function can be specified either analytically or algorithmically. In this case, the space can be set: analytically; algorithmically; analytically and algorithmically. Depending on the class of the structural model, various procedures for searching for the optimal structure variant can be used, differing from each other in the way of generating structure variants, the rules for checking analytically and algorithmically specified attributes, and the way to proceed to the next step.
In such a situation, it would be reasonable to use the optimization-simulation approach [3,16], based on the joint use of optimization and simulation models in the process of searching for optimal structure options.
Below are the main stages of the analytical-simulation procedure for the formation of a basic set of structures with a given set of attributes for multi-attribute selection of the best structure variant: Stage 1. Generation of a variant of the system structure i X , i = 1, …, m with a given set of attributes А j , j = 1, …, I. Stage 2. Checking the admissibility of the structure variant according to analytically specified attributes. If i X , then go to step 3, otherwise go to step 1.
Stage 3.
Conducting a computer experiment with a simulation model of the system i X for a variant of the structure that is admissible in terms of analytically given attributes i X . Model displays the functioning of the simulated system for various variants of the structure. Between stages 1 and 3, an information interface is organized to transfer data about the studied variant of the structure i X . Stage 4. Checking the validity of the structure option i X algorithmically defined attributes.
, then go to step 5, otherwise go to step 1. Stage 5. Memorizing a variant of the structure that is admissible according to the attributes specified analytically and algorithmically.
Stage 6. Checking the completeness of the analysis of all variants of the structure. If, as a result of the check, it turns out that not all variants of the structure have been analyzed, then go to stage 1. Otherwise, the output of the results obtained and the end of the algorithm.
This analytical and simulation procedure includes optimization and simulation models (stage 3) and allows in iterative mode to solve the problem of structural design of satellite geodetic networks using mobile monitoring stations.
Conclusion
Thus, based on the analysis of the problems of structural design of satellite geodetic networks using mobile monitoring stations, the need to search for rational options for the network structure using optimization and simulation models was revealed. This allows in the process of forming the structure to design, evaluate and select rational network options, including mobile monitoring points. Moreover, the proposed optimization and simulation procedure, including optimization and simulation models, allows solving the problem of structural design of satellite geodetic networks using mobile monitoring stations in an iterative mode.
In [20][21][22], applied aspects of the application of simulation modeling and optimizationsimulation approach to solving problems of structural design of space navigation systems, which include satellite geodetic networks using mobile monitoring stations, are presented. Such networks are distributed information and control systems and are multifunctional sets of stationary and mobile elements dispersed in space with advanced technical means for receiving, transmitting and processing information. | 2023-08-26T15:47:30.697Z | 2023-01-01T00:00:00.000 | {
"year": 2023,
"sha1": "357b5a1f646a78049f77c9dd35cdcf01c150591a",
"oa_license": "CCBY",
"oa_url": "https://www.e3s-conferences.org/articles/e3sconf/pdf/2023/54/e3sconf_geotech2023_06008.pdf",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "1233c26bcb0e629479a9ea4a19923e0814ddb97e",
"s2fieldsofstudy": [
"Engineering",
"Computer Science"
],
"extfieldsofstudy": []
} |
247406653 | pes2o/s2orc | v3-fos-license | Effect of Aromatherapy Massage on Depression and Anxiety of Elderly Adults: a Randomized Controlled Trial
Introduction This study investigated the effect of aromatherapy massage with lavender, chamomile, and rosemary oils on the depression and anxiety of elderly adults living in nursing homes. Methods This randomized controlled trail was conducted on elderly adults living in nursing homes in Kerman, Iran. Through convenience sampling, 38 elderly adults were recruited and assessed using demographic questionnaire and Hospital Anxiety Depression Scale (HADS), respectively. Then, elderly adults were randomly allocated to either a control (19) or an intervention (19) group through block randomization. Elderly adults in the intervention group received aromatherapy massage using lavender, chamomile, and rosemary. Each massage session lasted 20 min and was performed three times per week for two three-week periods with an intervening one-week break, while their counterparts in the control group solely received routine nursing homes care services. HADS Scale completed with repeated measurements before the intervention, at the end of the third week, at the beginning of the fifth week and at the end of the seventh week. Results According to the results, mean anxiety in the intervention group went from 11.9 ± 4 to 6.26 ± 3.38 (p <.0001), and the mean depression went from 9.94 ± 3.2 to 4.15 ± 2.14, indicating that anxiety and depression were significantly reduced compared with before intervention (p <.0001). Conclusion Aromatherapy massage with lavender, chamomile, and rosemary oils is effective in significantly reducing anxiety and depression of elderly adults living in the nursing homes.
INTRODUCTION
Aging is the result of the natural course of time that leads to physiological, possibly psychological, and social changes. (1) Today, the elderly population has become a major global phenomenon, and their number has increased due to increased life expectancy and better health outcomes. (2) The global growth of the elderly population has been predicted to increase from 9% to 16% during 1995-2030, and this figure is anticipated to be from 5.6% to 17.5% in Iran. (3) Older adults are generally considered as being among the most vulnerable groups of the population, (4) and providing psychophysical health for them requires special attention. (5) In addition to experiencing physiological changes, elderly adults face significant events such as retirement, the death of friends and family members, separation from social activities, and changes in social and economic relations, which affect their mental health, happiness, and compatibility. (6) Older people often like to live in their own homes, and their private life, f riends, and acquaintances are very important for them. (7) Admission to the nursing home will have psychological consequences, including a sense of rejection, mental stress, depression, in addition to loss of home. (8) The transition to a nursing home has been known as the most important relocation affecting an older person. (9) Therefore, the evaluation of health interventions should not be limited to the physical problems of the elderly, but involved in controlling stress and hormone balance. Depending on the type of aroma, nerve cells release different neurotransmitters. These neurotransmitters include enkephalins, endorphins, noradrenaline, and serotonin. On the other hand, according to the relationship between the sense of smell with the spirit and human feelings, essential oils can have an effect on the spirit and the body. In fact, odors are able to change the emotions in human. (17) Aromatherapy is used in conjunction with conventional treatments for modifying or treating diseases, and includes two methods: inhalation aromatherapy and aromatherapy massage. (16) In aromatherapy massage, aromatic oils get absorbed through the skin in to the blood and exert their therapeutic effects such as vasodilation, increased skin temperature, pain relief, body relaxation, reduced anxiety, and depression. (16) Lavender therapeutic benefits include sedative, pain reliever, antidepressant, antiepileptic, and antiheadache effects. (18) The benef its of rosemary include antispasmodic, analgesic, antimicrobial, antidepressant, anti-inflammatory, reduced menstrual bleeding, improve general weakness and fatigue. (19) Chamomile health benefits include antianxiety, antispasmodic, sedative, anti-inflammatory, antipeptic ulcer, antibacterial, antifungal, and antiviral. (18) Due to the limited clinical studies regarding the therapeutic effects of aromatherapy massage, especially on depression and anxiety in the elderly population, this study aimed to evaluate the effect of aromatherapy massage with lavender, chamomile, and rosemary oils on the depression and anxiety of elderly adults living in nursing homes.
DESIGN
This is a two-armed, multicenter, randomized controlled trial utilizing an herbal intervention with repeated measures. This study was conducted at the four elderly facilities (Mohammad Morsel, Baharestan, Amiralmomenin, and Mehrazin) in Kerman, the second largest city in southeast Iran. These facilities serve a total statistical population, and provide curative, preventative, and rehabilitative care.
it should include attention to all aspects of the health of an older person. (10) One of the important dimensions of health in the elderly is mental health, which requires attention and prevention of disorders such as depression and anxiety. (11) Tabei et al. (5) stated that depression is the most common psychological disorder of the elderly; as a result, depression in the elderly, after physical problems, is the second cause of disability in this period of life. (5) A person with depression manifests symptoms such as changes in appetite or weight, sleep, psychomotor activity, reduced strength, sense of worthlessness, difficulty in thinking, lack of concentration in decision-making, repeated thoughts of death, and suicide attempts. (2) Anxiety is also a common problem of aging because this period is associated with many disabilities. (11) Anxiety is a vague and unpleasant feeling that has always been accompanied by one or more physical signs such as tachycardia, sweating, headache, and dyspnea, and it is not specific to a particular time or culture. (12) It should be noted that the elderly adults are at higher risk of depression and anxiety due to reduced self-esteem, decreased motor activity, loss of friends and relatives, reduced physical and financial autonomy, and chronic diseases. (11) Treatment for depression and anxiety involves pharmacological and non-pharmacological interventions. (13) For example, benzodiazepines, the main treatments of anxiety, are associated with complications such as headache, drowsiness, lethargy, fatigue and ataxia, as well as risk of dependence. These symptoms are usually more prominent in the elderly populations because of the metabolic changes associated with normal aging. Other drugs are used to minimize the side effects of this category of medications and control anxiety, one of which is essential oils of aromatic herbal compounds. (14,15) Aromatherapy, as one of the complementary medicine methods, is the use of essential oils of aromatic herbs for the treatment and adjustment of mind, body, and spirit, which combines the physiological effects with the use of massage with specific oils. (16) The most obvious effect of the essential oils is their stimulation of the smell sense. Smells affect the brain. The olfactory system is connected to the limbic system, which is the emotional control center-hence anxiety symptoms and 7 for symptoms of depression). This instrument is based on a four-point Likert scale (0, 1, 2, 3), where score zero is related to the absence of a symptom in a person and a score of three is in perfect agreement with the target item. Twenty-one scores will be obtained from each part. In both scales, a score of 0-7 means healthy (without clinical symptoms), 8-10 means mild anxiety or depression, and a score of 11-21 is severe anxiety or depression (clinical). (22,23) In Iran, the validity and reliability of this questionnaire were determined by Kaviani and colleagues. (24) Its reliability coefficient using Pearson's correlation coefficient for anxiety and depression was 0.73 and 0.79, respectively. (24)
Intervention
The intervention included combined aromatherapy massage. Different procedures are used to deliver aromatherapy such as inhalation, massage, baths, and compresses. Aromatherapy massage is the most widely used complementary therapy. In the present study, aromatherapy was provided in the form of massage with essential oils of lavender, chamomile, and rosemary in a 4:3:2 ratio and diluted in jojoba oil at 3%. (20) The lavender, chamomile and rosemary essences and jojoba oil had been produce by Barij Essence Pharmaceutical Company (producer of herbal medicines in Iran) of Kashan, Iran. Concentration of essential oil used in this study was determined on the basis of a review of the literature (20,21) and in consultation with the traditional medicine specialist (third author). Ten drops of the above combination were poured over the back of the legs, foot, and knees. Each 20-min massage began with the foot, leg, and knee with light and heavy rubbing and vibration at a speed of 20 times per min. (20) Participants in the control group did not receive any intervention from the research team. The first researcher, who carried out the massage, trained under the supervision of a traditional medicine specialist who had obtained a certificate of massage therapy.
The fourth author studied aromatherapy massage theoretically based on a technique f rom a basic aromatherapy textbook, (25) and as a specialist in traditional medicine, trained the first author how to apply and massage the participants. Aromatherapy massage was performed for the intervention group in 20-min sessions
Sample
According to Rho et al. (20) (in study on elderly women) and Hur et al. (21) (in study on menopausal women), regarding the standard deviation of anxiety in the intervention group (5.99) and in the control group (3.60), the type 1 error of 0.05 and power of 80%, 17 people were included in each group to determine the betweengroup difference in anxiety variations (µ 1 -µ 2 = 5). (20,21) The goal was to recruit 20 participants in each group based on power analysis and attrition rate. Stratified randomization method was used to allocate the participants to the intervention and control groups, so that men and women were divided into groups separately; then in each group of men and women, block randomization design was used. R for Windows 3.5.1 software was used for random block design. Each block also included four people, two of whom were from the intervention group and two from the control group. Half of the men were entered in the intervention group and the other half in the control group. The same method was used for women. Statistician was responsible for randomization. It should be noted that, although he was involved in the entire study, he did not know who were in the intervention group; assigning people in the control and intervention group was noted as "A" and "B" by the statistician.
Inclusion criteria were ages between 65 and 85 years, ability to speak Persian, participation in the study design for seven weeks, and toleration of aromatic oils. Exclusion criteria include a history of skin allergy to aromatic substances, cognitive disorders (delirium & dementia), having mental disorder diagnosed by a psychiatrist, using antianxiety and antidepression drugs, as well as participants who were directly self-reporting and having a problem with the sense of smell (checked by a researcher with rose flower scent).
Instrument
Demographic questionnaires included variables such as age, gender, marital status, education, and satisfaction from the nursing home.
The Hospital Anxiety and Depression Scale (HADS) is a valid and reliable selfrating scale that measures anxiety and depression in both hospital and community settings. The HADS has 14 items (7 for three times per week for two three-week periods separated by a one-week break, (20) for a total of 18 treatments. Rationale for the one-week break between treatment periods was evaluation of longer duration effect of aromatherapy massage.
Data Collection
Sampling was started after receiving the code f rom the Ethics Committee of Kerman University of Medical Sciences and obtaining written informed consent from eligible elderly from the four nursing homes in Kerman. Data were collected by convenience sampling method f rom the four elderly facilities and the samples were randomly divided into two groups of intervention and control. Demographic questionnaire and HADS scale were used for data collection.
The first researcher presented the HADS for each time point and each participant completed the questionnaire individually through the self-report. Data were collected at four time points for two groups. The first point of time was before the intervention when the demographic questionnaire and HADS were administered to the participants in each group. The next points of time for data collection for two groups were the end of the third week, at the beginning of the fifth week, and the end of the seventh week after the intervention, respectively. (20) The dates of data collection were from 28.8.2019 to 16.10.2019, and all participants across the treatment and control groups completed the study at the same time.
Data Analysis
Data were analyzed using statistics (19) and descriptive and analytical statistics. For comparing demographic variables between two groups, independent t test, chi-squared test, and Fisher exact test were used. Repeated measures ANOVA were used to assess anxiety and depression between the two groups across times. Also, independent t test was used as a post hoc analysis to compare the mean scores between the intervention between the two groups. The a priori alpha level was set at 0.05.
Ethical Considerations
The code of ethics (IR.KMU.REC.1397.192) was received from the Ethics Committee of Kerman University of Medical Sciences. After obtaining written informed consent from each elderly individual, the research objectives were explained and they were assured that the information would be kept confidential. They were also informed that the participation was optional, and they were allowed to withdraw from the study. The results were reported to authorities and participants at the end of the study.
Descriptive Results
It should be noted that one participant in the control group was excluded due to stroke and one in the intervention group because of the discontinuation of the study, so the study continued with 19 participants in each group.
Based on the results, independent t test showed there is not a significant difference in age between the two groups. According to chi-squared test, there was no significant difference between the two groups in terms of gender, marital status, and level of education. Also, the Fisher exact test showed there was no significant difference between the two groups in terms of level of satisfaction f rom the nursing home (Table 1).
Depression & Anxiety
Results of the anxiety and depression HADS assessments are presented in Figures 1 and 2. At baseline, scores on the HADS anxiety and depression subscales for the control group were rated as mild and healthy, respectively. For the intervention group baseline scores were categorized as severe anxiety and mild depression. According to the results, depression (t = 2.80, p = .008) and anxiety (t = 2.80, p = .008) scores revealed a significant difference between the two groups at baseline (Table 1). Repeated measures ANOVA were used to compare the change in anxiety and depression scores across the seven weeks of the study (Figures 1 and 2). Repeated measures ANOVA results showed the anxiety mean score at the end of the third week (7.94 ± 3.70), at the beginning of the fifth week (7.94 ± 3.27), and at the end of the seventh week (6.26 ± 3.38) had decreased in the interven-
DISCUSSION
The results showed that elderly adults who received aromatherapy massage with lavender, chamomile, and rosemary in 20-min sessions three times per week for two three-week periods separated by a one-week break showed reduction in tion group, while it had increased in the control group (p < .0001). Also, repeated measures ANOVA results showed the depression mean score at the end of the third week (5.94 ± 2.81), at the beginning of the f ifth week (6.15 ± 2.69), and the end of the seventh week (4.15 ± 2.14) had significantly decreased in the intervention group, while it had increased in the control group (p < .0001). depression and anxiety than those in the control group. Similar studies that examined the effect of aromatherapy massage on depression in the elderly were not observed, but the results are consistent with the results of some studies that investigated the effect of similar scents, with or without massage, on depression reduction in menopausal women, (26) and cancer (27) and hemodialysis patients. Also, result of present study is consistent with the results of the Rho et al. (20) study that examined the aromatherapy massage with a combination of three oils similar to the present study (lavender, chamomile, and rosemary) and lemon oil on the anxiety and self-esteem of the elderly. They reported that anxiety of the elderly signif icantly decreased after the intervention. (20) The effect of lavender, chamomile, and rosemary oil on the depression and anxiety of elderly adults can be attributed to the stimulation of the olfactory system and neurobiological changes. The limbic system as the stress control center is connected to the olfactory system. Depending on the type of aroma, several neurotransmitters can be released by neurons. Neurotransmitters play a role in nearly every function in the human body. A balance of neurotransmitters is necessary to prevent certain health conditions, such as depression and anxiety. After topical application, the essential oils also get absorbed into the blood and exert their effect through the blood stream. (17) Aromatherapy massage has been shown to relieve self-reported symptoms of anxiety, and patients perceive aromatherapy massage as positive and beneficial. Aromatherapy oils administered by inhalation without massage do not appear to reduce anxiety. (28,29) Massage is a technique which provides both mental and physical relaxation. It is a direct form of non-verbal communication since it requires touching. Furthermore, it supports the communication between the nurse and the individual. (30) The main mechanism of massage includes increasing lymph flow throughout the body, shifting the autonomic nervous system from a state of sympathetic response to a state of parasympathetic response, and increasing blood lactate clearance. In general, massage potentially decrease anxiety, depression, and pain by increasing serotonin and endorphin. (31) According to the demographic questionnaire and Hospital Anxiety Depression Scale (HADS) results, the elderly had a high degree of depression before intervention. This is consistent with previous studies.
In a study by Ranjan (32) nearly half of the elderly, and in the study of Ahmed (33) more than one-third of the elderly, in nursing homes suffered from depression. Abbasi et al. (2) described depression as one of the major issues of aging in Iran. The number of elderly adults has increased in different societies because of the tremendous advances in the prevention and treatment of diseases in today's world and consequently, the incidence of mental and physical illnesses, including depression, has increased in this period. Praveen Kumar et al. (34) believe that a global increase in the elderly population causes health problems for them. The authors go on to say that mental illness is not usually identified and treated in the elderly, as the prevailing thinking in developing countries is that the psychiatric problems of the elderly due to chronic diseases, loss of family members or changes in a social situation are natural reactions.
The results of our current study also showed that the anxiety level of the elderly under study was high before intervention. This result is consistent with previous results.
Marsa et al. (35) reported higher anxiety levels among older people living in a nursing home than those living at home in Iran. Ahmed et al. (33) reported that anxiety levels among people living in nursing homes was 80%. The incidence of anxiety in the elderly is not unexpected because people in nursing homes are faced with factors such as dependency, sense of loneliness, financial dependence, chronic illness, fear of death or absence of someone at the time of death. Andreescu and Varon (36) wrote that anxiety in the elderly is usually chronic, and its diagnosis and treatment is not well done.
According to the results of our study, the important issue from the therapeutic point of view is the difference in the trajectory of the improvements between the elderly people in the two groups. Those receiving aromatherapy massage experienced a significant improvement in anxiety and depression after intervention. By contrast, the severity of anxiety in the control group went from subclinical to severe between the baseline measure and the second measured time point. Depression went from mild to moderate.
High appreciation of family support within the social and cultural structure of Massage is also one of the complementary therapies used in health care as an adjunct therapy. Massage can stimulate the central nervous system and create a sense of calm. Different procedures are used to deliver aromatherapy such as inhalation, massage, baths, and compresses. If massage is performed along with aromatic oils, it is called aromatherapy massage. Aromatherapy massage is the most widely used complementary therapy, in which case these oils can quickly get absorbed by skin and enter the bloodstream. Therefore, there is no proof that aromatherapy made the difference and more study is needed.
Other limitations include the small sample size, the impossibility of blindness of the researcher, and the research participants. In this study, the self-report approach was used to collect data. In this method, anxiety and depression may be reported to be less than they actually are. Therefore, for the results of the current study to be confirmed, it is suggested that further research be done with different designed method for gathering data, larger sample size, and provision of blindness conditions. Also, the lack of a control group receiving any kind of attention was a limitation. To overcome this limitation, the opportunity was given to the control group participants to receive intervention after the completion of the study, if they wanted. Another limitation with the study is the statistically significant baseline differences.
It is recommended that nurses who take care of elderly can use the aromatherapy massage with lavender, chamomile, and rosemary oils to reduce depression and anxiety. Also, that the managers of the nursing homes take an effective step in improving the mental health of the elderly with the inclusion of this intervention in the care program. Future clinical studies that include larger samples, equivalenttreatment control group, and long-term follow-up are needed to determine the effects of aromatherapy massage on other psychological variables in the elderly adults.
ACKNOWLEDGMENTS
The present article is the result of a Master thesis, which was registered by the research deputy at Kerman University of Medical Sciences with the code of research 97000227. Researchers would like to thank the elderly people who participated in this Iranian society has led to negative views of nursing homes within Iranian society. In Iranian culture, living in a nursing home is a stigma for the older person, and can cause a great amount of tension, feelings of isolation and rejection f rom family, and ongoing stress and reduced mental health. This stigma arises from the deeprooted culture of family cohesion and belonging, which is outlined in the Koran: "If one of them [parents] or both of them reached their old age while staying with you, do not utter a word of disrespect such as "Oh" nor irritate them, but address them in terms of respect and kindness and be humble out of compassion and pray. (37) Our results showed aromatherapy massage could significantly reduce depression and anxiety in the elderly, given that essential oils can be absorbed into the body via the skin or the olfactory system. Olfactory stimulation produces immediate changes in physiological and psychological parameters. (20) According to the relationship between the sense of smell with the spirit and human feelings, essential oils can have an effect on the spirit as well as the body. (17) Ventola (38) wrote that many patients find it easier to use aromatherapy than conventional medicine. Since the elderly are looking for solutions to their problems due to losing or reducing their physical and cognitive activities, training the elderly in aromatherapy massage as an easy way to reduce anxiety and depression, and also involves them in health-related activities. (39) Frequent measurements during the intervention can be considered as the strengths of this study compared with other studies.
One limitation of this research was the lack of an equivalent-treatment control group to estimate the superior effectiveness of aromatherapy massage. Therefore, it is not clear whether the positive effects were due to the aromatherapy, the massage, or both (e.g., identical results may have been achieved by using only massage, or by using aromatherapy and massage). According to the previous evidence given, massage alone showed benef it for both anxiety (28,40) and depression, (30) and inclusion of the massage without aromatherapy would have allowed for aromatherapy comparison while potentially decreasing symptomology in the control participants. | 2022-03-13T05:13:43.574Z | 2022-03-01T00:00:00.000 | {
"year": 2022,
"sha1": "71bdfdb9907e8de2517df81991365b8d554b5742",
"oa_license": "CCBYNCND",
"oa_url": "https://ijtmb.org/index.php/ijtmb/article/download/645/783",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "71bdfdb9907e8de2517df81991365b8d554b5742",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
53001705 | pes2o/s2orc | v3-fos-license | INTEGRAL long term monitoring of 4 U 1722 – 30 : evidence for spectral state variations
We report on the 2003-2005 INTEGRALobservations of the Neutron Star Low Mass X-ray Binary 4U 1722–30 (also known as GRS 1724–30) located in the Globula r Cluster Terzan 2. The JEM-X and IBIS light curves show the source with a persistent yet va riable flux. The Hardness-Intensity diagrams highlight the behaviour of a typical Atoll source: 4U 1722–30 repeatedly moves in the diagrams from the Banana (Soft state) to the Island (Hard sta te). We report on the detailed spectral analysis of Soft and Hard states and, for the first time, also o f an Intermediate state. The Hard spectra reveal a Comptonised corona emission up to 200 keV ch ara terized by a high temperature of 40 keV and optical depth of 0.5. In the Soft state the main em ission is from the accretion disk (with kTin ∼ 0.5 keV) whereas the Comptonised emission decreases showin g an optically thick and cold corona ( τ ∼ 9, kTe ∼ 2 keV). During the hardening there is an increase of the inner radius of the accretion disk suggesting a system expansion d uri g the spectral transition. This behaviour is reminescent of the Soft X-ray transient source s though 4U 1722–30 never reaches a real "quiescent" state.
and IBIS light curves show the source with a persistent yet variable flux.The Hardness-Intensity diagrams highlight the behaviour of a typical Atoll source: 4U 1722-30 repeatedly moves in the diagrams from the Banana (Soft state) to the Island (Hard state).We report on the detailed spectral analysis of Soft and Hard states and, for the first time, also of an Intermediate state.The Hard spectra reveal a Comptonised corona emission up to 200 keV characterized by a high temperature of 40 keV and optical depth of 0.5.In the Soft state the main emission is from the accretion disk (with kT in ∼ 0.5 keV) whereas the Comptonised emission decreases showing an optically thick and cold corona (τ ∼ 9, kT e ∼ 2 keV).During the hardening there is an increase of the inner radius of the accretion disk suggesting a system expansion during the spectral transition.This behaviour is reminescent of the Soft X-ray transient sources though 4U 1722-30 never reaches a real "quiescent" state.
Introduction
4U 1722-30, also known as GRS 1724-30, is a bright LMXB located in the Globular Cluster Terzan 2 [1].The observed Type 1 X-ray bursts indicate that the compact object is a weakly magnetized neutron star [1], [2].The timing properties outlined with RXTE observations suggest that its timing properties are typical of an atoll source [3].4U 1722-30 is a persistent though variable source, and it is one of the first neutron star systems from which hard X-ray emission (E >35 keV) was detected by SIGMA with a power law spectrum with photon index Γ ∼1.65 extending above 100 keV [4].Previous EXOSAT observation didn't reveal flux above 10 keV [5].BeppoSAX and RXTE allowed a broad band observations, detecting the source with a Comptonized spectrum extending up to 200 keV, plus an additional soft componenet (below 3 keV), described by a blackbody emission.
The INTEGRAL monitoring
We monitored the sources with INTEGRAL during the period October 2003 -April 2005, collecting a total of 883 pointing for IBIS and 256 pointings for JEM-X.Because of this long term monitoring, INTEGRAL detected the source in different spectral states and for the first time a detailed spectral analysis of different spectral states has been performed.The colors identify the spe1 (purple), spe2 (green) and spe3 (red) data sets (see Table 1).
The light curves and hardness-intensity diagrams
The JEM-X and IBIS light curves in different energy bands are shown in Figure 1.Each INTEGRAL point correspond to a single pointing lasting about 2000 seconds.In the top panel are reported the ASM/RXTE one-day monitoring during the same period.The source reveals a flux variation (similar to outbursts) in the soft band (<20 keV), while in the hard ones (>20 keV) it is more rarely significantly detected and the flux changes to a minor extent.
We constructed two hardness-intensity diagrams, the first one with the JEM-X data and the second one with IBIS and JEM-X simultaneous data.Both are shown in Figure 2, with indication on the Intensity and Color bands used.The source moves through the diagram showing spectral changes.We show with different color the different spectral data sets for which observations log is reported in Table 1.The purple data set refers only to the JEM-X/JEM-X hardness intensity diagram because of the lack of high energy detection by IBIS.This data set correspond in fact to a Soft (banana) spectral state.The green and red pointings correspond to a hardening of the sources that enters the Hard (island) spectral state.These pointings are reported in both the hardnessintensity diagrams.The mass accretion rate increases during the softening with a corresponding decreasing of the inner accretion disk radius (see discussion for details).
The spectral evolution
Spectral analysis was performed by collecting the data corresponding to the same spectral state as shown in the Hardness-Intensity diagram of Figure 2.
Soft state
For the spe1 data (see Table 1) the best fit model is represented by a black body model (or also simple black body model) [6] plus a Comptonization model [7] with parameters showed in Table 2. Changing the diskbb model with the simple black body model the fit doesn't change.
The source shows this spectral state during the soft "outbursts" clearly evident in the JEM-X light curve, when it isn't detected above 30 keV.
Hard and Hard/Intermediate state
The green and red spectral data sets corresponds to the hardening of the source.The Intermediate state is detected just after the soft "outburst" shown in the light curve, and the Hard state follows soon after this.
The Hard/Intermediate state is well represented by the diskbb model plus a Comptonisation model.For the Hard state, the best fit is a simple Comptonisation and the fit doesn't improve by adding a diskbb component.The plasma temperature rises with the hardening, while the optical depth decreases, as indicated in Table 2.
In the Hard state the plasma temperature is not constrained very well and there is the indication of a lack of cut-off, similarly to the Hard state of the atoll 4U 1608-522 [9].A reflection component is not necessary.The spectra, model and residuals are shown in Figure 3 (panel spe2, spe3).The bolometric luminosity of the Intermediate state corresponds to 1.2×10 38 ergs s −1 , that yields a 0.6 L Edd .The bolometric luminosity of the Hard state corresponds to 1.4×10 37 ergs s −1 , i.e. a L Edd ratio of 0.07.2.
Discussion and conclusion
The INTEGRAL observations of 4U 1722-30 allow us to follow the X-ray behaviour of this source that is very similar to a X-ray transient, though a real "quiescent" state is never reached. of transient sources as also confirmed by the color-intensity diagrams, such as the ones for the transient source 4U 1608-522 [9].
At high soft flux level the source was in Soft state, followed by a hardening.During the Soft state the source doesn't show emission above 30 keV, and the spectrum is well described by a cold and optically thick Comptonized corona (τ ∼ 9 and kT e ∼ 2 keV) plus a soft black body emission (kT in ∼ 0.46 keV) coming from either the accretion disk or the neutron star.
During the hardening (at low accretion rate) the contribution of the soft component decreases, with a corresponding increase of the hard X-ray emission (up to 200 keV) described by a hot and optically thin Comptonizing corona (τ ∼ 0.5 and kT e ∼40 keV), without any evidence of an energy cut-off.
We estimated the inner radius of the accretion disk in the soft and hard/intermediate state and derived an increase from 5 to 20 km, suggesting an extension of the inner radius during the hardening as also shown by the transient source 4U 1608-522 [9].
a
Fixed parameters bThe Fluxes are in units of erg s −1 cm −2
Figure 3 :
Figure 3: IBIS and JEM-X energy spectra of 4U 1722-30, with the residuals of the data to the model.Spe1 corresponds to the Soft spectral state; spe2 to the Hard-Intermediate; spe3 to the Hard.The models applied are the blackbody plus a Comptonization for spe1 and spe2 data sets and only a Comptonization for spe3.The best fit parameters are reported in Table2.
Table 1 :
The log of the INTEGRAL/IBIS and JEM-X data used for spectral fitting of 4U 1722-30.
Table 2 :
Spectral fitting results for the JEM-X and IBIS broad-band spectra of 4U 1722-30.The model is CompTT for spe3 and CompTT + diskbb for spe1 and spe2.
The outbursts are clearly visible in the INTEGRAL light curves, with spectral changes typical | 2018-08-10T00:25:25.502Z | 2009-07-08T00:00:00.000 | {
"year": 2009,
"sha1": "fe34b8ab5817b56e8f46d07472529c947d4ef9a8",
"oa_license": "CCBYNCSA",
"oa_url": "https://pos.sissa.it/067/046/pdf",
"oa_status": "HYBRID",
"pdf_src": "Anansi",
"pdf_hash": "fe34b8ab5817b56e8f46d07472529c947d4ef9a8",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Environmental Science"
]
} |
244282273 | pes2o/s2orc | v3-fos-license | Challenges in the Implementation of Peer-to-peer Recognition
This paper aims to explore potential issues in implementing peer-to-peer recognition to motivate employees. A qualitative inquiry is applied to enhance our understanding of peer-to-peer recognition schemes in practical situations. Sixteen participants, including HR researchers and managers from different organizations where this program has been implemented, attended semi-structured interviews. From which twelve cases are chosen for this study. Using thematic analysis on the collected responses, the researcher finds that challenges to peer-to-peer recognition are to be found in four categories: the impact of personal relationship, the unsuitability of working climate, the confusing nomination criteria, and the lack of an appropriate rewarding policy. Practical solutions to deal with those challenges are also proposed, which include building the culture and assigning monitor, to assist employers to effectively make use of a peer-to-peer program in motivating employees. This study therefore raises the awareness of optimizing motivation program for HR practitioners, as well as contributes a qualitative approach toward the non-cash recognition evaluation.
Introduction
Peer-to-peer recognition in the workplace is defined as a HR management program that allows employees to be recognized by their peer and gives recognition to the co-workers (Hastings, 2009;Vranjes, 2014). It provides a bunch of benefits for both employees and employers. As for employees, peer-to-peer recognition firstly generates more chances for an employee to be recognized. Compliments from peers create as much motivation as those from managers (Maslow, 1954;Smith, 2014;Achor, 2016). Unlike manager's appraisal that normally takes place quarterly or at the end of project and is based on overall performance or given for mostly high-visible tasks, peer recognition happens in daily interaction and is based on small achievements and good behavior. Therefore, the latter is more equal and valuable and makes employees proud more of their efforts (White, 2017). By increasing the number of positive interactions among workers, the program also enhances trust, strengthens relationship and working spirit (Gostick and Elton, 2008;Vranjes, 2014;Byam, 2015).
As for employers, peer-to-peer recognition, at first, creates a supportive working environment which leads to the increase of employee engagement and commitment (Nelson, 2004;Globoforce, 2015). Second, it helps reduce the mistakes in the work. The "Hawthorne effect" indicates that people work harder and more carefully under supervision (McCarney et al., 2007), even when the supervision comes from peers. At the same time, an employee would reflect to herself and attempt to make equivalent progress as she nominates her peers (Byam, 2015). Third, the results of peer-to-peer recognition efficiently assists managers in terms of performance appraisal, particularly for talent discovery and long-distance management (Vranjes, 2014;Steele, 2016).
Recently, HR researchers worldwide have noticed an increasing trend in the usage of peer-to-peer recognition since 2010 (GloboForce, 2015;TINYpulse, 2015). Worldatwork (2015) indicates that 48 percent of 489 surveyed organizations apply the recognition method of peer-to-peer, and the number is forecasted to increase. Among the surveyed sample, there are big names including Harvard Law School, Nestle, Zappos and Mindvalley. Despite the upward trend in using peer-to-peer recognition, this method, particularly its implementation process, still receives lack of attention from academic researchers (Smith, 2014;Gribben et al., 2018). Acknowledging the need, this study looks for organizations where peer-to-peer recognition are put on action, either who have gained or failed from it, to collect intelligence and answer the question: What are the difficulties when implementing peer-to-peer recognition and how to resolve?
From the interview results, a qualitative inquiry is applied to enhance our understanding about peer-to-peer recognition schemes in practical situation. This study demonstrates a way to use semi-structured interview and thematic analysis to explain in depth a current recognition method of peer-to-peer, exploring its characteristics and implications.
The research determined four challenges during the implementation of peer-to-peer recognition and four related solutions within the experiences of the respondents. The first concern is about how to achieve a bias-free environment for recognition, as recognition results are seriously influenced by employee personal relationship at work. The riddle could be solved if managers facilitate a culture of complimenting before starting the program. The ambience, or climate, within the organization itself is the second most mentioned issue by respondents. For example, a high-competitive working environment can diminish the overall efficiency as peers are reluctant to recognize others. The suggestion is to elaborate a campaign that communicate clearly about the purpose and benefits of peer-to-peer recognition, which is of more intangible than tangible values. The third challenge is about setting the nomination criteria. Employees might find it unfair or misunderstand the criteria, while implementers struggling to make it right. Encouraging labor representatives to participate in discussing the rule before establishment is the best choice to reduce this risk. The last challenge is about maintaining the program in the long term, as employee engagement is likely to decrease harshly overtime. This is the consequence of not having proper rewards from the organization, as peer recognition alone is not enough to motivate.
Methodology
From March 20 th , 2017, to June 20 th , 2017, the researcher contacted 31 people from many different organizations and industries around the world, who have proof of experience about peer-to-peer recognition. Among those who were contacted, sixteen agreed to participate in the interview. These participants were found and contacted online using social networks, such as Linked, Quora, Facebook, Skype, and email, as Internet is the most popular way nowadays to connect research initial informants (Merriam and Tisdell, 2016:127).
Semi-structured interview is chosen as the core of this research design. This type of interview has flexible questions but still follows a guidance of topic boundary. It allows researchers, who already have moderate understanding about the topic through personal reflection and literature review, to maintain the control over the interview (Melia, 2000;Whiting, 2008;Merriam and Tisdell, 2016:111). With semi-structured style, the researcher can narrow the interview topic to a specific area, but still set an equal role with interviewees to adopt new thoughts and feelings.
The purposive sampling, also known as judgement sampling, is chosen for this study. Answering this research question requires only a limited number of participants who experience the peer-to-peer recognition, as a manager or an implementor, in their organizations (Tongco, 2007) studies qualitative researchers and concludes that purposive sampling is very effective for studying specific knowledge, case studies and practices comparison. In addition, this sampling method allows the researchers to base on the research questions to choose the best sample that can serve the research objectives (Saunders et al., 2015:301).
As for the sample size, Saunders et al. (2015:297) recommends having 5 to 25 interviews in a semi-structured interview. Unlike quantitative research, the quantity of responses in qualitative research is not as important as the quality contributed to the understanding of subject by those responses (Merriam and Tisdell, 2016). The online interview session was operated in 3 phases: warm-up, discussing and closing. As for warm-up phase, Merriam and Tisdell (2016:125) and Whiting (2008) recommend starting an interview session with greeting, making informal and neutral conversation, asking descriptive question to build the rapport with the interviewee. In the discussing phase, the participants were asked about their experiences and opinions about peer-to-peer recognition program of their organization. As the interview is semi-structure, there was no predetermined script and the respondents were free to share, or extend their personal ideas (Saunders et al., 2015:393), but every session was consciously controlled to be a "systemic, structured and with-purpose conversation" (Brinkmann and Kvale, 2015:5). During this phase, the researcher needs to keep in mind general instructions while asking questions in clear language, remaining the neutral and respect attitude toward interviewees' answers by not giving personal beliefs or values and not arguing with the interviewees (Merriam and Tisdell, 2016:128). Finally, the ending phase involves asking if the participants have any concerns, questions, or thoughts that they want to share, thanking them and finishing the session.
For the analysis of collected qualitative data in this research, the researcher uses inductive approach, together with thematic analysis. Inductive approach is for researchers who start with collecting data, then explore new theory from the reality observation (Collis and Hussey, 2014:7). From the collected answers of interviewees who have experience in peer-to-peer recognition, the researcher explores the issues in its implementation process and the solutions. This can be considered as a fair contribution of knowledge to the research topic of an inductive researcher Thematic analysis is the core method to explore the data in this research. In thematic analysis, raw data is coded into themes, or patterns, or categories (Merriam and Tisdell, 2016:204) and interpreted as answer for the research question. The thematic analysis process includes four overlapping activities: data reduction, data display, conclusion, and verification (Collis and Hussey, 2014). In the first stage, the researcher gets familiar with the raw data by reading it all over again before coding it into themes. After being coded, themes are discussed, linked, and refined to develop propositions for testing.
Theme One: Peer-to-peer recognition is significantly biased towards personal relationships
This theme is not only mentioned the most by the interviewees but also widely discussed in business literature. It is a significant key finding of this research supporting the consensus. According to the result, the most common challenge of peer-to-peer recognition correlates with the personal relationship among co-workers. Sami, who launched this program in his company, said: If you are my colleague and I like you, then I will submit a very positive evaluation for you, otherwise I will submit a negative one. (LinkedIn, 2017) where employees would endorse each other basing on a whim, and so it became meaningless. (David) In another organization, employees used to take peer-to-peer recognition as a favor and tend to repay the giver: In the beginning, it was somehow similar to a kind of repay. When I received a thank-you note from a particular colleague, I have an urge to give back another note to that very person... I would look for a proper occasion, but it is possibly biased. (Linh) Personal bias could make peer-to-peer recognition meaningless, from a motivation program to become a flattery channel. It is still recognition but neither satisfies an Esteem need (Maslow, 1954), nor is a motivator factor (Herzberg, 1966) as the recipients value it just as much as a handshake. In one's defense, it might create reinforcing motivation (Skinner, 1953), but only among closed-relationship employees, or favored toward popular employees (Creelman, 2016). A recognition program only works well if it serves the true purpose of appreciating achievements and create fair opportunities for everybody (White, 2017). For the solution, many interviewees admitted that personal bias is something inevitable, but not unchangeable. It is recommended that manager should be patient and use bottom-up approach to gradually eliminate the personal bias. To be success, a peer recognition program may take as long as building a new culture. Maarten suggested to start the program bottom-up: Involve employees from the start. (...). We also develop the culture of how to use those feedbacks afterward. (Phuong) Anh Duc suggested role-changing to make co-workers understand others' tasks, and from that to create a culture of sympathetic: We switched the role of employees through projects: coders, testers, account managers, etc. In this way, employees not only learn more skills but also put themselves in others' shoes, to understand others' difficulties and advantages. It sets a good cultural base for peer-to-peer recognition. (Anh Duc) According to Creelman (2016), peer recognition is not seriousness but culture and celebrating other's accomplishment with joy. Previous studies also stress the point of involving the employees to build the peer recognition culture (Vranjes, 2014;Burke, 2016). Building recognition giving culture can provide peer-to-peer recognition program with a supporting context (Nelson, 2016), with high awareness of different work responsibilities among employees and high program engagement. Strong recognition culture can play a role of a strong "reinforcer" (Skinner, 1953;Daniels, 1989) to motivate employees.
Theme Two: The current climate within the organization does not support this program
According to some respondents, peer-to-peer recognition may encounter difficulties in eastern organizational culture, large and hierarchal organizations, competitiveness, and working environment in which unskilled workers are the majority. Daphne, Phuong and Son, who share the Asian culture, agreed that organization with a hierarchical structure is a challenge: The existing workplace culture (in Asia) which may not be supportive of such programs. Many Asian workplaces are leader led, hierarchical and not good at praising. People in these environments are more likely not to say anything or acknowledge good work as this is an example of 'just doing your job'. (Phuong) From another aspect, Dinh Duc and Sami thought highly competitive working environment can make peer-to-peer recognition ineffective. It even raises the level of jealousy and internal conflict.
Since the BOD (board of directors) made teams to bid on projects, they have competed harshly even on little things. G1 (a team) just give recognitions among themselves. The team leader did nothing about that… (Dinh Duc) In another example, Mai believed the work nature itself and workers' level of education can create challenges for applying peer-to-peer recognition: It is easier to apply the peer-to-peer recognition for companies with high-skilled workers such as a law company, a university, or an educational organization, also for small and medium enterprises than big-mature ones. (Mai) Daphne had a concern that work nature can become an issue of unfairness in a peer-to-peer recognition program: Often it may be the same person all the time because their work nature and personal styles are framed that way, for example the boss secretary or customer service agents or the IT help desk people. Individual contributor roles, or cross functional roles are rarely nominated or recognized. (Daphne) These opinions are correlated with the context criteria of a recognition program, which was also a concern of Luthans (2000), Nelson (2016), and Steele (2016). Some of the responses supported the argument of Gostick and Elton (2008), Vranjes (2014) and Mazzei (2016) that peer-to-peer recognition works better for small team and high-visible job positions, and less effective in hierarchal organization. For the solution, new implementors need to carefully prepare the communication campaign before launching. Dang believed a peer-to-peer recognition scheme should be marketed like a product, by which the benefits are highly communicated. (Vranjes, 2014;Byam, 2015;White, 2017) themselves. Therefore, it is the responsibility of the organization to make it clear. When employees clearly understand the program benefits, they will engage contingently and frequently (Nelson, 2016). Furthermore, organizations should communicate their expectations of employees in the program, such as being honest, fair and carefully writing nomination (Steele, 2016).
Theme Three: Nomination criteria are confusing or not applicable to all groups/departments
One third of respondents agreed that creating recognition criteria can be a challenge. Basing on its own core values and goals, each organization will create a unique set of nomination criteria when it comes to peer-to-peer recognition. As a result, one cannot affirm her set of criteria as the best fit. In addition, for something as complicate and abstract as recognition (Smith, 2014), a set of guidelines or criteria can be perceived differently by different teams or departments. Daphne and Sami were worried that some criteria can become confusing and difficult to be measured: Of course, each one has his own interpretation, but I think sometimes my company had the wrong values or behaviors recognized or appreciated. For example, should a person who works late be recognized? Is this a culture that we want in our office? (Daphne) Linh and Dang reported the situation when criteria are too difficult to understand and can be interpreted inconsistently: We used to have so many difficulties with the criteria, such as some criteria received little or zero recognition. (Dang) From what I know, the HR department, the Customer Service department, and the Training department understand and proceed the evaluation method in totally different ways. (Linh) Mai complained that her organization's peer recognition marking form cannot adapt to every project:
Criteria need to be modified to suit my working responsibility and experience. Many times, I found it does not reasonable to mark my team performance. (Mai)
Because of inapplicable criteria, employees from different functions may have difficulties recognizing each other, such as when a salesperson reviews an accountant, or when a receptionist reviews a marketing executive (Steele, 2016). When a confusing recognition criterion is left out by employees, the expected behavior or action will not be encouraged to happen (Hansen et al., 2002;White, 2017). It does not meet neither the frequency and context criteria of giving recognition (Nelson, 2016) nor the reinforcement theory (Skinner, 1953;Daniels, 1989). Setting easy-to-understand recognition criteria can be a real challenge. However, the best choice is to make the criteria as short and simple as possible, by linking it with company and department missions and goals, or just having no criteria at all: Identify criteria for recognition and tie them back to organization's goal and mission. You can seek for input from all layers of your organization. (Simon) Why make it complicated? We let our employees to make the choices on their own whether to give a thank-you or not. We just remind them to write down the reasons clearly so that the receiver feels recognized and motivated. (Linh) A couple of respondents suggested passing the ball to the employees themselves. Dang suggested embedding employees' opinions: We interviewed to find out what our employees need about recognition and combine it in the program design. For example, we took the idea of attaching peer-to-peer recognition into KPI (Key Performance Indicator). By this way, it can cover all kinds of performance and behavior. (Dang) From the literature, Creelman (2016) suggested that instruction should be given, and criteria should be gradually changed using best-practice approach. As for the issue of fairness among job positions and personalities, Steele (2016) agreed with creating forms and criteria that are applicable to all departments or job positions, also giving a blank textbox for adding comments. In the end, having a well-designed nomination rule and criteria would help meet contingency, frequency, formality, and context criteria of a good recognition program (Vranjes, 2014;Nelson, 2016).
Theme Four: Maintain employee engagement -The absence of appropriate rewards
Finally, yet importantly, without appropriate accompanied rewards, peer-to-peer recognition is likely to lose employee engagement and become a wasted program. Half of the respondents agreed on this point. Phuong and Dang complained about the limited numbers of tangible rewards that his company offered: In my case, the reward as for receiving peer recognition is too little. It was granted quarterly by the company for the department with the most nominations. I do not mean greedy, but more budget for rewarding will certainly boost the engagement (for the program). (Dang) Daphne had her experience when an inappropriate reward not only fails to motivate the employee but also becomes a cynicism: It did not go as well when a marketing executive felt that the secret gift, an 89$ spa voucher, was not commensurate with her effort. She had put in many late hours to secure the client's renewal of contract and in Australia, which is not an Australian common working manner. After winning the prize, she quitted because basically, she felt insulted rather than inspired to do more. (Daphne) Linh admitted that the peer-to-peer recognition in her department has happened not as effective as another department, due to offering no tangible reward: One thing that HR department did better than us (Training department) is that they have monthly reward, while we did not. I noticed the engagement has drop since the fourth month. What a shame!... I had to sit down to create a policy for the reward and how lucky, the CEO approved it straight away. (Linh) Maarten warned that the absence of rewards or an equal distribution of rewards will completely knock-out the peer recognition goodwill: Most important, reward of behavior must be perceived as genuine and authentic for these programs to work. It's proof that people "see" and care about efforts. Any tendency of "the next round should go to X because X hasn't gotten it yet" undermines the whole initiatives. (Maarten) Inappropriate reward is a concern that was mentioned in previous studies. Irvine (2012) calls it "an unfortunate trend" as there was an increasing number of organizations equip their peer-to-peer recognition program with zero-value reward. To be effective, recognitions should be valued by the recipient (Smith, 2014;Nelson, 2016). Inferably, tangible, or cash reward from the company, will enhance the effectiveness of peer-to-peer recognition. Without real rewards, a peer-to-peer recognition program has a very high chance to be looked down by employees and would collapse in the short-term (Steele, 2016). This last theme is linked closely with the literature. Data from both primary and literature emphasized that rewards, accompanied by recognition, must be practical, genuine to the individual, specific to the behavior. Simon talked about real-word rewards and specific compliment: Organizations should leverage peer-to-peer recognition by rewarding acclaimed workers within departments and teams with real-word rewards. Rewards should be tied closely to the effort to make the employee aware of the behavior or attitude he should replicate. Plus, the more specific and detailed your recognition is, the sincerer and more honest the recipient will feel it. (Simon) The rewards are recommended to be specified not only for the behavior but also for individuals' needs. Sami, Son, Mai and Dang all agreed on the point that small rewards, both tangible and intangibles are essential add-ons, and they should be grant more frequently.
Rewards should be available for short-term recognition, granted more than quarterly and annually, and for individual apart from for the departments and teams. (Dang) Non-cash reworks can be special treatments, like a priority parking lot, the right to access top secret resources, or a high-end working tool. (Son) Former researchers also recommended having flexibility of choices so that the recognition winners can pick the option that suits them the most. According to Achor (2016), the selection of rewards not only grant employees the chance to choose the most meaningful gift but also increases platform engagement. Irvine (2012) strongly suggest that any type of recognition scheme should have supported rewards. Long and Shields (2010) recommend combining both non-cash and cash reward. Reward strengthens the reinforcement effect of recognition (Skinner, 1953). In addition, well-designed rewards can help a peer-to-peer recognition program achieve the criteria of contingency, formality, importance of recognizer and being valued by recipient (Nelson, 2016).
Conclusion
The increasing trend in adopting employee recognition demonstrates that in general, organizations have raised concerns about the application of the HR management theory. Under this circumstance, the findings of this research contributed to HR researchers and companies who intend to or in the process of implementing employee peer-to-peer recognition programs. This research argues if the core purpose of giving recognition, which is to motivate people to continue a good action or behavior, is successfully performed. Hence, if a recognition program fails to motivate employees, it needs to be investigated and intervened. Researchers and organizations should be aware of the challenges that might happen during the implementation of a peer-to-peer recognition program. Employee relationships and understandings, organizational climate, and reward system are the sources that can cause troubles toward the program success. To the wider business world, this research offers practical solutions to deal with those challenges, such as building the culture and assigning monitor, to assist implementers to effectively make use of a peer-to-peer program in motivating employees. | 2021-11-18T16:22:48.480Z | 2021-01-01T00:00:00.000 | {
"year": 2021,
"sha1": "7c6d3e5ad9225f15007edc665f1380d2c94b7141",
"oa_license": "CCBY",
"oa_url": "https://www.shs-conferences.org/articles/shsconf/pdf/2021/35/shsconf_icmesh2020_08007.pdf",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "fd6169ddc12b0fbb61850c5c3a6a255efd260ac0",
"s2fieldsofstudy": [
"Business",
"Computer Science"
],
"extfieldsofstudy": []
} |
34799134 | pes2o/s2orc | v3-fos-license | Non-traditional melanoma prevention strategies in the young adult and adolescent population
Malignant melanoma is the most common cancer among women and men aged 25–29 [1]. Unfortunately, the incidence of melanoma far surpasses the rate of any other malignancy in this demographic. Due to the growing trends of tanning and advanced nail art, the incidence of melanoma may increase in the coming years. Less than a quarter of patients examine their own skin on a regular basis [2]. An Austrian study concluded that patients deemed physicians as the fourth most reliable source for skin health information behind print media, television, and family members [3]. With this in mind, physicians must develop non-traditional melanoma prevention strategies to increase early detection.
In a society where the majority of the population uses the internet, non-traditional screening tools for melanoma recognition include implementation of an electronic health system designed for regular self-evaluation of skin and nevi surveillance. E-health tools may detect melanoma at an earlier stage, reducing overall morbidity and mortality from disease [4]. However, experts are concerned that inaccurate readings may lead patients to neglect a physician consult, which could thwart early diagnosis of treatable melanocytic lesions [4].
Other non-traditional strategies for early detection of melanoma in young women and men are also noteworthy. For example, hair stylists may aid in early detection of scalp lesions since young women and men frequently visit the same stylist. Training stylists to recognize the signs of skin cancer in the scalp may increase rates of early detection of head and neck lesions that would have otherwise been discovered in later stages due to their inconspicuous locations [5]. These locations represent 6% of all melanomas, but are responsible for 10% of all deaths from melanomas [5]. Similarly, massage therapists who see the same client on a regular basis may be able to report new neoplasms and changes in nevi on the body.
Nail artists represent another demographic that can be trained to recognize nail melanoma between appointments and alert their clients of pathologic nail changes. Only 66% of patients address melanocytic changes observed in nails [6]. Women will often use nail polish, various gel techniques, and wrapping procedures to conceal nails for prolonged periods of time. Training these professionals may be an effective population-based strategy to increase rates of early stage melanoma detection.
A survey of 206 hair professionals found that only 28.1% had received formal skin cancer education, but almost fifty percent of hair professionals were interested in a skin cancer education program [7]. About 37% looked at greater than 50% of their customers’ scalps, 29% looked at greater than 50% of their customers’ necks, and 15% looked at greater than 50% of their customers’ faces for concerning lesions during the preceding month.7 Hair professionals’ personal health practices corresponded with frequency of observation of customers’ lesions (P < .001) [7]. This study led to development of The Skinny on Skin, an educational program designed to aid in the early detection and prevention of melanoma by beauty industry professionals.
Counseling the young demographic can be difficult due to the pressure the media places on young women and men to enhance their appearance. With this in mind, appeals to the negative cosmetic impact of sun and indoor tanning may be more effective than health-based appeals [8]. It is important to explain to patients that intermittent periods of excessive sun exposure early in life may lead to an increase in fine lines, wrinkles, lentiginous proliferations, and rapidly evolving nevi that may develop into melanoma [9]. Although counseling against tanning bed use may be successful in some patients, it has been shown that even after learning the associated risks, patients continue to tan [10].
The use of tanning beds has been associated with addictive behavior and may contribute to a compulsive desire to tan [11]. The brain of a person who habitually tans exhibits activity similar to that of a substance abuser and can experience tolerance, dependence, and withdrawal [11]. The brain responds to ultraviolet radiation (UVR) and can differentiate UVR from non-UVR tanning beds.11 When counseling these patients, it may be beneficial to use screening tools including the CAGE questionnaire (Table 1) to determine the motivation and goals associated with excessive tanning [12]. Understanding these desires will improve the ability of a provider to suggest appropriate alternatives to tanning [12]. For example, individuals who tan for relaxation may substitute yoga as an appropriate alternative, while those who tan for aesthetic purposes may choose to use dihydroxyacetone, the active ingredient in sunless tanners, as an alternative. It is also pertinent to discuss tanning in the pediatric and teenage populations to fully understand familial beliefs about tanning. Children whose parents tan indoors are more likely to do the same compared to children whose parents do not tan; a population based survey found that indoor tanning was 30% in the twelve to eighteen year old age group when the caregiver personally tanned compared to only 10% when the caregiver did not tan [13].
TABLE 1
In conclusion, primary care providers and dermatologists should continue to educate women and men about the major modifiable risk factors for melanoma, including unprotected sun exposure and monitoring of existing nevi. Education of hair stylists, nail artists, and massage therapists may expedite a referral to health-care professionals. Through the institution of both traditional and non-traditional melanoma prevention strategies, patients and physicians may achieve greater awareness, early detection, and prevention of disease.
of melanoma far surpasses the rate of any other malignancy in this demographic. Due to the growing trends of tanning and advanced nail art, the incidence of melanoma may increase in the coming years. Less than a quarter of patients examine their own skin on a regular basis [2]. An Austrian study concluded that patients deemed physicians as the fourth most reliable source for skin health information behind print media, television, and family members [3]. With this in mind, physicians must develop non-traditional melanoma prevention strategies to increase early detection.
In a society where the majority of the population uses the internet, non-traditional screening tools for melanoma recognition include implementation of an electronic health system designed for regular self-evaluation of skin and nevi surveillance. E-health tools may detect melanoma at an earlier stage, reducing overall morbidity and mortality from disease [4]. However, experts are concerned that inaccurate readings may lead patients to neglect a physician consult, which could thwart early diagnosis of treatable melanocytic lesions [4].
Other non-traditional strategies for early detection of melanoma in young women and men are also noteworthy.
For example, hair stylists may aid in early detection of scalp lesions since young women and men frequently visit the same stylist. Training stylists to recognize the signs of skin cancer in the scalp may increase rates of early detection of head and neck lesions that would have otherwise been discovered in later stages due to their inconspicuous locations [5]. These locations represent 6% of all melanomas, but are responsible for 10% of all deaths from melanomas [5]. Similarly, massage therapists who see the same client on a regular basis may be able to report new neoplasms and changes in nevi on the body. Nail artists represent another demographic that can be trained to recognize nail melanoma between appointments and alert their clients of pathologic nail changes. Only 66% of patients address melanocytic changes observed in nails [6].
Women will often use nail polish, various gel techniques, and wrapping procedures to conceal nails for prolonged periods of time. Training these professionals may be an effective population-based strategy to increase rates of early stage melanoma detection.
A survey of 206 hair professionals found that only 28.1% had received formal skin cancer education, but almost fifty percent of hair professionals were interested in a skin cancer education program [7]. About 37% looked at greater than Non-traditional melanoma prevention strategies in the young adult and adolescent population 50% of their customers' scalps, 29% looked at greater than 50% of their customers' necks, and 15% looked at greater than 50% of their customers' faces for concerning lesions during the preceding month.7 Hair professionals' personal health practices corresponded with frequency of observation of customers' lesions (P < .001) [7]. This study led to development of The Skinny on Skin, an educational program designed to aid in the early detection and prevention of melanoma by beauty industry professionals.
Counseling the young demographic can be difficult due to the pressure the media places on young women and men to enhance their appearance. With this in mind, appeals to the negative cosmetic impact of sun and indoor tanning may be more effective than health-based appeals [8]. It is important to explain to patients that intermittent periods of excessive sun exposure early in life may lead to an increase in fine lines, wrinkles, lentiginous proliferations, and rapidly evolving nevi that may develop into melanoma [9]. Although counseling against tanning bed use may be successful in some patients, it has been shown that even after learning the associated risks, patients continue to tan [10].
The use of tanning beds has been associated with addictive behavior and may contribute to a compulsive desire to tan [11]. The brain of a person who habitually tans exhibits activity similar to that of a substance abuser and can experience tolerance, dependence, and withdrawal [11]. The brain responds to ultraviolet radiation (UVR) and can differentiate UVR from non-UVR tanning beds.11 When counseling these patients, it may be beneficial to use screening tools including the CAGE questionnaire (Table 1) to determine the motivation and goals associated with excessive tanning [12].
Understanding these desires will improve the ability of a provider to suggest appropriate alternatives to tanning [12]. For example, individuals who tan for relaxation may substitute yoga as an appropriate alternative, while those who tan for aesthetic purposes may choose to use dihydroxyacetone, the active ingredient in sunless tanners, as an alternative. It is also pertinent to discuss tanning in the pediatric and teenage populations to fully understand familial beliefs about tanning. Children whose parents tan indoors are more likely to do the same compared to children whose parents do not tan; a population based survey found that indoor tanning was 30% in the twelve to eighteen year old age group when the caregiver personally tanned compared to only 10% when the caregiver did not tan [13].
In • Have you ever felt the need to Cut down on your tanning?
• Have you ever felt Annoyed by criticism of your tanning?
• Have you ever felt Guilty about your tanning?
• Have you ever felt the need to tan Every morning?
of both traditional and non-traditional melanoma prevention strategies, patients and physicians may achieve greater awareness, early detection, and prevention of disease. | 2017-06-17T15:31:24.524Z | 2014-10-01T00:00:00.000 | {
"year": 2014,
"sha1": "55bb6094fc7793b2ce5ec8cfcbc07a5381c5124e",
"oa_license": "implied-oa",
"oa_url": "https://europepmc.org/articles/pmc4230265?pdf=render",
"oa_status": "GREEN",
"pdf_src": "PubMedCentral",
"pdf_hash": "55bb6094fc7793b2ce5ec8cfcbc07a5381c5124e",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
89444077 | pes2o/s2orc | v3-fos-license | Growth , Yield and Nutrient Uptake Capacity of Rice under Different Sulphur Levels
Article history: Received 24 March 2016 Accepted 23 July 2016 Available online, ISSN: 2148-127X Because of using sulphur free fertilizer, rice crop facing tremendous sulphur (S) deficiency throughout the world as well as in Bangladesh. However, farmers of Bangladesh often overlook the importance of sulphur fertilization on rice yield. Therefore, the present study was conducted to investigate the influence of different levels of sulphur on growth, yield attributes and uptake of N, P, K and S by grain and straw of BRRI dhan41 rice variety. The experiment was laid out in randomized complete block design with three replication and eight treatments including control. The result of the study revealed that the highest plant height, panicle length, filled grain was found in T1 (50% RFD of S) treatment, while the utmost number of effective tiller and straw yield was recorded in T5 (150% RFD of S) treatment. Interestingly, maximum grain yield as well as uptake of N, P, K and S by grain and straw was significantly higher in T6 (175% RFD of S) treatment. The highest grain yield of T6 treatment might be due to maximum nutrient use efficiency. However, in all cases control treatment (T0) gave the worst result. The present study clearly indicates that higher rates (175% of the recommended dose) of S along with other fertilizers may be recommended for better growth and yield of BRRI dhan41 in Bangladesh soil.
Introduction
Rice is the most important crop within the sector contributes for about 92% of the total food grains produced in the country.In respect of area and production of rice, Bangladesh ranks fourth following China, India and Indonesia (FAO, 2013).Currently, rice occupies 11.42 million hectares of land and overall production of rice is about 34.23 million tons (BBS, 2013).Nevertheless, the population of Bangladesh is growing by two million every year and may increase by another 30 million over the next 20 years.Thus, Bangladesh will require about 27.26 million tons of rice for the year 2020 (BRRI, 2011).During this time, total area under rice will also shrink to 10.28 million hectares.Under this circumstance, horizontal expansion of rice area is not possible (Abdullah et al., 2015).Hence, only one avenue left is to increase production of rice by vertical means through introduction of high yielding varieties and appropriate management practices (Islam et al., 2015;Sohel et al., 2016).Among the management practices, judicious use of sulphur fertilizer is the most important one.
Sulphur is an essential macronutrient nutrient for plants ranked 4 th after nitrogen, phosphorus and potassium because of its indispensable role in proteins synthesis, vitamins, enzyme and flavoured compounds in plant (Bera and Ghosh, 2015;Islam et al., 2016).About 90% of plant sulphur present in amino acid (methionine and cysteine) and a variety of metabolites (thiamine, thiamine, pyrophosphate, glucosinolates, glutathione and phytochelatins), which play a pivotal role in building blocks of protein, formation of chlorophyll, activation of enzymes etc. (Tewari et al., 2010;Hoefgen and Nikiforova, 2008).Furthermore, deficient supply of S in soil causing lower uptake of nitrate hence retards the activity of nitrate reductase as well as N metabolism in plants (Prosser et al., 2001;Abdallah et al., 2010).In Bangladesh, sulphur deficit soils ranked third among problem soils, which is prominent in light-textured soils.About 3.95 million hectares of land are sulphur deficient, which represents approximately 16 % of the total problem soils in Bangladesh (Huq and Shoaib, 2013).This area gradually increasing due to intensive agriculture practices, following multiple cropping systems in association with using of sulphur free fertilizer and no use of organic manures.Unfortunately, most of the farmers of our country often overlook to maintain an optimum S level in their field; even sometimes, they are confused with S deficiency symptoms with P or N deficiencies or Al toxicity.It is worthy to mention that the critical level of sulphur for Bangladesh soil has been determined as 10 µg g -1 soil.To obtain higher yield, farmers currently use excess amount of gypsum, ammonium sulphate, zinc sulphate, etc. as sulphur fertilizers to the soils to replenish sulphur deficiency but it is not a good practice for the soils as well as environments in the long run.However, farmers of Bangladesh transplant rice seedlings without maintaining the optimum level of S. Therefore, the present study was undertaken to study the effect of different levels of S on nutrient uptake by BRRI dhan 41 and to find out the optimum level of sulphur for profitable rice production.
Experimental Site, Design and Management Practices
The experiment was conducted at the Soil Science Field Laboratory of Bangladesh Agricultural University, Mymensingh, during July to December 2007.Thirty days old seedlings of BRRI dhan41, a high yielding variety of rice, was transplanted in experimental field with the spacing of 20 cm x 20 cm.The seedlings were collected from a rice seedbed.The experiments were comprised of eight treatments including control (Table 1).The experiment was laid out in a randomized complete block design (RCBD) with three replications.Each replication was sub-divided into eight unit plots.The treatments were randomly distributed to the unit plots therefore the total numbers of plots were 24.The unit plot size was 4m x 2.5m.The distance between plot-to-plot and replication to replication were 0.25 and 0.50 m, respectively.Gap filling was done with the even aged seedling within two weeks after transplanting.The full doses of TSP, MOP, gypsum and zinc oxide were applied according to treatments as a source of P, K, S and Zn as kg ha -1 before transplanting (Table 1).Urea was applied as a source of N (kg ha -1 ) in three equal splits as per treatments.The first split of urea was applied after 12 days of transplanting, the second split was applied as top dressing after 30 days of transplanting and third split was applied after 58 days of transplanting.Irrigation, weeding and other cultural management practices were done whenever necessary.
Data Collection
Soil properties: Initial soil sample was collected before land preparation from the plough depth layer (0-15cm).The samples were taken by means of an auger from different locations covering the whole experimental plot and mixed thoroughly to make a composite sample.The composite samples were then air dried, pulverized, sieved and used for chemical analysis to determine the soil health in terms of different physical (percentage of sand, silt, clay, and textural class) and chemical properties (soil pH, soil organic carbon, total nitrogen, available phosphorous, exchangeable potassium, CEC and available sulfur) (Table 2).The soil analysis for the above parameters was done by using the following methods: Physical properties: Among physical properties, percentage of sand, silt, clay, and textural class was determined by Hydrometer method as described by Bouyoucos, 1962.Chemical properties: Soil pH was measured by Glass Electrode pH meter method with soil-water ratio being maintained at 1: 2.5 (McLean, 1982).
Organic matter: At first organic carbon (%) in soil sample was determined by wet oxidation method of Walkley and Black (1934) and then it was converted into organic matter (%) multiplying by 1.72 (Edwards et al., 1999) conversion factor.
Total nitrogen: Total nitrogen content of soil was determined by Kjeldahl Method following concentrated sulphuric acid digestion and steam distillation with 40% NaOH.The ammonia evolved was collected in a beaker containing 4% boric acid with an indicator composed of 95% alcohol, methyl red and bromocresol green and was titrated against 0.02 N H 2 SO 4 as described by Jackson (1973).
Exchangeable potassium: Exchangeable potassium was extracted with neutral 1N NH 4 OAc (pH 7) as described by Jackson (1973) and was measured by Atomic Absorption Spectrophotometer (Model No. 170-30, HITACHI, Japan) and readings were calibrated with a standard K curve (Page et al., 1982).
Available phosphorous: Extraction for available phosphorous in the soil samples was made with 0.5 M NaHCO 3 solution at a nearly constant pH of 8.5 following the method described by Olsen et al. (1954).Spectrophotometer (660 nm wave length) was used to measure the intensity of the colour developed by ascorbic acid method as outlined by John (1970) and readings were calibrated with a standard P curve (Page et al., 1982).
Available sulfur (ppm): Available sulfur was determined turbidimetrically according to Bardsley and Lancaster (1965).
Plant Samples
Plant samples for instance grain and straw were collected from the field and dried in an oven at about 65 0 C for 48 hours.Then the samples grinded properly and the grinded materials were stored in small paper bags and placed in desiccators.The grain and straw samples were the analyzed for the determination of N, P, K and S by using following methods: Total nitrogen (%): Total nitrogen contents were determined by modified Kjeldajl digestion (colorimetric) method (Cataldo et. al., 1974).
Total Phosphorus (P), potassium (K), Sulphur (S) content in plant (%): Oven-dried plant materials were digested with concentrated HNO 3 and HCLO 4 mixture as described by Piper (1966) for determination of P, K, S content both in grain and straw samples.
After chemical analysis of straw and grain samples, the nutrient uptake was calculated from the nutrient content and yield of rice crop by using the following formula as described by Sharma et al. (2012).
Growth and Yield Contributing Characters
Ten plants were randomly selected from each unit plot for collection of data on the following parameters viz.plant height (cm), panicle length (cm), filled grains per panicle, grain yield (t ha -1 ), straw yield (kg ha -1 ) and biological yield (t ha -1 ).
Statistical Analysis
Finally, the data was analyzed by using statistical package program (MSTAT-C).The analysis of variance for crop characters and for the nutrient content and nutrient uptake by the grain and straw were done following by the Duncan's Multiples Range Test (DMRT) (Gomez and Gomez, 1984).
The Growth and Yield Contributing Characters
The plant height of BRRI dhan41 was not significantly affected by the application of different levels of sulphur.The effect of different levels of S on plant height was inconsistent (Table 3).The highest value (120.5 cm) was recorded in the treatment T 1 (50% RFD of S), the second highest value (118.8 cm) was recorded in the treatment T 3 (100% RFD of S), while the lowest plant height (113.8 cm) was found in the control treatment (T 0 ).All treatments produced statistically similar plant height.The results also indicated that addition of S applied insignificant effect on plant height.There was a significant effect due to the application of different levels of S on the production of effective tillers per hill of BRRI dhan41 (Table 3).The number of effective tillers per hill due to different treatments varied from 7.87 to 9.33.The highest number of effective tillers per hill (9.33) was found in the treatment T 5 (150% RFD of S) which was statistically similar to those recorded in the treatments T 7 (200% RFD of S) and T 3 (100% RFD of S) with the values of 9.20 and 9.07.The treatment T 0 (control) produced the lowest number of effective tillers per hill (7.87).It indicated that S application at higher rates over the recommended dose has significant effect on the number of effective tillers per hill.An increase in the number of tillers per hill due to application of S was also reported by several workers (Patra et al., 1998 andChandel et al., 2002).Panicle length of BRRI dhan41 was not significantly influenced by different treatments (Table 3).The length of panicle varied from 22.83 cm to 24.20 cm.The highest panicle length (24.20 cm) was observed in T 1 (50% RFD of S) treatment.The lowest panicle length of 22.83 cm was recorded in the treatment T 0 (control).All treatments produced statistically similar panicle length.Apostol (1989) observed that organic and in organic fertilizers increased the panicle length of rice.Similar results were also found by Ahmed and Rahaman (1991).Results presented in the Table 3 showed a significant effect of different S levels on the number of filled grains per panicle of BRRI dhan41.The number of filled grains per panicle due to different treatments ranged from 98.00 to 135.6.The highest number of filled grains per panicle (135.6) was obtained from the treatment T 1 (50% RFD of S).The lowest number (98.00) of filled grains per panicle was obtained from the treatment T 0 (control).The significant effects of S application on the number of grains per panicle were also reported by Hque and Chowdhury (2004).The positive responses of sulphur application on the number of grains per panicle were also reported by Uddin et al. (1997).There was significant effect on 1000 grain weight of BRRI dhan41 by the application of different levels of sulphur fertilizer.The 1000 grain weight varied from 27.16 to 29.30 gm (Table 3).The maximum 1000 grain weight (29.30g) was found in treatment T 3 (100% RFD of S) and minimum (27.16 g) value was obtained in the treatment T 0 (control).Choudhury and Bodiuzzaman (1992) reported that sulphur fertilizer increased the 1000 grain weight.
Grain and Straw Yield
Data on grain and straw yields of BRRI dhan41 as affected by different S application are presented in Table 3. Results in Table 3 showed that grain yield of BRRI dhan41 was significantly influenced due to different treatments.The grain yield due to various treatments ranged from 3.72 to 5.22 t/ha.The highest grain yield (5.22 t/ha) was obtained in the treatment T 6 (175% RFD of S).The lowest grain yield (3.72 t/ha) was obtained in the treatment T 0 (control) which was statistically different from all other treatments.The increase in grain yield over control ranged from 16.40 to 28.74% (Table 3).The highest percentage (28.74%) of increased grain yield over control was recorded in the treatment T 6 (175% RFD of S).The lowest percentage (16.40%) of increased grain yield over control was recorded in the treatment T 2 (75% RFD of S).Deng et al. (1992) observed 23.9% yield increase of rice due to application of S as gypsum.Bhuiya and Portch (1992) also reported the superiority of gypsum over SSP as S source.Gupta and Soi (1992) reported that gypsum increased grain yield of rice by 28% over control.The present findings are in agreement with the results of the previous study.Tandon (1995), Gupta et al. (1996), Oliveira et al. (2003) who reported increased yield of rice due to sulphur application irrespective of S source and rates.Similar results were also reported by Islam et al. (1996) and Raju and Reddy (2001).Results presented in the Table 3 show that straw yield of BRRI dhan41 was significantly influenced by different treatments under study.The straw yield obtained from different treatments ranged between 4.69 to 6.67 t/ha.All the treatments gave higher straw yield over control.It was observed that the treatment T 4 (125% RFD of S) gave the highest straw yield (6.67 t/ha) which was statistically similar to those recorded in the treatment T 7 (200% RFD of S) and T 5 (150% RFD of S) with value of 6.63 and 6.60.The lowest straw yield (4.69 t/ha) was recorded in the treatment T 0 (control).The percent increase of straw yield over control ranged from 18.69 to 29.69% and the highest and lowest value was recorded in the treatment T 4 (125% RFD of S) and T 1 (50% RFD of S) respectively.The straw yield due to different treatments ranked in the order of T 4 > T 7 > T 5 > T 6 > T 2 > T 3 > T 1 >T 0 .The results revealed that the application of higher dose of sulphur exerted pronounced effect in producing higher straw yields of rice.It also indicates that application of S fertilizer at rates higher than the recommended dose significantly increased the straw yield of BRRI dhan41.Hoque and Khan (1981), Idris and Jahiruddin (1983), Tapatkar and Sonar (1995) reported that S application significantly increased the straw yield of rice.Table 3 shows that biological yield of BRRI dhan41 responded significantly.The biological yield varied widely among the treatments, ranging from 8.41 to11.75 t/ha.The highest biological yield (11.75 t/ha) was obtained in the treatment T 6 (175% RFD of S).The lowest biological yield (8.41 t/ha) was obtained in the treatment T 0 (control).The increase in biological yield over control ranged from 22.56 to 28.43% (Table 3).The highest percentage (28.43%) of increased biological yield over control was recorded in the treatment T 6 (175% RFD of S) and the lowest percentage (22.56%) of increased biological yield over control was recorded in the treatment T 1 (50% RFD of S).The biological yield obtained from different treatments ranked in order of T 6 > T 7 > T 4 > T 5 > T 1 > T 3 > T 2 >T 0 .The results revealed that the application of higher dose of S on the biological yield of rice was more pronounced than that of other treatment under study.Azim et al. (1999) reported a significant increase in biological yield of rice due to application of fertilizer.
Nutrient Content and Uptake by Grain and Straw
The grain and straw samples of BRRI dhan41 were analyzed for the determination of N, P, K and S contents and the nutrient uptake was calculated from the yield and nutrient concentration of grain and straw.Data presented in Table 4 indicate that the various treatments had significant influence on N content in grain and straw of BRRI dhan41.Nitrogen nutrition and metabolism in plants is intimately associated with the sulphur status and their metabolic process (Abdallah et al., 2010).However, the N content in grain ranged from 1.16% to 1.25%.The highest N content (1.25%) in grain was found in treatment T 7 (200% RFD of S).The lowest N content (1.16%) in grain was observed in treatment T 0 (control).In the case of straw, the N content ranged from 0.90% to 1.03% (Table 4).The highest N content (1.03%) was recorded in treatment T 5 (150% RFD of S) which was statistically identical with the treatments T 7 (200% RFD of S), T 6 (175% RFD of S), T 4 (125% RFD of S), T 2 (75% RFD of S) and T 1 (50% RFD of S) with values of 0.98%, 0.95%, 0.93%, 0.93% and 0.95% respectively.As expected the lowest N concentration (0.90%) in straw was observed in the treatment T 0 (control).The higher amount of N in grain in compare to straw might be attributable to translocation of the N occur from vegetative part to the reproductive organ or senescence of leaves.Corroborate findings also found by Sachdev et al. (1983) on rice.Results in Table 4 show a wide variation in N uptake by grain and straw due to different treatments.The nitrogen uptake by rice grain ranged from 43.22 to 63.32 kg/ha.The highest amount of N uptake (63.32 kg/ha) by grain was observed in the treatment T 7 (200% RFD of S) which was statistically different from other treatments.The lowest amount of N uptake (43.22 kg/ ha) by grain was observed in treatment T 0 (control).In case of straw, the nitrogen uptake ranged from 41.34 kg/ha to 68.41 kg/ha.The highest nitrogen uptake by rice straw was recorded in T 7 (200% RFD of S) which was statistically similar with the treatment T 5 (150% RFD of S) with the value of 68.10 kg/ha.The lowest amount of N uptake (41.34 kg/ha) by straw was observed in treatment T 0 (control).The total nitrogen uptake by grain and straw of BRRI dhan41 ranged from 84.56 kg/ha to 131.7 kg/ha.The highest amount of total N uptake (131.7 kg/ha) was observed in treatment T 7 (200% RFD of S) which was statistically identical with the treatments T 6 (175% RFD of S), with the value of 126.7 kg/ha.The lowest amount of total N uptake (84.56 kg/ha) was recorded in treatment T 0 (control).The total nitrogen uptake by BRRI dhan41 due to different treatments ranked in order of T 7 > T 6 > T 5 > T 4 > T 1 > T 2 > T 3 > T 0 .The higher N uptake might be associated with exploration of greater soil volume by root exacerbated by sulphur application (Mandal et al., 2003).More precisely we can say that root represent a portion of aerial biomass, which greatly heighten in response to S fertilization.The concomitant increase of root biomass allowing capturing of more N from the soil, and thus increasing N uptake by the plant (Salvagiotti et al., 2009).Furthermore, synergism between N and S also ameliorate N uptake (Shamima and Imamul Hug, 2002;Salvagiotti et al., 2009).The result is in consistent with the result of Singh and Singh (2002), Sharma and Gangwar (1997) who noted that N uptake in rice plant increased with increasing rates of S application.
Data in Table 4 showed that the P content in grain and straw of rice (var.BRRI dhan41) varied significantly due to the application of different doses of sulphur fertilizer.Phosphorus content in grain varied from 0.22% to 0.30%.The highest amount of P concentration (0.30%) was found in the treatments T 6 (175% RFD of S) and T 1 (50% RFD of S), which was statistically identical with the treatments T 5 (150% RFD of S), T 4 (125% RFD of S), T 3 (100% RFD of S), T 2 (75% RFD of S) and T 0 (control), with the values of 0.28%, 0.28%, 0.28%,0.25%and 0.24%, respectively.The lowest amount of P concentration (0.22%) in grain was found in the treatment T 7 (200% RFD of S).In case of straw, there was a significant effect of various treatments on P content in straw (Table 4).The P content in straw ranged from 0.08 to 0.14%.The highest amount of P concentration (0.14%) in straw was observed in the treatment T 1 (50% RFD of S) which was statistically different from other treatments.The lowest amount of P concentration (0.08%) was recorded in the treatment T 2 (75% RFD of S).The result indicated that P content in grain increased by the application of S. It also indicated that the concentration of phosphorus in grain is always higher than that of straw.Padihar et al. (1985) reported that phosphorus content in grain was increased with increasing level of S application.Phosphorus uptake by grain and straw was significantly influenced due to application of different levels of sulphur (Table 4).Phosphorus uptake by the grain of BRRI dhan41 by grain ranged from 9.20 to 15.48 kg/ha.The highest amount of P uptake (15.48 kg/ha) by grain was recorded in the treatment T 6 (175% RFD of S).The lowest amount of P uptake (9.20 kg/ha) by grain was observed in the treatment T 0 (control).The P uptake by straw of BRRI dhan41 ranged from 5.02 to 8.33 kg/ha (Table 4).The highest amount of P uptake (8.33 kg/ha) by straw was found in the treatment T 1 (50% RFD of S) which was statistically identical with the treatment T 5 (150% RFD of S) with the value of 7.70.The lowest amount of P uptake (5.02 kg/ha) by straw was observed in the treatment T 0 (control).The total phosphorus uptake by grain and straw varied from 14.22 to 22.47 kg/ha due to different treatments.The highest amount of total P uptake (22.47 kg/ha) was found in treatment T 1 (50% RFD of S) which was statistically similar with the treatment T 6 (175% RFD of S) with the value of 22.45.The lowest amount of total P uptake (14.22 kg/ha) was obtained in T 0 (control).The total P uptake due to different treatments ranked in the order of T 1 > T 6 > T 5 > T 4 > T 3 >T 7 > T 2 > T 0 .
Potassium content in grain and straw of rice was significantly influenced due to different levels of S application (Choudhury and Badiuzzaman, 1992).The potassium content in grain varied from 0.206 to 0.240% (Table 5).The highest amount of K concentration (0.240%) in grain was observed in T 1 (50% RFD of S) which was statistically different from all other treatments.The lowest amount of K concentration (0.206%) in grain was found in the T 2 (75% RFD of S).On the other hand, K concentration in straw ranged from 2.10 to 2.83%.The highest amount of K concentration (2.83%) in straw was observed in T 7 (200% RFD of S) which was statistically identical with the treatments T 6 (175% RFD of S), T 4 (125% RFD of S), T 3 (100% RFD of S), T 2 (75% RFD of S) and T 1 (50% RFD of S) with the values of 2.68%, 2.63%, 2.63%, 2.73%, and 2.73%, respectively.The lowest amount of K concentration (2.10%) in straw was found in T 5 (150% RFD of S).The results indicated that the K uptake by grain and straw of rice (var.BRRI dhan41) was also significantly influenced by the different treatments (Table 5).Dobermann and Fairhust (2000) also found the same results with the increasing of S level K content and K concentration was increased in rice grain and straw.The K uptake by rice grain varied from 7.82 to 11.48 kg/ha.The highest amount of K uptake (11.48 kg/ha) by grain was recorded in the treatment T 6 (175% RFD of S) which was statistically similar with the treatment T 1 (50% RFD of S) with the value of 11.42 kg/ha.The lowest amount of K uptake (7.82 kg/ha) by grain was observed in T 0 (control).Potassium uptake by straw ranged from 108.1 to 187.7 kg/ha.The highest amount of K uptake (187.7 kg/ha) by straw was observed in treatment T 7 (200% RFD of S) which was statistically similar with the treatments T 6 (175% RFD of S), T 4 (125% RFD of S), T 3 (100% RFD of S), T 2 (75% RFD of S) and T 1 (50% RFD of S) with the values of 174.8, 175.3, 165.2, 172.7 and 166.6 respectively.On the other hand the lowest amount of K uptake (108.1 kg/ha) by straw was obtained in the treatment T 0 (control).The results also revealed that K uptake by straw was higher than that by grain.In general, potassium uptake by leaf and petiole increased up to reproductive stage and decreased at maturity level, which implies that translocation of potassium to developed grain (Shamima and Imamul Hug, 2002).Uptake of potassium by stem, on the other hand is associated with increased of stem dry matter.It is worthy to mention that the stem served as a largest reservoir for potassium throughout the growth cycle.Nevertheless, we found contrasting result as higher potassium in straw and lower in grain.This result what we found might be due to the uptake potassium by the above ground part is mostly translocate to root instead of grain for their profuse growth to facilitate higher N uptake.This is phenomenon is termed as luxury consumption or overconsumption (Saleque et al., 1998).Kabir et al. (2011) support the findings of our study.Total potassium uptake by grain and straw varied from 115.9 kg/ha to 198.4 kg/ha.The highest amount of total K uptake (198.4 kg/ha) was found in treatment T 7 (200% RFD of S) which was statistically similar with the treatments T 6 (175% RFD of S), T 4 (125% RFD of S), T 2 (75% RFD of S) and T 1 (50% RFD of S) with the values of 186.2, 184.8, 181.8 and 178.0 respectively.The lowest amount of total K uptake (115.9 kg/ha) was observed in T 0 (control).The total K uptake due to different treatments can be ranked in the order of T 7 >T 6 >T 4 >T 2 >T 1 >T 3 >T 5 > T 0 .
Sulphur content in grain and straw of rice was significantly influenced due to different levels of S application.The S concentration in grain ranged from 0.122 to 0.175% (Table 5).The critical level of S in rice straw and grain is 0.5% (Mukit, 2009).The highest amount of S concentration (0.175%) in grain was obtained in the treatment T 1 (50% RFD of S) which was statistically different from all other treatments.The lowest amount of S concentration (0.122%) in grain was found in the T 4 (125% RFD of S).On the other hand, S content in straw varied from 0.122 to 0.178% (Table 4.6).The highest S concentration (0.178%) was observed in treatment T 7 (200% RFD of S) which was statistically different from all other treatments.The lowest amount of S concentration in straw (0.122%) was found in T 4 (125% RFD of S).The results indicated that S content in grain and straw increased by the application of higher rates of S. It also indicated that S content in grain is greater than that of straw.Chandel et al. (2002) observed concentration of S in rice plant increased with increasing rates of S application.Result in Table 5 indicated that the S uptake by grain and straw of rice was significantly influenced by the different treatments.Sulphur uptake by grain varied from 5.36 to 8.34 kg/ha.The highest amount of S uptake (8.34 kg/ha) by grain was observed in the treatment T 1 (50% RFD of S) which was statistically identical with the treatments T 6 (175% RFD of S) and T 5 (150% RFD of S) with the values of 7.34 and 7.56 respectively.The lowest S uptake (5.35 kg/ha) by grain was observed in T 0 (control) treatment.Sulphur uptake by straw varied from 7.42 to 11.81 kg/ha.The highest amount of S uptake (11.81 kg/ha) by straw was recorded in treatment T 7 (200% RFD of S) which was statistically identical with the treatment T 6 (175% RFD of S) with the value 10.60.The lowest amount of S uptake (7.42 kg/ha) by straw was obtained in T 0 (control).The total S uptake by grain and straw ranged from 12.78 to 18.57 kg/ha.The highest amount of total S uptake (18.57kg/ha) was observed in treatment T 7 (200% RFD of S) which was statistically similar with the treatments T 6 (175% RFD of S) and T 5 (150% RFD of S) with the value of 17.94 and 17.85, respectively.The lowest total S uptake (12.78 kg/ha) was observed in T 0 (control).The total S uptake due to different treatments ranked the in order of T 7 > T 6 > T 5 > T 1 > T 3 > T 2 > T 4 > T 0 .Sakal (1995) and Poongothi et al. (1999) reported that application of S increased S uptake by rice plant.
Conclusions
Deficiency of sulphur has become a great concern over the past decades in most of the agricultural areas of Bangladesh.Therefore, to replenish the nutrient deficiency, farmers use gypsum as a source of sulphur without knowing their perilous consequences to the soil.Optimum doses of sulphur along with its interaction effect with other nutrients is crucially needed for successful crop production.The results obtained from the present study demonstrated that different levels of sulphur fertilizer significantly influenced the growth and yield attributes of BRRI dhan41 rice variety.Among the growth characteristics, plant height and panicle length were not significantly influenced by S fertilization likewise number of effective tillers.On the other hand, yield attributes viz.number of filled grains per panicle, 1000-grain weight, biological yield and grain yield were considerably varied by different gradient of S fertilizers.Regarding nutrient analysis, in all cases, the highest amount of nutrients found in grain instead of straw except for potassium.Furthermore, nutrient uptake by grain and straw differed in different treatment.However, based on the above fact, in spite of different interaction effect with other nutrients, T 6 (175% RFD of S) treatments gave the highest yield might be due to the highest nutrient use efficiency.Therefore, further study with given priority to improvement of nutrient use efficiency with nutrient uptake should be conducted to boost yield of the new variety.
Table 1
Treatments and fertilizer doses used in the study
Table 4
Effect of different levels of S on the Nitrogen and Phosphorus of BRRI dhan41
Table 5
Effect of different levels of S on the Potassium and Sulphur of BRRI dhan41 Different letters within the column indicate statistically significant differences between treatments according to Duncan's Multiples Range Test (DMRT) (P<0.05).Letters were not shown in columns, which were statistically non-significant.Values are means ± SE of three independent replications (n = 3). | 2019-01-02T04:08:24.323Z | 2016-07-15T00:00:00.000 | {
"year": 2016,
"sha1": "9fd08733ba25018d71b9827360dc19d4ba34a3e5",
"oa_license": "CCBYNC",
"oa_url": "http://www.agrifoodscience.com/index.php/TURJAF/article/download/709/336",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "9fd08733ba25018d71b9827360dc19d4ba34a3e5",
"s2fieldsofstudy": [
"Agricultural And Food Sciences"
],
"extfieldsofstudy": [
"Biology"
]
} |
7070180 | pes2o/s2orc | v3-fos-license | Longitudinal change of selected human milk oligosaccharides and association to infants’ growth, an observatory, single center, longitudinal cohort study
Background Human milk is the recommended and sole nutrient source for newborns. One of the largest components of human milk is oligosaccharides (HMOs) with major constituents determined by the mother genotype for the fucosyltransferase 2 (FUT2, secretor) gene. HMO variation has been related with infant microbiota establishment, diarrhea incidence, morbidity and mortality, IgE associated eczema and body composition. Objectives We investigated the (i) dependence of several major representative HMOs on the FUT2 status assessed through breast milk 2’Fucosyllactose (2’FL) and (ii) the relation of the 2’FL status with infant growth up to 4 months of life. Design From an open observatory, single center, longitudinal cohort study with quantitative human milk collection at 30, 60, and 120 days postpartum from 50 mothers, who gave birth to 25 female and 25 male singleton infants, we collected a representative sample of human milk. We quantified the following 5 representative HMOs: 2’FL, Lacto-N-tetraose (LNT), Lacto-N-neotetraose (LNnT), 3’Sialyllactose (3’SL) and 6’Sialyllactose (6’SL). We grouped the milk samples and corresponding infants according to the measured milk 2’FL concentrations at 30 days of lactation, which clustered around low concentrations (95% CI of mean 12–42 mg/L) and high concentrations (95% CI of mean 1880–2460 mg/L) with the former likely representing Secretor negative mothers. Infant anthropometric measures were recorded at birth, 1, 2 and 4 months of age. Relations among the quantified HMOs and the relation of the high and low 2’FL HMOs groups with infant growth parameters were investigated via linear mixed models. Results The milk samples with low 2’FL concentration had higher LNT and lower LNnT concentrations compared to the samples with high 2’FL. The milk 3’- and 6’SL concentrations were independent of 2’FL. Over lactation time we observed a drop in the concentration of 2’FL, LNT, LNnT and 6’SL, especially from 1 to 2 months, while 3’SL remained at relatively constant concentration from 1 month onwards. Up to 4 months of age, we did not observe significant differences in body weight, body length, body mass index and head circumference of the infants who consumed breast milk with low or high FUT2 associated HMO concentrations and composition. Conclusions Our findings on HMO concentrations over time of lactation and clusters based on 2’FL concentrations confirm previous observations and suggest that LNnT and LNT are ‘co-regulated’ with the FUT2 dependent 2’FL concentration, with LNnT showing a positive and LNT a negative relation. Further, our findings also suggest that the relatively substantial variation in HMOs between the high and low 2’FL clusters do not impact infant growth of either sex up to 4 months of age. The study was registered in www.ClinicalTrial.gov (NCT01805011).
Introduction
Human milk is the sole nutrient source for newborn infants, adapted by evolution likely to provide both nutrition and protection [1]. Yet, in maternal milk, some nutrients such as vitamins and fatty acids vary depending on the dietary intake of the lactating mother, while others are primarily under genetic control [2]. To the best of the available knowledge, the major human milk oligosaccharides (HMOs) are determined primarily by the mother's genotype.
HMOs represent one of the largest compound groups by mass after lactose and fat, with slightly higher concentrations than protein [3]. These generally non-digestible oligosaccharides are extensions of the milk sugar lactose and are brought about by the action of a series of glycosyltransferases such as those transferring N-acetyl-glucosamine, galactose, sialic acid or fucose. For the latter, fucosyltransferases FUT2 (secretor gene), FUT3 (Lewis gene) and probably FUT9 or others are implicated [4,5]. Of these, the former two are polymorphic with different alleles being responsible for the Secretor negative (FUT2-/-) and Lewis negative (FUT3-/-) glycosylation phenotypes not only in milk, but also on epithelial cell and mucosal surfaces.
The prevalence of functional FUT2 enzyme activity within the population varies strongly depending on geography [5]. In most geographic regions functional FUT2 alleles are predominant by over 70%, while in some regions of Asia and Africa non-functional FUT2 alleles were reported equally prevalent. Human microbial and viral interactions, in particular with pathogens but possibly also commensals or symbionts, are likely evolutionary forces that could have driven FUT2 dependent glycan loss and diversification in general [6]. Loss of functional FUT2 activity was associated with better resistance to some pathogens such as specific Norovirus genotypes and Helicobacter pylori, with the trade-off to have a higher risk for infections by other pathogens affecting respiratory, urinary tract or the gastro-intestinal systems.
Furthermore, non-functional FUT2 genotypes were also associated with higher risk for type 1 diabetes, Crohn's disease and neonatal sepsis (for review see [5]).
In women with a functional FUT2, 2'Fucosyllactose (2'FL) is the major milk oligosaccharide, followed by Lacto-N-fucopentaose I (LNFP I) and di-Fucosyllactose (diFL). These three HMOs depend on FUT2 and represent a good proxy to assess secretor status or the presence and absence of a functional FUT2 [7][8][9][10]. This manifests in measurable variation of specific human milk oligosaccharides and profiles [10,11]. Recently, the maternal milk FUT2 dependent oligosaccharide status was shown by association to affect the early establishment of a bifidobacteria dominated microbiota in breastfed infants [12,13] and possibly also the onset of IgE associated eczema in C-section born infants [14]. Further, other fucosyl-HMO might be related to allergic morbidity such as cow milk protein allergy [15]. The early bifidobacteria establishment and possibly also the breast milk FUT2 dependent oligosaccharides per se might lead to enhanced infant immune development and protection from infection [16][17][18]. Further, early life microbiota establishment was also shown to drive long-term morbidities such as asthma [19]. Recently, the concentration of a specific FUT2 dependent HMO, LNFP I, but not diFL or 2'FL, was reported to be associated with infant weight and body composition at 6 months of age [20].
Here, we set out to better understand the quantitative FUT2 dependent changes in representative HMOs over time of lactation and the impact of those changes for the early growth of the infants. To this end, we analyzed and quantified representative HMOs in relation to FUT2 dependent 2'FL from quantitative breast milk at 1, 2 and 4 months postpartum from a cohort of 50 mothers, who gave birth to 25 female and 25 male singletons. And, we explored if FUT2 dependent breast milk composition has an influence on growth, namely body weight, body length, head circumference and body mass index, in females and males up to 4 months of age.
Study design, participants and milk sampling
The study was conducted in Singapore at the National University of Singapore. The participants provided written informed consent to participate in the study after receiving explanation and having read and understood the purpose of the study. The informed consent forms were approved by the National Healthcare Group (NHG) Domain Specific Research Boards (DSRB). This ethics board is made up of a group of independent reviewers and the processes had received AAHRP (Association for the Accreditation of Human Research Protection Programs) certification. Participants who did not understand English were provided with Chinese translated informed consent form as approved by the DSRB. The ethics approval reference for this study is NHG DSRB (domain B) 2011/00376. The protocol for this exploratory cohort study with quantitative collection of breast milk was reviewed and approved by NHG DSRB. The study was registered in www.ClinicalTrial.gov (NCT01805011).
We previously reported on lipid composition, which was the primary objective measurement of the breast milk samples from this same study cohort together with the secondary objectives of measuring lactose and energy content [21].
The study flow chart is depicted in Fig 1. Fifty mother-infant pairs were enrolled into the study between the infant's birth and 1 month of age. Because of the exploratory nature of this trial, no formal sample size calculation was performed. Inclusion criteria were, birth at gestational age 37 to 42 weeks, mothers from 18 to 40 years old, mothers' pre-pregnancy BMI (body mass index) between 20 kg/meter square (kg/m 2) and 29 kg/m 2 , mothers willing to breastfeed at least up to 4 months. Exclusion criteria were, pre-eclampsia, gestational diabetes, arterial hypertension above 140/90 mm Hg, and smoking. Visits were planned in the morning between 8:00 and 10:00 am to allow mothers to collect first morning expression either at home before the visit or during the visit at the hospital. Recruitment was from July 2011 to January 2012. Ninety six mother-infant pairs were screened for eligibility and 46 were excluded as 6 had gestational diabetes, 9 delivered premature at less than 37 weeks, 16 had a BMI below 20 kg/m 2 , 1 was over 40 years old, 5 declined to participate due to long distance from home to hospital, 9 were not suitable as they were unlikely to be motivated to continue breastfeeding until 4 months. For the visit at 4 months one mother was not able to provide breastmilk and for one breastmilk sample from the 4 months visit, we could not detect several of the HMOs that we intended to analyze due to unknown technical problem with the milk sample preparation. A total of 49 mothers provided 3 milk samples and a total of 48 milk samples was analyzed for all 5 representative HMOs at the 3 time points. As representative for the FUT2 dependent HMOs, we measured 2'FL, as representative of the sialylated HMOs, we measured 3'SL and 6'SL and as representative of the HMO core structures that can be further decorated with sialic acid or fucose, we measured LNT and LNnT.
A customized secure electronic Case Report Form database was used for data capture. Mothers were instructed to maintain a stable usual diet throughout the study.
For mothers' demographic, dietary habit and health related data as well as data related to their infants' birth (mode of delivery and parity) and breastfeeding habits (ever mixed feeding) were collected at enrollment.
For the infants, birth date, gender and gestational age were recorded at enrollment and anthropometry namely weight (kg), height (cm) and head circumference (cm) were recorded at birth and after enrollment to this study at the 1, 2 and 4 months visits. BMI was calculated.
The mothers provided breast milk samples (approximately 30 mL; at 30, 60, and 120 days postpartum). Samples were collected after full expression from one breast using a milk pump (Symphony Breastpump, Medela), while the baby was fed on the other breast to produce a satisfactory let-down. We made all efforts to collect complete feed that included fore-, mid-, and hind-milk as a representation of one feed and to avoid within feed variation of lipid and other nutrient contents. Approximately 30 mL aliquot was separated into two conical 15 mL polypropylene tubes for analysis and the rest was returned to the mother to feed the infant. Samples collected for research were stored at −80˚C and shipped on dry ice for analyses to the Nestlé Research Center, Lausanne, Switzerland.
From a representative sample of quantitatively expressed breast milk we used the 2'FL concentrations measured in 30 day postpartum milk samples to group the mother infant pairs into those with low (considered Secretor negative) and high 2'FL concentrations.
HMO analysis
For analysis by liquid chromatography, milk samples were defrosted, mixed and centrifuged for 20 min at 1700 × g. An aliquot of the skimmed milk was diluted 10 times with MilliQ water. All samples were analyzed thereafter by high performance anion exchange chromatography (HPAEC) coupled to a pulsed amperometry detector (ICS3000, Thermo Fischer Dionex, Sunnyvale, USA) in duplicate. Some samples were further diluted to allow for quantification. Briefly, samples were loaded onto a CarboPac PA1 analytical column (4x250mm; Thermo Fischer Dionex) without guard column at 25˚C and a flow rate of 1 mL/min. Milk oligosaccharide separation and elution was achieved using the following gradient: For 18 min from 50mM NaOH to 135mM NaOH followed by a 7 min isocratic elution with 135mM NaOH and 7.5mM sodium acetate followed for the next 10 min with 135mM NaOH and a gradient of sodium acetate from 7.5mM to 12.5mM followed by a 10 min isocratic elution with 125mM NaOH and a gradient from 12.5 to 100mM sodium acetate followed by a 10 min wash step with 150mM NaOH and 250mM sodium acetate and a 10 min equilibration step at 50mM NaOH. For identification and quantification standard curves with authentic external standards were employed at concentrations of 1.25, 2.5, 5, 10, 30, 70 and 100 mg/l. Oligosaccharide standards were purchased from Dextra (UK). Eluents were prepared from 50% NaOH (J.T.Baker, UK) and sodium acetate (Fluka, CH). To control the HPAEC and to quantify the oligosaccharides the Chromeleon software package 6.8 was employed.
Statistical analysis
Descriptive statistics and statistical analyses were done using an open source statistical software R version 3.0.1 (The R Foundation for Statistical Computing, Vienna, Austria). A linear mixed model was used to assess the impact of 2'FL status (Low or High 2'FL) on the concentration of the other HMOs. A logarithmic transformation was performed on the HMO concentrations. 2'FL status, lactation stage, their interaction, mode of delivery and gender were declared as fixed effects and within subject variability was taken into account by declaring the subject as a random effect. No Imputation methods were used for missing values as there were very few such cases. Inclusion of parity, ethnicity, mothers' BMI and age in the model were considered as they may have confounding effects, but these factors were discarded as no statistically significant effect was observed. Contrast estimates were then provided for each lactation stage using the contrast library to compare the differences between Low 2'FL and High 2'FL subgroups. The same model was used to assess the lactation stage effect comparing the consecutive stages for each of the HMOs. A linear mixed model was also used to assess anthropometric differences between infants who were breastfed Low and High 2'FL breast milk. Lactation stage, 2'FL status, gender and their interaction were declared as fixed effects and within subject variability was taken into account by declaring the subject as a random effect. A similar model was done based on zscores. Inclusion of breastfeeding habit (ever mixed feeding) in the model was considered as this may have confounding effect and was discarded as no statistically significant effect was observed.
Baseline characteristics of the study cohort
The baseline characteristics of mothers and infants by low and high concentrations of 2'FL concentrations in breast milk are depicted in Table 1. Based on the 1 month post-partum milk samples we grouped the cohort into low 2'FL (mean 27 mg/L; 95% CI of mean 12-42 mg/L) and high 2'FL concentrations (mean 2170 mg/L; 95% CI of mean 1880-2460 mg/L). We had 16 mother infant pairs (32% of the study population) in the group with low 2'FL and 34 (68%) in the high 2'FL group. Both groups had an equal distribution between females and males and a similar distribution of gestational age, mode of delivery, parity and ethnicity. Only one mother was vegetarian. At birth, the anthropometry findings of both the male and female infants were similar from mothers with low 2'FL compared with those with high 2'FL. These anthropometry findings were within the means of the WHO child growth standards for each sex.
Concentrations and interdependence of representative human milk oligosaccharides
We summarized the quantification of the analyzed HMOs over time of lactation by groups with low and high 2'FL concentrations in Table 2 and Fig 2. The group with low 2'FL had median concentrations between 15 and 10 mg/L from 1 to 4 months of lactation, while the high 2'FL had a median of 2053 mg/L at 1 months, 1652 mg/L at 2 months and 1306 mg/L at 4 months ( Table 2). As expected, in the group with high 2'FL concentrations, 2'FL was the most prominent HMO spanning a 95% CI of the mean at 1 months of lactation from 1880 to 2460 mg/L, at 2 months from 1543 to 1986 mg/L and at 4 months from 1165 to 1587 mg/L. With low 2'FL in milk, LNT became the most prominent HMO with a median concentration of 1248 mg/L at 1 month that dropped to 888 and 753 mg/L at 2 and 4 months respectively (Table 2). Over the 4 months lactation period milk LNT concentrations were significantly higher in the samples with low 2'FL as compared to those with high 2'FL. On the other hand, milk LNnT concentrations were lower in the samples with low 2'FL and higher in the high 2'FL milks. No such dependence on 2'FL was observed for the sialylated HMOs 3'SL and 6'SL (Table 2).
Over the 4 months lactation period, we found 3'SL at relatively constant concentrations both in the samples with low and high 2'FL ( Fig 2). All other measured HMOs decreased over time in breast milk. In the group with low 2'FL, 6'SL reduced from a median of 487 mg/L at 1 month to 263 and 150 mg/L at 2 and 4 months respectively, while LNnT and LNT reduced only from 1 to 2 months postpartum. In the group with high 2'FL, 6'SL as well as LNnT and LNT reduced from 1 to 2 months and also from 2 to 4 months postpartum. The connections between these different HMOs are also illustrated in the correlation matrix depicted in S1 Fig. 2'FL inversely correlates with LNT concentrations and positively correlates with LNnT. The sialylated HMOs 3'SL and 6'SL correlate between them and 3'SL also correlates to LNT.
Together, the FUT2 status with low and high 2'FL as well as the observed relationships of 2'FL with LNT (negative correlation) and LNnT (positive correlation) show that breast milk varies substantially both in HMO composition and amounts of specific oligosaccharide structures over the 4 months lactation studied here.
Effects of breast milk oligosaccharide composition on infant growth
We analyzed the effect of FUT2 dependent HMO differences on infant growth of breastfed infants in terms of body weight and body length (Fig 3), body mass index (Fig 4) and head circumference ( Fig 5). As illustrated, we plotted those growth parameters separated by females and males who were breastfed by mothers with low and high 2'FL type milk against the WHO growth curves for infants. Body length, weight, BMI and head circumference were not significantly affected by the breast milk types over time until 4 months. Likewise, based on z-scores of body weight, body length, BMI and head circumference, we also did not observe any significant differences in the growth parameters of females and males who got either low or high 2'FL breast milk (data not shown).
Globally, we did not observe any significant impact of the low or high 2'FL breast milks on the measured infant growth parameters up to 4 months of age.
Discussion
Here we show from quantitative, longitudinally collected breast milk samples, (i) a time dependent reduction of the major FUT2 dependent HMO 2'FL, the core structures LNT and LNnT and the sialylated HMO 6'SL, but not 3'SL and (ii) a relation of LNT and LNnT concentrations with 2'FL or FUT2 dependent HMOs. Since these FUT2 dependent breast milk composition changes are quite substantial, and because the affected HMOs were reported to affect early microbiota establishment [12,13], a key driver for infant health, and infant growth and body composition [20] we assessed in our cohort of breastfed infants whether their growth is affected by the breast milk FUT2 dependent HMO composition. Although preliminary due to the relatively low number of infants that we studied here, our findings suggest that early infant growth, at least up to 4 months, is independent of the qualitative and quantitative differences in the breast milk HMOs described here.
FUT2-dependent HMO changes over time in breastmilk
The mother FUT2 (Secretor) status is probably the most important genetic factor affecting the breast milk glycosylation pattern and in particular the glycosylated lactoses commonly known as HMOs. In Secretor negative individuals the FUT2 encoded enzyme activity is strongly or completely reduced [22]. In lactating mothers the FUT2 polymorphism leads to missing or strongly reduced concentrations of 2'Fucosylated HMOs like 2'FL and LNFP I for example [7,23]. Interestingly, small amounts of 2'FL were seen in all our milk samples and small amounts of 2'fucosylation were also identified in milk of serologically tested Secretor negative women from Africa [23]. The majority of Secretor negative humans carry loss of function FUT2 mutations, while some mutations lead to missense and retain some enzyme activity [22,24,25]. Alternatively, the small amounts of 2'FL in milk of presumed Secretor negative mothers might also be due to some pathway redundancy. Interestingly, mothers in the high 2'FL group with the highest 2'FL concentrations and those with the lowest remained either highest or lowest throughout the 4 months of lactation. This indicates that the FUT2 expression levels are under tight control. Previous studies had found that 6'SL in colostrum (1 to 3 days of lactation) was reported at about 370 mg/L and 3'SL at 300 mg/L [26], and in milk at 2-4 week of lactation 6'SL was about 380 mg/L and 3'SL 270 mg/L [27]. In our study, for milk at 1 month of lactation, we found 6'SL around 500 mg/L and 3'SL around 220 mg/L. These latter 3'SL values are similar to the reported concentration in colostrum and 2-4 week milk, indicating that 3'SL is expressed at relatively constant concentrations in very early and later milk. Further, in milk of mothers who gave preterm birth 3'SL concentrations were also relatively similar during the first month of lactation in the range of about 230 mg/L, while 6'SL was in the range of 500 to 700 mg/L [28]. As for 3'SL and 6'SL, our quantitative data on LNT and LNnT and 2'FL match very well those previously reported from different ethnic groups [7,26,29,30].
Depending on low or high 2'FL concentrations we saw significant differences in the concentrations of LNT and LNnT in milk confirming previous observations [31], but not seen for LNnT in one recent study [32]. While higher LNT concentrations were seen when 2'FL was low, we observed the contrary with LNnT, which was lower when 2'FL was low. This indicates that synthesis of LNT is independently regulated from 2'FL synthesis or FUT2 activity. On the other hand, formation of LNnT appears to be co-regulated with 2'FL synthesis or FUT2 activity. After lactose, LNT is the major acceptor substrate for FUT2 during HMO synthesis, leading to the formation of LNFP I [7]. After 2'FL, LNFP I is the second major 2'-fucosylated HMO [9,29]. Therefore, higher LNT in milk in the absence of 2'FL, or FUT2 activity, indicates that synthesis of the HMOs 2'FL and LNT is not co-regulated. On the other hand LNnT was lower in our study, when 2'FL or FUT2 activity was low. This might indicate co-regulated synthesis of 2'FL and LNnT or it could also be that LNnT is more readily used as acceptor substrate by other glycosyltransferases such as FUT3 [7], when FUT2 is low and GDP-fucose remains available as donor substrate. In line with this, Chaturvedi et al. [29] and Austin et al. [33] reported decreasing 2'FL and LNFP I over time of lactation concomitant with increasing 3-Fucosyllactose, a major 3'fucosylated HMO.
Differences of breastmilk HMOs in relation to breastfed infants
As a consequence of the FUT2 polymorphism of mothers, infants naturally receive milk with different amounts and types of oligosaccharides. This leads to the evident question whether such a difference in breast milk composition is associated with specific infant morbidity, growth and developmental phenotypes. In a Mexican population higher concentrations of the FUT2 dependent oligosaccharides 2'FL and LNFP I in breast milk were associated with lower incidence of infectious diarrhea [18], indicating that the presence of FUT2 dependent HMOs might provide better protection from specific pathogens like Campylobacter jejuni, E. coli and norovirus [34]. Similarly, higher breast milk 2'-fucosylated HMOs, 2'FL and LNFP I, as well as non-2'-fucosylated HMOs were associated with lower infant mortality in HIV-exposed infants during lactation, but not thereafter [35]. In part, such observations might have been mediated via the establishing intestinal microbiota, which was also reported to be affected by the FUT2 dependent HMOs in breast milk [12,13]. FUT2 dependent variations of HMOs in breast milk, the sole and recommended nutrition for infants, might also associate with growth and development of the breastfed infant, for example through effects on the establishing gut microbiota. Alderete et al. [20] reported on an inverse relation between the concentrations of the FUT2 dependent LNFP I, but not 2'FL, in breast milk at 6 months and body weight, lean and fat mass of infants at 6 months. Further, LNnT also inversely related to body fat mass at 6 months of age.
In our study cohort we did not see any consistent effects of FUT2 positive or negative breast milk on infant growth up to 4 months of age. Notably, we had 33% of the infants who received breastmilk with low 2'FL throughout the studied 4 months indicating that these were Secretor negative milk and, although we did not quantify LNFP I in our study, milk with low or no 2'FL basically have no LNFP I, because both HMOs are synthesized by the same enzyme FUT2 (see also [9,20,29,36]). Therefore, 2'FL is a good proxy to group milks into low (Secretor negative) and high 2'Fucosyl-HMOs (secretor) that includes 2'FL, LNFP I and also diFL among other FUT2 dependent HMOs. Samples in our cohort with low 2'FL concentrations remained in the low concentration group at the 3 time points indicating that the low concentration is due to a genetic defect in FUT2 (e.g. Secretor negative genotype) and equally high 2'FL concentrations were also consistently high at 1, 2 and 4 months of lactation. Consequently, the categorical testing as done here for possible relations of low and high 2'FL milks with infant anthropometry is a stringent approach that is very unlikely to be biased by total breastmilk intake per day or variation of HMO concentration due to sampling and analytical variations. Based on infant growth curves and z-scores (body weight, body length, BMI and head circumference) we did not spot any indication that breast milk FUT2 dependent HMOs (including 2'FL, LNFP I, diFL) influence infant growth over the first 4 months of life. Yet, some variation in growth parameters was observed in our cohort. Although not statistically significant, but numerically visible, males from mothers with low 2'FL milk appeared to have a slightly higher BMI at 1 month, which was not seen any more at 4 months of age when they rather had a smaller BMI and body weight gain. Possible effects of FUT2 dependent HMOs on infant growth may not be apparent here, because our cohort represents an apparently healthy group of term-born infants. Further, we cannot exclude that changes in growth curves might only appear beyond 4 months of age.
Conclusion
Together, although not powered to assess infant growth and only based on anthropometry and not body composition, we conclude from our exploratory observation study that FUT2 related alterations of breast milk HMOs composition as assessed through 2'FL concentrations, does not impact growth of breastfed infants during the first 4 months of life. | 2017-11-02T05:46:04.478Z | 2017-02-09T00:00:00.000 | {
"year": 2017,
"sha1": "f28652459803359dac5dffa5e4d15a23d1649f44",
"oa_license": "CCBY",
"oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0171814&type=printable",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "3eff131fa46a993f1ae9677ef643e0330655031d",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
247558449 | pes2o/s2orc | v3-fos-license | Intravascular large B‐cell lymphoma in renal cell carcinoma incidentally detected by robot‐assisted partial nephrectomy
Introduction Intravascular large B‐cell lymphoma is a rare and aggressive type of extranodal large B‐cell lymphoma. Although intravascular large B‐cell lymphoma can invade various organs, renal involvement has been rarely reported. Synchronous occurrence of intravascular lymphoma with renal cell carcinoma is extremely rare. We herein report a case of intravascular large B‐cell lymphoma in a renal cell carcinoma incidentally detected by robot‐assisted partial nephrectomy. Case presentation A 69‐year‐old female with recurrent fever lasting 4 years underwent robot‐assisted partial nephrectomy for small renal cell carcinoma. Histological findings led to the diagnosis of intravascular large B‐cell lymphoma, which involved the normal tissue of right kidney as well as clear cell renal cell carcinoma. She received six cycles of chemotherapy without major complications and achieved complete remission. Conclusion We encountered a rare case of synchronous intravascular lymphoma with renal cell carcinoma.
Introduction
IVLBCL is a rare and aggressive type of extranodal large B-cell lymphoma, with an estimated annual incidence of 0.5-1 per 100 thousand population. Most cases occur in adults, with a median age of 67 years. Men and women are equally affected. It often involves the brain and skin, however it can invade almost any organ. 1 Renal involvement of IVLBCL is rarely reported, with <50 previous cases. 2 Furthermore, the synchronous occurrence of intravascular lymphoma with RCC is extremely rare. There have been only four cases of IVLBCL involving the microvasculature of RCC. [3][4][5][6] We present the case of a 69-year-old female who underwent RAPN for a clear cell RCC sized 1.7 cm, in which IVLBCL was incidentally detected. To our knowledge, this is the first case of synchronous IVLBCL with RCC detected by RAPN.
Case presentation
A 69-year-old female with a surgical history of maxillary sinus cancer and uterine fibroids visited the Department of Infectious Diseases of a nearby general hospital for examination of severe pitting edema and recurrent fever of 39°C. Although the recurrent fever lasted 4 years, she had not previously undergone detailed examinations.
Infectious diseases or autoimmune diseases were not detected by close examinations. However, eCT revealed an enhanced mass measuring 17 9 17 9 11 mm in her right kidney, which was suspected to be an RCC (Fig. 1). She also had a right adrenal adenoma (28 9 20 9 21 mm), which was revealed to be nonfunctioning. No metastatic mass was observed, then she was referred to the Urology Department of our hospital for surgical resection of the renal mass. At the time of admission, the physical findings were as follows: height, 148.5 cm; weight, 43.6 kg; body temperature, 37.3°C; blood pressure, 98/73 mmHg; heart rate, 110/min; and oxygen saturation at rest on room air, 98%. Laboratory findings at admission are presented in Table 1.
She underwent RAPN along with right adrenalectomy at the same time. After surgery, she experienced a continuous fever of over 38°C. Therefore, eCT and echocardiography were performed, however there were no abnormal findings.
Two weeks after surgery, histological examination revealed clear cell carcinoma and adrenal adenoma. The RCC was pT1a, G1, INFa, lymphatic invasion (À), vascular invasion (À), resected margin (À). Vessels of clear RCC are packed with atypical lymphoid (Fig. 2a). Atypical cells were positive for CD20 expression. (Fig. 2b) The Ki-67 index increased to 80%. Atypical cells also infiltrated the vessels of the normal tissue of the kidney, however they did not infiltrate the adrenal sinusoid. Histological findings led to the diagnosis of IVLBCL, which involved the right kidney and clear cell RCC.
She underwent bone marrow and skin biopsies to examine the clinical stage of IVLBCL to determine whether IVLBCL invades other sites in her body. Although the bone biopsy did not show evidence of lymphoma invasion or any other hematological abnormalities, including hemophagocytic syndrome, the skin biopsy showed atypical lymphoid cells involving the vessels in subcutaneous tissue. Therefore, she was diagnosed as stage IVB IVLBCL with an International Prognostic Index of 4 (high risk) due to advanced age, clinical stage IV, two extranodal sites, and elevated serum lactate dehydrogenase above the upper limit of normal. Treatment with oral steroids (predonisone, 50 mg/day) was initiated to mitigate the symptoms associated with IVLBCL. Her body temperature turned to be normal, and she was tentatively discharged 1 month postoperatively.
Two months after the surgery, chemotherapy with R-CHOP was initiated. She had received six cycles of R-CHOP without any major complications and achieved complete remission. She has no recurrence of RCC or IVLBCL 1.5year after RAPN.
Discussion
IVLBCL can invade various organs and manifest diverse symptoms according to the types of organ involved. A definitive diagnosis of IVLBCL requires histopathological examination. Random skin biopsies, including the abundant vascular structures of adipose tissue, could be helpful for diagnosis.
Chemotherapy is the first option of treatment for patients with IVLBCL. As an aggressive form of lymphoma, it has been classically treated with cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP regimen). The introduction of the R-CHOP regimen has significantly improved the therapeutic effects. 7 It has recently been reported that autologous stem cell transplantation following high-dose chemotherapy can improve the outcome of patients with IVLBCL. However, this aggressive treatment approach appears to be practical only in a limited proportion of patients. 8 The recently updated World Health Organization classification of tumors of hematopoietic and lymphoid tissues has suggested considering variants of IVLBCL according to their clinical characteristics: classical variant, cutaneous variant, and hemophagocytic syndrome-associated variant. 1 The classical variant IVLBCL often presents with various nonspecific symptoms, such as fever, fatigue, and altered consciousness. In cutaneous variant, involved lesions are limited to the skin and disease progressions are less aggressive. The hemophagocytic syndrome-associated variant showed the most rapidly aggressive onset and progression, with a median survival time of 2-8 months. We diagnosed this case as a classical variant due to IVLBCL in her kidney and skin with recurrent fever for a long time, and complete remission was achieved with R-CHOP chemotherapy.
The occurrence of synchronous IVLBCL with RCC is very rare, with only four cases reported to date ( Table 2). IVLBCL was incidentally detected by radical nephrectomy in all the four previous cases. The reported prognoses for these cases are poor; they died at 1, 6, and 18 months, respectively. Our case is unique in two aspects. Firstly, the progression of disease was less aggressive. She had recurrent fever for 4 years before the diagnosis, the clinical response to R-CHOP treatment was excellent without recurrence for 18 months. Secondly, this is the first case of synchronous IVLBCL with RCC detected by RAPN. Compared with the four reported cases, in this case, the renal mass was small and resected with RAPN.
To diagnose IVLBCL at an early stage, it is essential that we suspect lymphoma when facing with constitutional symptoms. It is also important that if we find a patient with a renal mass and recurrent fever or fatigue for a long time, we suspect the possibility of IVLBCL.
Conclusion
We herein report a case of IVLBCL in an incidentally detected RCC by RAPN. To our knowledge, this is the first case of synchronous IVLBCL with RCC detected by RAPN. | 2022-03-20T15:10:59.334Z | 2022-03-18T00:00:00.000 | {
"year": 2022,
"sha1": "c1091c6a6df8b36687253ce05a7572b529dfc439",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Wiley",
"pdf_hash": "b1bc358af7b0dc969de2bd41736cdc132aaa7321",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
206310731 | pes2o/s2orc | v3-fos-license | Inversion of Qubit Energy Levels in Qubit-Oscillator Circuits in the Deep-Strong-Coupling Regime
We report on experimentally measured light shifts of superconducting flux qubits deep-strongly coupled to LC oscillators, where the coupling constants are comparable to the qubit and oscillator resonance frequencies. By using two-tone spectroscopy, the energies of the six lowest levels of each circuit are determined. We find huge Lamb shifts that exceed 90% of the bare qubit frequencies and inversions of the qubits' ground and excited states when there are a finite number of photons in the oscillator. Our experimental results agree with theoretical predictions based on the quantum Rabi model.
According to quantum theory, the vacuum electromagnetic field has "half-photon" fluctuations, which cause several physical phenomena such as the Lamb shift [1].A cavity can enhance the interaction between the atom and the electromagnetic field inside the cavity and enables more precise measurements of the influence of the vacuum.Cavity/circuit-quantumelectrodynamics systems are usually well described by the Jaynes-Cummings Hamiltonian [2,3].In the strong-coupling regime, when the cavity's resonance frequency ω is on resonance with the atom's transition frequency ∆, the vacuum Rabi splitting [4][5][6] and oscillations [7,8] have been observed.In the off-resonance case, the Lamb shift [9][10][11] caused by the vacuum fluctuations and the ac-Stark shift proportional to the photon number in the cavity were observed [10][11][12][13].In the so-called ultrastrong-coupling regime [14,15], where the coupling constant g becomes around 10% of ∆ and ω, and the deep-strong-coupling regime [16,17], where g is comparable to or larger than ∆ and ω, the rotating-wave approximation used in the Jaynes-Cummings Hamiltonian breaks down, and the system should be described by the quantum Rabi Hamiltonian [18][19][20].In these regimes, the light shifts of an atom could nonmonotonically change as g increases, and the amount of the shift is not proportional to the photon number in the cavity [21,22].
In this work, to study the light shift in the case of g ∼ ω, we investigated qubit-oscillator circuits that each comprise a superconducting flux qubit [23] and an LC oscillator inductively coupled to each other by sharing a loop of Josephson junctions that serves as a coupler [Figs.1(a) and 1(c)].By using two-tone spectroscopy [24,25], the energies of the six lowest energy eigenstates were measured, and the photon-number-dependent qubit frequencies were evaluated.We find Lamb shifts over 90% of the bare qubit frequency and inversions of the qubit's ground and excited states when there are a finite number of photons in the oscillator.
The qubit-oscillator circuit is described by the Hamiltonian The first two terms represent the energy of the flux qubit written in the basis of two states with persistent currents flowing in opposite directions around the qubit loop, | q and | q .
The operators σx,z are the standard Pauli operators.The parameters ∆ and ε are the tunnel splitting and the energy bias between | q and | q , where ε can be controlled by the flux bias through the qubit loop Φ q .The third term represents the energy of the LC oscillator, where ω = 1/ (L 0 + L c )C [see Fig. 1(a)] is the resonance frequency, and â † and or "e" the excited state and the number of real photons in the oscillator is n.The arrows indicate transition frequencies between energy eigenstates and also mean that the transitions are allowed.
â are the creation and annihilation operators, respectively.The fourth term represents the coupling energy.
At ε = 0, the Hamiltonian in Eq. ( 1) reduces to that of the quantum Rabi model ĤRabi .
In the limit ∆ ω, the energy eigenstates are well described by Schrödinger-cat-like entangled states between persistent-current states of the qubit and displaced Fock states of the oscillator D(±α)|n o [21,22]: Here, D(α) = exp(αâ † − α * â) is the displacement operator, and α is the amount of the displacement.The energy eigenstates on the left-hand side are expressed as |in (i = g, e), where "g" and "e" denote, respectively, the ground and excited states of the qubit and n the number of real photons in the oscillator.On the right-hand side, |n o denotes the oscillator's n-photon Fock state.Note that the displaced vacuum state D(α)|0 o is the coherent state The photon-number-dependent qubit frequency ∆ n (g/ω) ≡ ω en − ω gn is defined as the energy difference between the energy eigenstates |gn and |en , and it can be expressed as (see the solid lines in Fig. 4): Here, L n is a Laguerre polynomial: and so on.The difference between ∆ n and the bare qubit frequency ∆ can be considered as the n-photon ac-Stark shift |∆ n − ∆|.In particular, |∆ 0 − ∆| is referred to as the Lamb shift.Note that the Bloch-Siegert shift [26,27], the contribution from the counterrotating terms, is included in the n-photon ac-Stark shifts.Since L 0 = 1, a considerable Lamb shift is expected when g becomes comparable to ω.A similar suppression of transition frequencies because of coupling to other degrees of freedom is well known in polaron physics and other fields.For example, such an effect was recently discussed for an Andreev-level qubit [28].
Considering that L n has n zeros, i.e., points where L n (x) is equal to zero, ∆ n (x) also has n zeros, and, hence, in general alternates between positive and negative values.In other words, the qubit's ground and excited states exchange their roles every time when ∆ n = 0.
The bare qubit frequency ∆ is the tunnel energy between the states | q and | q .Taking either one of these two states and a finite value of g, the oscillator is populated by virtual photons even in the ground state, and the virtual photon states for the qubit states | q and | q are different from each other.As a result, the qubit has to "drag" the oscillator every time it flips its state [21], which can be seen as an effective reduction of ∆ by a factor that is determined by the overlap integral between the interaction-caused displaced n-photon Fock states of the oscillator [29] as described by the second line of Eq. (3).One way to understand negative values of ∆ n is to think of them as describing a situation where the antibonding state of | q and | q is more stable than the bonding state.Note that here the displaced states D(±g/ω)|0 o contain only virtual photons, while the states D(±g/ω)|n o for n ≥ 1 contain a mixture of real and virtual photons.
Although Eqs. ( 2) and (3) are not exact for general values of the circuit parameters, they remain reasonably good approximations as long as ∆ < ω.Furthermore, the symmetry of ĤRabi is independent of the circuit parameters, which means that certain transitions will remain forbidden even if the corresponding states do not have simple forms.These two considerations allow us to easily identify the energies of the different eigenstates from the experimental spectra [29].
To determine the parameters of the qubit-oscillator circuits (∆, ω, and g), spectroscopy was performed by measuring the transmission spectrum through the transmission line that is inductively coupled to the LC oscillator [Fig.1(a)].In total, nine sets of parameters (A-I in Table I) in five circuits were evaluated.The shared inductance of the circuit for set A is a superconducting lead [Fig.1(b)], while that of the circuits for sets B-I is a loop of Josephson junctions [Fig.1(c)], where eight flux bias points in four circuits were used [29].
Therefore, much larger g is expected for sets B-I.When the frequency of the probe signal ω p matches the frequency ω kl of a transition |k → |l , where |0 stands for the ground state and |k with k ≥ 1 stands for the kth excited state of the coupled circuit, the transmission amplitude decreases, provided that the transition matrix element k|(â + â † )|l is not 0.
Note that because ε is now generally nonzero, we have labeled the energy eigenstates using a single integer k instead of the label |in used above.Figure 2 shows the amplitudes of the transmission spectra |S meas 21 (ε, ω p )/S bg 21 (ω p )| for sets A and H. Here, ω p is the probe frequency, and S meas 21 (ε, ω p ) and S bg 21 (ω p ) are, respectively, the measured and background transmission coefficients [29].
The circuit parameters are obtained from fitting the experimentally measured resonance frequencies to those numerically calculated by diagonalizing Ĥ with ∆, ω and g treated as fitting parameters.In Fig. 2, the right panels show the calculated transition frequencies superimposed on the measured spectra.In Fig. 2(a), one can see the characteristic spectrum of the strong-and ultrastrong-coupling regimes.From the fitting, the parameters are obtained as ∆/2π = 1.246GHz, ω/2π = 6.365GHz, and g/2π = 0.42 GHz.The spectrum shown in Fig. 2(b) looks qualitatively different from that in Fig. 2(a) as discussed in Ref. [17].The parameters are obtained as ∆/2π = 1.68 GHz, ω/2π = 6.345GHz, and g/2π = 7.27 GHz.
Here, g is larger than both ∆ and ω, indicating that the circuit is in the deep-strong-coupling regime [g max(ω, √ ∆ω/2)] [21, 30,31].The parameters from all the sets are summarized in Table I.
To obtain the photon-number-dependent qubit frequency ∆ n (n = 0, 1, 2), at least five transition frequencies out of seven allowed transitions [Fig.1(d)] are necessary.However, in each spectrum at ε = 0, we see only two signals at frequencies ω g0,g1 and ω e0,e1 corresponding, respectively, to the transitions |g0 → |g1 and |e0 → |e1 , which were also observed in our previous experiments [16,17].There are two main reasons behind this limitation on single-tone spectroscopy, where only a single-frequency weak probe signal is applied to the circuit.First, only transition frequencies in the range of the measurement setup (in our case 4 to 8 GHz) can be measured.Second, the signal from transitions that do not start from the lowest two energy levels will be weak because of the small thermal population of higher energy levels (in our case, the thermal population decreases by 2 orders of magnitude for each step up in the value of n).
To access transitions other than |g0 → |g1 and |e0 → |e1 , two-tone spectroscopy was used, where a drive signal with frequency ω d is applied while the transmission of a probe signal with frequency ω p around the frequency ω g0,g1 or ω e0,e1 is measured.When ω d is equal to the frequency of an allowed transition involving at least one of the states |g0 , |g1 , |e0 , and |e1 , an Autler-Townes splitting [32] takes place and is observed in the probe transmission signal.Figure 3 shows the measured two-tone transmission spectra from and the emission of one photon to the drive field in Fig. 3(c).Together with the frequencies numerically calculated from ĤRabi , the corresponding transitions are identified as shown in the right-hand side of each spectrum.The spectrum in Fig. 3(c) demonstrates that the energy of |g1 is higher than that of |e1 , and, hence, ∆ 1 is negative.In other words, the qubit's energy levels are inverted.
Moreover, from these three two-tone transmission spectra, five transition frequencies, ω g0,g1 , ω g0,g2 , ω e0,e1 , ω e0,e2 , and ω g0,e1 , can be evaluated.In Fig. 3 the diagonal line is at ω p = ω g0,e1 +ω d .From these five transition frequencies, all the eigenenergies up to the fifth excited state can be determined up to an overall energy shift.One thing is worth emphasizing here.In the two-tone spectroscopy of a deep-strongly-coupled qubitoscillator circuit, the states of the qubit are doubly dressed: one is the conventional dressing by the classical drive field, while the other is in the quantum regime due to deep-strong coupling to the oscillator, where the oscillator's states are displaced.The experimental results demonstrate that the two kinds of dressed states coexist.
From Eq. ( 3), the normalized photon-number-dependent qubit frequencies ∆ n /∆ are expected to depend solely on the normalized coupling constant g/ω.We, therefore, plot ∆ n /∆ as functions of g/ω for all nine parameter sets together (Fig. 4).The parameters ∆, ω, and g are obtained from the transmission spectra.These results demonstrate huge Lamb shifts |∆ 0 − ∆|, some of them exceeding 90% of the bare qubit frequencies ∆.These results also demonstrate that one-photon and two-photon ac-Stark shifts are so large that ∆ 1 and ∆ 2 change their signs depending on g/ω.The solid lines are theoretically predicted values given by Eq. (3).Table I shows a comparison between the measured and the numerically calculated ∆ n [29] using ĤRabi and the parameters ∆, ω, and g.In many circuits, the measured ∆ 2 is smaller than the numerically calculated one, while the agreement of ∆ 0 and ∆ 1 is good, with the deviations being at most 10 MHz.Since ∆ 2 given by Eq. ( 3) is an approximation that becomes exact in the limit ∆/ω → 0 while the numerically calculated ∆ 2 is based on the exact ĤRabi for any set of parameters, the agreement of ∆ 2 in Fig. 4 is a coincidence.In this way, our results can be used to check how well the flux qubit-LC oscillator circuits realize a system that is described by the quantum Rabi model Hamiltonian, which is the basis for several important applications, e.g., ultrafast gates [33] and quantum switches [34].A possible source of the deviation in ∆ 2 is higher energy levels of the flux qubit.As discussed in Ref. [17], the second or higher excited states can shift the energy levels of the qubit-oscillator circuit, even though there is an energy difference of at least 20 GHz between the first and the second excited states.Consideration of higher energy levels is necessary to identify the origin of the deviation in ∆ 2 .
In conclusion, we have used two-tone spectroscopy to study deep-strongly-coupled flux qubit-LC oscillator circuits.We have determined the energies of the six lowest energy eigenstates of each circuit and evaluated the photon-number-dependent qubit energy shifts.
We have found Lamb shifts that exceed 90% of the bare qubit frequency and inversions of the where H n is the Hermite polynomial; and so on.Note that φ n (x, g/ω) is real.The overlap integral, which appears in the second line of Eq. ( 3) in the main text, can be calculated as To be concrete, in the following, we consider the case of n = 2 as an example.Figures S1(a the states and operators that appear in the quantum Rabi Hamiltonian, is summarized in Table SI.Here, ∆ is the qubit's transition frequency.Because both σx and (â + â † ) have negative parities, their product has a positive parity, meaning that all three terms in ĤRabi have positive parities.Therefore, [ ĤRabi , P ] = 0, and hence, the energy eigenstates are also eigenstates of P , and have well-defined parities.Note that this property does not depend on the values of ∆, ω, and g.
Although the energy eigenstates and their eigenenegies of ĤRabi cannot be described analytically for arbitrary values of ∆, ω, and g, the symmetry allows to define energy eigenstates and their eigenenegies of ĤRabi as |in and ω in , where i (= g, e) indicates that the qubit is in "g" the ground or "e" the excited state and n is the number of real photons in the oscillator.Since the parity of (â + â † ) is odd, the transition matrix elements im|(â + â † )|jn may have non-zero values when the parities of the energy eigenstates |im and |jn are opposite, whereas they always vanish when the parities are same.
From the transition frequencies alone, the energy eigenstates and the eigenenergies cannot be determined uniquely.However, by using the parity symmetry discussed above, energy eigenstates and eigenenergies are recursively determined as long as ∆ < ω in the following parity even odd qubit state
S3. COUPLER INDUCTANCE AND FLUX BIAS POINTS
The coupler inductance for sets B-I is a dc superconducting quantum interference device (SQUID) consisting of two parallel Josephson junctions as shown in Fig. 1(c) in the main text.Its Josephson inductance is given as where and we assumed that a background transmission coefficient S bg 21 (ω p ) is independent of energy bias ε and is written by a polynomial of the probe photon frequency ω p .Eq. (S7) can be applied to a transmission line that is inductively and capacitively coupled to an LC oscillator [35], where Q L is the total quality factor of the oscillator, Q e is the external quality factor due to the coupling to the transmission line, and φ is a phase factor that accounts for the asymmetry of the resonance line shape.Note that |S 21 (ε, ω p )| may become larger than 1 depending on the value of φ.
S5. AVOIDED CROSSINGS IN TWO-TONE SPECTROSCOPY
In this section, we discuss the physical origin of the avoided crossings observed in Fig. 3 of the main text.The Hamiltonian of a three-level system under the application of a drive field with frequency ω d can be described by ∆ n (g/ω) ∆ exp(−2g 2 /ω 2 )L n (4g 2 /ω 2 ), (S10) which is also given in the main text.The dotted lines are numerically calculated values from ĤRabi and the parameter ∆/ω = 0.933.Here, the eigenenergies and the energy eigenstates
FIG. 1 .
FIG. 1.(a) Circuit diagram.A superconducting flux qubit (red and black) and a superconducting LC oscillator (blue and black) are inductively coupled to each other by sharing an inductance (black).(b), (c) Scanning microscope images of the qubit and the shared inductance located at the orange rectangle in diagram (a).Josephson junctions are represented by magenta rectangles.The shared inductance is a superconducting lead (b) or a loop of Josephson junctions (c).(d) The diagram of the six lowest energy levels of a qubit-oscillator circuit.The energy eigenstates are expressed as |in (i = g, e and n = 0, 1, 2, • • • ), which indicates that the qubit is in "g" the ground
®FIG. 3 .
FIG. 3. (Left) Measured two-tone transmission spectra as functions of drive frequency ω d and probe frequency ω p .The color scheme is chosen such that the lowest point in each spectrum is red and the highest point is blue.The white dotted lines are calculated transition frequencies considering dressed states due to the drive signals.The right panels show the energy-level diagrams.The thin green arrows indicate transitions scanned by the probe signal, while thick magenta arrows indicate transitions scanned by the drive signal.
FIG. 4 .
FIG.4.Photon-number-dependent normalized qubit frequencies ∆ n /∆ as functions of g/ω.The parameters ∆, ω, and g are obtained from the single-tone transmission spectra.The black, red, and blue solid circles are, respectively, the qubit frequencies ∆ 0 , ∆ 1 , and ∆ 2 obtained from the two-tone transmission spectra.The solid lines are ∆ n obtained from Eq. (3).
qubit's ground and excited states caused by the one-photon and two-photon ac-Stark shifts.The results agree with the quantum Rabi model, giving further support to the validity of the quantum Rabi model in describing these circuits in the deep-strong-coupling regime.constant, ω is the resonance frequency of the oscillator, and |n o is the n-photon Fock state of the oscillator.The wave function of D(g/ω)|n o in the coordinate basis is given by φ n (x, g/ω) = o x| D(g/ω)|n o , where |x o is an eigenstate of the coordinate operator x = (â + â † )/2, and â and â † are the annihilation and creation operators.Besides the normalization factor, it is given by Fig.S1(f)].When the positions of peaks or dips of φ 2 (x, ±g/ω) coincide, the product φ 2 (x, −g/ω)φ 2 (x, g/ω) is mostly positive, and hence I 2 (g/ω) becomes maximal [Figs.S1(a)and (e)].On the other hand, when the peak positions of a wave function coincide with the dip positions of the other, the product φ 2 (x, −g/ω)φ 2 (x, g/ω) is mostly negative, and hence I 2 (g/ω) becomes minimal [Fig.S1(c)].
S5) where I c is the critical current of each Josephson junction, n φc is the normalized flux bias through the loop in units of the superconducting flux quantum Φ 0 = h/(2e), and I b is the current flowing through the SQUID.An external superconducting magnet produces a uniform magnetic field, and flux biases are applied to the qubit and the coupler proportionally to the areas of their loops.The area ratio of the loops r c = A coupler /A qubit is approximately 0.05.The flux bias through the coupler loop n φc = r c n φ depends on the normalized flux bias of the qubit n φ , which in most cases is around the symmetry point of the qubit, i.e. n φ = ±0.5, ±1.5, and so on.
FIG. S2.(a) (c) The energy-level diagram when (a) ω a < ω b < ω c and (c) ω a < ω c < ω b .Here, the states |i, k (i = a, b, c, and k is the number of drive photons) are energy eigenstates of H ω b <ωc and H ωc<ω b when the off-diagonal terms are ignored.We have kept only the states with drive photon numbers N − 1 and N in (a) and N and N + 1 in (c).The gray arrows indicate allowed transitions in the three-level system.The magenta arrow indicates the drive frequency ω d .The dotted and solid green arrows indicate the transitions with the transition frequencies around ω ab when Ω bc = 0 and Ω bc = 0, respectively.(b) (d) Transition frequencies as functions of ω d when (b) ω a < ω b < ω c and (d) ω a < ω c < ω b .The dotted and solid green lines correspond to the transition frequencies when Ω bc = 0 and Ω bc = 0, respectively.
TABLE SI .
The parity symmetry in the states and operators that appears in ĤRabi .way.(i) The ground and the first excited states of a coupled circuit are respectively |g0 and |e0 , since there is no energy-level crossing between them.The corresponding eigenenergies are respectively ω g0 and ω e0 .(ii) Between the (2n+2)th and (2n+3)th excited states (n ≥ 0), the state having nonzero transition matrix element with the state |gn is |gn + 1 and the other is |en + 1 .The corresponding eigenenergies are respectively ω gn+1 and ω en+1 .In this way, photon-number-dependent qubit frequency ∆ n ≡ ω en − ω gn can be uniquely determined for all the parameter sets in this work. | 2018-05-07T05:40:31.000Z | 2017-12-13T00:00:00.000 | {
"year": 2017,
"sha1": "16a597fd5f690e291bbe6cf15f2902c8d0aceacd",
"oa_license": "publisher-specific, author manuscript",
"oa_url": "https://link.aps.org/accepted/10.1103/PhysRevLett.120.183601",
"oa_status": "HYBRID",
"pdf_src": "Arxiv",
"pdf_hash": "16a597fd5f690e291bbe6cf15f2902c8d0aceacd",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Medicine",
"Physics"
]
} |
247462887 | pes2o/s2orc | v3-fos-license | Effect of Annealing Temperature on the Interfacial Microstructure and Bonding Strength of Cu/Al Clad Sheets with a Stainless Steel Interlayer
To explore the influence of annealing temperatures on the interfacial structure and peeling strength of Cu/Al clad sheets with a 304 stainless steel foil interlayer, an intermediate annealing treatment was performed at temperatures of 450 °C, 550 °C, and 600 °C, separately. The experimental results indicate that the interfacial atomic diffusion is significantly enhanced by increasing the intermediate annealing temperature. The average peeling strength of the clad sheets annealed at 550 °C can reach 34.3 N/mm and the crack propagation is along the steel/Cu interface, Cu-Al intermetallic compounds layer, and Al matrix. However, after high-temperature annealing treatment (600 °C), the liquid phase is formed at the bonding interface and the clear Cu/steel/Al interface is replaced by the chaotic composite interfaces. The clad sheet broke completely in the unduly thick intermetallic compounds layer, resulting in a sharp decrease in the interfacial bonding strength.
Introduction
Cu/Al clad sheets have been widely applied in many fields, such as power electronics, aerospace, and electronic communication [1]. The common methods to produce Cu/Al clad sheets are rolling bonding [2], twin-roll casting bonding [3], explosive bonding [4], and diffusion bonding/TLP bonding/diffusion brazing [5]. Different from these traditional techniques, the powder-in-tube method exhibits dominant advantages to fabricate metallic clad sheets with high interfacial bonding strength, which can easily achieve the closure of local defects around the bonding interface and the regulation in thickness and structure of intermetallic compounds (IMCs) [6]. It is a promising and environmentally friendly method to fabricate metallic clad sheets.
The annealing treatment has a significant influence on the interfacial structure and mechanical performance of metallic clad sheets [7,8]. Cu-Al IMCs are easily formed at the bonding interface due to the strong chemical affinity between Cu and Al [9]. As such, a considerable amount of research has been carried out to regulate the structure of Cu-Al IMCs. In the annealing process, Gao et al. [10] found that the CuAl 2 phase was in possession of the priority formation rather than other Cu-Al IMCs, whose formation sequence was in the order of CuAl 2 , Cu 9 Al 4 , CuAl, and Cu 4 Al 3 (or Cu 3 Al 2 ) [2]. The mathematical relationship between the IMCs layer thickness, annealing time, and annealing temperature was built by Pelzer et al. [11]. Similarly, Lee et al. [12] indicated that the growth of Cu-Al IMCs could be controlled by a diffusion mechanism. Li et al. [13] showed that a thicker IMCs layer induced by the high annealing temperature usually led to a decrease in interfacial bonding strength. Mao et al. [14] indicated that the thickness of the IMCs layer in Cu/Al clad sheet could be controlled within 550 nm after annealing at 250 • C and the peeling strength could reach 39 N/mm. Therefore, effectively controlling the thickness and structure of IMCs is a critical factor to obtaining a high-performance Cu/Al clad sheet.
Our previous research found that the introduction of a 304 stainless steel (SUS304) interlayer could significantly improve the interfacial strength of the Cu/Al clad sheet by optimizing the type and structure of IMCs [15]. In this paper, Cu/Al clad sheets with a SUS304 interlayer were prepared by the powder-in-tube method. The influence of annealing temperature on the interfacial structure and bonding strength was investigated. Furthermore, the interfacial strengthening mechanism was systematically discussed.
Materials and Methods
The initial materials are the commercial pure copper tube (99.9%, Guangfeng Metal Materials Co., Ltd., Dongguan, China) with an outer diameter of 10 mm and a wall thickness of 1 mm and atomized aluminum powder with an average diameter of 10 µm. The cold-rolled SUS304 foils (Guangfeng Metal Materials Co., Ltd., Dongguan, China) with a thickness of 30 µm are chosen as the interlayer materials. The mechanical parameters of the SUS304 foils have been illustrated in our previous research [15]. Before roll cladding, the oxide layer on the inner surface of the Cu tube is removed by the wire brush. The detailed preparation process of the Cu/Al clad sheets with the SUS304 interlayer is illustrated in Figure 1. Firstly, the Cu cube, SUS304 foil, and Al powder (Hunan Jinhao New Material Technology Co., Ltd., Miluo, China) are manually assembled together to form the composite tube billet. Secondly, the Cu/Al clad sheets are rolled to 1.5 mm after multi-pass cold rolling and then annealed at different temperatures of 450 • C, 550 • C, and 600 • C, which are designated as IFR-450, IFR-550, and IFR-600, respectively. Finally, these annealed samples are further rolled to 0.5 mm. Cu-Al IMCs could be controlled by a diffusion mechanism. Li et al. [13] showed that a thicker IMCs layer induced by the high annealing temperature usually led to a decrease in interfacial bonding strength. Mao et al. [14] indicated that the thickness of the IMCs layer in Cu/Al clad sheet could be controlled within 550 nm after annealing at 250 °C and the peeling strength could reach 39 N/mm. Therefore, effectively controlling the thickness and structure of IMCs is a critical factor to obtaining a high-performance Cu/Al clad sheet. Our previous research found that the introduction of a 304 stainless steel (SUS304) interlayer could significantly improve the interfacial strength of the Cu/Al clad sheet by optimizing the type and structure of IMCs [15]. In this paper, Cu/Al clad sheets with a SUS304 interlayer were prepared by the powder-in-tube method. The influence of annealing temperature on the interfacial structure and bonding strength was investigated. Furthermore, the interfacial strengthening mechanism was systematically discussed.
Materials and Methods
The initial materials are the commercial pure copper tube (99.9%, Guangfeng Metal Materials Co., LTD., Dongguan, China) with an outer diameter of 10 mm and a wall thickness of 1 mm and atomized aluminum powder with an average diameter of 10 μm. The cold-rolled SUS304 foils (Guangfeng Metal Materials Co., LTD., Dongguan, China) with a thickness of 30 μm are chosen as the interlayer materials. The mechanical parameters of the SUS304 foils have been illustrated in our previous research [15]. Before roll cladding, the oxide layer on the inner surface of the Cu tube is removed by the wire brush. The detailed preparation process of the Cu/Al clad sheets with the SUS304 interlayer is illustrated in Figure 1. Firstly, the Cu cube, SUS304 foil, and Al powder (Hunan Jinhao New Material Technology Co., LTD., Miluo, China) are manually assembled together to form the composite tube billet. Secondly, the Cu/Al clad sheets are rolled to 1.5 mm after multipass cold rolling and then annealed at different temperatures of 450 °C, 550 °C, and 600 °C, which are designated as IFR-450, IFR-550, and IFR-600, respectively. Finally, these annealed samples are further rolled to 0.5 mm. To observe the microstructure of the bonding interface and peeling surface, scanning electron microscopy (SEM) was performed using an FEI Quanta 250F (FEI Company, Hillsboro, OR, USA) device with an acceleration voltage of 20 kV and equipped with energy dispersive spectrometry (EDS, Oxford Instruments Group, London, UK). The Al-Cu To observe the microstructure of the bonding interface and peeling surface, scanning electron microscopy (SEM) was performed using an FEI Quanta 250F (FEI Company, Hillsboro, OR, USA) device with an acceleration voltage of 20 kV and equipped with energy dispersive spectrometry (EDS, Oxford Instruments Group, London, UK). The Al-Cu phase diagram is plotted by the software 'Binary Alloy Phase Diagrams' (ASM International, Almere, the Netherlands). The crystalline phases on the peeling surface are detected by D8 ADVANCE X-ray diffraction (XRD, ceramic X-ray tube, Brooke company, karlsruhe, Germany), using Cu K α radiation (λ ≈ 1.54 Å), equipped with a Linx one-dimensional array detector (Liwei wisdom international Co., Ltd., HongKong, China). The angle step, time step, and scanning range are 0.01 • , 0.1 s, and 10-100 • , respectively. T-peel tests on a TH5000 universal testing machine (Xintianhui Electronic Technology Co., Ltd., Yangzhou, China) are adopted to test the interfacial bonding strength of Cu/Al clad sheets under a crosshead speed of 1 mm/min.
Interfacial Microstructure
Prior research indicated that the peeling strength of Cu/Al clad sheets was mainly determined by the bonding strength between the SUS304 interlayer and the Cu/Al matrix [15]. Figure 2 exhibits the interfacial microstructure of IFR-450 and the corresponding EDS mapping results. A flat bonding interface without visible cracks between the SUS304 interlayer and Cu/Al matrix is formed (Figure 2a), and the IMCs are seldom formed due to the lower intermediate annealing temperature [16], which is proved by the EDS mapping results of the Al/SUS304/Cu bonding interface (Figure 2b-d). The position of the SUS304 interlayer is marked by the distribution of the Fe and Cr elements (Figure 2e,f). In the rolling process, the SUS304 fragments are squeezed into the Cu/Al matrix. Under this situation, the interfacial bonding strength is mainly contributed by the mechanical joggles [17]. phase diagram is plotted by the software 'Binary Alloy Phase Diagrams' (ASM International, Almere, the Netherlands). The crystalline phases on the peeling surface are detected by D8 ADVANCE X-ray diffraction (XRD, ceramic X-ray tube, Brooke company, karlsruhe, Germany), using Cu Kα radiation (λ ≈ 1.54 Å), equipped with a Linx one-dimensional array detector (Liwei wisdom international Co., LTD, HongKong, China). The angle step, time step, and scanning range are 0.01°, 0.1 s, and 10-100°, respectively. T-peel tests on a TH5000 universal testing machine (Xintianhui Electronic Technology Co., LTD., Yangzhou, China) are adopted to test the interfacial bonding strength of Cu/Al clad sheets under a crosshead speed of 1 mm/min.
Interfacial Microstructure
Prior research indicated that the peeling strength of Cu/Al clad sheets was mainly determined by the bonding strength between the SUS304 interlayer and the Cu/Al matrix [15]. Figure 2 exhibits the interfacial microstructure of IFR-450 and the corresponding EDS mapping results. A flat bonding interface without visible cracks between the SUS304 interlayer and Cu/Al matrix is formed (Figure 2a), and the IMCs are seldom formed due to the lower intermediate annealing temperature [16], which is proved by the EDS mapping results of the Al/SUS304/Cu bonding interface (Figure 2b-d). The position of the SUS304 interlayer is marked by the distribution of the Fe and Cr elements (Figure 2e,f). In the rolling process, the SUS304 fragments are squeezed into the Cu/Al matrix. Under this situation, the interfacial bonding strength is mainly contributed by the mechanical joggles [17]. interlayer and Al matrix [18], as shown in Figure 3a, which is also strongly proved by the corresponding EDS mapping results (Figure 3b). A small increase in the annealing temperature may result in a significant increment in the diffusion coefficient [19] due to their exponential relationship ( Figure 3c). In contrast, the chemical compound type formed between the SUS304 interlayer and Cu matrix [20] is the solid solution ( Figure 3d). The position of the SUS304 interlayer is marked by the distribution of Fe and Cr elements (Figure 3e,f). For Cu/Al clad sheets without an interlayer, the Cu-Al IMCs are broken into fragments in the rolling process [6]. Nevertheless, the IMCs formed between the SUS304 interlayer and Al matrix can retain their continuity. The existence of an SUS304 interlayer with a weak deformation capacity can significantly inhibit the crush of Al-SUS304 IMCs and enhance the interfacial bonding strength. When the intermediate annealing temperature increases to 550 °C, obvious IMCs with an average thickness of 10 μm are formed at the bonding interface of the SUS304 interlayer and Al matrix [18], as shown in Figure 3a, which is also strongly proved by the corresponding EDS mapping results (Figure 3b). A small increase in the annealing temperature may result in a significant increment in the diffusion coefficient [19] due to their exponential relationship (Figure 3c). In contrast, the chemical compound type formed between the SUS304 interlayer and Cu matrix [20] is the solid solution ( Figure 3d). The position of the SUS304 interlayer is marked by the distribution of Fe and Cr elements ( Figure 3e,f). For Cu/Al clad sheets without an interlayer, the Cu-Al IMCs are broken into fragments in the rolling process [6]. Nevertheless, the IMCs formed between the SUS304 interlayer and Al matrix can retain their continuity. The existence of an SUS304 interlayer with a weak deformation capacity can significantly inhibit the crush of Al-SUS304 IMCs and enhance the interfacial bonding strength. The interfacial microstructure of IFR-600 is exhibited in Figure 4. As the intermediate annealing temperature reaches 600 °C, the clear Cu/SUS304/Al interface disappears and is replaced by the chaotic composite interfaces (Figure 4a). According to the Al-Cu phase diagram ( Figure 5), the liquid phase will be formed at the bonding interface with the an- The interfacial microstructure of IFR-600 is exhibited in Figure 4. As the intermediate annealing temperature reaches 600 • C, the clear Cu/SUS304/Al interface disappears and is replaced by the chaotic composite interfaces (Figure 4a). According to the Al-Cu phase diagram ( Figure 5), the liquid phase will be formed at the bonding interface with the annealing temperature of 600 • C (red line in Figure 5). In this case, the Cu/SUS304/Al interface is destroyed by the disturbing force induced from the formation process of the liquid phase, resulting in the exfoliation of SUS304 fragments from the Cu matrix and being involved in the liquid phases. Due to the relatively high annealing temperature, the diffusional degree of IFR-600 is much larger than those in IFR-450 and IFR-550 (Figures 2 and 3), which is also proven by our prior research [21]. Based on the EDS mapping results (Figure 4b-d), it can be observed that the SUS304 fragments are surrounded by thick Cu-Al IMCs. The position of the SUS304 interlayer is marked by the distribution of Fe and Cr elements (Figure 4e,f). Moreover, the hardness of Cu-Al IMCs is higher than the Cu/Al matrix, leading to the increase in the thickness reduction of the SUS304 interlayer in the further cold rolling process. For IFR-600, the residual thickness of the SUS304 interlayer is just 5.9 µm, around a third of those in IFR-450 and IFR-550.
nealing temperature of 600 °C (red line in Figure 5). In this case, the Cu/SUS304/Al interface is destroyed by the disturbing force induced from the formation process of the liquid phase, resulting in the exfoliation of SUS304 fragments from the Cu matrix and being involved in the liquid phases. Due to the relatively high annealing temperature, the diffusional degree of IFR-600 is much larger than those in IFR-450 and IFR-550 (Figures 2 and 3), which is also proven by our prior research [21]. Based on the EDS mapping results (Figure 4b-d), it can be observed that the SUS304 fragments are surrounded by thick Cu-Al IMCs. The position of the SUS304 interlayer is marked by the distribution of Fe and Cr elements (Figure 4e,f). Moreover, the hardness of Cu-Al IMCs is higher than the Cu/Al matrix, leading to the increase in the thickness reduction of the SUS304 interlayer in the further cold rolling process. For IFR-600, the residual thickness of the SUS304 interlayer is just 5.9 μm, around a third of those in IFR-450 and IFR-550.
Elemental Diffusion across the Bonding Interface
The EDS line results across the Cu/SUS304/Al interface are provided in Figure 6. After roll cladding, the thickness reduction in the SUS304 interlayers in IFR-450, IFR-550, and IFR-600 are 11.6 μm, 10.2 μm, and 24.1 μm, respectively. In general, the thickness reduction in the SUS304 interlayers should decrease with the increment of intermediate
Elemental Diffusion across the Bonding Interface
The EDS line results across the Cu/SUS304/Al interface are provided in Figure 6. After roll cladding, the thickness reduction in the SUS304 interlayers in IFR-450, IFR-550, and IFR-600 are 11.6 µm, 10.2 µm, and 24.1 µm, respectively. In general, the thickness reduction in the SUS304 interlayers should decrease with the increment of intermediate annealing temperature due to the aggravate softening of the Cu/Al matrix. Nevertheless, the formation of liquid phases at the bonding interface leads to the SUS304 interlayer being surrounded by the Cu-Al IMCs and a large thickness reduction in the SUS304 interlayer ( Figure 4). On the other hand, the interfacial atomic diffusion is also enhanced by the increase in annealing temperature, which may also increase the corrosion resistance of the clad sheets [22]. Compared with IFR-450, the width of the diffusional layer for the Al/SUS304 interface and Cu/SUS304 interface in IFR-550 is increased from 2.2 µm to 20.1 µm and from 1.1 µm to 3.5 µm (Figure 6a,b). As for IFR-600, the SUS304 fragments are totally surrounded by the thick Cu-Al IMCs, which is strongly proven by the EDS line results (Figure 6c).
Peeling Surface of the Clad Sheets
The XRD patterns of the peeling surface of IFR samples are provided in Figure 7. It can be seen that the intermediate annealing temperature has a significant impact on the type and content of crystalline phases on the peeling surface. The main crystalline phases in the peeling surface of the Cu side for IFR-450 and IFR-550 are Cu, Al, Cu9Al4, and CuAl2. The interfacial elemental diffusion is enhanced by the increase in the intermediate annealing temperature, resulting in the increment of Cu-Al IMCs, especially the Cu9Al4 phase (Figure 7a). In addition to the Al, Fe, CuAl2 phases, a new Al7Cu2Fe phase is observed at the Al side of IFR-550 ( Figure 7b). As the intermediate annealing temperature reaches 600 °C, the main crystalline phases in the peeling surface of the Cu side for IFR-600 are transformed into CuAl2, Al, Cu9Al4, and FeAl. As to the Al side of IFR-600, the main crystalline phases are CuAl2, Fe3Al, and Al.
Peeling Surface of the Clad Sheets
The XRD patterns of the peeling surface of IFR samples are provided in Figure 7. It can be seen that the intermediate annealing temperature has a significant impact on the type and content of crystalline phases on the peeling surface. The main crystalline phases in the peeling surface of the Cu side for IFR-450 and IFR-550 are Cu, Al, Cu 9 Al 4, and CuAl 2 . The interfacial elemental diffusion is enhanced by the increase in the intermediate annealing temperature, resulting in the increment of Cu-Al IMCs, especially the Cu 9 Al 4 phase (Figure 7a). In addition to the Al, Fe, CuAl 2 phases, a new Al 7 Cu 2 Fe phase is observed at the Al side of IFR-550 ( Figure 7b). As the intermediate annealing temperature reaches 600 • C, the main crystalline phases in the peeling surface of the Cu side for IFR-600 are transformed into CuAl 2 , Al, Cu 9 Al 4, and FeAl. As to the Al side of IFR-600, the main crystalline phases are CuAl 2 , Fe 3 Al, and Al.
in the peeling surface of the Cu side for IFR-450 and IFR-550 are Cu, Al, Cu9Al4, and CuAl2. The interfacial elemental diffusion is enhanced by the increase in the intermediate annealing temperature, resulting in the increment of Cu-Al IMCs, especially the Cu9Al4 phase (Figure 7a). In addition to the Al, Fe, CuAl2 phases, a new Al7Cu2Fe phase is observed at the Al side of IFR-550 ( Figure 7b). As the intermediate annealing temperature reaches 600 °C, the main crystalline phases in the peeling surface of the Cu side for IFR-600 are transformed into CuAl2, Al, Cu9Al4, and FeAl. As to the Al side of IFR-600, the main crystalline phases are CuAl2, Fe3Al, and Al. The morphologies of the peeling surface for IFR-450 and the chemical compositions of corresponding crystalline phases are shown in Figure 8 and Table 1. There are two entirely different peeling morphologies on the Cu side and Al side. The ridge-shaped mixture of the Al matrix and Cu-Al IMCs and invaginated Cu matrix can be observed on the Cu side (Figure 8a), which is identified by the EDS results (Table 1). Moreover, many distinct wrinkles are formed at the invaginated Cu matrix, which resulted from the severe shear deformation induced by the plastic difference of the Cu matrix and SUS304 interlayer ( Figure 8c). As to the Al side, many SUS304 fragments with uneven sizes are tightly embedded into the Al matrix, which are formed through the random fracture of SUS304 foil in the roll cladding process (Figure 8b). Moreover, these SUS304 fragments' gaps are filled with a mixture of reticulated Al and Cu-Al IMCs (Figure 8d). The morphologies of the peeling surface for IFR-450 and the chemical compositions of corresponding crystalline phases are shown in Figure 8 and Table 1. There are two entirely different peeling morphologies on the Cu side and Al side. The ridge-shaped mixture of the Al matrix and Cu-Al IMCs and invaginated Cu matrix can be observed on the Cu side (Figure 8a), which is identified by the EDS results (Table 1). Moreover, many distinct wrinkles are formed at the invaginated Cu matrix, which resulted from the severe shear deformation induced by the plastic difference of the Cu matrix and SUS304 interlayer ( Figure 8c). As to the Al side, many SUS304 fragments with uneven sizes are tightly embedded into the Al matrix, which are formed through the random fracture of SUS304 foil in the roll cladding process (Figure 8b). Moreover, these SUS304 fragments' gaps are filled with a mixture of reticulated Al and Cu-Al IMCs (Figure 8d). The morphologies of the peeling surface for IFR-550 and the chemical compositions of corresponding crystalline phases are shown in Figure 9 and Table 2. The crystalline phases in the peeling surface of the Cu side for IFR-550 are folded Cu matrix, reticulate Al matrix, and the Cu9Al4 phase with obvious cracks (Figure 9a), which is consistent with the XRD results. Compared with IFR-450, the content of the Cu9Al4 phase exhibits a significant increase. The increase in the intermediate annealing temperature leads to the formation of The morphologies of the peeling surface for IFR-550 and the chemical compositions of corresponding crystalline phases are shown in Figure 9 and Table 2. The crystalline phases in the peeling surface of the Cu side for IFR-550 are folded Cu matrix, reticulate Al matrix, and the Cu 9 Al 4 phase with obvious cracks (Figure 9a), which is consistent with the XRD results. Compared with IFR-450, the content of the Cu 9 Al 4 phase exhibits a significant increase. The increase in the intermediate annealing temperature leads to the formation of a thick Cu-Al IMCs layer, which is brittle and will be broken in the further rolling process (Figure 9c). SUS304 fragments can still be clearly observed on the peeling surface of the Al side for IFR-550 and their height is smaller than the Cu-Al IMCs (Figure 9b). Furthermore, obvious wrinkles appear on the surface of the SUS304 fragments and a new Al 7 Cu 2 Fe phase is detected on the edge of the SUS304 fragments (Figure 9d). Those above phenomena indicate that the interfacial elemental diffusion has been enhanced by increasing the intermediate annealing temperature. The morphologies of the peeling surface for IFR-600 and the chemical compositions of corresponding crystalline phases are shown in Figure 10 and Table 3. The main crystalline phases on the peeling surface of the Cu side for IFR-600 are the CuAl2 phase with a coarse surface and the vitreous FeAl phase (Figure 10a,c). Compared with IFR-450 and IFR-550, the Cu matrix is undetectable for IFR-600, which implies the transformation of the crack propagation path from the initial Cu/SUS304 interface to the Cu-Al IMCs and Fe-Al IMCs layers. High intermediate annealing temperature results in the formation of an unduly thick IMCs layer and Kirkendall voids, which promotes crack propagation and reduces the interfacial bonding interface [23,24]. Similarly, the SUS304 fragments cannot be observed on the peeling surface of the Al side, only some castle peaks of the CuAl2 phase and rock-like Fe3Al phase (Figure 10b,d). Furthermore, the reticular Al matrix also disappears. The above results also prove that the interfacial cracks are propagated along the Cu-Al IMCs and Fe-Al IMCs layer and the bonding strength of the Al matrix is higher than these IMCs. The morphologies of the peeling surface for IFR-600 and the chemical compositions of corresponding crystalline phases are shown in Figure 10 and Table 3. The main crystalline phases on the peeling surface of the Cu side for IFR-600 are the CuAl 2 phase with a coarse surface and the vitreous FeAl phase (Figure 10a,c). Compared with IFR-450 and IFR-550, the Cu matrix is undetectable for IFR-600, which implies the transformation of the crack propagation path from the initial Cu/SUS304 interface to the Cu-Al IMCs and Fe-Al IMCs layers. High intermediate annealing temperature results in the formation of an unduly thick IMCs layer and Kirkendall voids, which promotes crack propagation and reduces the interfacial bonding interface [23,24]. Similarly, the SUS304 fragments cannot be observed on the peeling surface of the Al side, only some castle peaks of the CuAl 2 phase and rock-like Fe 3 Al phase (Figure 10b,d). Furthermore, the reticular Al matrix also disappears. The above results also prove that the interfacial cracks are propagated along the Cu-Al IMCs and Fe-Al IMCs layer and the bonding strength of the Al matrix is higher than these IMCs. Figure 11 shows the peeling strength curves and average peeling strength of the Cu/Al clad sheets with the SUS304 interlayer annealed at different temperatures. To highlight the strengthening effect of the SUS304 interlayer, the peeling curve of the clad sheets without the SUS304 interlayer (SR-0) is also introduced. With the increase in the intermediate annealing temperature, the peeling strength of the clad sheets firstly increase and then decrease (Figure 10a). Compared with IFR-450, the peeling strength of IFR-550 is increased by 11%, from 30.9 N/mm to 34.3 N/mm. Moreover, the fluctuation of the peeling strength curve for IFR-550 is larger than that of IFR-450, indicating the difference in the bonding strength among the SUS304/Cu interface, Cu-Al IMCs layer, and Al matrix being enlarged. Figure 11 shows the peeling strength curves and average peeling strength of the Cu/Al clad sheets with the SUS304 interlayer annealed at different temperatures. To highlight the strengthening effect of the SUS304 interlayer, the peeling curve of the clad sheets without the SUS304 interlayer (SR-0) is also introduced. With the increase in the intermediate annealing temperature, the peeling strength of the clad sheets firstly increase and then decrease (Figure 10a). Compared with IFR-450, the peeling strength of IFR-550 is increased by 11%, from 30.9 N/mm to 34.3 N/mm. Moreover, the fluctuation of the peeling strength curve for IFR-550 is larger than that of IFR-450, indicating the difference in the bonding strength among the SUS304/Cu interface, Cu-Al IMCs layer, and Al matrix being enlarged.
Discussions
According to the above results in Figures 7-10, it can be deduced that the interfacial cracks propagate along the Cu/SUS304 interface, Al matrix, and Cu-Al IMCs layer for IFR-450 and IFR-550. While for IFR-600, the propagation path of interfacial cracks is changed, which is along the Al/SUS304 interface, Al matrix, and Cu-Al IMCs layer. The high intermediate annealing temperature leads to an excessive formation of Al-Fe IMCs and Cu-Al IMCs, resulting in a decrease in the bonding strength for the Al/SUS304 interface and Cu/Al interface. Moreover, Chen et al. [25] indicated that the bonding strength of the Cu/steel interface would be enhanced by increasing the annealing temperature in the range of 400 • C-1000 • C. Thus, the interfacial cracks are more likely formed and prolongated along the Al/SUS304 interface instead of the Cu/SUS304 interface for IFR-600.
The diffusion process across the bonding interface of Cu/Al clad sheets has been verified to conform to the vacancy diffusion [26]. The energy input for interfacial atomic diffusion from two aspects: one is the thermal energy from the intermediate annealing treatment; the other is the shear strain energy provided by the ductility difference between the SUS304 interlayer and Cu/Al matrix, as shown in Figure 12. Our previous research has indicated that increasing the thickness of the SUS304 interlayer can improve the interfacial bonding strength by enlarging the shear strain energy at the bonding interface [15]. In this study, the interfacial atomic diffusion, as well as the interfacial bonding strength, is enhanced by raising the intermediate annealing temperature. While for IFR-600, the peeling strength sharply decreases to 8.3 N/mm, which is even lower than the clad sheets without the SUS304 interlayer, as shown in Figure 11. A high intermediate annealing temperature leads to the formation of unduly thick IMCs layers, which promote the crack propagation and decrease the interfacial bonding strength.
Discussions
According to the above results in Figures 7-10, it can be deduced that the interfacial cracks propagate along the Cu/SUS304 interface, Al matrix, and Cu-Al IMCs layer for IFR-450 and IFR-550. While for IFR-600, the propagation path of interfacial cracks is changed, which is along the Al/SUS304 interface, Al matrix, and Cu-Al IMCs layer. The high intermediate annealing temperature leads to an excessive formation of Al-Fe IMCs and Cu-Al IMCs, resulting in a decrease in the bonding strength for the Al/SUS304 interface and Cu/Al interface. Moreover, Chen et al. [25] indicated that the bonding strength of the Cu/steel interface would be enhanced by increasing the annealing temperature in the range of 400 ℃-1000 ℃. Thus, the interfacial cracks are more likely formed and prolongated along the Al/SUS304 interface instead of the Cu/SUS304 interface for IFR-600.
The diffusion process across the bonding interface of Cu/Al clad sheets has been verified to conform to the vacancy diffusion [26]. The energy input for interfacial atomic diffusion from two aspects: one is the thermal energy from the intermediate annealing treatment; the other is the shear strain energy provided by the ductility difference between the SUS304 interlayer and Cu/Al matrix, as shown in Figure 12. Our previous research has indicated that increasing the thickness of the SUS304 interlayer can improve the interfacial bonding strength by enlarging the shear strain energy at the bonding interface [15]. In this study, the interfacial atomic diffusion, as well as the interfacial bonding strength, is enhanced by raising the intermediate annealing temperature. While for IFR-600, the peeling strength sharply decreases to 8.3 N/mm, which is even lower than the clad sheets without the SUS304 interlayer, as shown in Figure 11. A high intermediate annealing temperature leads to the formation of unduly thick IMCs layers, which promote the crack propagation and decrease the interfacial bonding strength.
Conclusions
In this work, the influence of an intermediate annealing temperature on the interfacial microstructure, elemental diffusion, and peeling strength of Cu/Al clad sheet containing a SUS304 interlayer is investigated. The main conclusions are as follows: (1) The interfacial atomic diffusion is significantly enhanced by increasing the intermediate annealing temperature. Nevertheless, after a high-temperature annealing treatment (IFR-600), a liquid phase is formed at the bonding interface and the clear Cu/SUS304/Al interface in IFR-450 and IFR-550 is replaced by the chaotic composite interfaces.
(2) For IFR-450 and IFR-550, the interfacial crack is propagated along the Cu/SUS304 interface, Cu-Al IMCs layer, and Al matrix. Compared with IFR-450, the interfacial bonding strength for IFR-550 is improved by 11%, from 30.9 N/mm to 34.3 N/mm, which is proven by the obvious wrinkles on the surface of SUS304 fragments and the formation of a new Al7Cu2Fe phase.
Conclusions
In this work, the influence of an intermediate annealing temperature on the interfacial microstructure, elemental diffusion, and peeling strength of Cu/Al clad sheet containing a SUS304 interlayer is investigated. The main conclusions are as follows: (1) The interfacial atomic diffusion is significantly enhanced by increasing the intermediate annealing temperature. Nevertheless, after a high-temperature annealing treatment (IFR-600), a liquid phase is formed at the bonding interface and the clear Cu/SUS304/Al interface in IFR-450 and IFR-550 is replaced by the chaotic composite interfaces. (2) For IFR-450 and IFR-550, the interfacial crack is propagated along the Cu/SUS304 interface, Cu-Al IMCs layer, and Al matrix. Compared with IFR-450, the interfacial bonding strength for IFR-550 is improved by 11%, from 30.9 N/mm to 34.3 N/mm, which is proven by the obvious wrinkles on the surface of SUS304 fragments and the formation of a new Al 7 Cu 2 Fe phase. Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement:
The data presented in this study are available on request from the corresponding author. | 2022-03-16T15:24:22.303Z | 2022-03-01T00:00:00.000 | {
"year": 2022,
"sha1": "62c75b27450c68feb7695731e8d604859dee0d47",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/1996-1944/15/6/2119/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "4810fe0cf8ca7bfe69ccda11b0b897d2ffb80313",
"s2fieldsofstudy": [
"Materials Science"
],
"extfieldsofstudy": [
"Medicine"
]
} |
257792210 | pes2o/s2orc | v3-fos-license | Hepatocellular carcinoma: State of the art diagnostic imaging
Primary liver cancer is the fourth most common malignancy worldwide, with hepatocellular carcinoma (HCC) comprising up to 90% of cases. Imaging is a staple for surveillance and diagnostic criteria for HCC in current guidelines. Because early diagnosis can impact treatment approaches, utilizing new imaging methods and protocols to aid in differentiation and tumor grading provides a unique opportunity to drastically impact patient prognosis. Within this review manuscript, we provide an overview of imaging modalities used to screen and evaluate HCC. We also briefly discuss emerging uses of new imaging techniques that offer the potential for improving current paradigms for HCC characterization, management, and treatment monitoring.
INTRODUCTION
Hepatocellular carcinoma (HCC) is the most common liver malignancy, accounting for 90% of liver tumors, and a leading cause of mortality worldwide [1,2]. Global risk factors for HCC include cirrhosis, existing in up to 90% of new cases [3], or patients with long-standing liver infections such as viral hepatitis B and C [4]. East Asia and sub-Saharan Africa account for greater than 80% of cases, and incidence within the United States continues to rise [3,4]. Unfortunately, many patients are diagnosed in the advanced stages of the disease, which emphasizes the importance of early detection and surveillance [5]. Surveillance and early tumor detection using ultrasonography is recommended by the American Association for the study of Liver Diseases (AASLD) for high-risk populations driven by ultrasoundguided imaging. Unlike other malignancies, HCC expresses distinctive characteristics that can be diagnosed based on imaging features alone, without the need for confirmation from tissue sampling [6]. In this review, we provide a summary of diagnostic criteria and imaging modalities used to detect and stage HCC, as well as emerging methods to further assist in the surveillance and characterization of the disease.
Conventional ultrasound
Tumor burden can significantly impact management, where patients with small, localized tumors can receive curative methods such as liver transplantation, resection, or locoregional therapies. On the other hand, treatment options are limited in patients with HCC displaying more aggressive features (e.g., extrahepatic metastases, multifocal tumors, and vascular invasion). As a result, early detection may offer a significant benefit in select patients. Patient populations recommended for HCC surveillance differs between AASLD[7], the European Association for the Study of the Liver (EASL)[8] and the Asian Pacific Association for the Study of the Liver (APASL) [9], but largely consist of adults with cirrhosis or patients with hepatitis B virus (HBV) [10]. Across surveillance recommendations, biannual abdominal ultrasonography (Figure 1) is the standard modality for HCC detection with major advantages including accessibility, cost-effectiveness, and safety [10].
Surveillance programs typically consist of ultrasound examination performed at either 6-or 12-mo intervals. Across all stages of HCC, ultrasound detection carries 84% sensitivity. A number of investigations have found a mortality and cost-benefit of biannual ultrasonography for imaging surveillance of HCC [10][11][12]. For example, a study in HBV-infected patients in China found a 37% reduction in mortality in those receiving biannual ultrasonography examinations compared to the control group [12]. Suboptimal visualization poses a major limitation for ultrasound screening. Poor visualization and sonographic sensitivity to HCC lesions can be caused by a number of different extrinsic factors such as morbid obesity, patient inability to suspend respiration, obscured portions of the liver by bowel gas or rib shadowing, and intrinsic factors such as hepatic steatosis or fibrosis causing parenchymal heterogenicity [13]. A recent investigation reported approximately 20% of scans were inadequate to exclude liver lesions [14]. Detecting smaller nodules (< 2 cm) appears to be a major limitation of ultrasound, with studies reporting detection rates as low as approximately 28% [15]. In response, the American College of Radiology (ACR) screening guidelines have recommended the use of systematic documentation and scoring for visualization (A: No or minimal limitations; B: Moderate limitations; C: Severe limitations). The use of tumor biomarkers, such as alpha-fetoprotein (AFP), in concert with ultrasound examination appears to have an additive effect on detection rate [16]. For example, a recent meta-analysis of prospective studies found the sensitivity for ultrasound alone to detect any stage of HCC was 78% compared to 97% when adding AFP [17]. Interestingly, the synergistic impact of combining ultrasound with AFP also exists for detection of earlier, smaller nodules (45% vs 63% sensitivity, respectively) [17] which is particularly salient given the limitations of ultrasound and earlier stages of HCC. This is not without controversy, however. The 2018 AASLD guidelines do not designate any preferences between adjunctive use of AFP while the 2017 APASL guidelines recommend the combination of ultrasound and AFP [18]. In contrast, the 2018 EASL guidelines discourage the use of ultrasound with AFP for 6-mo HCC surveillance, citing concerns of false-positives in the setting of active liver inflammation with infection[18]. While the objective of this review is to elucidate the latest advancements in technological imaging for the screening and diagnosis of HCC, it is important to note the efficacy of detection can be limited due to multifactorial screening challenges. In fact, less than 1 in 5 patients with cirrhosis receive surveillance screening for HCC [19]. Previous reviews have extensively examined the numerous challenges encountered during the screening process, including the inability to properly stratify high-risk patients, the presence of socio-economic and logistical impediments to accessing healthcare, as well as training and detection limitations using conventional imaging techniques, as previously discussed. One of the most common attributable factors to surveillance underuse includes lack of surveillance orders or unrecognized cirrhosis [5]. Therefore, strategies to improve education and integrating primary care providers in surveillance efforts can have a drastic and meaningful effect on rates of patients undergoing HCC screening [20]. The implementation of patient-centered outreach programs such as reminder protocols or embedding best-practice advisories within the electronic health record may be solutions to improve barriers of patients undergoing surveillance [5,20]. The decision to select which patients to screen also been discussed and studies have developed scoring systems across different risk factors (e.g. hepatitis or cirrhosis) to refine and improve risk stratification have been proposed [21]. Other methods have also focused on improving surveillance outcomes and detection rates, such as utilizing serological biomarkers (e.g. AFP) either as a single screening modality or in concert with imaging to improve sensitivity, at the potential cost of increased rates of false positivity. The use of biomarkers may also be especially helpful for smaller HCCs, not easily visible with ultrasound[21].
Liver imaging reporting and data systems diagnostic reporting system
Established by the ACR, standardized methods for imaging interpretation and reporting are defined using the liver imaging reporting and data systems (LI-RADS). The application of the LI-RADS diagnostic algorithm was initially developed for computed tomography (CT)/magnetic resonance imaging (MRI). LI-RADS is subdivided into 8 categories and ranges in greater probability of malignancy from LR-1 to LR-5 with additional categories including LI-RADS M (LR-M) (probably or definitely malignant but not HCC specific), LR-definite tumor in vein, and LR-cannot be categorized (NC) [13,22]. Observation of lesions categorized in LR-5 is designated as almost certainly HCC, with a systematic review of 454 studies reporting 94% of LR-5 lesions confirmed to be HCC and 97% malignant[23]. The Organ Procurement and Transplantation Network (OPTN) is another diagnostic criterion established by National Organ Transplant Act, with subcategories ranging from class 0-5. Specificities are similar between LI-RADSv2018 LR-5 and OPTN class 5[24]. However, inter-reader agreement and sensitivity of LI-RADS (sensitivity: 63.9%) is higher than that of OPTN (sensitivity: 53.6%)[24].
Cross-sectional imaging: Multiphase CT and MRI
Standard recommendations for HCC diagnosis include multiphase CT or MRI which are beneficial modalities for highlighting unique features of HCC (Figures 2 and 3). Physiological differences in blood perfusion between hepatocarcinogenic lesions and non-neoplastic tissue display distinguishing differences in imaging characteristics using multiphasic contrast examinations [25]. Phases consist of late hepatic arterial (20-40 s), portal venous (60-90 s), and delayed (3-5 min). Late arterial phase is useful for detecting hypervascular lesions with HCC lesions characteristically enhancing relative to surrounding liver parenchyma. Arterial lesion enhancement can be appreciated within lesions as small as 1 cm. Within the portal venous and delayed phases, washout or hypointensity is commonly observed for HCC lesions [26]. During the delayed or equilibrium phase, other characteristics of HCC such as capsule features (e.g. lesion washout with pseudocapsule enhancement) and mosaic architecture can be visualized [26]. The introduction of gadolinium-based contrast agents (Gadobenate dimeglumine and gadoxetate acid) may aid LI-RADS categorization. These agents are taken up by hepatocytes of normal liver parenchyma and there is little uptake in non-functioning or dysfunctional hepatocytes, such as the case for HCC. Gadolinium-based agents function similarly to extracellular agents, but can aid in the diagnosis of lesions with atypical features (e.g. without washout, arterial hyperenhancement) or distinguish HCC from pseudolesions [27,28]. For example, these agents permit an additional postcontrast hepatobiliary phase, which will display a majority of HCC lesions (90%-95%) as hypointense relative to surrounding hyperintense liver parenchyma [27,29]. MRI is recommended for staging of HCC disease given that some reports have estimated CT to underestimate 52% of cases[1,2]. MRI also has superior diagnostic efficiency to CT in the detection of small (≤ 3cm) lesions [30]. However, CT is more readily available than MRI, and limitations to using MRI including greater costs and technical complexity make CT a complementary diagnostic alternative [31,32]. A report showed that the combined use of CT/MRI provides better diagnostic accuracy in characterizing liver lesions using LI-RADS (91.29%) than MRI (85.37%) or CT (67.6%) alone, but combined protocols should be limited to difficult or uncertain cases in order to warrant use[32].
Contrast-enhanced ultrasound
In recent years, there has been an emerging use of contrast-enhanced ultrasound (CEUS) for the evaluation of focal liver lesions (Figure 4). CEUS combines the benefit of accessible, non-invasive assessment without ionizing radiation as well as improvements in temporal resolution. Given some of
CT and MR perfusion
The acquisition of sequential imaging combined with IV contrast administration by multidetector can permit the assessment of tissue perfusion. The distribution of contrast media between the intravascular and interstitial compartments depends on the extent of blood flow and capillary permeability. Using kinetic modules, perfusion parameters across liver tissue can be calculated (blood flow, volume, permeability, hepatic arterial perfusion, portal venous perfusion, perfusion index, slope of increase/ decrease). Quantitative color-graded perfusion maps can be used to localize lesions with abnormal tissue perfusion with a high degree of spatial resolution [39][40][41][42]. Tumor angiogenesis mediates differences in blood supply between normal liver parenchyma and HCC [42]. Quantitative parameters from CT perfusion, such as hepatic perfusion index, carries high sensitivity and specificity (≥ 99%) for HCC detection in patients with cirrhosis[43]. Early HCC lesions or hepatocellular nodules will demonstrate increased arterial supply as hepatocarcinogenesis progresses, reflected in increased hepatic arterial perfusion and perfusion indices [40]. Other common focal liver lesions can exhibit unique CT perfusion behavior and therefore be used to identify HCC from hemangiomas, liver metastases, and arterioportal shunts[44]. Bai et al [45] reported histopathological features such as microvascular density, a poor prognostic factor for HCC, is correlated with multiple perfusion parameters[45]. A major prognostic factor for HCC, microvascular density has been shown to be correlated with multiple CT perfusion parameters.
Similar to CT perfusion, dynamic contrast-enhanced MRI (DCE-MRI) permits the quantification of perfusion characteristics of liver lesions [46]. This approach is accomplished using gadolinium IV contrast administration followed by image acquisition with high temporal resolution and kinetic modules to quantify contrast distribution perfusion to reflect focal perfusion differences [47]. DCE-MRI perfusion parameters have highlighted unique physiological characteristics in HCC lesions, including increased arterial hepatic blood flow, arterial fraction, and lower portal hepatic blood flow, compared to normal liver parenchyma [46]. Pahwa et al [48] reported, arterial fraction and distribution volume are high in HCC and metastatic lesions compared to normal liver parenchyma [48]. Metastatic lesions may also be distinguished from HCC and normal liver parenchyma by the perfusion parameter mean transit time [48]. By evaluating tumor vascularity pre-and post-treatment, DCE-MRI has shown efficacy for treatment monitoring (e.g. following administration of anti-angiogenic agents, transarterial chemoembolization or radiotherapy) and predicting survival outcomes for HCC patients [49].
Elastography
Elastography is an imaging method to quantify mechanical properties, notably stiffness, to evaluate focal fibrotic cirrhotic changes ( Figure 5). Either MRI or ultrasound, coupled with a device that generates low frequency vibrations (i.e. shear waves) and wave propagation, can be quantified in order to calculate levels of stiffness in a focal area of interest [50]. First introduced with ultrasound, there are multiple elastography methods which include transient elastography, point shear wave, twodimensional sheer wave, and quasi-static elastography. Ultrasound elastography has shown to provide satisfactory sensitivity and specificity for identifying histological stages of severe fibrosis (sensitivity: 81.9%, specificity: 84.7%) and cirrhosis (sensitivity: 84.8%, specificity 87.5%) [51]. MRI elastography may also aid to differentiate focal liver lesions. For example, malignant tumors have been reported to have greater levels of stiffness relative to benign lesions, focal fibrotic regions, and normal liver parenchyma [52].
Evaluation of liver stiffness may also offer a prognostic biomarker for determining risk of HCC development and survival. A meta-analysis of 9 studies by Singh et al [53] reported increased liver stiffness is associated with an elevated risk of HCC [53]. A more recent metanalysis of 1735 patients reported a varied sensitivity (31%-100%) and high specificity (81%-94%) for predicting HCC development. The use of MR elastography can also be used as a biomarker to predict treatment response and tumor recurrence [54][55][56]. A prospective investigation assessed 192 patients undergoing HCC treatment (e.g. transarterial chemboembolization, ablation, or resection) found liver parenchymal stiffness to be an independent predictor of early recurrence [54]. A recent investigation also reported efficacy for predicting both early and late recurrence in 180 patients with HBV-related HCC prior to undergoing hepatectomy[56]. Qayyum et al [57] reported the use of MR elastography to evaluate stiffness changes in patients treated with immunotherapy (i.e. Pembrolizumab). HCC tumor stiffness significantly correlated with survival outcomes, including overall survival and time to disease progression, as well as intratumoral T-lymphocyte abundance [57].
T1 mapping
Similarly, T1 mapping is an MR method by which T1 relaxation time is measured and can be useful for identifying liver fibrosis. In the setting of inflammation and fibrosis, T1 relaxation time will be increased and has been used extensively to evaluate myocardial edema and scarring [58]. In fact, there is a moderate correlation between T1 relaxation time and elastography-measured stiffness [59]. T1 relaxation time can also be measured before and after administering hepatobiliary contrast agents such as GD-EOB-DTPA (Eovist ® ) to provide a more reliable, quantitative evaluation of contrast media uptake within the liver parenchyma [60]. Given that T1 relaxation is influenced by intrinsic properties of tissue, T1 mapping can overcome some of the traditional limitations of conventional MR signal intensity, which can be influenced by technical factors and imaging parameters. Combining T1 mapping and GD-EOB-DTPA has shown to be useful for identifying and classifying differentiated lesions (e.g. reduced uptake associated with increased HCC grade), and can be used to distinguish HCC from other focal liver lesions, including hepatic cysts, focal nodular hyperplasia, and hemangiomas [61][62][63]. HCCs with microvascular invasion have also shown reductions in T1 relaxation relative to lesions without evidence of microvascular invasion [64], showing promise of another imaging method for predicting prognosis.
Diffusion weighted imaging
Diffusion weighted imaging (DWI) is a non-invasive MR sequence which can characterize focal liver lesions without the need for contrast media by measuring diffusion properties of water molecules within tissues (Figure 6). Gradations of diffusion are measured using b-values, with greater values denoting more sensitivity to diffusion and higher signal intensity. These values are used to calculate and generate apparent diffusion coefficient (ADC) maps, used clinically to assess local changes in liver tissue diffusion [65]. Changes in cellularity, cell architecture and extracellular space, combined with necrosis and vascularization, can restrict diffusion and thus, HCC will appear hyperintense on DWI [65,66]. DWI has exhibited exceptionally high sensitivity and specificity for a single HCC lesion (100%) and moderate sensitivity and high specificity for multiple lesions (75% and 100%, respectively) early, smaller HCC (≤ 2 cm) is a clinically useful feature of DWI, especially when combined with contrast use. Generally, poorly differentiated lesions will exhibit lower ADC compared to well-differentiated lesions [66]. 1.01 × 10 −3 mm 2 /s yields the highest sensitivity (80%) and specificity (100%) for detecting residual HCCs 3 mo following treatment [73]. However, differentiation between benign and malignant lesions are difficult in the setting of cirrhosis, as the ADC between both lesion types exhibit considerable overlap [65]. Further, standardization for DWI sequences may be needed since different study protocols can alter ADC calculation [74].
MR spectroscopy
MR spectroscopy is an analytical technique that permits the characterization and quantification of tissue metabolite composition in vivo. For each given voxel, a plot of signal intensity and metabolites/ chemicals are expressed by their frequencies [75]. Malignant hepatic lesions have been found to have elevated choline levels relative to normal liver parenchyma. Changes in metabolite frequencies exist between healthy and cirrhotic livers, namely choline and lipid levels [76]. Choline is a component of phospholipid membranes, which increases in states of cell proliferation and carcinogenesis. Zhang et al [77] reported the diagnostic efficacy of measuring choline-containing compounds using MR spectroscopy is high for discriminating malignant and benign tumors (sensitivity: 94.3% and specificity: 93.3%) [77]. Determining ratios of choline and lipids within a given lesion has also been used to monitor treatment responses after locoregional therapy [78]. For example, prospective investigations have found choline levels to decline following TACE therapy [79,80].
Radiogenomics/radiomics
The integration of artificial intelligence and diagnostic imaging modalities has led to an exponential rise in radiogenomics or radiomics, which collectively refers to processes that aim to bridge quantitative radiologic data with immunobiological or clinical characteristics to inform prognosis or predict treatment outcomes [81]. The computing process consists of extracting quantitative features from medical images (CT, MRI or positron emission tomography) into large analyzable databases [82]. This process is carried out in multiple steps, including: (1) Defining volumes or regions of interest; (2) image segmentation (performed manually, semi-automatic or automatic tools); (3) image processing used to normalize grey-level intensities, denoise and improve data quality (e.g. signal intensity normalization, motion correction, filtering, image interpolation, and bias field correction); (4) feature extraction; and (5) statistical model building [83,84]. The final process, specific to oncology, includes generating association maps to correlate radiomic-based models with clinical outcome data, microvascular invasion, histological grade, or genomic/molecular data (e.g. immune marker expression).
Multiple studies have utilized either radiogenomic or radiomic models to diagnose and differentiate liver tumors or predict treatment efficacy for HCC [85][86][87]. For example, Lewis et al [88] reported greater prediction accuracy when combining LI-RADS and DWI-derived radiomics models using the ADC values than LI-RADS alone for distinguishing HCC from other primary liver cancers [88]. Banerjee et al [89] evaluated radiomic-models using contrast-enhanced CT to predict prognostic factors, such as microvascular invasion [89]. Radiogenomic venous invasion, an imaging biomarker, can predict microvascular invasion and correlates with lower overall and recurrence-free survival [89]. A recent meta-analysis/systematic review including 4947 patients showed promising predictive potential of radiomic models for microvascular invasion, reporting a pooled area under the curve (AUC) of 0.85, 0.87, and 0.74 across studies using CT, MR, and ultrasound-based models, respectively [90].
Given the link between gene expression and immunotherapy response, the ability to predict HCC immunoprofiles can be critical for delineating appropriate treatment. Hectors et al [91] retrospectively evaluated the relationship between MRI radiomic features (e.g. models using tumor size, enhancement ratios, fat content, ADC, texture features) and HCC immunhistochemical and genomic makers[91]. In particular, multiple relationships were found to exist between radiomic features and immunotherapy targets, cytotoxic T-lymphocyte-associated antigen 4 and programmed death 1[91]. Radiomic models have also shown promise for predicting treatment response and potential adverse outcomes after locoregional therapy use [92][93][94][95][96]. For example, MRI-based radiomics models have a reported AUC of 0.861 and 0.884 for predicting tumor response at 3-mo post TACE, evaluated using the mRECIST criterion. While larger, multi-center cohort models should be investigated in the future, the use of radiogenomics or radiomic models provide a novel method that can improve tumor grading, predicting prognosis and clinical decision-making strategies.
CONCLUSION
In summary, the use of imaging is an essential component to the diagnosis and management of HCC. Ultrasound is cost-effective modality for the screening. The utility of diagnostic modalities such as MR or CT for differentiation and grading of HCC continues to expand, especially with the advancement of new techniques and image analyses. The implementation of techniques, including elastography, T1 mapping, perfusion imaging and CEUS provide multiple unique benefits to further aid in the characterization of HCC. Other methods, such as radiomics/radiogenomics, which seek to integrate imaging data to predict prognostic risk factors and determine treatment response probability, will be a new frontier | 2023-03-29T15:08:34.383Z | 2023-03-28T00:00:00.000 | {
"year": 2023,
"sha1": "5e24fefea51e51a0ab4830feca77e4854009a991",
"oa_license": "CCBYNC",
"oa_url": "https://doi.org/10.4329/wjr.v15.i3.56",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "1698597aa73ec921d9bb31442662f9a1fd25994f",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
216702098 | pes2o/s2orc | v3-fos-license | The Impact of Glass Ceiling on Career Development of Executive Level Female Employees in Financial Sector in Kandy District
In Sri Lanka even though women represent more than half of the population, their labour force participation rate is less than men. Men's participation in the labour force is twice as women's. In recent years women and gender issues have become a major area of concern. Even seminars, workshops and conferences are being held over the world to discuss women issues and women advancement in all spheres of life, still it exists the barriers for women career development. This study was entirely design by centering the focal problem of impact of glass ceiling on career development. This study also attempts to identify the contribution factors which create the glass ceiling effect on women career development. The study was conducted to find out the impact of glass ceiling on women career development of executive level female employees in Financial Sector in Kandy district and hypotheses were developed to find out whether there is a significant impact of individual, family, organisational and cultural factors on women career development. The data were collected from 144 executive level female respondents out of 157 executive level female Original Research Article Victor and Shamila; AJARR, 2(4): 1-11, 2018; Article no.AJARR.44467 2 employees in Financial Sector at Kandy district with a structured questionnaire. For presenting and analysing the data both descriptive and inferential statistics were used through SPSS 19.0 version and the results were derived. The major finding of this study revealed that the glass ceiling and women career development has small negative relationship and individual, family and cultural factors have a significant impact on women career development while organisational factors have an insignificant impact on women career development. Further, in this study Adjusted R 2 of women career development is 0646. It is explained that 64.6% of women career development is explained by glass ceiling. The conclusion was finally made that there are significant impact of the glass ceiling on women career development of executive level female employees in the financial sector at Kandy district.
INTRODUCTION
Human is the most valued resource and the contribution of the human resource is immense for the success and the growth of an organisation. Hence every organisation attempts to acquire and retain competent and motivated employees [1]. As [2] state career development enables the organisations to attract and retain efficient and effective employees who would contribute to achieve the goals and objectives of the organisations. Similarly, this enables the employees to achieve their individual career goals and objectives.
Thus, the key for organisation be succeed is to maximise its human potential. Evidence showed that managerial and professional women are not moving up to top managerial position at the comparable rates as compared to their male counterparts [3]. Some of the reasons of this problem for women in their career are clearly connected to the idea of a glass ceiling. "Glass ceiling" refers to invisible barriers that impede the career advancement of women. It also refers to situations where the advancement of a qualified person within the hierarchy of an organisation is halted at a particular level [4]. Women at the highest levels of business are still rare [5].
For successful career development of women employees, a concerted effort of the organisations, employees, their families and the wider society is essential. In the absence of these efforts, these employees face many challenges as they advance through careers. Furthermore, women have reported greater barriers than men and greater difficulties in getting career development opportunities for them [6]. Even though women represent more than half of the population in Sri Lanka, their labour force participation rate is less than men. Men's participation in labour force is twice as women's [7]. Table 1 shows the contribution of the female employees to the total employment by each occupational group in Sri Lanka. It clearly shows that 65.8 percent contribution of female to the total employment through "professionals". The mentioned reasons for that, because the occupation group of "professionals" includes teachers, nurses, etc. In plant / machine operators and assemblers occupational group, physical ability of men may be reason for showing high rates by them than women. However, it is not applicable to the senior ranks as well, since that levels work not purely because of their physical ability but also their mental ability. However, noted women's rate on Senior Officials & Managers and Proprietors & Managers of Enterprises occupational groups. They show different rates than men and also their contribution to the employment also less [7]. In 2015 4.6 percent of female employees employed in "Managers, Senior Officials and Legislators" occupation compared with 7 percent of males in Sri Lanka.
Labour force participation of women in the world has significantly increased during the last few decades. However the increment has primarily occurred in relevant to the proportion of women at entry level and in lower and middle level management positions, while the proportion of women reaching senior and top management positions has remained relatively low [7]. The reason for this state may be the glass ceiling or the invisible barriers women face when developing their careers [7].
An investigation into the relative under representation of women in senior management positions in UK travel and tourism showed that the majority of respondents cited career breaks for child birth, child care and the pressures of combining work with family and personal commitments as the key issues [8].
A Malaysian study showed that women middle managers face a glass ceiling in their working environment which inhibits the promotion of female managers and entails a barrier to the career development opportunities for women and entails that women do not have enough organisational support [9]. In Singapore organisations the glass ceiling barriers are studied from the aspects of corporate culture, corporate climate and corporate practices finding that the common precepts of the glass ceiling theory are supported in the Singapore context [9].
On the other hand findings indicated that Malaysian respondents viewed family and personal responsibilities as their greatest impediment to attaining senior management positions; this view is strongly supported by the Australian women as well [7]. The paper has explored the relationship between women, management and globalisation in the Middle East. It has shown that women face social and organisational barriers in the labor market and through gendered organisations and cultural practices [7].
According to [7] there is a negative impact of the dimension of the Glass Ceiling on Women Career Development. It was found that there is a negative relationship between corporate practices and women career development in state commercial banks in Sri Lanka and the barriers from corporate practices on women career development are very low. There is a negative relationship between family commitment and women career development and the barrier from family commitment on women career development are low. Also stereotypes about women were found to have a negative impact on women career development and the barriers from stereotypes about women on women career development are very low. Overall there is a weak negative impact of glass ceiling on the career development of women branch managers in state commercial banks in Sri Lanka.
According to the [7] the Glass Ceiling influences on the Women Career Development among the female executive level employees working in private sector organisation in Sri Lanka. There is a moderate negative correlation between glass ceiling and women career development among the executive level female employees working in private sector organisations and there is a weak negative correlation between family factors and women career development. At the same time, there is a moderate negative correlation between individual, organisational, cultural factors of glass ceiling and women career development.
In the Sri Lankan context, although many research had been done about different topics or the problems relating to women career in Sri Lanka, prevalence of the research done on the glass ceiling and women career development are limited. This study is based on female executive level employees working in the financial sector in Kandy district in order to fill the empirical research gap. Therefore this research examines the certain questions related to how glass ceiling impacts on career development of executive level female employees in the financial sector.
The key research problem of the study is,
Glass Ceiling
The glass ceiling is a phenomenon which argues that a barrier is so subtle and transparent yet so strong that it prevents from moving up in management hierarchy [10]. Further, glass ceiling refers to invisible barriers that impede the career advancement of women. The barriers that women employees and minority face when developing through their careers are termed as glass ceiling. It also refers to situations where the advancement of a qualified person within the hierarchy of an organisation is halted at a particular level because of some form of discrimination, most commonly sexism or racism [4].
Gendered organisational structure (GOS) model emerged to describe the basic career barriers that created by glass ceiling. There is a significant body of literature support this model. The four factors of this models are individual factors, family factors, organisational factors and cultural factors. These factors are used as the dimensions of the glass ceiling in this study.
Individual factors
Individual constraints can directly affect the person's ability to function fully as a human being and in relation to his/her quality as a leader or manager. Individual factors that can affect career progression includes; education, self-perception and motivation. Low career advance of women has been explained in terms of personality traits [11].
Personal skills or competencies were considered to be the most important factor that attributed to successful promotion to a senior management role.
And also communication skills, leadership skills, hard work, dedication, Selfconfidences are also considered to reach the top management position [12].
Family factors
Work-family commitment is the factor to which relation causes affect to the performance of female employees and family responsibilities such as home and child care play a much bigger part in female employees' lives [13].
Work whilst having family responsibilities was the most significant barrier by career women [13]. Women still take on most of the responsibilities of homemaking, which makes the issue of how to balance work and family life without sacrificing either a big concern for women. Also, the increase in dual -career couples has made the work -life balances an important factor for both women and men's career decisions [8].
Organisational factors
Within the organisation, there were certain factors, mainly subtle which served to hamper women's career advancement [3]. [14] Mention that, when it comes to the structure and culture of the organisation, organisations are often not structured to accommodate women's values.
Organisation culture is a system of shared meanings, values, beliefs, practices, group norms of the members to produce behavioural norms with regard to the working conditions of the organisation [15]. Organisational structure impedes women's entry to and advancement in the workplace. That structure can include job recruitment, job assignment, mentoring, retention, and training, how work and family are balanced by employees, and promotion and reward systems [16].
Cultural factors
The socialisation of the girl child in many societies is to blame for perceived inabilities on the part of women [11]. The life of passage of women are sacramental, celebrated or even acknowledged illustrative of the position ascribed to women, right from the birth of the girl child, in comparison to the boy child.
Female employees can perceive their professional dreams only after fulfilling their culturally accepted roles, an expectations nearly impossible considering at what age this should be happening. Women who perform in a positive, antagonistic, self-governing mode are seen as performing outside of societal norms.
Career Development
Career development is the total constellation of psychological, sociological, educational, physical, economic and chance factors that combine to influence the career of any given individual over the life span. Career development is a "continuous lifelong process of developmental experiences that focuses on seeking, obtaining and processing information about self, occupational and educational alternatives, life styles and role options". Career development involves concerted efforts directed towards assessing a workers' potentials identifying likely career paths for that employee and designing and implementing "various forms of training and experience to prepare that person for more advanced job" [17].
Career focused (CF)
Employee interested in advancing or achieving their career. So, People with high career focused attempt to develop their profession [7].
Family support (FS)
Family support is the support given by family members to continue success of person's profession. Person who has good or high family support achieve a situation of balancing work-family life and develop their profession [7].
Attitudes towards organisation (AtO)
Attitude towards organisation refers the opinions of employee as regards working organisation. People with positive attitudes concerning the organisation worked stay with origination and develop their career within it [7].
Relationship between Glass Ceiling and Career Development of Female Employees
The proportion of women in lower and mid-level management positions has increased dramatically, while the proportion of women reaching top management positions or climbing up the corporate ladder has remained relatively small [18]. It is commonly believed that women have fewer opportunities in advancing through their career development than men. Some of these obstacles that lead to the underrepresentation of women in leadership and upper management positions are directly relevant to the phenomenon of "Glass Ceiling" [4]. [3] also indicates that the glass ceiling issue has a considerable impact on women's careers internationally. Despite, the increased number of women both participating in the workforce and reaching management positions, the evidence demonstrates that, for the majority, advancement to the very highest levels is rare and that the 'glass ceiling' is still in corporate culture, corporate practices, and corporate climate [5].
The GC is defined as, the barriers that prevent women from ascending to senior management positions in large corporations have often been described by the metaphor "glass ceiling", a transparent barrier which prevents women from moving up the corporate ladder past a certain point. It is visible that women traditionally encounter a "glass ceiling" and men are more likely to be accelerated into management positions by means of a "glass escalator". Some of them propose barriers such as attitudinal, behavioral and structural that hinders career advancement for many women [8].
Women's awareness toward their career paths is an important factor [16]. There is a common perception that men are leaders and women are the supportive followers. This perception of women as followers is strengthened by a lack of strong female role models. According to [16] the barrier to women career progression is the role stress due to multiple role demands inherent in running a career while also running a home and family. Due to male imposed pressure women are also stressed throughout the day and indirectly it affects their performances and then it would be easy to find excuse to restrain them from promotion. It has been seen that a woman, in her prime time in organisation or at the peak of her career, suddenly bounds to drop her career to meet social obligations and responsibilities and never think of returning back due to the unwelcoming situation of the organisation.
The scarcity of female leaders is linked to ongoing prejudice and discrimination against women in the workplace. This refers although women are now capable of moving to upper levels, at some point they are halted by an invisible barriers. It applies to women as a group who are kept from advancing higher because they are women. According to [19] explanations for the extremely low number of women in the position of CEO and other senior management positions go beyond corporate policies and practices and also embrace the impact of gender-based behavioural dynamics.
[20] mentioned that many jobs are still seen as men's or women's jobs and this influences the initials of a particular gender to organisations.
According to the [21] Glass ceiling inequality represents a gender or racial inequality in the chances of advancement into higher levels, not merely the proportions of each gender or race currently at those higher levels. And the Glass ceiling inequality represents inequality that increases over the course of a career. Women middle managers face a glass ceiling in their working environment which inhibits the promotion of female managers and entails a barrier to the career development opportunities for women and entails that women do not have enough organisational support, such as: networking, mentoring and family friendly initiatives [9]. Women are under-represented in top management positions because of fear that they are not confident enough to act as managers. The reason is because society generally perceives that women do not and cannot function effectively as managers. It is believed that management is best suited to men [22].
According to [22], that women have a desire to advance their careers, but are limited by the following key factors, gender role, work-family commitments, personal factors, work and family conflict, stereotype of women as managers, male-dominated environment, organisational structure and culture, gender discrimination, job characteristics and lack of support systems at work, sexual harassment and lack of mentoring and coaching called glass ceiling. The certain biographical characteristics such as age and marital status play a role in career advancement. It was noted that women in different marital groups (single, married, divorced or widowed) have a desire and willingness to advance their careers, but are deterred by particular marital concerns and/or commitments.
Study Population
Study population for this research identified as executive level female employees in financial sector in town area of Kandy District. Main bank branch network of commercial bank in Sri Lanka: there are major five banks which are Bank of Ceylon, People's Bank, Commercial Bank, Hatton National Bank, and Sampath Bank. Based on that article particular five (5) banks and nine (9) other financial institutions including insurance companies, leasing and financial companies' executive level female employees at Kandy district was considered as study population. The population of financial sector was taken as one hundred and fifty seven (157) executive level female employees in town area at Kandy district.
Sampling Technique
According to Sekeran [18]. every element in the population has a known and equal chance of being selected as a subject and it has the least bias and offers the more generalisability. There were one hundred and fifty seven (157) executive level female employees in selected five (5) banks and other nine (9) financial institutions in town area at Kandy district. For the study total number of one hundred and fifty seven (157) executive level female employees was selected without any further filtering.
Sample Size
According to [23] that the sample size larger than thirty (30) and less than five hundred (500) are appropriate for most research and where samples are to be broken into sub samples, a minimum sample size of thirty (30) for each category is necessary. The current study purposively selected one hundred and fifty seven (157) female executives from fourteen (14) financial institutions in Kandy District in order to collect required data for the study. Table 2 shows the sample distribution of the study. The questionnaires are issued to 157 executive level female employees in Financial Sector at Kandy district. However, the data were collected from 144 executive level female respondents out of 157 executive level female employees in Financial Sector at Kandy district with a structured questionnaire.
Survey Instrument
A questionnaire is a pre-formulated written set of questions to which respondents record their answers. In this study questionnaire is developed with the support of previous studies carried out by expert in the relevant subject areas.
Reliability Analyses
In this study, result of Cronbach's Alpha is 0.837 of glass ceiling and Cronbach's Alpha is 0.753 of women career development. All items considered in this study are to be reliable. Whole questionnaire was considered reliable because reliability is responsibly acceptable since Cronbach's alpha coefficient is greater than 0.70.
Analyses of Demographic Variable
In this study, out of 144 respondents of executive level female employees in selected financial institutions, 16
Multiple Regression Analyses
In this study, R, 0.810 is the Multiple Correlation Coefficient of the independent variable with the dependent variable after all the inter correlations are taken into account. Adjusted R square 0.646 is the explained variance in career development by the combination of these variables. F value 66.372 is significant at 0.000 level. The results indicate that 64.6% of the variance of career development has been significantly explained by glass ceiling. Hence it can be clearly said that other factors have 35.4% influences on career development. Coefficients table helps to identify which among the independent dimensions influences most the variance of career development.
The strength of the influence that each of the independent variable had on the dependent variable (Career Development) were determined by the use of multi regression coefficients of the independent variables. The influence of each independent variable is show in the Table 7.
The Impact of Personal Factors on Women Career Development
A one way between subjects ANOVA was conducted to compare the Personal Factors on WCD.
The results of the comparison of Designation of female executive level employees with their career development represent the p value of 0.048 which is less than 0.05. This indicates that there is a significant impact of current working capacity on WCD at the p<0.05 level. This result suggests that according to the position female employees held, the GC influence can be different. There is no significant effect on age, marital status and number of children of female executive employees on their career development and P values are 0.251, 0.098 and 0.543 respectively which is greater than 0.05 (p>0.05). When it considering the education qualifications, work experience and salary level of the female employees, educational qualifications has the p value of 0.516 while work experience and salary represents the p values of 0.780, 0.305 which is p>0.05 respectively.
Hence it can conclude that except the Designation of the respondents, other personal factors have no significant impact on WCD and suggest these Personal factors have no significant different in GC influence. | 2019-11-22T01:00:57.077Z | 2018-12-12T00:00:00.000 | {
"year": 2018,
"sha1": "68845f1bb82b6dee43756622af2e814e8a5b1b77",
"oa_license": null,
"oa_url": "https://www.journalajarr.com/index.php/AJARR/article/download/29773/55954",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "2fbaa7d61184dce8e5d114ba077414f1bd1efe95",
"s2fieldsofstudy": [
"Economics"
],
"extfieldsofstudy": [
"Business"
]
} |
12623126 | pes2o/s2orc | v3-fos-license | Persistence of Leishmania antigen in C 57 Bl / 6 j inbred mice infected with Leishmania ( Leishmania ) amazonensis
PURPOSE. To develop an animal model for studying mucocutaneous leishmaniasis. METHODS. The hind footpad of C57Bl/6j inbred mice was experimentally infected with 107 Leishmania (Leishmania) amazonensis promastigote and the skin was studied through light and electron transmission microscopy and immunohistochemistry (PAP) techniques. RESULTS. There were morphological evidences of cellular immune mechanisms and hypersensitivity reaction after eight weeks of infection and metas-
INTRODUCTION
Increasing in tourism and labors migration disseminated leishmaniasis throughout the world 1 .In the New World there is a particular expression of the disease called American Tegumentary Leishmaniasis, comprising three clinical features: Localized Leishmaniasis (LCL), Diffuse Cutaneous Leishmaniasis (DCL) and Mucocutaneous Leishmaniasis (MCL) 2 .Some authors 3 suggested that DCL and MCL could be both originating from Leishmania (Leishmania) amazonensis 4 , but the mechanisms by which Leishmania leads to MCL remains obscure, owing mainly to the lack of an appropriate animal model [5][6] .
The great majority of papers concerning to animals' models with different susceptibilities to L. (L.) amazonensis do so from an immunological point of view.Morphologic studies about the inflammatory response of the inoculation site are scarce and brief in follow up periods of infection.The aim of this study is to propose this model to study the human disease, according to the histopathologic features of the inoculation site and the clinical aspects, in the course of one-year infection.maintained in plastic cages and received proper food and water ad libitum.
Parasites.Leishmania (Leishmania) amazonensis (WHOM/BR/00LTB0016) -G.Grimaldi, Fiocruz -RJ.The strain was sustained in the laboratory through sequential passages in BALB/c mice, culture in NNN / BHI medium and reinoculation at hind footpad in mice 7 .
Infection.Experimental groups, each with three animals were inoculated under ether anesthesia with 50 µl of saline solution containing 10 7 Leishmania (Leishmania) amazonensis promastigote forms in the stationary phase of growth in the right hind footpad of mice.There was one control animal in each experimental group, inoculated under the same conditions with 50 µl of sterile saline solution.Three inoculated animals and one control were killed 2, 6, 8, 10, 20 and 51 weeks post inoculation (WPI).
Techniques.Fragments from right hind footpad skin were fixed in buffered 10% formaldehyde solution, pH 7.2 and processed by usual histology technique and stained with Hematoxilyn and Eosin.Fragments from the same tissue were processed for immunohistochemistry (PAP) demonstration of Leishmania antigen in tissues 18 and for Transmission Electron Microscopy 9 .
The experiment was repeated twice.
Clinical features.
C57Bl/6j mice put on weight along the experiment.The animals displayed a torpid disease with an apparent healing of hind footpads' cutaneous lesions around 20 th WPI.By 50 th WPI there was relapse of the lesions at the inoculated footpad and appearance of lesions at the tail and in the nose of all animals.
Histopathology.The C57Bl/6j hind footpad dis-closed a rather organized tissue response with granuloma formation after the 8th WPI and fibrosis in the dermis, together with hyperplasia of epidermis, after the 38th WPI.Eosinophils were commonly seen.The nerve endings of the dermis were surrounded by dense macrophage infiltrate very close to the perineurial sheet.Late in the course of the infection (51st WPI) there were grouped lymphocytes surrounding the vessels of the dermis that showed fibrinoid degeneration.Necrosis was extensive and confluent, coming soon after 2nd WPI and followed by granuloma formation and collagen deposition in the dermis.By the time the dermis of C57Bl/6j mice disclosed granuloma formation and collagen deposition, its epidermis displayed exocytosis of mononuclear cells, acanthosis and transepidermal elimination of parasites (evident by the 20 th WPI).The microscopic aspects of control animals were quite normal.
Immunohistochemistry (PAP).This technique displayed well-shaped amastigotes forms of Leishmania stained in golden brown in the inoculated hind footpad skin.There was no evidence of amorphous antigenic material inside the cytoplasm of macrophages, as seen in human lesions 8 .By the 20 th WPI there was elimination of cellular debris and parasites through the epidermis (Fig. 1).
Transmission electron microscopy.The inoculation site disclosed epidermal keratinocytes mitochondria crystolysis (Fig. 2) throughout the experiment.Some eosinophils showed phagocytised parasites (Fig. 3).There were preserved parasites within macrophages (Fig. 4) and in the extracellular space among collagen fibers of the dermis by 51st WPI.
DISCUSSION
The epidermis of C57Bl/6j mice was clearly involved in the inflammatory process since the beginning of the infection and showed areas of ulceration and hyperplasia.In spite of the presence of parasites in the right hind footpad until the 51 st WPI, this animal seems to be able in circumscribing the infection 10,11 .Necrosis was intense with elimination of parasites through the epidermis and the granuloma surrounded necrosis foci, which were partially replaced by collagen.
C57Bl/6j mice showed a well-developed tissue response (granuloma formation) by the 8th week post infection, in spite of macrophage parasitism.This was coincident with the appearing of morphological evidences of cellular immune mechanisms (granuloma) and of hypersensitivity reaction (fibri-
Fig. 2 -Hind footpad -epidermal keratinocytes mitochondria crystolysis (→). Esperimental cutaneous leishmaniasis, C57Bl/6j inbred mice, inoculation of 10 7 L. (L.) amazonensis promastigote forms at right hind footpad (1 st week post infection, Transmission Electron Microscopy, Bar = 1µm).
LEISHMANIA ANTIGEN IN C57BL/6J MICE noid degeneration of vessel's walls and epidermal aggression by the inflammatory infiltrate).This histologic aspect is similar to that of human Mucocutaneous Leishmaniasis 12,13 .Besides, by the final period of the experiment, cutaneous lesions reappeared mucous lesions appeared and well-preserved parasites could be seen at the ultrastructural level.Thus, such an animal model presented here may be used for experimental studies of human mucocutaneous leishmaniasis.formas promastigotas da Leishmania (Leishmania) amazonensis e a pele foi estudada através da microscopia óptica e eletrônica e de técnica imunohistoquímica (PAP).
tasis and well shaped parasites at ultrastructural level by fifty-one weeks post infection. Relapse of infection with mucosa lesions occurred around the 50 th week after inoculation. CONCLUSION. The use of this animal model in long term follow up could be an useful experimental model for human mucocutaneous leishmaniasis.
MATERIAL AND METHODAnimals.Female 40-day-old inbred mice C57Bl/ 6j strain from the University of São Paulo Medical School General Colony was used.The animals were | 2017-07-13T19:33:41.970Z | 1999-07-01T00:00:00.000 | {
"year": 1999,
"sha1": "8b5bcfefa68a7c92d752554f2a138cfac21ee8ba",
"oa_license": "CCBY",
"oa_url": "https://www.scielo.br/j/ramb/a/KdwnpjSCyRz3mgDSBkfGNBH/?format=pdf&lang=en",
"oa_status": "HYBRID",
"pdf_src": "Anansi",
"pdf_hash": "8b5bcfefa68a7c92d752554f2a138cfac21ee8ba",
"s2fieldsofstudy": [
"Biology",
"Medicine"
],
"extfieldsofstudy": [
"Biology",
"Medicine"
]
} |
254780500 | pes2o/s2orc | v3-fos-license | The Whole of Community Engagement initiative: Interculturality in remote Aboriginal education
It is generally accepted by researchers, policy-makers and practitioners that progress in Indigenous education depends on working in partnership with Indigenous people, and that programs and services are best provided in partnership. The 2014–2016 Whole of Community Engagement initiative built a partnership of non-Indigenous researchers with researchers, teachers, education leaders and elders from six remote Northern Territory Aboriginal communities. In this paper we describe the features that led us to characterise the initiative and the remote community and school context as intercultural and complex. The former included methodology, staffing, meeting procedure and interpersonal communication, negotiation of meaning and decision-making. On the basis of this approach, we found that intercultural complexity was strongly evident in schooling in Galiwin’ku, Elcho Island, North East Arnhem Land, which was the community most closely studied by the authors. The paper contributes to the recognition of intercultural complexity in remote Aboriginal schooling, and the potential benefit that its recognition can have for educational outcomes in those contexts.
Introduction
The Whole of Community Engagement (WCE) initiative 1 was conceived in response to issues of Indigenous higher education identified in the Aboriginal-led Behrendt Report (Behrendt et al., 2012). The initiative sought to understand the factors influencing student engagement with school and aspiration to further education in those communities, from local and social scientific perspectives. It approached six remote communities, identified on the basis of their capacity, to gauge their interest in participating. All six-Yuendumu and Tennant Creek in central Australia, Yirrkala and Galiwin'ku in North East Arnhem Land, and Gunbalanya and Maningrida in West Arnhem Land-agreed. The initiative took place in the context of Indigenous mistrust associated with colonial history and a history of research of doubtful benefit to remote communities (de Crespigny et al., 2004, p. 8;Hall, 2014), as well as long governmental neglect and recent intervention (the Northern Territory Emergency Response). The remote schooling context has also been powerfully shaped by the notion of "both-ways" education, which emerged in Dagaragu (d'Arbon et al., 2009, p. 19) and was further developed in Yirrkala (Frawley & Fasoli, 2012, pp. 311-313), one of the communities in which the WCE worked.
The initiative employed one or more Aboriginal researchers from each community, most of whom had formal teaching qualifications and had worked as teachers in the 1980s era of responsiveness by tertiary institutions to the needs of remote trainees. These researchers were based for much of the time in their own community, where they worked in tandem with a visiting campus-based researcher, most of whom were non-Indigenous. Each team conducted ethnographic research into barriers to and attitudes towards school and further education in its community over a three-year period. We return below to outline how the members of these teams, notably the Indigenous members, were intercultural people (Imahori & Cupach, 2005).
The initiative itself was an intercultural endeavour. Each aspect of the partnership within each research team and with research participants implicated cross-culturality, which flags cultural boundedness, difference and statis (e.g. see Ober & Bat, 2007), and interculturality, which flags cultural intersection, identity multiplicity and dynamism (as outlined below). In this context, interculturality existed in the relations between individuals whose primary background was in one or other of the major cultural groups involved (Aboriginal and Anglo-Australian), but who at the same time bridged the differences between them, in part on the basis of other shared cultural affiliations. The project required the researchers to negotiate the intersections of culturally conditioned values, knowledges and habits of interpretation. Its aims and design were initially outlined in the communities, and the research conducted, in local languages and English. Its conduct demanded intercultural communication. Its governance involved intercultural decision-making. Communication in team meetings generated an intercultural relationship. Findings were reached collaboratively and feedback relevant to each community conveyed in locally appropriate and academic forms.
A result of intercultural partnership at various levels was trust between staff and with schools and communities. Trust-which, according to Dai (2009, p. 8) and Ting-Toomey (2005, p. 218, p. 220), is in intercultural encounters an indicator of successful communication-resulted in strong commitment to the research by staff and satisfaction among Aboriginal stakeholders with the initiative's address of longstanding frustrations. The relationships revealed that the Aboriginal researchers and community educators were aware of the growing complexity of their lives. Their attempts to deliver cultural and
Interculturality
The concept of interculturality is infrequently used to understand Aboriginal lives, policy or education. Before contact, Aboriginal groups were diverse, relatively discrete cultural blocs, while interacting with neighbouring groups and others such as the Macassans (Macknight, 1976), and with some commonalities (Moore et al., 2016, p. 46). Local distinctions continued with colonisation, alongside a developing notion of a single imagined community and political Aboriginality in the longer-settled southern and coastal regions. The extent of intersection has increased with urbanisation, globalisation, superdiversity (Vertovec, 2007), symbolic technologies (Nowotny, 2005, p. 18) and the emergence of a single world system (Urry, 2005b). These phenomena impact on the smallest and most remote of communities. They mean that people identify with large cultures like the "Aboriginal" and the "mainstream" Australian, within which they learn values, expectations and behaviours, and form "master" identities. The context means that "Aboriginal" identity is often held with particular intensity and salience (see Collier and Thomas, 1988). They also identify with small cultures (Holliday, 1999), such as clan and language group, or profession, which also condition their values, perceptions and behaviours. That is, the same people identify with supposedly contradictory large cultures like religion or nation (as suggested by Noel Pearson's "tea, cricket and the rule of law", 2015, p. 16), and smaller cultures like those of sports clubs and organisations. In addition to affiliating with multiple cultures, they are categorised as belonging to, and treated by others as having the characteristics of, certain population sectors, and are ascribed the properties of a gender, sexuality, generation and other categories critical to their life chances.
Remote communities are intersected in these ways, even where obligatory kin relations, traditional languages and responsibilities to country continue. As difference from others continues, overlapping and intersecting socio-cultural meanings coexist, interact with and affect each other, putting all differences into perspective against others, incorporating and modifying them, and becoming more nuanced. Some of the meaning of this is captured in global theory (Martin, 2015, p. 7), and some in the Yolŋu metaphor of ganma, the lagoon where salt and fresh waters blend and change (Bat & Guenther, 2013, pp. 128-129;Frawley & Fasoli, 2012, p. 312). Some of the meaning is captured in Nakata's (2002; "cultural interface", which seems, though, to suggest "a thin layer that has properties that differ from [those] associated with the bulk on either side [and] gives rise to properties or effects that are the result of what happens between the two faces" (Nowotny, 2005, p. 20). The interface is of long standing and is now comprehensive, and (even as they also live in other separate worlds) all people live in one world, as Sullivan (2016) alluded to when he said: The ABC recently published an article on Aboriginal child suicide titled "Lost between two worlds". [But] we should remind ourselves that "two worlds" is only a metaphor. There is only one world-they are part of our world, we are part of theirs. (n.p.) This points to the increasing interpenetration of the Aboriginal and wider "worlds" and the fact that Aboriginal people are, after around 200 years, not mere sojourners in modern Australia. Moreover, it is not cultures that do things at interfaces, but people, acting from the logic of their many cultural attachments and individual psychologies. Everyone now lives at multiple "interstitial" (Bhabha, 1994) zones, and works within and across socio-cultural borders to negotiate their social location and maintain their own interests. At the borders, individuals interpret the influences shaping them and others' meanings, respond according to cues, reveal and hide what they wish of their own meanings, and otherwise manipulate the subtleties of shifting meanings to move interactions in directions they wish. They learn to see themselves as others perceive them and expect them to be, to act as they are expected to act (see Anzaldua, 1987), and to mimic, mock, unsettle or deny expectations. This sociality, which following Goffman (1959Goffman ( , 1967Imahori & Cupach, 2005;Ting-Toomey, 2005) is known as "facework", is done so routinely that it easily goes unnoticed.
In summary, all who belong to a group are similar in some ways to others who belong, and different from them in other ways. They share much with fellow members while also being distinct from them. They also share much with non-members. Interculturality implies multiple, simultaneous and shifting sameness and difference. Aboriginal people are the same in this as others, and the nature of their samenesses with and differences from other Aborigines and other Australians varies around the country, and according to context. Given the prolonged, extensive and growing nature of interactional experiences between Aboriginal and cultural others, they can be known as insiders and outsiders, and as different from and sharing much with other Australians. They are "us" and "we" are them, and differences are often either indeterminable (as in Canada; see Ermine, 2007, p. 197) or exist in nuances of emphasis. The continuous negotiation of positioning can bring tension, as discursive, structural and institutional power saturates interactions (Shi-Xu, 2001). All this has to be negotiated, and the negotiation over time "jointly produces new meanings, new realities and new futures" (Shi-Xu, 2001, p. 288).
The WCE initiative was a site of these dynamics, and its success depended on their careful management. Remote schools are sites of similar dynamics, and students' learning depends similarly on their careful management. Staff and students share a world of fluidly interacting social and cultural influences and personalities. The "intercultural" refers to the intersection of multiple, notionally separable cultures or "worlds" at their interfaces, to individuals' interactions across the borders, and to the social, cultural and subjective transformations that result. Teachers in remote schools must manage this interactional context to generate "common meaning" (Leigh, 2004) and, so, learning. We present one such context-that of Galiwin'ku-as an exemplary case of an attempt to provide guidance for teachers.
Complexity
Interculturality implies overlaid and interpenetrating cultural perspectives, social orders, knowledge domains, visions of past and future, and selfhoods. It acknowledges a complexity that exceeds that which has been explained with cross-cultural and both-ways approaches to Indigenous education (see Bat & Guenther, 2013). In the sense of complexity theory adopted in the social sciences, those approaches and concepts can be seen as consistent with the "positivistic research tradition [that seeks] control through deliberate [attempts to] unambiguously … isolate cause and effect relationships and [pursue] generalizable laws of behaviour" (Cooksey, 2001, p. 82). The attempt in those metaphors to reduce life to essentials and binaries has been useful in elucidating causality and in guiding policy and practice, but it is not sufficient. A complexity-oriented approach understands reality as "composed of complex open systems and emergent properties and transformational potential [with the] crucial corollary [that] knowledge is inherently local rather than universal" (Byrne, 2005, p. 97;Richardson & Cilliers, 2001). Rather than seeking explanation by focusing on the variables per se, a complexity-oriented approach focuses on the actions of agentic individuals, their motivations, choices and actions as they interact with others and negotiate the forces of history, social structure and natural environment (Byrne, 2001, p. 63). A focus on everyday, lived reality can identify the array of forces that constitute that reality.
Any complex system (like remote schooling) is "nested within, [has] complex systems nested within [itself], and intersect[s] with, other complex systems" (Byrne, 2005, p. 105). It is open to external influences (e.g. community conflict, cyclone), nested in other systems (e.g. the education department, provincial and national social policy), intersected by still others (e.g. the health system), and by those in the wider environment (e.g. post-colonial human and Indigenous rights discourses). At the interactions of the interfaces: New properties or phenomena emerge, that belong to [wholes] that have no recognised boundary.
[They] … increase complexity by creating difficulties in understanding where [the] constituent parts come from, how they are interconnected, or how they can be traced. It is unclear where the phenomenon belongs and where it is heading. Since it is unclassified, it usually is also difficult to describe, since the appropriate language is not (yet) available, or different mutually incomprehensible languages are spoken on the other side of the divide. (Nowotny, 2005, p. 21) The interfaces produce complexity and attempt to reduce it to make it manageable. The forces of production and reduction are not in equilibrium, and that, along with human agency interacting with infinite combinations of factors, ensures that systems are dynamic and nonlinear. They are adaptive to complex contingent causation, and local-level perturbations can have vast effects at unpredictable times elsewhere in the system (Urry 2005a, p. 7). Inter-connections between parts of the system and minimal central hierarchical structure mean that "determination [of system trajectory] runs in all possible directions, not just top down" (Byrne, 2005, p. 105). Iterative local level adaptations (Urry 2005b, p. 243) can feed back to affect the top and other levels, giving complex systems a self-organising or autopoietic capacity (Urry 2005a, p. 7), and relative order that can flip into dramatic change. Bat and Guenther (2013) identify these dynamics, and attempts to tame them, in remote education.
Partnership
Gaining more than partial knowledge of any one perspective on intercultural complexity requires the plural perspectives (Richardson & Cilliers, 2001, p. 17) to be defined and understood in partnership. We consider partnership as a relationship characterised by equality of power and trust, where trust implies that neither partner can or will apply coercion to have the other adopt certain positions or behaviours (Seligman, 2011). Though the initiative was managed by a non-Indigenous academic, it was initiated by Aboriginal interests. Also, the manager and other researchers demonstrated their commitment to equality in their efforts to maximise active community participation in all phases of the research, respect community meanings and directions, and embrace community voices, alongside and in collaboration with the academic (see de Crespigny et al., 2004, pp. 11-12;Frawley & Fasoli, 2012). As to be expected (Cargo et al., 2008, p. 905, p. 912), the balance of power was fluid, in the fact of leadership but also the sense that one partner had greater control in one context and the other in another. For instance, the project manager had institutional power while seeking to maximise community input. At the same time, the community organisation Yalu Marngithinyaraw (Yalu) was led by a Yolŋu manager with greater power than her non-Indigenous employees, while seeking academic input.
The WCE as intercultural partnership
The WCE initiative was an intercultural encounter in its methodology, staffing, governance, management, communication and ethics. It took a place-based, ground-up, participatory and "mutually respectful" approach (Moore et al., 2018). It was "critical Indigenous qualitative research" (Denzin et al., 2008, pp. 1-20, and passim) in the sense that it was: • Localised: grounded in the meanings, customs, social relations, struggles, aspirations and politics of the particular time and place in each setting.
• Participatory and action-oriented: initiated by (wider) Aboriginal concerns and sought action of benefit to the participating communities.
• Collaborative: sought dialogue between Indigenous and critical perspectives, and coconstructed knowledge that was both "reflexively consequential" for participants, and had wider purchase.
• Interpretive/constructivist: respected the existence of multiple intersected perspectives on lived realities associated with culture and social location, and encouraged dialogue between researchers and researched to shape data analysis and knowledge construction (Schwandt, 1998).
• Critical: informed by social justice concerns, deconstruction and redress of structured power and knowledge disparities; privileged Aboriginal perspectives and self-determination; and cautious in generalisation.
This framework is akin to others informed by critical race, culturally responsive, Indigenous standpoint and decolonising theories. It emphasised, for example, the cultural humility and critical self-reflexivity of researchers, the rebalancing of the power of the dominant discourse and the idea that research should be for the researched (e.g. Foley, 2006;Guenther et al., 2016;McDonald, 2003;Milner, 2007;Paradies, 2018). These and other features are evidenced in the following description of the initiative.
Most of the community researchers were senior in cultural terms, career educators and education leaders in their communities. They participated fully in the cultural life of their communities, as well as in roles beyond the community. In this project, they mentored junior community members as researchers. Most of the campus-based researchers were of Anglo-Australian cultural background. Some had non-Anglo migrant backgrounds. Two were urban Aboriginal women. Several had established relationships with their Aboriginal colleagues, and were currently or had previously lived in the communities concerned. All researchers came to the project, then, with several positionalities (Caretta, 2015) and identities. Each local partnership bridged "insider" and "outsider" perspectives. The outsiders generally came to the project with greater knowledge of social scientific research and, while seeking community meanings, sought to translate research meanings to their community partners. The insiders came with expert knowledge of local meanings and translated them to their research partners, while adopting a social scientific approach. All shared aspects of each other's cultural outlook, including, for example, the value of formal education and research. In addition, they came with intercultural skills; they had, for instance, knowledge of their own and others' cultures and empathy for their partners.
From whatever starting point, each team undertook ethnographic/auto-ethnographic (see Frazer & Yunkaporta, 2019, pp. 90-91) field work, observing local protocols and adopting methodologies that suited local preferences for relationship and storytelling (Hall, 2014;Ober, 2017). They participated in everyday life, collected relevant documents and artefacts, "yarned" (Bessarab & Ng'andu, 2010) and conducted semi-structured interviews where it suited . The community researchers interpreted cultural metaphors and other knowledge for their campus-based colleagues. The latter listened "carefully and deeply" (Shore et al., 2014) to their community colleagues' and participants' perceptions, concerns, needs and visions, and took direction from them about how the research was to be acted on in each community. Each team respected local ways of knowing and shared ownership of the research.
While the community researchers' role as cultural gatekeeper, broker, informant, translator and mediator was critical (see Caretta, 2015), the partners took advantage of each other's strengths across their several identities. The campus-based researchers working with their Aboriginal partners in the communities, and the community-based researchers working with the campus-based in Darwin, negotiated the "facework", avoided "freezing" each other into unitary identities, and negotiated a relational culture within which they came to celebrate and transcend their cultural differences (see Imahori & Cupach 2005, pp. 197-205).
Each team came to conclusions collaboratively and delivered them jointly at whole-of-team meetings and conferences. An explicit expectation of the project's ethics approval was that those meanings would be accurately reported back to communities, and that practical outcomes would be achieved. Accordingly, the teams took great care to ensure that community members understood, approved and found their reports helpful. Equally, the community researchers accepted the validity of, and sought to better understand, what the social scientific perspective offered them. They engaged with new meanings. In this reciprocal relationship, Indigenous voices were empowered and power differentials minimised. The initiative was phronetic in placing "a premium on Aboriginal insights" into education issues and in promoting "responsive and innovative ethical research practice linking action and change" (Bainbridge et al., 2013, p. 277).
Given the intercultural macro context for the project, it also demanded critical interpretation of what had been learned (see Nakata, 2012). It drew on Western knowledge systems and research methodologies that had synergies with the local, notably grounded research (Bainbridge et al., 2013;Guenther, 2015). Indigenous and non-Indigenous team members negotiated their interpretations of Indigenous cultural metaphors and other insights. They made the research accountable to its subjects by repeated "memberchecking" of their interpretations against local perceptions. Only when they were agreed upon did they relay them to whole-of-team meetings for comparative analysis. Through this iterative process, knowledge that retained the fidelity of both knowledge systems was "co-constructed" (Bainbridge et al., 2013) for the community reports and the whole project.
Communication in whole team interactions was intercultural. The team (manager, campus-based coordinators and research teams) met for four day meetings at different times in Darwin, Tennant Creek and Maningrida. In the meetings, each team presented their research to date and progress made toward action plans. Projects were discussed and analysed. Community staff negotiated intersections of cultural norms, such as brother and sister working in close proximity, and intra-cultural Aboriginalities. Everyone negotiated the discourse of colonial and recent policy histories that influenced perceptions and interactions. Aboriginal staff explicitly educated their non-Aboriginal colleagues about the sensitivities, but the discourse underlay all interactions, and moments of tension did arise. On one occasion a community researcher felt that he was not being treated by campus-based researchers with the respect due to him as the local custodian. Some excited discussion by staff members-who were aware of the danger of identity stereotyping, and had begun to see each other more as individuals-had interrupted him in his Welcome to Country. By "not sufficiently support[ing] his cultural identity" (Imahori & Cupach, 2005, p. 200), they had momentarily upset the delicate balancing act of recognising his cultural identity (i.e., his positive face) and his individuality (i.e., negative face), and in so doing breached the politeness system that had been built up (see Economidou-Kogetsidis, 2004, p. 27). However, all participants were aware of the communication challenges, and the tension was soon resolved, in this case through the intervention of one of the community researchers (RG). The earlier, mutually appropriate, effective and satisfactory intercultural communication was restored.
Over the life of the project, participants negotiated many intersections to co-construct knowledge (Collins & Markova, 1999). As in the Galiwin'ku case below, success required mutual compromise of values, attitudes and practices, and tolerance of small breaches of etiquette. As one of the community researchers said at the final whole-of-team meeting, "You [non-Indigenous staff] have taught us well, and we have taught you well." In terms of data analysis, copies of the community presentations were taken and annotated. The dynamics and discussions were observed, noted, recorded and transcribed (other than discussions in language). NVivo software was used to manage the data, and assisted researchers to identify key themes. Analysis was based on an interpretive close reading of the everyday social reality that was appearing. This was informed by the interculturality and complexity framework, knowledge of history, power structures, discourse and ideology. Conclusions were member-checked with community representatives.
Governance of the initiative was intercultural, though less than was hoped. The steering group sought to bring together senior representatives of partner organisations (Batchelor Institute of Indigenous Tertiary Education (BIITE), Charles Darwin University (CDU) and Menzies School of Health) and representatives of the Aboriginal communities. Unfortunately, challenges of timeframe, design and implementation limited the effectiveness of this aspect. Delays in communication of data from the communities meant the steering group often had little information on which to base its work, and conflicts of interest made community representatives sometimes reluctant to discuss challenges faced by projects in which they were invested, on the basis that it might jeopardise the projects. Despite these limitations, which may be aspects of structural inequality, the initiative was managed on a day-to-day basis in a highly consultative and responsive way. Meetings were dominated by community researchers' use of cultural metaphor, humour and diagrams or sketches to convey meanings. One community researcher began to explain her recent work in English, but abandoned it, saying that words (in English) were no good for her, and continued by talking to sketches in the artistic language of her community. Though sometimes difficult for non-Indigenous staff to immediately understand, re-direction or requests for clarification were kept to a minimum in the expectation that the relevance to the research would emerge. The freedom contributed to the trust that pervaded the initiative.
The initiative's investment in intercultural partnership was important in eliciting the perspectives of community members and researchers, and their preparedness to engage in dialogue with a critical social scientific perspective. The investment uncovered three key interlinked desires: for schools to assist in building strong cultural foundations, acknowledge cultural difference, and provide standard training focused on numeracy and English literacy (Moore et al., 2018, pp. 17-35). Interpretation through an interculturality lens suggested that the concurrence of the three indicated that the people involved were aware at some level of interculturality, and trying to get control of it. It indicates, too, that they had definite ideas about how to go about gaining that control.
The address of interculturality in the Galiwin'ku case
The features of the WCE partnership noted thus far were exemplified in that between staff of WCE and Yalu (including the authors and others acknowledged) and of Shepherdson College (the local school) in Galiwin'ku. The approach facilitated the revelation of the depth of efforts being made to negotiate emerging intercultural complexity. Shepherdson College is a P-12 school with around 700 students. It provides standard syllabus and a bilingual program in English and the major local dialect. Formal English reading and writing start in year 4. It describes itself as a "Learning on Country" school that endeavours to bring cultural meaning to the curriculum. The usual arrangement is for a Yolŋu assistant teacher (AT) to work in a team-teaching arrangement with a qualified, visiting Balanda (non-Aboriginal) teacher, with rare exceptions. Yalu was created in 2000 by community members with the intention of providing developmental workshops in education and health and wellbeing. Before the WCE project began, Yalu had begun providing cultural awareness training for Balanda teachers in response to perceived problems in their interactions with ATs, and mentoring for students in response to disengagement, misbehaviour and bullying, which were understood as outcomes of differences between teachers' and students' cultural norms. In the mentoring, a senior mentor, retired qualified teacher and cultural senior (EM), and younger mentors provided calming activities and cultural activities centred on Yolŋu Matha (language) concepts, norms and literacy that helped the students learn traditional ways of behaving centred on respect for their elders, teachers and other students. They also provided remedial teaching in maths and English literacy. The program was the first of a series of attempts to manage the intersection of schooling and local culture, aiming to improve student engagement and learning. In 2015, a service agreement that envisaged a structured cultural awareness program and induction manual for new teachers was signed between Yalu, CDU (via WCE) and Shepherdson College.
Unfortunately, the workshops ceased with a period of instability that occurred when Shepherdson College had several short-term principals. A more stable period began in 2016 with the appointment of a principal and deputy who accepted the possibility that student behaviour problems were related to cultural difference. This allowed the WCE team to work with Yalu and the school to explore the difficulties Yolŋu teachers were experiencing with the earlier program. At Yolŋu staff meetings, ATs discussed their concerns, notably the quality of their relationships with Balanda teachers, and the relationships the latter had with students. A standout issue was the tendency of Balanda teachers to speak loudly when disciplining students, rather than the quiet non-confrontational discussion style that is normal Yolŋu practice. The teachers were violating local communicative norms. The Yolŋu staff also discussed how they might work more collaboratively with school leaders, Balanda teachers and local parents.
These discussions, and others with the Yalu manager (RG), her Balanda development advisor (SH) and a resident Bible translator (MM), established a receptive intercultural climate, and led to a number of "modest" (Cilliers, 2005, p. 256) changes that reinvigorated the cultural awareness and mentoring programs. Yalu began providing weekly workshops for ATs, who took the opportunity to air the challenges they faced in their roles and search for strategies to overcome them. All staff were invited, more Yolŋu became involved in cultural awareness workshops, and the earlier set structure was abandoned in favour of following emerging situations of interest. Three presenters were used to ensure consistency, provide a range of opinion on cultural interpretations and give each other confidence. Two culturally senior advisors oversaw activities and provided support. It was also decided that, as many teachers were part of local family networks, knowledge of gurrutu (kinship systems) was an ideal basis on which to build Balanda teachers' cultural connection with ATs and students. It was hoped that the knowledge would lead to better communication and improved relationships between the teaching partners and with students, and so to better learning.
In addition, the student mentoring program was refined. The manager of Yalu (RG) and her predecessor (ELM) believed that Yalu had a leadership role in improving the quality of Yolŋu teaching after the loss of a cohort of older Yolŋu graduate teachers who had left teaching to pursue work in the community. In their view, current Yolŋu ATs needed support to become good teachers, so they set about developing a program based on the Yolŋu code of conduct known as raypirri. Raypirri is a set of guiding principles covering all aspects of Yolŋu life in East Arnhem Land. It is an expression of core cultural values and personal identity. The version developed for Shepherdson College sought to help local Yolŋu teachers commit fully to the role and model the behaviours expected of them as Yolŋu teachers. Through this, it aimed to ensure that they could provide students, particularly those at risk of disengaging, with support and guidance via their local cultural perspectives, to come back to the "right path". The following translations were provided by ELM and RG, though not numbered in the original as here.
1.
Yaka ŋaramurryi; yaka ŋarrtjur yothuny. (As a role model, you should not bring your outside troubles into the classroom; you must be in control of your emotions, like anger.)
2.
Ngayaŋuy manymakthu guŋga'yurr djamarrkuliny ga ŋayaŋuy dap'maram yuwalkkum. (Talk to the children with a good feeling, in a way that makes them feel close to you, not in ways that make distance; this is about having an emotional connection with the children.)
3.
Gurrupul yolŋu matha ga gurrupul nhanŋu dharanŋanmirr, nha ŋunhi mayali. (Use our language so that they understand the full meaning.) The authors suggest that numbers 1 and 2 are expressions of Yolŋu values and expectations, but also of universal principles of teaching, as are others in raypirri, though with added intensity, as called for by the intercultural context. Number 1 refers to the need to separate home and work, that is, to adopt the impartial professional role at work in which every child is deserving of equal nurturance. It suggests that clan favouritism is not appropriate in school. Number 2 refers to the centrality of the teacher-student relationship to Yolŋu learning, and it wants Yolŋu teachers to compensate for an area in which Balanda teachers have some difficulty. Number 3, again, wants Yolŋu to know their Balanda colleagues' shortcomings and complement them.
The Yalu raypirri tells Yolŋu teachers not to be shy about their teaching, but to be forthright. It addresses some of the dilemmas that arise from interculturality, such as distinguishing between their obligations at home and those at work:
4.
Yaka favouritism -dhu go ŋorra teacher-wal; guŋga'yun walalany level-kum. (Have no favouritism; treat all the children equally.) This refers again to the challenge of negotiating the divergent obligations of kinship and profession, and implies that, in school, the professional should prevail. Though it may not fully address the true nature of the challenge (in maybe interpreting the roles as more discrete than they can be), this dilemma stands as a metaphor for the challenge of bothness in remote communities. 2 The dilemma exists full time for Yolŋu teachers, and for Balanda teachers who are adopted into a family/clan and respectful of the implications. Even such contradictions can, though, be negotiated where there is forbearance on the part of others involved, as there invariably is when good faith is evident. The document finishes with another reminder of the complex standpoint demanded of a Yolŋu teacher in the context:
5.
Gumurr-watjarrmirr. (You have chosen to be a teacher, and should stand in the middle, be balanced, not take sides, as a role model to the community and school; you have a role in both.) In this context, raypirri is a structured engagement with interculturality. It is an attempt to help local teachers develop the intercultural skills to successfully negotiate their relationships with Balanda teachers and Western learning. It gives guidance from respected cultural and educational leaders in how children should behave when their teachers are talking. It complements the cultural awareness programs that work to develop Balanda teachers' cultural knowledge, with the intercultural skills needed by Yolŋu teachers to work at the cultural interface (and help students learn at it). It is a strong basis for a philosophy of intercultural education in this context. Importantly, it suggests that an equivalent program is needed for visiting Balanda teachers who work at other facets of the same interface.
Yalu has claimed that the success of the program was evident in improved student engagement and behaviour, and a reduced incidence of bullying. To develop an additional element, Yalu appointed a retired teacher who had worked at Shepherdson College for many years, spoke Yolŋu Matha and was known to the teaching staff. She provided support for Yolŋu ATs completing a teaching degree with BIITE, and for activities shaped in consultation with Yalu and Balanda teachers. With EM, she led a series of workshops for all staff called "Learning Together". It was a biliteracy program that aimed to develop participants' capacity to appreciate both cultures (i.e., Yolŋu and Western), starting with explorations of Yolŋu Matha language terms and English idiom, looking to appreciate Yolŋu and other worldviews, and the nuances of meanings carried by different terms.
This case study recounts not a simple Yolŋu desire to secure their cultural heritage from schooling, nor to switch between notionally separate cultural worlds. It recounts a sophisticated attempt to be both Yolŋu and educated, to negotiate the intersections of Yolŋu cultural traditions and those of the West, in the knowledge that both are necessary to wellbeing. This is an established Yolŋu standpoint (Shore et al., 2014). The attempt is to secure threatened cultural foundations, and use them to build the ability of students and local teachers to engage with the Western world from a position of equality. In acknowledging the importance of Yolŋu and Balanda culture, and their interdependence, the program engages with the challenge of intercultural realities.
Like the WCE initiative, the program is an outcome of intercultural partnership. It is an accommodation by Yolŋu to Balanda worldview, culture and practices that depends for its effectiveness on reciprocal Balanda accommodation to Yolŋu. The success of student mentoring guided by raypirri depends on Balanda teachers' ability to accommodate to Yolŋu, which implies their knowledge of, and ability to integrate gurrutu in their classrooms. This is to build a better relational environment, which grounds learning in the Yolŋu world (and in education philosophy generally). For Yolŋu, a flourishing studentteacher relationship and, so, student learning, depend on connectedness; the prioritisation of curriculum outcomes, time segmentation and Western-style discipline undermines students' connectedness with them. That is, an approach that does not find a cultural synthesis can trouble Yolŋu students' emotional connection, relationship, and engagement and learning. Mentors shared with researchers that, without emotion and connection, there can be little respect for the adult and little learning. In remote Yolŋu contexts, Yolŋu culture is necessary to school learning. This is paralleled by the reality that numeracy and English literacy, and critical political literacy (as Nakata, 2012, argues for), are also necessary to cultural continuity. Culture and school learning are mutually dependent, and gurrutu and raypirri engage their interface.
The Galiwin'ku case illustrates interculturality and complexity in other ways. The program evolved as those involved adapted it in response to changes in the environment, as when large (e.g., Yolŋu) and small (e.g., school) cultural imperatives intersected, government policy changed, funding opportunities arose, education system requirements changed, and people came and went with their varied relational abilities. As a complex system it was always dynamic, vulnerable to the consequences of small changes, and nonlinear (see also Bat & Guenther, 2013). The larger partnership depended on others within each organisation, governed by internal structures. In Shepherdson College, this meant changing relationships with different principals, teachers and trainers, and, in Yalu, those of manager, development adviser, translator, mentor and campus-based researcher. Managing the relationships required that each partner watch for relational shifts, listen respectfully and mindfully, and judge how strongly to press for and give ground, ask for trust, or claim past loyalties, etcetera. Yolŋu trusted that Balanda acted not as archetypical Balanda but as individuals, and vice versa. Relational stability was produced by collaboration but always temporary. For instance, the early cultural awareness program stalled when the balance of cultural style appeared to Yolŋu to favour Balanda logic, then flourished when Yolŋu logic prevailed to a greater extent. The program had initially taken for granted what was felt to be Balanda logic of education, and Yolŋu ATs found it to be non-negotiable. To them, it went too far in the Balanda direction, and they disengaged. The partnership recovered when a more fluid structure more amenable to local exigencies was adopted. (Presumably the program would also falter were the terms to be too favourable to Yolŋu, and the school would withdraw.) This fluidity had to be constantly worked at.
Further, the balancing act persisted over time because of the chance alignment of people with cultural knowledge and intercultural sensibility and skill-three criteria which Ting-Toomey presents as key to identity negotiation competence (2005, pp. 226-227). Key outsiders had long experience in the community and Yolŋu Matha fluency. They respected the paramountcy of Yolŋu control and the school's equally inescapable responsibilities. They worked in trusting relationships with Yolŋu partners attuned to the intercultural complexities. Each was aware of the other's categoric characteristics that suggested potential strengths and weaknesses, knew the biases that colonial and recent history introduce, yet related to each other as individuals.
Conclusion
It has been suggested that the Galiwin'ku program enhanced student interest in school and led to improvements in behaviour and learning. That success was portrayed as an outcome of successful management of an environment in which individuals' multiple, overlaid, intertwined and subtly different imperatives dynamically intersect. The reference to subtle differences is important, as it is the contingent expressions of universal human concerns that often go unnoticed, but must be noticed and managed. For example, Yolŋu say that emotion and connection are critical to Yolŋu students' learning, and so is discipline, as in raypirri. But the Western heritage of school culture also recognises emotion, connection and discipline. There is no radical departure in this. It is the variance in emphasis and application of universal human concerns that demand attention. Once a basic appreciation of cultural difference is gained, it is the fluidity and nuance of coexistent sameness and difference that most trouble intercultural relations, intersubjective connection and learning.
This paper suggests that research in remote settings such as Galiwin'ku requires equal respect for the meanings of the Indigenous and other research partners, and participants in the research, but that, for the full benefits to accrue, that respect must be critical. That is, neither can be accepted without critical appraisal. Equal respect is necessary to the subjects' preparedness to reveal their everyday lived perceptions and to researchers' fully engaged listening. It is necessary to the capacity to examine the local perceptions in the light of the social scientific, and so to the creation of productive new knowledge. These conditions allowed the revelation of the Galiwin'ku program.
The paper also suggests that interculturality must be addressed if school and further education is to be effective for remote Aboriginal students. In this respect, it reiterates calls in the work of a number of scholars, including d'Arbon and colleagues (2009), Frawley and Fasoli (2012), Bat and Guenther (2013), Bat and Shore (2013), Shore and colleagues (2014), and Guenther and colleagues (2016). Their work suggests changes to accommodate intercultural complexity, beginning with the adoption of a "red dirt" way of thinking (Bat & Guenther, 2013;Guenther et al., 2016) and listening carefully to community voices, so favouring a ground-up collaborative approach over the usual top-down linear approach produced by metropolitan imaginaries. This paper adds the notion that a valid vision for schooling and guide for reform would be to invest in producing interculturalists like the Yalu mentors (RG and ELM), WCE researchers and others involved in Galiwin'ku, since they have the attitudes and skills to celebrate their cultural particularity and to exceed it, to transcend ethnocentrism.
This establishes grounds for more research into interculturality in remote life and schooling, the impacts of inappropriate communication on student learning and the ways in which educators can negotiate intercultural communication more effectively. It also implies training that can lead them to become something like Watkins's "therapeutic selves" (1978), who watch their students intently, listen to their every utterance, observe their every nuance of postural and gestural change, perceive slight inflections of voice and changes in associations, listen to the unconscious implications underlying their overt statements, and record their tinges of anxiety, insinuations of hostility and subtle denials of stated belief (adapted from Watkins, 1978, pp. 251-252). That implies training to have educators become interculturalists, insofar as they can manage the facework dynamics of the intercultural encounter that is remote schooling.
About the authors
Terry Moore is a sociologist. His first career was in Indigenous education in Far North Queensland, and he has been involved in tertiary teaching and research in Indigenous issues since 1998. His work focuses on the dilemmas of superdiversity that face Indigenous Australians in negotiating their increasing social, cultural and subjective complexity, and the state in its role in Indigenous education and governance. He is interested in the challenges posed by difference for social cohesion. He is an adjunct researcher at the University of Tasmania.
Eliani Boton is an eLearning professional with extensive experience and qualifications in the design, development and evaluation of education programs, involving blended, hybrid and fully online course environments. She is skilled in identifying pedagogically effective strategies that align online learning and assessment activities with a focus on higher student engagement.
Cat Street has worked for 10 years in education, health and housing settings, internationally and in the Northern Territory. She is now Research, Monitoring and Practice Manager at Curijo. She has expertise in research, evaluation, strategy and program management. Her interests include cultural safety in research and evaluation and the use of evidence to inform decision-making in Indigenous policies and programs.
Rosemary Gundjarranbuy is a senior Yolŋu educator and researcher in school-based and health education. She has worked on a range of community projects related to youth wellbeing, aged care and | 2022-12-17T16:20:21.224Z | 2022-12-14T00:00:00.000 | {
"year": 2022,
"sha1": "a706c993cb4da360d468dd8f1776d9c2377e0d47",
"oa_license": "CCBY",
"oa_url": "https://ajie.atsis.uq.edu.au/ajie/article/download/38/560",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "813ff83df490ec63a92df77b8d13e48529f73416",
"s2fieldsofstudy": [
"Education"
],
"extfieldsofstudy": []
} |
119178233 | pes2o/s2orc | v3-fos-license | Correlation effects during liquid infiltration into hydrophobic nanoporous mediums
Correlation effects arising during liquid infiltration into hydrophobic porous medium are considered. On the basis of these effects a mechanism of energy absorption at filling porous medium by nonwetting liquid is suggested. In accordance with this mechanism, the absorption of mechanical energy is a result expenditure of energy for the formation of menisci in the pores on the shell of the infinite cluster and expenditure of energy for the formation of liquid-porous medium interface in the pores belonging to the infinite cluster of filled pores. It was found that in dependences on the porosity and, consequently, in dependences on the number of filled pores neighbors, the thermal effect of filling can be either positive or negative and the cycle of infiltration-defiltration can be closed with full outflow of liquid. It can occur under certain relation between percolation properties of porous medium and the energy characteristics of the liquid-porous medium interface and the liquid-gas interface. It is shown that a consecutive account of these correlation effects and percolation properties of the pores space during infiltration allow to describe all experimental data under discussion.
INTRODUCTION
Energetics of "nanoporous medium-nonwetting liquid" systems is one of the new directions in basic and applied research (see e.g. [1][2][3][4][5][6][7][8][9][10][11][12]). In the simple model of a porous medium in the form of cylindrical channels, threshold pressure is described by the Laplace-Washburn equation p = 2σ/R |cosα|, where σ is the surface energy of the liquid, R is the pore radius, and α is the contact angle (for a nonwetting liquid, α > 90 • ). For filling nanometer-sized pores by a nonwetting liquid with a surface energy of 0.05 ÷ 0.5 J/m 2 , this threshold pressure is 10 2 ÷ 10 3 atm. When the liquid passes from the bulk state to a dispersed state in pores of the nanoporous medium with a specific volume of ∼ 1 cm 3 /g, the absorbed and accumulated (returned when the liquid flows out) energy can reach 10 ÷ 100 kJ/kg. This value is one order of magnitude higher than the energy observed for widely used materials, such as polymer composites and alloys with the shape memory effect [10][11][12].
Among the systems under investigation are silochromes, zeolites with liquid metals, hydrophobized silica gels, and zeolites with water and aqueous solutions of organic compounds and salts. In recent years, hydrophobized nanoporous media have become available owing to the development of the method used for modifying the surface of nanoporous media, generally, with alkyl chlorosilanes [3,6,7,[12][13][14][15][16][17][18][19][20]. To date, nanoporous media with different pore shapes, porosities, specific surface areas, specific volumes, average pore radii, and pore size distributions have been studied [1-7, 9-12, 17-50]. The investigations performed thus far have been concerned primarily with equilibrium properties. Experiments have been carried out at a low compression rate of the system when the rate of increase in the pressure (ṗ) in the liquid-porous medium system is (10 −3 − 1) atm/s. In the infiltration-defiltration cycle, there is a hysteresis, so that the threshold pressure of infiltration is higher than the pressure of defiltration. Moreover, the majority of the systemsare characterized by the phenomenon of nonoutflow of a nonwetting liquid when a part of this liquid remains in the porous medium as the excess pressure decreases to zero. The absorbed energy is determined by the product of the volume of filled pores and the difference between the infiltration and defiltration pressures. In frameworks the model of cylindrical channels, these pressures are described by the Laplace-Washburn equation with different contact angles.
The revealed difference between the infiltration and defiltration pressures and the absorption of the mechanical energy observed in the infiltration-defiltration cycle due to the pressure hysteresis, as a rule, have been explained by the hysteresis of the contact angle; however, the mechanism responsible for the appearance of the latter hysteresis has remained unclear [4,6,7,[19][20][21][22].
It has been established that the infiltration and defiltration pressures depend on the temperature and that, for the porous medium with a disordered structure of pores, the defiltration pressure increases (by several factors) with an increase in the temperature from 280 to 350 • K, whereas the infiltration pressure decreases only slightly (by less than 10%) or remains constant [21,22,24]. This means that, during infiltration and defiltration, the phenomenological contact angles differently depend on the temperature. For zeolites, the revealed temperature dependences exhibit a more complex behavior; moreover, the volume (V ) memory effect can be observed with an increase in the temperature and its subsequent decrease. It is worth noting that this effect is one order of magnitude (in ∆V /V ) stronger than that observed for known alloys and composites [10,11].
At present, there exist several hypothesis regarding the nature of the contact angle hysteresis. This hysteresis has been attributed to the rough surface of pores, the chemical inhomogeneity of the surface, and the dependence on the direction of the liquid motion [19,20].
However, these conceptions cannot explain the different dependences on the temperature of infiltration and defiltration pressures, and hence the temperature dependence of contact angle.
In the framework of the concept that the porous medium is a system of cylindrical channels, the absorbed energy is expended for forming a liquid-porous medium surface, which appears in the course of infiltration and disappears during defiltration at different pressures due to different contact angles [2, 6, 7, 19-22, 46, 47]. In case of the closed hysteresis loop in the infiltration -fulldefiltration process initial and finite states of the system are similar, the internal energy change ∆E cicle = dE = 0 and the work done to perform the filling of the porous medium should be equal to the thermal effect dA = dQ. The measurements carried out in [19,51] showed that rise of temperature in the hydrophobic silica gel-water systems under investigation during infiltrationdefiltration cycle was <10 −3 K. On the other hand, it was found in [43] that when one-third of the porous medium volume of a similar system was filled and the liquid-porous medium interface area change was not equal to zero, the temperature did not increase within the limits of error (≤ 0.1K), while the estimation provides the temperature increase by ∆T = 0.8K.
In [2] the dependence of the thermal effect ∆Q, which is accompanied by the thermal effect due to elastic compression of water and the porous medium, on the filling degree in the hydrophobic silica gel-water system was measured. The author found that during infiltration this porous medium with ϕ ≈ 0.22, the heat generation took place (∆Q < 0), whereas dσ dT < 0 for water [51] and, consequently, in case of independence of the contact angle from temperature during the formation of liquid-solid interface with area ∆S one would expect the ∆Q = − dσ dT ∆S value to be positive, i.e. heat absorption shall be observed. In [2], it was experimentally found that, as the degree of infiltration increases, the quantity |∆Q| reaches a maximum at θ ≈ 0.6 and decreases to zero after the complete infiltration of the porous medium: ∆Q = 0. This result led the author of [52] actually to the conclusion that the first law of thermodynamics is violated in the process under investigation. Thermal effect associated with the infiltration of the modified porous medium was also observed [47,48]. It was found that for the investigated porous media with ϕ = 0.4 and ϕ = 0.6 the value ∆Q < 0 reduces with increasing degree of filling, reaching a minimum at complete filling.
Note that the energy absorption can not be to all appearance explained by viscous dissipation, since, as it ascertained in [59], change in viscosity of nonwetting liquid (aqueous solution of CaCl 2 ) by 7 times does not alter the dependence of the filled volume on time, the threshold infiltration pressure and rate of filling hydrophobic silicagel.
Statistical theory of fluctuations allows us to generalize the Laplace-Washburn equation in case of filling the pores in the system of interconnected pores [4,5]. The proposed approach makes it possible to analyse the probability of the fluctuation infiltration pores with the size R near the threshold pressure. If the probability of w ≈ exp(−δA/T ) increases continuously with growing pressure, then the pore can be filled with liquid (δA is work expended for infiltration of one pore, δA(p, σ, δσ, R)). The value δA in takes into account a contribution of formation a liquid-porous medium interface and a surface of menisci at the mouths of the filled pores. Threshold infiltration pressure is determined from the condition δA in (p out , σ, δσ, R) = 0, and the defiltration pressure is determined from the condition δA out (p out , σ, δσ, R) = 0. This makes it possible to calculate contact angles during infiltration and defiltration, to describe the dependences of the volume of system liquid-porous medium on the pressure and calculate the volume of confined liquid after defiltration.
In [53] the energy absorption during the infiltration-defiltration process within the framework of fluctuation model is associated with the energy of formation of menisci. However, in this model is assumed, that the work of filling the pores (liquid defiltration) and, consequently, the probability of infiltration (defiltration) does not depend on the degree of filling of the porous medium. Therefore, if during infiltration and defiltration the contact angle is independent of the filling degree, the area of menisci which are formed should be equal to the area of menisci disappeared, and therefore the formation energy at full infiltration (defiltration) should be equal to zero. It also does not allow to explain the results of [2,46,47].
Thus, the experimental data currently known seems to be self-contradictory and well-known traditional mechanisms of energy absorption do not make it possible to explain the infiltrationdefiltration hysteresis, temperature dependences of the infiltration and defiltration pressures and special features of heat generation during the filling of nanoporous media with nonwetting liquid which are observed.
As shown in this paper (Section 2), to describe the discussed phenomena we should take into account the structure of pores in disordered porous medium.
During infiltration into porous medium with an increase in the number of filled pores, surrounded by empty filled pores a formation meniscus in pore mouths is changed by theirs disappearance on the average ensemble of pores with increasing number of full filled pores surrounded by empty loadable pores. In this case expenditure of work on formation meniscus are replaced in the gain energy during theirs disappearance. Thus, the correlation effect of the relative position of filled and empty pores must be taken into account at describing the fluctuation infiltration. In addition, the porosity of the medium determines the possible neighbors number of pores and possible macroscopic set of filled pores in the percolation cluster, through which liquid flow from the surface of the porous medium to fill the pores is possible. This means necessity of taking into account another correlation effect -the spatial arrangement of pores in the medium, arising as a result of probabilistic system realization of interconnected pores in the infinite percolation cluster.
Consequently, it is possible to sugges the mechanism of energy absorbtion during filling of the nanoporous medium with nonwetting liquid is proposed (Sections 3, 4). In accordance with this mechanism, the mechanical energy absorption is a result of expenditure of energy for the formation of menisci in the pores on the shell of the infinite cluster and expenditure of energy for the formation of liquid-porous medium interface in the pores belonging to the infinite cluster of filled pores. It has been demonstrated (Section 4), that the infiltration-defiltration cycle can be closed with the complete defiltration of the liquid and the reproducibility of the cycle when the specific relationship between the percolation properties of the porous medium and the energy characteristics of the liquid-porous medium and liquid-gas interfaces is satisfied. It turned out that, depending on porosity and, consequently, the number of nearest neighbors of the filled pores, the thermal effect can be either positive or negative. This makes it possible to explain the known experimental data [43,46,47] mentioned above and to describe the temperature dependences of the infiltration and defiltration pressures [21,22,24]. The description of experimental data on the basis of the proposed approach is given in sections 5, 6.
THE MODEL OF A POROUS MEDIUM. INFILTRA-TION FLUCTUATIONS
Let us consider a disordered porous medium infiltrated with a nonwetting liquid. It is assumed that the half-width of the pore size distribution δR satisfies the inequality δR/R < 3, so that the fulfillment of this inequality ensures the independence of the percolation threshold from the radii of pores [54].
Filling of the porous medium is a process of liquid infiltration into the disordered porous medium which contains pores of different size. It is assumed that the size of the porous medium a is much more than the maximum size of the pores R max (a ≥ 10 3 R max [55]) so that the porous medium can be regarded as infinite. Obviously, the infiltration of all the pores can take place only when the pores are connected with the surface and form a connected system. Consequently, the filling of the porous medium can take place only when the pore system in it is far beyond the percolation threshold (ϕ > ϕ c ), ϕ is its porosity equal to the ratio of pores to the porous medium volume, ϕ c is the percolation threshold, which is the characteristic of the porous medium. For 3D systems, the percolation threshold ϕ c = 0.18 [56]. At the same time, the connectivity of pores with one another is the result of the formation of infinitely large clusters of pores at ϕ = ϕ c . Figure 1 shows the dependence of the probability normalized to unity of a pore belonging to the infinitely large cluster on porosity ϕ [56,57].
It can be seen from Fig. 1 that, in the vicinity of the percolation threshold ϕ c , only a small number of pores (∼ 1%) belong to the infinite cluster; therefore, in this case, only a small fraction of these pores, as well as pores that belong to the finite clusters connected with the boundary of the porous medium, can be infiltrated. At increasing porosity and for ϕ ≫ ϕ c P (ϕ) → 1 and, consequently, the pore space becomes homogeneous due to the growth of the infinitely large cluster of pores. Under these conditions, the infiltration of the porous medium can be described as the infiltration of an infinite cluster of pores. It is this infiltration that will be considered in the present paper.
It is assumed that thermal fluctuations at the pressure p in the vicinity of the infiltration threshold of the porous medium bring about the formation of macroscopically small regions in the form of clusters consisting of N pores filled with a liquid. Each cluster arises at the boundary of the porous medium and, in the view of the boundedness of the pore volume, grows through the attachment of the other filled pores to it. We believe that, at the beginning of the growth, each cluster can be considered a system of branched chains consisting of filled pores. In the course of infiltration of the porous medium with the liquid, the external pressure does the work. This φ P Figure 1: The dependence of the probability normalized to unity of a pore belonging to the infinitely large cluster on porosity ϕ process is accompanied by the formation of energetically unfavorable surfaces of both the menisci of the liquid in pores and the liquid-porous medium interfaces. Moreover, the state of the gas in the pores and the elastic state of the porous medium change as well. If the adiabatic work of formation of an infiltration fluctuation is δA(N) and the energy of dissipation due to the friction can be disregarded (see [55]), the probability of the formation of a fluctuation can be written as w ∼ w 0 exp(δS) [58], where δS = −δA/T is the fluctuation of the entropy. Therefore, an increase in the quantity δA (δA > 0) with an increase in the number of pores N in the cluster leads to a decrease in the fluctuation probability. This corresponds to the thermodynamic stability of the initial state of the system. For δA ∼ T , the infiltration fluctuation can increase. In this case, the system becomes unstable and the liquid begins to infiltrate the porous medium.
The infiltration of a nonwetting liquid under the pressure p in a porous medium requires a certain amount of work to be done to fill the pores of the porous medium. For this purpose, it is necessary to overcome a certain critical pressure, which is the Laplace pressure p c (R) ∼ δσ R for an isolated pore with characteristic size R. δσ = σ sl − σ sg , where σ sl is the surface energy of the solid-liquid interface, σ lg = σ is the surface energy of the solid-gas interface, δσ = σ sl |cos α|, α is the contact angle.
At a pressure lower than the critical value p in , the adiabatic work satisfies the inequality δA(N) > 0 at any value of N and the fluctuation probability decreases with an increase in N. Therefore, the fluctuation probability is equal to zero for any macroscopically large number of pores. Fluctuations of finite length arise, but no infiltration of the porous medium occurs. At a pressure in the vicinity of the critical value p in , the work is δA ∼ T and thermal fluctuations in the system can lead to the formation of clusters from N pores. At a pressure p > p in , the infiltration of individual pores becomes energetically favorable because the quantity δA(N) is negative. Since the work is δA ∼ N, the fluctuation probability at a pressure p > p in is w ∼ 1. The pressure difference p − p in causes the liquid to move in the porous medium. Now, we consider a porous medium immersed in a nonwetting liquid under an external pressure p, which does the work in the course of infiltration of the porous medium. Let δA (p) be the work expended for providing the fluctuation infiltration of one pore. According to [5], the expression for the work δA (p) with due regard for the formation of menisci can be written in the form: Here, V is the volume of the pore, S is the surface area of the pore, S m is the surface area of the menisci, and σ is the surface energy of the liquid. For a spherical pore with the radius R, the work δA (p) can be represented in the form [5]: where η is the ratio of the meniscus surface area to the pore surface area. A similar expression can be written for a cylindrical pore. For example, we write the following expression for the work expended for providing the fluctuation emptying of a cylindrical pore by nucleation with the nucleus length L and the radius R: where x = L/2R. It follows from relationship (2) that the infiltration of the pore becomes energetically favorable at a pressure higher than the critical value p in . The critical pressure is determined by the equality of the work on the fluctuation infiltration of the pore to zero. An analysis of the critical pressure for a spherical pore was performed in [5]. Expression (3) implies that, devastation of the pores becomes energetically favorable at a pressure less than the critical pressure p out , which depends on the nucleus length L and the nucleus radius R and, at L → ∞, transforms into the Laplace pressure p out ∼ 2δσ R . This means that, in the work expended for infiltrating the pore, the change in the surface energy of the pore dominates over the energy expended for forming menisci.
For characteristic values of the parameters of the porous medium and the liquid σ ∼ δσ ∼ 0.5 kJ/m 2 , R ∼ 10 nm, and, χ ∼ 10 −4 atm −1 , when the pressure deviates from the value p I by δp = −10 −2 p I , the work changes from δA ∼ T to ∼ 1 eV. In this case, the change in the probability of infiltration fluctuation at the temperature T = 400K is equal to ten orders of magnitude. Therefore, for systems with the aforementioned characteristic parameters, the probability w changes abruptly from 0 to 1 in a narrow pressure range (δp/p ∼ 10 −2 ) in the vicinity of the pressure p in . It should be noted that the inclusion of the gas filling the pores in the analysis leads to the appearance of an additional contribution to the work δA g . The value of this contribution δA g during filling one pore with the volume V can be estimated under the assumption that it is equal in the order of magnitude to the specific evaporation energy of the liquid µ 0 multiplied by the density of the gas ρ g at the pressure p, i.e., δA g ∼ µ 0 ρ g V . For water, we can write µ 0 ∼ 2.2 · 10 3 J/kg and ρ 0 ∼ 1.0 kg/m 3 . Assume, that the filling of the porous medium occurs at a pressure of the liquid p in ∼ 1 · 10 7 J/m 3 . In this case, the additional contribution to the work due to the presence of the gas in the pores is given by δA g ∼ µ 0 ρ g V ∼ 10 −2 · δA. Nominally, accounting of the influence of gas reduces to the appearance of additional summands in equations (2) and (3), each of them proportional to the gas pressure at a given temperature. The value of these summands A gin and A gout for typical infiltration pressure p in ∼ 200 atm and p out ∼ 1 − 10 atm, for pore radius R ≈ 10 nm is A gin ∼ A g ∼ A gout ∼ 10 −2 eV in order of magnitude. Therefore, the influence of the gas in the pores on the infiltration of the porous medium can be ignored at considering the filling of the pore, but accounting of these influence may be important at considering the effects of the defiltration liquid the cylindrical pores.
Since the pore can be either filled (probability w ∼ 1, δA (p) < 0) or empty (probability w = 0, δA (p) > 0), the normalized probability can be written in the form [5] Note, that relations (3) and (4) explain the obtained in [6,7] experimental data for the dependence of infiltration pressure and defiltration pressure on the pore size.
Pores are not isolated in a porous medium, but they are connected with one another by throats (mouths), in which menisci are formed during infiltration of a certain pore. Thus we can assume that the medium subjected to infiltration is the heterogeneous medium which consists of full and empty pores playing the role of white and black spheres, respectively, in the percolation theory [56]. Such medium can experience percolation transition occurring via the formation of clusters of accessible pores and followed by infiltration of nonwetting liquid into such formations. [55]. In addition to that, filling of the macroscopic volume of a porous medium occurs by infiltration in the infinitely large cluster of accessible pores [55].
Below, we will consider the infiltration of pores located on the shell of an infinite cluster consisting of filled pores. In this case, the condition δA = 0 determines the pressure necessary for the infiltration of a pore on the shell of the infinite cluster of filled pores.
WORK AND THERMAL EFFECT IN THE INFILTRATION-DEFILTRATION CYCLE
Let us calculate the work δA and the thermal effect ∆Q for an arbitrary degree of infiltration θ of the porous medium. The thermal effect ∆Q in filling of a porous medium by nonwetting liquid comprises the thermal effect ∆Q p due to the formation of the liquid-solid interface , the thermal effect ∆Q w related to formation-disappearance of menisci and the thermal effect ∆Q u related to the compressibility of the nonwetting liquid-nanoporous medium system.
The ∆Q p , ∆Q w and ∆Q u values can be calculated using thermodynamic relations [58], which determine the thermal effect in formation of the surface ∆Q s : Here, ∆S is the change in the system surface.
To calculate the thermal effect ∆Q s let us suppose that each pore in a porous medium has z nearest neighbours and pores contact each other by throats, each of which has an area S z . If an empty pore contacts a full one, the meniscus is formed in the throat. A pore in a porous medium can be filled only if liquid can reach it. In compliance with above-mentioned assumptions this condition can be satisfied by formation of an infinitely large cluster of filled pores. In this case, only those pores which belong to the shell of the infinite cluster will be filled. It is possible to show that the contribution of filled finite-size clusters (which liquid can reach via filled clusters contacting with the surface of the porous medium)to the filled volume is small. Distribution f (N) of the number of pores in clusters of finite size near the percolation threshold is determined by the scaling dependence f (N) ∼ 1 N τ , τ = 2.2 [56]. Therefore the bulk of the cluster contains one or more pores, which are mostly not associated with the surface of the porous medium. Liquid cannot reach such pores and, consequently, they are not filled at θ ∼ θ c . Taking it into consideration, we can represent the thermal effect δQ related to filling of one pore as: S = 4πR 2 is the area of the surface of a pore with radius R, z is a number of nearest neighbours, W (θ) is the difference (averaged over the ensemble of pores) between the numbers of menisci before and after the infiltration of the pore per the nearest neighbor of the infinite cluster.
Considering that filling of a porous medium is the result of formation of the infinitely large cluster of filled pores and taking into account the normalized probability P (θ), we obtain that the quantity of heat per a pore released in the process of filling the porous medium to the degree of filling θ can be written as: Here, f (R) is the function normalized to unity of the size distribution for pores. For a disordered porous medium average values in (8) can be calculated in the framework of a specific model of a porous medium. We will use the model of randomly arranged spheres in which pores represent randomly arranged spherical holes [59]. This model does not take into account correlations in location of pores with different radii in accordance with assumption made about the narrowness of size distribution for pores (δR) ≪R. In accordance with the model, the average number of nearest neighboursz, associated with porosity of medium ϕ, and the area of a throat can be written in forms [59]: Using expression (8) and taking into account that (δR) ≪R, we obtain from (9): To calculate W (θ) we consider an empty pore located on the perimeter of the infinite cluster of filled pores. Let us suppose that this pore contacts the infinitely large cluster of filled pores via n throats. Thus, menisci are formed in all n mouths and menisci are absent in the remaining z − n throats. After filling this pore, menisci which were there at the beginning of infiltration disappear and the number of menisci will be equal to z − n. In this case we can write W (θ) as: The first factor under the summation sign determines the probability that an empty pore contacts the infinite cluster of filled pores n times, the second factor is the probability of finding the empty pore close to the infinite cluster, provided that this pore is surrounded by z − n empty pores and therefore has z −n throats. The third factor determines the difference between a relative number of menisci after (z −n) and before (n) the filling of the pore. The binomial coefficient takes into account variants of allocation of n menisci on number of pore nearest neighbours. Note that the obtained expression coincides with the full perimeter of the infinitely large cluster calculated in [60], if the third factor is substituted for unity.
The sum in Eq. (11) can be calculated analytically: Relations (9), (10), (12) determine the thermal effect during filling a porous medium with porosity ϕ to the degree of filling θ.
The work ∆A expended for infiltrating the porous medium to the volume determined by the fraction θ and the corresponding work expended for infiltrating one pore δA in (θ) can be calculated from the thermodynamic relationship A = σdS [58]. By using expression (2) for spherical pores, we obtain: The sum of heat (10) and work (13) determines the change in the energy of isothermal infiltration of the porous medium: It follows from (14) that the change in energy of the system during filling the porous medium is determinded by the specific surface energies σ and δσ, geometric properties of the porous medium and the evolution of the infinite cluster of filled pores, which depends on the properties the disordered porous medium (Fig. 1).
For the calculation of the work δA(θ) and the thermal effect ∆Q v arising upon the defiltration of the liquid from the porous medium, it should be noted that, in the infiltrated porous medium, the defiltration of the liquid leads to the formation of empty pores surrounded by at least one filled pore connected through other filled pores with the surface of the porous medium. As in case of filling, the formation of an empty pore goes with the change in surface energy of the liquid-solid and liquid-gas interfaces as well. The change in surface energy is associated with formation-disappearance of menisci [5].
Taking into account this fact, the work expended for emptying the pore in the porous medium δA(θ) with the degree of infiltration θ, the work for defiltration liquid ∆A v from 1 to θ the degree of infiltration, and the thermal effect δQ v associated with the defiltration of the liquid from one pore in the porous medium can be written in the form: Relations (7) and (15) differ in sign of the last term and the functions W (θ) and W 1 (θ) which determine the difference per one nearest neighbour between the number of menisci before and after infiltration (defiltration) in pores.
In contrast to the case of infiltration, the defiltration of the liquid occurs initially through the formation of individual empty pores and clusters of empty pores with a decrease in the pressure and, after the infinite cluster of empty pores is formed, through the formation of pores on the shell of this cluster. Upon the defiltration when the low degree of infiltration is reached, the liquid can be retained in the porous medium if it is contained in individual pores or clusters of filled pores surrounded by empty pores with smaller sizes from which the liquid defiltrated at higher pressures. However, as the number of neighbors of empty pores increases, according to relationships (15) (see below Figs. 1-3), the defiltration of the liquid becomes energetically more favorable; i.e., it should proceed at higher pressures. Therefore, the quantity W 1 (θ) should be defined as the difference (averaged over the ensemble of pores) between the numbers of menisci before and after the emptying of the pore on the shell of the system of empty pores. Taking it into consideration, calculation of W 1 (θ) gives: The thermal effect, the work, and the change in the energy during the defiltration, when the degree of infiltration varies from 1 to θ, can be written in the form similar to relationships (10), (13), and (14): Expressions (17), like relationships (10), (13), and (14), are valid for the case of an isothermal process. This implies that they can be used for describing experiments in the case where the characteristic time of heat transfer (removal) τ Q is considerably shorter than the characteristic time τ V of change in the volume of the nanoporous medium-nonwetting liquid system: τ Q ≪ τ V . At τ Q ≥ τ V , the temperature and, correspondingly, the quantities σ, δσ, dσ/dT , and dδσ/dT become dependent on the time and, hence, on the degree of infiltration θ. In this case, they should be introduced under the integral sign in relationships (14) and (17). The inequality τ Q ≪ τ V impose constraints on on velocity of compression system under the study of its equilibrium properties.
CONDITIONS FOR THE CLOSED CYCLE
It can be seen from relationships (14) and (17) that, if after the increase in pressure and the infiltration of all pores of the nanoporous medium with a liquid and the subsequent decrease in pressure and the complete defiltration, the system reverts to its original state, the following relationship should hold true: Here, ∆E is the change in the internal energy of the system upon the transition from the initial state to the final state in the course of infiltration and defiltration. Expression (18) relates the energy parameters of the liquid-solid and liquid-gas interfaces and the macroscopic characteristics of the porous medium, such as the porosity ϕ, the structure of the percolation cluster P (θ), and the quantities W and W 1 , which determine the dependence on θ for the surface of menisci at the mouths of filled pores on the shell of an infinite cluster of filled pores in the case of infiltration and for the surface of menisci in pores on the shell of all clusters of empty pores (including the infinite cluster) in the case of defiltration. It follows from relationships (18) that the absorbed energy in contraction and expansion of the system in the closed cycle is dependent on the quantities σ and δσ and equal to the total heat released upon the formation and disappearance of the liquid-solid and liquid-gas surfaces. This heat is determined by the independent quantities, namely, the derivatives of the quantities σ and δσ with respect to temperature. The integrals in expression (18) account for the different paths of the system in the course of infiltration and defiltration. In the closed cycle, according to relationships (10), (13), (17), and (18), during the infiltration and defiltration, the system undergoes different sequences of equilibrium states that differ in macroscopic sets of filled pores.
In particular, the infiltration of the porous medium according to expression (18) is accompanied by an increase in the number of pores that belong to the infinite cluster of filled pores and by a change in the number of menisci in pores on the shell of this cluster. The defiltration of the porous medium is accompanied by an increase in the number of pores in all clusters (including the infinite cluster of empty pores and single pores) and by a change in the number of menisci on the shell of the entire system of empty pores. As follows from relationships (9), (10), and (16), these sequences of states depend on the porosity ϕ and the number of neighbors z in the system of connected pores. Therefore, in terms of the percolation theory and the model under consideration, the contact angle hysteresis is associated with different (in infiltration and defiltration) macroscopic properties of systems of filled and emptied pores that manifest themselves as different spatial distributions of the liquid in the connected pores. If the sets of macroscopic equilibrium states characterized by the distributions of filled and empty pores in the course of infiltration and defiltration were identical, the total thermal effect in the closed cycle would be equal to zero. In this case, the thermodynamic relationship (18) for the closed cycle is not satisfied.
The closed cycle and, consequently, the complete transformation of the work into the heat was observed for a number of water-hydrophobized silica gel systems [7,17,18,21,22]. In particular, the complete emptying of pores after the infiltration and the subsequent defiltration with a decrease in the excess pressure to zero was observed in [17,18] for the KSK-G silica gel modified by n-alkylsilane molecules (n = 8, 16) grafted to the silica gel surface with a surface density higher than 2 nm −2 . Before the modification, the specific surface area, the pore volume, and the average pore radius for this silica gel were equal to 310 ± 20 m 2 /g, 0.95 cm 3 /g, andR = 5.2 nm, respectively. The values of these quantities after the modification are not presented in [17,18], which complicates the analysis of the results obtained in these works. The closed cycle was also observed in [21] for the water-C8W silica gel (Waters) system in which the silica gel was modified by n-alkylsilane with the chain length n = 8, the average pore radiusR = 4.2, and the specific pore volume of 0.53 m 2 /g in the temperature range 287 ÷ 333 K. The authors established that the small nonoutflow (< 1%) takes place only at a temperature of 278 K. For the systems containing water and the Fluka 60 C8 silica gel, as well as the Zorbax Z4, Z8, Z18, and PEP10C18 silica gels, the closed cycle was observed in [7,22]. In [7], the authors investigated the infiltration and defiltration in four porous media MCM41 with pores in the form of cylindrical channels. These media were also modified by n-alkylsilane with n = 8 and a surface density of 2.1 nm −2 and had the average pore radiiR = 1.3, 1.6, 2.0, 5.4 nm. The phenomenon of nonoutflow was observed only for the porous medium withR = 5.4 nm, whereas the other porous media withR = 1.3, 1.6, 2.0 nm underwent a closed infiltration-defiltration cycle.
For the porous medium with specified parameters of the macroscopic structure of the pore space, the closed cycle with the complete defiltration, according to relationship (18), is possible only when the values of the quantities σ, δσ, dσ/dT and dδσ/dT fall in particular ranges. Therefore, relationship (18) requires a separate detailed quantitative analysis, which will be performed elsewhere. It should be noted that this analysis, in turn, necessitates the knowledge of the values of the quantities σ, δσ, dσ/dT and dδσ/dT ; the porosity; the pore size distribution (which is changed after the modification [19]); and the quantities ∆Q and ∆A. Here, we restrict our consideration to the qualitative analysis of the available experimental data for which relationship (18) is not satisfied and the sum of the work and the heat in the cycle is equal to the change in the internal energy of the system. In this case, the mechanical work dissipated by the system is not equal to the total heat release, as it was described in [44]. The phenomenon of nonoutflow associated with the change in the internal energy is characteristic of the majority of the studied hydrophobic porous media and liquids, namely, water [7,10,19,20,23,24,[33][34][35]37], aqueous solutions of salts [12,27,28], organic compounds, ethylene glycol [5], alcohol [26], and glycerol [45], as well as systems with the liquid metal [1], Wood's alloy [4], and mercury [61,62].
For water, the derivative involved in relationship (18) has the value dσ/dT = 1.5 · 10 −4 J/m 2 K [63]. As the temperature changes from 293 K to 353 K, the surface energy σ in accordance with this value of the derivative changes by ≈ 5%. Such a small change of the surface energy σ in the system containing water and the modified Fluka 100 C8 silica gel appears to be sufficient that the closed cycle will transform into the cycle with a nonoutflow of more than 80% of water with a decrease in the temperature from 353 to 293 K [24]. This cycle is characterized by a change in the internal energy of the system (∆E = 0). The quantity ∆E reflects both the reversible and irreversible changes in repeated infiltration-defiltration cycles. The reversible change can be associated with the adsorption of water on the modified surface [19,20]. This water evaporates already at room temperature (and more rapidly at an elevated temperature), and the first cycle with nonoutflow is reproduced [7,24,25]. The differential thermal analysis performed in [19] demonstrated that water evaporates at a temperature T = 373 K. The irreducible change of the internal energy ∆E is governed by the interaction of the liquid with the surface of the porous medium and depends on the maximum pressure during the infiltration [25], the temperature and time of heating of the system [23], and the procedure used for preparing the surface after the modification [12]. The change of the internal energy ∆E also depends on the length (n) of the grafted n-alkylsilane molecule. It is worth noting that, for water and silica gels, an increase in n from 4 to 18 leads to a increase δσ, a decrease in the nonoutflow and a decrease in the change of the internal energy ∆E [7,8,[16][17][18][19][20], whereas for water and the modified medium, the closed cycle is observed for n = 1 and the complete nonoutflow takes place for n = 8 [32]. An increase in the concentration of the NaCl or CaCl 2 salt in the aqueous solution in specific ranges leads to an increase in the quantity δσ and a decrease in the change of the internal energy ∆E for the Fluka 100 C8 silica gel [25,27,28,32]. A decrease in the concentration of ethylene glycol in the aqueous solution and the hydrophobic silica gel L23 results in the transition from the complete nonoutflow at a concentration c > 15% to the closed cycle (the nonoutflow is less than <5%) [5]. Unlike ethylene glycol, ethanol wets the modified surface of the Fluka 100 C8 silica gel. This brings about the adsorption of ethanol and an increase in the change of the internal energy |∆E| [26].
In recent years, experimental data have been published on the infiltration of hydrophobic zeolites with water and aqueous solutions of salts [10,11,[34][35][36][37][38][39][40]. It has been revealed that, for systems containing water and silicalite 1 (OH), silicalite 1 (F − ), and ZSM-5 zeolite, the hysteresis is not observed and the pressure dependences of the volume of the system measured for the infiltration and defiltration coincide with each other. This means that such systems exhibit properties of an elastic spring without dissipation at ∆E = 0 in the cycle. With a cyclic change in the temperature from 358 to 318 K, these systems manifest a volume memory effect [10,11]. For the water-MFI zeolite system, the dependences of the pressure on the volume and temperature during the infiltration remain unchanged when the rate of decrease in the volume changes by three orders of magnitude [38]. This implies that the Laplace-Washburn and Poiseuille equations are not applicable to the system under investigation. The infiltration-defiltration hysteresis was observed for a KCl aqueous solution and the Zeolyst CBV-901 (HY) zeolite treated with SiCl 4 [11]. This hysteresis depends on the nature of the anion [40] and, for the Y(ZY)-zeolite, on the nature of the cation (Li, Na, K, Cs); in this case, the infiltration pressure decreases with an increase in the cation radius [39].
In the analysis of experimental results obtained for zeolites, it is necessary to take into account that, for a channel (pore) diameter smaller than 1 nm, the liquid acquires properties of onedimensional systems [38,64,65], which differ qualitatively from the properties of the liquid in channels of larger sizes. This problem requires a separate analysis.
In conclusion of the discussion of relationship (18), we should note that it was derived under the assumption that δR/R < 3 in the absence of correlations in the mutual arrangement of pores of different sizes and with the use of the model of pores as a system of randomly arranged spheres. Relationship (18) for the closed cycle does not contain the average pore radiusR; however, the value ofR affects the porosity ϕ and, hence, the number of neighbors z, the structure of the percolation cluster, and the quantities W (θ) and W 1 (θ).
TEMPERATURE DEPENDENCES OF THE INFILTRA-TION AND DEFILTRATION PRESSURES
Now, we analyze the signs of the derivatives dσ/dT and dδσ/dT involved in relationship (18) and the integrals of the quantities P (θ) and [W (θ) + W 1 (θ)]. It follows from relation (10) that the sign of the total thermal effect during infiltration is determined by signs of ∆Q p , ∆Q w values. The sign of the thermal effect due to the formation of the liquid-solid interface ∆Q p depends on the sign of dδσ/dT , which can be ascertained using known dependences of pressure at the beginning of infiltration p in and defiltration p out on temperature. The experiments carried out showed that for all investigated systems (modified silica gel -water, aqueous solutions of salts) the pressure at the beginning of infiltration changes not much as temperature increases dp in /dT ∼ = 0, while the defiltration pressure increases as temperature rises [21,22,24]. Therefore, in order to determine the sign of the derivative dδσ/dT , we calculate the pressures required for the infiltration of one pore on the shell of the infinite cluster and the defiltration of the liquid from an arbitrary pore under the conditions δA in = 0 and δA out = 0, which follow from relationship (4). By assuming that the quantities η,R, and W do not depend on the temperature and taking into account that the infiltration and defiltration begin in the vicinity of the corresponding percolation threshold from relationships (13) and (15), the derivatives dp in /dT and dp out /dT can be written in the following form: It follows from eq. (12), (16) Since dp in /dT ∼ = 0, it follows from eq. (19), (20) that Hence, the sign of dδσ/dT is opposite to the sign of dσ/dT . The coefficient of the surface tension of the liquid-gas interface decreases as temperature increases, and vanishes at critical point so that dσ/dT < 0 [66]. For water, value dσ/dT is −1.5 · 10 −4 J/m 2 K [66]. In this case, from eq. (20) find that for pressure at the beginning of infiltration dp out /dT > 0, which corresponds to experimental data [21,24].
Since the probability that a pore belongs to an infinite cluster is P (θ) ≤ 1 and , it is necessary to analyze the dependences W (θ) and W 1 (θ) in order to determine the sign of the thermal effect in the course of the infiltration and defiltration. The behavior of the quantities W 1 (θ) and W (θ) and, hence, the sign of the integral 1 0 (W 1 (θ) + W (θ))dθ depend substantially on the porosity ϕ. The figures shows the dependences W (θ) and W 1 (θ) calculated from relationships (12) and (16) for different values of the porosity ϕ. W (θ) = 0 for θ < θ c = 0.18. At low porosity ϕ ∼ 0.22 the structure of pores being filled is close to the fractal structure of the low-density infinite cluster near the percolation threshold, so that the growth of its surface is compensated by the decrease in the difference between the number of menisci in final and initial states of the filled pore at θ ≤ 0.3, since in concordance with (9) there are few nearest neighbors of this porez < 3 (Fig. 2). During further infiltration only the reduction in number of emerging menisci occurs (Fig. 2), which reaches minimum at θ ≈ 0.68. If θ = 1 the value W (1) is zero, since the filled porous medium meniscus is absent. For media with high porosity ϕ > 0.3 in which the number of nearest neighbours isz > 3, as the degree of filling θ ≥ θ c increases, the infinite cluster of filled pores grows, accompanied by the growth of its surface and increase in the number of contacts of an empty pore with its neighbours on the shell of the infinite cluster filled with liquid. It leads to decrease in the difference between the number of menisci in final and initial states of the filled pore, so that as θ increases the W (θ) value reaches maximum and then decreases (Fig. 3, 4). In accordance with (12) it will continue until decrease in the difference between the number of menisci in final and initial states of the filled pore compensate the growth of area of the infinite cluster, which will lead to vanishing of W (θ) at θ ∼ 0.5 (Fig. 2, 3). The further growth of the degree of filling leads to further decrease in the difference between the number of menisci in final and initial states of the filled pore while the growth of the surface of the infinite cluster of filled pores slows down. As a result W (θ) reaches maximum at θ ∼ 0.7. The surface of the infinite cluster of filled pores will vanish at θ = 1, which will lead to vanishing of W (1) (Fig. 3, 4). The dependence of the W 1 (1−θ) value, which determines defiltration, is similar to the dependence of W (θ), which determines infiltration. The difference in behavior of W (θ) and W 1 (1 − θ) is connected with the fact that liquid defiltration from the porous medium does not require the percolation cluster formation. Figures 2-4 The calculation of the integral 1 0 (W 1 (θ) + W (θ))dθ shows that, with an increase in the porosity, this integral increases from negative values at ϕ = 0.25 to positive values at ϕ = 0.6 and vanishes at ϕ = 0.3.
Expressions (13), (15), (19), (20) and conditions δA in = 0 and δA out = 0 allow one to calculate the temperature dependences of the infiltration pressure p in (T ) and the defiltration pressure p out (T ). Figure 5 shows the experimental data for silica libersorb 23 (silica gel KSK-G with the modification of 8-tier alkynesilan-C8), Fluca100 C8 [24], and S8W [21] filled by water and calculated according to the considered correlation effects. Experimental data within the measurement error are described by linear dependences with different slopes. With increasing temperature, the pressure of infiltration decreases at about 10% and the defiltration pressure increases in the times for all the porous media. For these environments values of pressure p in and p out also differ at initial temperature. In concordance with relationships (13) and (15), dependences p in (T ) and p out (T ) are described by the quantities W (θ = θ 0 > θ c ), W 1 (θ = 1 − θ 0 ), and by dependences σ(T ) and δσ(T ). The dependence of σ(T ) is known [66]. The value δσ in our experiments for libersorb 23 was δσ = 22 ± 1 J/m 2 . The same value was taken for the other two porous media because of their similar modifications. The value dδσ/dT was calculated from the condition (21). Quantities W and W 1 calculated according to expressions (9), (12), (16). The porosities estimated from the specific pore volumes according to the data taken from [21,24] were as follows: ϕ = 0.33 for Libersorb 23, ϕ = 0.46 for Fluka 100, and ϕ = 0.53 for C8W. The quantities W (θ = θ 0 > θ c ) and W 1 (θ = 1 − θ 1 , θ 1 << 1) corresponded to the maximum values in the curves W (θ) and W 1 (1 − θ) for ϕ = 0.33, ϕ = 0.46, and ϕ = 0.53, respectively.
As is also seen from Fig. 5, the calculated temperature dependences of the infiltration and defiltration pressures for the porous medium Libersorb 23 C8, Fluka 100 C8, and C8W infiltrated with water satisfactorily fit the experimental data.
It follows from relationships (13) and (15) that, in the framework of the proposed model, the pressures of the beginning of the infiltration and defiltration p in and p out are proportional to 1/R if the quantities η, W , and W 1 are independent ofR. This is consistent with the known experimental data [6,7,19,43,50,52]. Such dependence is a consequence of the lack of correlations in the mutual arrangement of pores of different sizes in a model of randomly arranged spheres with a narrow pore size distribution δR/R < 3, characteristic for silica gels. For the porous medium MSM 41 more than the strong dependence of the average radius [6,7] due to the peculiarities of the fluctuation formation of the nucleus in a cylindrical channel (see (3)). (13) and (15). Points are the experimental data taken from [24,43]
THERMAL EFFECT
Expressions (10) and (12) allow one to calculate the thermal effects observed during the infiltration of a porous medium with a nonwetting liquid in different cases as functions of the porosity and surface energies of the liquid and the porous medium.
It was found in [43] that when one-third volume of the porous medium from Sigma-Aldrich was filled by water, the temperature did not increase within the limits of error (≤ 0.1K). A maximum value of the elastic energy in performed tests was about 1 J. Authors [43] estimated the possible increase of temperature, in condition that the work (A) of filling is to raise the temperature. In experiments [43] at defined value A as 2.9 J, the temperature increase should be ∆T = 0.8K [43]. Figure 5 shows the calculated (from relationships (10) and (17)) dependences of the thermal effect on the degree of infiltration of the porous medium (in relative units) due to the formation of the liquid-porous medium surface ∆Q p (θ)/A (Fig. 5a, lower curve) and menisci ∆Q w (θ)/A (Fig. 5a, upper curve) and the corresponding dependence of the total effect ∆Q/A = (∆Q p + ∆Q w )/A (Fig. 5b). In accordance with (10), (12), the value of the thermal effect depends on the value of the integrals appearing in these relations, which, (see (9), (12) and (16)) depend on the porosity. Estimates show that in these experiments the porosity of silica gel after modification was ϕ ≈ 0.68.
The parameter T · dδσ/dT was calculated from the condition (21). Note that W (θ) is equal to zero at θ < θ c . Therefore, in Fig. 6 for comparison with experimental data, degree of infiltration is delayed on the horizontal axis, X shifted by an amount θ c = 0.18 so that the value θ = 0 corresponds θ = θ c in the equation (12).
It can be seen from Fig. 6 that the thermal effect associated with the infiltration of the porous medium with the above parameters is small: at the maximum, it reaches ∼ 0.15A for the degree of infiltration X ∼ 0.17. For X ≈ 0.27, the thermal effect vanishes. This is explained by the different origins of the contributions from the menisci and the pore surface to the total thermal effect, so that each contribution is one order of magnitude larger than the total thermal effect.
Thus, in the performed experiments [44], upon the infiltration of the porous medium with water (the heat capacity is 4.2 J/g K [66]) for the heat release Q ∼ 0.15A = 0.45 J, the maximum increase in the temperature is ∆T ∼ 0.2K. In this case, it should be expected that, according to relationship (18), the change of the internal energy of the system ∆E in the performed experiments upon the transition from the initial state to the final state during the infiltration and defiltration differs from zero and is comparable in the order of magnitude to the work expended for infiltrating the porous medium: ∆E ∼ A. This internal energy is equal to the difference between the energy of menisci in the pores on the shell of the infinite cluster of filled pores and the surface energy of the porous medium-liquid interface in the filled pores. The positive energy of system compression also give contribution to ∆E.
Relationships (9) and (12) make it possible to compare the values of the heat release observed in the experiments on the infiltration of the KSK-G (modification C16) porous medium with water [2]. According to the estimates, the medium porosity in the experiments performed in [2] is ϕ ≈ 0.22. For this medium, we calculated the thermal effect associated with the infiltration. Figure 7 shows experimental data and the dependence of the thermal effect (in relative units) ∆Q(θ)/ |Q 0 | (where |Q 0 | is the maximum heat release upon the infiltration of the porous medium with water; according to [2], Q 0 = 4 J/g) calculated from relationships (9) and (12). The elastic energy of compression of the porous medium and water is determined by water ∆Q u = −T ∆V · dK/dT , where ∆V is the specific decrease in the volume of the water-porous medium system and K = 0.166 mN/mK is the compressibility of water [2]. The maximum value of |∆Q u | is approximately equal to 6 J/g. Figure 7: Dependences of the heat release Q = ∆Q(θ)/Q 0 on the degree of infiltration according to the measurements performed in [2] (points) and calculations from equations (9), (12) with allowance for the compressibility of the porous medium with ϕ ∼ 0.22 Figure 7 also presents the components of the thermal effect measured in [2]: the dependence of the heat release associated with the compressibility of the system (curve 1) and the dependence of the heat release ∆Q p due to the change in the surface energy of pores (curve 2) and menisci ∆Q w (curve 3). It should be noted that the porous medium studied in [2] has a porosity ϕ ∼ 0.22 and, hence, its system of pores is located in the vicinity of the percolation threshold ϕ c = 0.18. Consequently, the infinite cluster of empty pores, which is required for the infiltration of the porous medium, is very sparse [10] and the probability (involved in relationship (10)) that a pore belongs to an infinite cluster is low P ≪ 1, therefore term contribution ∆Q p to the thermal effect is small. In this case, it follows from relationship (10) that, in the infiltration of this medium, the decisive role will be played by the second term in expression (10), which containing integral from W (θ) and corresponds to the contribution of menisci to the thermal effects (∆Q w < 0). It is this circumstance that is responsible for the unusual thermodynamic properties of the system used in [2]. It can be seen from figure 7 that calculated dependences of total thermal effect is in good agreement with the experimental data. Relationships (9)-(12) make it possible to explain the thermal effect observed in the experiments [46,55] on KSK-G (modification C16) with porosity ϕ ∼ 0.4, on PEP100 (modification C18) and PEP300 (modification C18) with porosity ϕ ∼0.65 and ϕ ∼0.58 correspondingly. The average radii of the investigated porous medium are 6.5 nm, 5 nm and 15 nm for the KSK-G (C16), PEP100 and PEP300, respectively. The infiltration of the KSK-G silica gel was carried out at 308 • K and the infiltration of the PEP100 and PEP300 were carried out at 298 • K. Figures 8-10 show experimental data [46,55] and dependences of the thermal effect (in relative units) calculated from relationships (9)-(12) for the KSK-G (C16), PEP100(C18) and PEP 300(C18). The filled volume was normalized and transformed to the degree of filling in calculations for dependences from relations (9)- (12). The elastic energy was estimated from the experimental data of dependence of pressure on the volume, taking into account the compressibility of the container and liquid. The parameter T · dδσ/dT was calculated from the equation (21) as previously. Thus, for the investigated porous media with ϕ = 0.4, ϕ ∼ 0.58 and ϕ ∼ 0.65 value of the thermal effect is negative and decreases with increasing the degree of filling, reaching a minimum at complete infiltration. It is indicates, that menisci give a significant contribution to thermal effects observed in [46,47]. It can be seen from figures 7-10 that calculated dependences of total thermal effect is in good agreement with the experimental data.
The measurements performed in [9,19] during multi-cycle infiltration-defiltration process (number of cycles ∼ 1200) demonstrated that the increase in temperature in the systems containing hydrophobized silica gels and water per cycle is less than < 10 −2 K. The analysis of the experimental data reported in [19] showed that, in the course of the cyclic infiltration-defiltration process, the hysteresis and, hence, the heat release in the first cycle are two times larger than those observed in the 1200th cycle. Thus, in the experiments conducted in [19], there occurs a partial nonoutflow of the liquid from the porous medium, which leads to changes in parameters and characteristics of the porous medium, such as the porosity, the average number of nearest neighbors, and the interfacial energy. In this case, it is necessary to perform a detailed analysis of the changes in the parameters of the porous medium in each cycle with inclusion of the thermal diffusivity of the porous medium and water in terms of the above-derived relationships, which is beyond the scope of our present work. Figure 9: Dependence of the heat release on the degree of infiltration according to the measurement performed in [46] (points) for porous medium PEP300 C18 and calculations from equations (9)- (12) with allowance for the compressibility of the porous medium ϕ ∼ 0.58
CONCLUSIONS
Thus, in this work, we have established the relation between the energy properties of the interface and the macroscopic properties of the pore space in a disordered porous medium and proposed the mechanism of energy absorption during the infiltration of the nanoporous medium with a nonwetting liquid. It turned out that thermal effects observed during the infiltration of the porous medium and the defiltration of the liquid from it can be positive and negative depending on the porosity and, hence, on the number of nearest neighbors of filled pores.
The proposed mechanism is based on the inclusion of correlation effects during percolation infiltration of an infinite disordered porous medium. It is assumed that the infiltration of the porous medium is the result of growth of a percolation cluster consisting of filled pores through the attachment of empty pores (accessible to infiltration) to the shell of this cluster. Menisci appear and disappear in pores on the shell of the percolation cluster in the course of its growth, and these processes depend on the degree of infiltration.
The above analysis is based on the representation of a system of pores in a porous medium in terms of the model of randomly arranged spheres. However, in this model, the correlations in the mutual arrangement of pores of different sizes are ignored and, hence, it is impossible to adequately describe the effects of blocking of the liquid in pores with large radii that are surrounded by pores with smaller radii. Therefore, in the framework of the model of randomly arranged spheres, we can describe the infiltration of disordered porous media only with a narrow distribution of pores over the sizes (δR) ≪R. We assumed that pores in the porous medium have a regular (either spherical or cylindrical) shape, ignored the coordinate dependence of the surface energy of a pore, and operated actually with the average values of this energy on the surface of the corresponding pore.
One of the consequences of the proposed mechanism is a condition that determines the class of systems for which a closed infiltration-defiltration cycle can exist. According to this condition, the initial and final states of the system coincide, the change of the internal energy is equal to zero, and the work expended for infiltrating the porous medium, which is determined by the area of the hysteresis loop, is equal to the thermal effect. Another consequence of the proposed approach is [46] (points) for porous medium PEP100 C18 and calculations from equations (9)-(12) with allowance for the compressibility of the porous medium ϕ ∼ 0.65 the dependence of the effective contact angle on the degree of infiltration of the porous medium with a liquid (see relationships (13) and (15)). This dependence results from different paths on the way to the final state of the disordered porous medium during the infiltration (defiltration) with a liquid. Thus, the proposed approach makes it possible within a unified context to describe the temperature dependences of the infiltration and defiltration pressures for a porous medium with a disordered structure and the thermal effects associated with the absorption of the energy by "disordered porous medium-nonwetting liquid" systems.
ACKNOWLEDGMENTS
In conclusion, it is our duty to thank Professor Yu Qiao for additional information on the experiments described in the publications of his research group. This work was supported by the analytical departmental target program "Development of scientific potential of higher school for 2009-2010" and the Federal target program "Scientific and scientific-educational personnel of innovative Russia" | 2010-03-09T11:03:59.000Z | 2010-03-09T00:00:00.000 | {
"year": 2011,
"sha1": "e6ccaa16a6878fe9f71441ad5b1eb7d7422aa1f2",
"oa_license": null,
"oa_url": "http://arxiv.org/pdf/1003.1841",
"oa_status": "GREEN",
"pdf_src": "Arxiv",
"pdf_hash": "e6ccaa16a6878fe9f71441ad5b1eb7d7422aa1f2",
"s2fieldsofstudy": [
"Materials Science",
"Physics"
],
"extfieldsofstudy": [
"Materials Science",
"Physics"
]
} |
211757492 | pes2o/s2orc | v3-fos-license | Research on the Innovation Path of Logistics Formats Based on 5G Technology
At present, with the commercialization of 5G, all walks of life are affected, and logistics industry, as a supporting industry of the national economy, is no exception. This paper studies the innovation path of logistics formats based on 5G. This paper establishes an intelligent logistics traceability system through the integration of 5G and Internet of things, artificial intelligence, and realizes fully automated transportation through the integration of 5G and Internet of things, which accelerates the development of smart logistics. In addition, the integration of 5G and blockchain can maintain logistics security, reduce the risk of information leakage, and provide ideas for logistics finance.
Introduction
In recent years, with the popularization of artificial intelligence and the rapid development of communication technologies, the logistics industry is constantly improving and optimizing. In order to improve delivery efficiency and reduce transportation costs, from warehousing, transportation to delivery, all logistics links are integrated with advanced technology, which greatly improves the quality of logistics and level of services. According to the data released by the China Federation of Logistics and Purchasing, China's total logistics cost to GDP ratio has been declining from 2014 to 2017, which shows that China's logistics industry has achieved certain results in reducing costs and increasing efficiency. However, compared with developed countries, China's total logistics cost to GDP ratio in 2017 is 14.6% 1 , still much higher than the US 7.2% 2 in the same pe-riod, and also higher than the global level of 11.7% 3 , so we can see that Chinese logistics industry has a huge space in reducing costs and increasing efficiency.
With the deep integration of the traditional logistics industry and the technical means represented by the IoT, cloud computing, blockchain and artificial intelligence, the era of smart logistics has arrived. Smart logistics is a feasible and effective development mode of modern logistics, which can greatly reduce the cost of manufacturing industry, logistics industry and other industries and promote industry upgrading [1]. The development of the above technologies is inseparable from the support of high-level communication technologies, and the development of 5G has provided support for the implementation of products based on the above technologies. At present, the predecessors studied the concept, basic functions and implementation framework of smart logistics [2], as well as the promotion effect of IoT, big data and cloud computing on the development of smart logistics [3] [4] (Figure 1).
Research Design
For the logistics industry, how to use the data efficiently and quickly to coordinate each link of supply chain, so as to make the whole supply chain system operate with low cost and high efficiency, has been the focus.
According to the characteristics of ultra-low delay, high-speed broadband and mass access, 5G can help each link acquire relevant data more quickly, intuitively and accurately. In this way, the data such as transportation, commodity loading and collection can reach the client side, management side and operation side more quickly, and the data sharing and connection between these three sides are also closer. By taking advantage of the high bandwidth of 5G network, the scope and efficiency of the IoT can be further expanded. Besides, the acquisition of environmental information by devices is also faster and richer, providing more valuable reference for logistics companies. However, the research on the deep integration of technologies such as 5G and the logistics industry is not enough. Therefore, from the perspective of 5G, this paper explores the path of deep integration of 5G and logistics format, thereby enhancing the core competitiveness of the logistics industry and providing ideas for the development of logistics finance.
Research Process
The term "format" comes from Japan and was originally dedicated to the "retail formats" [5]. Retail format refers to the different element formed by retail enterprises in order to meet the different needs of different consumers [6]. By referring to the more mature retail formats, Chinese experts and scholars have studied the development path of logistics formats. As the world's second largest economy, the development of China's real economy has driven huge logistics demand, and the logistics industry has become a supportive industry for the national economy [7]. The development of China's logistics industry has passed the stage of mechanization and automation, and now entered the stage of smart logistics. In this process, mutual cooperation in the fields of media, telecommunications and information services has gradually deepened, and there has been a phenomenon of intersection and integration between industries, which can be divided into internal integration and external integration [8].
Logistics is a fusion of distribution, transportation, warehousing and other industries compound service industry [9], in order to meet the demand of "one-stop" logistics services, different functions of the logistics enterprises between strategy and cooperation continued to deepen the reform, aiming at expansion capacity to conduct horizontal integration development, and gradually formed at the beginning of the traditional logistics industry, which is logistics industry of internal integration.
The external integration of logistics industry is also known as the cross-industry integration, producing a new industrial that is different from the original industry and providing additional functions and stronger competitiveness for the logistics industry [10]. With the integration of logistics industry and other industries, more and more new logistics industries keep emerging. Cold-chain logistics, agricultural logistics parks are all the products of the integration betwen logistics industry and the primary industry. In addition, the integration of logistics industry and manufacturing industry improves the industrial chain and value chain of enterprises. Besides, the logistics industry and tertiary industry have also been integrated, such as Logistics finance is the product of the integration of logistics industry and financial industry.
Thanks to the leapfrog development of logistics industry in our country, Chinese logistics industry has gradually integrated with the three major industries, greatly enriching the additional functions of the logistics industry. Logistics finance is the combination of logistics and finance, giving logistics enterprises the role of regulator, to provide supervision services for financial institutions such as Banks. Since 1999, when Zhongchu started the first inventory financing business in China, China's logistics finance business has achieved a breakthrough from scratch [11]. During this period, logistics and technical means were gradually integrated. The application of rf scanning code technology and IoT made the logistics process visible, traceable and controllable, so as to im-prove the problem of information asymmetry between Banks and enterprises and reduce the loan risk of Banks or other financial institutions. At present, the emergence of 5G energizes the logistics format, and realizes the transition and upgrading of the logistics industry again.
Analysis on the Innovation Path of China's Logistics Format in the 5G Era
Nowadays, although the traditional logistics system has made remarkable achievements, with the rapid economic development, the logistics industry can no longer meet the needs of economic development, and the logistics industry innovation is extremely urgent.
As an inseparable important part of the 5G industrial chain, logistics will undergo great changes due to the generation of 5G. Therefore, the value of 5G for logistics is self-evident [12]. The reason why 5G can be widely applied in the logistics industry is that there is a close connection between logistics and the IoT.
5G has three features: ultra-low delay, high-speed broadband and mass access.
Each of these features is likely to bring leapfrog technological progress in artificial intelligence, Internet of things, automated driving and other fields, and technological progress in these fields is exactly the key to the development of smart logistics [13]. Therefore, it is possible to realize the intellectualization of logistics transportation, the automation of logistics warehousing and the networking of logistics information. In addition, the integration of 5G and block connection is applied in the field of logistics finance, providing real-time data feedback and prevent tampering, so as to improve the reliability of logistics supervision [14].
1) Establishing an intelligent logistics traceability system based on 5G
Smart logistics tracing system is the use of the IoT and Internet to realize tracking and traceability of products, Suppliers can conduct controllable query on the forward logistics of goods and report analysis, the consumers can reverse query information through the platform or software after receipt of the goods. It usually used in agricultural and sideline products or cold-chain logistics system [15]. Its essence is a bottom-up, multi-level, distributed and multi-node infor- The intelligent logistics traceability system uses a large number of connected devices of perception layer for data recording by using 5G's mass connection feature, and edge server using 5G network for data collection, information gathered the data center. The application layer draws the visual scene through these data, providing accurate decision for logistics management.
2) Implementing fully automated logistics transportation based on 5G
The establishment of the fully automated logistics transportation system depends on the multi-function unmanned intelligent robot. Therefore, the development of unmanned driving technology and unmanned distribution system greatly limits the development of the fully automated logistics industry. In recent years, unmanned driving technology has been a hot topic in various industries, but it have not been a major breakthrough due to its high technical threshold, legal constraints and other issues. With the appearing of 5G, the characteristics of high bandwidth, low latency and wide connection not only make the unmanned driving industry more "hot", but also provide a feasible solution from the technical level, making full automatic logistics transportation no longer just "a paper idea".
Compared with the traditional transportation, the fully automatic logistics transportation is controlled by the computer. The terminal equipment is connected to the network through the Internet of vehicles technology, and the data obtained by the control center is used to make the path decision, so as to realize the complete interaction between people, vehicles and roads, making the logistics transportation more safe and efficient. Any unmanned driving technology needs the support of Internet of vehicles, and powerful data communication technology is the cornerstone of Internet of vehicles technology. The development of 4G to 5G not only improves the speed of data transmission, but also effectively solves the problem of "Shared sensing" between vehicles, greatly avoiding accidents and enhancing the safety and reliability of transportation.
3) Maintaining logistics security by 5G+ blockchain
Traditional logistics systems are generally based on large-scale and scalable mass data storage technology, which requires the analysis and safe storage of multi-party data [15]. In recent years, logistics security has become the focus of the industry due to the frequent occurrence of security problems such as lost bag and wrong collar or information leakage. Therefore, the exchange of logistics data and information based on block chain technology emerges at the historic moment, realizing the safe transmission of physical information. Due to the characteristics of block chain technology such as distribution and sharing mechanism, the recorded items have a strong traceability and can realize the capitalization of commodities. Therefore, block chain technology can be widely applied in the field of logistics security. 5G, as its communication mode, is the foundation of block chain technology, ensuring the real-time and high efficiency of information transmission process.
In addition, these true-fidelity and untamable data can reflect the real trade situation of small and medium-sized enterprises, ensure the authenticity of the flow of credit certificates, and realize the credit penetration of core enterprises in the supply chain. The judgment based on actual transaction data can help financial institutions effectively evaluate credit risks, reduce the default cost in the logistics financial process, and maintain the security of the whole logistics process.
Conclusions
Smart logistics tracing system is the use of the 5G+ IoT and Internet to realize tracking and traceability of products, which implemented the data flow of information chain; 5G also provides a feasible solution for unmanned driving technology because of the characteristics of high bandwidth, low latency and wide connection. In addition, with the integration between logistics industry and other industries, the security of logistics has become increasingly important. So the exchange of logistics data and information based on block chain technology emerges at the historic moment. All of them are the products of 5G and the development of smart logistics.
In a word, 2019 is a year of rapid rise of 5G. The emergence of 5G is affecting all walks of life. Logistics industry, as the first level of service industry, must be prepared to lead the popularization of 5G. With the rapid growth of the logistics industry, the traditional logistics industry can no longer meet the diverse needs of people. The emergence of 5G communication technology is bound to promote the innovation of the traditional logistics industry.
In recent years, with the rise of technological means such as IoT+, big data and cloud computing, scientific and technological innovation has triggered a new round of logistics competition, and cross-border integration ability has become the core competitiveness of new logistics industry. The birth of 5G not only optimizes the communication technology in the logistics system, but also serves as the technical support and media for many hot technologies to help them to deeply integrate with the logistics industry. At the same time, 5G supports blockchain to truly achieve real-time data capturing and tracking, making blockchain "better". The massive connectivity of 5G makes the Internet of everything possible. The flexibility of using 5G technology is improved by the on-demand networking, which enables the IoT to be closely integrated with the logistics industry. These things together make it possible to be monitored at any time and tracked everywhere. In the future, 5G will definitely promote the split development of the logistics industry and eventually form the 5G ecosystem.
Conflicts of Interest
The author declares no conflicts of interest regarding the publication of this paper. | 2019-10-17T08:58:48.476Z | 2019-08-06T00:00:00.000 | {
"year": 2019,
"sha1": "44e52d455c279893aa670628553cbbf0b6a7a7fd",
"oa_license": "CCBY",
"oa_url": "http://www.scirp.org/journal/PaperDownload.aspx?paperID=95766",
"oa_status": "GOLD",
"pdf_src": "Adhoc",
"pdf_hash": "238d9f65152cf757bb8bbf79d5f6574034bff352",
"s2fieldsofstudy": [
"Engineering",
"Computer Science",
"Business"
],
"extfieldsofstudy": [
"Business"
]
} |
257073432 | pes2o/s2orc | v3-fos-license | Analysis of Costs Associated with the Use of Personalized Automated Dosing Systems versus Manual Preparation in a Residential Center for the Elderly in Extremadura
Introduction: During the SARS-CoV-2 pandemic, there has been a decrease in the supervision of the medication of subjects with chronic diseases. Customized automated dosing systems (SPDA) are devices that allow medication to be dispensed and administered, which have proven to be safe and effective for the patient and cost-effective for the healthcare system. Methods: an intervention study was carried out on patients from January to December 2019 in a residential centre for the elderly with more than 100 beds. The economic costs derived from manual dosing were compared with those of an automated preparation (Robotik Technology®). Results: Of the 198 patients included, 195 (97.47%) of them were polymedicated. Of the total of 276 active substances of registered medicinal products, it was possible to include them in the process of automating the preparation of the SPDA 105 active pharmaceutical ingredients. A cost reduction of EUR 5062.39 per year was found using SPDA. Taking into account the active ingredients of emblistable and non-emblistable medicines, the use of SPDA resulted in savings of EUR 6120.40 per year. The system contributed to the detection of cases of therapeutic duplication and reduced the time to prepare the medication. Conclusions: the use of SPDA is a useful and economically profitable strategy for its use in residential centres for the elderly.
Introduction
In the SARS-CoV-2 pandemic situation, the need to improve the continuity of care and ensure access to medicines in a safe and efficient manner became evident [1][2][3][4]. The pandemic has greatly affected the care of chronic pathologies [5,6]. In the case of polymedicated, non-institutionalised patients with multiple pathologies, there is also a decrease in the supervision of medication by relatives or direct caregivers, as visits have been reduced due to periods of restricted mobility [5,7].
The model of personalized, integrated pharmaceutical care becomes more relevant in this pandemic context [8][9][10], giving greater importance to the role of the community pharmacist after the act of dispensing medicines, participating in the administration of medicines with tools such as personalised dosage systems (PDS) that allow the medication to be organised using special blister packs, on set days and at set times. SPDs are an aid to pharmaceutical care integrated into the continuity of care for polymedicated patients while ensuring better adherence to the pharmacological treatment of these polymedicated patients and, therefore, saving the health system money by avoiding therapeutic non-compliance and admissions due to medication errors [11][12][13][14].
The patient receives his or her medication organised in a personalised system, which improves adherence to treatment, thereby optimising the control of consumption as it allows for the safekeeping of the remaining medication not dispensed at the pharmacy [15]. At the same time, it helps polymedicated patients to maintain their independence and safety and improves the quality of care and the efficiency and sustainability of the healthcare system [16][17][18].
In a residential care home for the elderly, patients have multiple pathologies and the average consumption of medicines is higher than for those who continue to live at home [18]. The manual administration of these medications is a common practice in residential centres without the intervention of the community pharmacist, and this practice has repercussions on patient safety because it is associated with dosage errors, non-compliance with dosage, incorrect interpretation of medical indications and lack of organisation of inventories [9]. In response to this, the use of personalised dosing systems (PDS) performed by community pharmacists in the Pharmacy Office has been proposed. Personalised dispensing systems can be prepared by the pharmacist manually or automatically by a blister robot, and in this case, we are talking about personalised dispensing automation systems (PDA). These robotic blistering systems constitute a new technology capable of reducing the time required for the preparation and distribution of medicines in individualised doses.
In general, only solid oral preparations with good physical, chemical and pharmaceutical stability are suitable for inclusion in SPDs, provided that they are stable outside the original primary packaging at room temperature for the period of unblistering, preparation, delivery and use. Non-oral dosage forms, such as ova or suppositories, and non-solid oral dosage forms, such as sachets, should be excluded from SPD. Thermolabile (refrigerator storage between 2 and 8 • C) medicinal products should also be excluded [18].
Extremadura has a population that is older and more dependent than the national average and usually resides in rural areas [19]. This has been associated with the implementation of both public and private resources to improve the care of the adult population residing in residential centres or day centres [20]. In order to overcome deficiencies in the supply and dispensation of medicines, the use of SPDA in these facilities has been proposed [21,22]. However, it is not known to what extent these systems are associated with a reduction in financial expenditure. For this reason, the main objective of this study is to compare the costs of manual drug administration with SPDA. Performed by pharmacists in a Pharmacy Office in a residential center for the elderly in Extremadura.
Materials and Methods
An intervention study was carried out in which a model of pharmaceutical care integrated into the care model of the resident population of the Ciudad Jardín Residential Centre in Cáceres was designed as the object of study, using personalised automated dosage systems (SPDA).
This study was carried out from January to December 2018 and January to December 2019 on a closed multi-pathological and elderly population in the Ciudad Jardín residential centre, with a private management model and a mixed public-private medication supply system of the Extremadura Health Service/Pharmacy Office. As it is a residential centre of more than 100 beds, part of the supply of medicines for patients is provided by the Hospital Pharmacy of the referral hospital and the medicines that the hospital does not have in stock are supplied by a Pharmacy Office. This residential centre places fortnightly orders for medicines from the Hospital Pharmacy of the Hospital San Pedro de Alcántara according to the patient's treatment schedule for 15 days. Residential centre does not have any medicine management programme to order the exact consumption of medicines to be consumed in the next 15 days, so the nurses make an estimate according to the existing stock and with reference to previous orders accumulating in the residential centre to create a large stock of medicines. The Ciudad Jardín Residential Centre has 240 beds. The two groups of patients, both in 2018 and 2019, were made up of the same patients in order to avoid selection bias.
During 2018, the Garden City Residential Centre used the manual pill dispenser method by the staff. Starting in 2019, throughout the year, the production of SPDA blister packs using the medicine blistering robot (RobotikTechnology ® ) was introduced at the Pilar Leal Pharmacy in Alcuéscar, Spain.
Pharmaceutical dosage forms that can be included in both manual and automated personalised drug delivery systems include only solid oral medicinal products with good physical, chemical and pharmaceutical stability, provided that they are stable outside the original primary packaging at room temperature for the period covering unblistering, preparation, delivery and use. Non-oral dosage forms, such as ova or suppositories, and non-solid oral dosage forms, such as so-bres, should be excluded from SPD. Thermolabile (refrigerator storage between 2 and 8 • C) medicinal products should also be excluded.
This modification in the way medication is administered was used to compare the costs of manual and automated production (year 2018 compared to 2019). The medication supply system was the same in both 2018 and 2019. The Garden City Residential Centre has a supply of public and private medication. Public management corresponds to the Extremadura Health System, and the private management corresponds to the Pharmacy Office.
In order to implement the model, project management techniques, process management, training for the people involved in the process and a study to assess the previous situation in terms of financial and patient safety were used in the different phases of the project's implementation.
A homogeneous population was considered in terms of the main variables: age over 64 years of age and without a direct reference person for medication management.
The following variables were chosen as study variables: age, gender, co-payment of the individual health card (IHC), pharmaceutical expenditure on medicines, medical devices and medicines not financed supplied by the Extremadura Health Service and patient co-payment at the Alcuéscar Pharmacy Office. The number of medicines consumed by each patient was also taken into account, and polymedication was considered to be the consumption of three or more medicines/per day.
The information was obtained from the electronic and paper prescriptions from the geriatric centre's doctor, as well as from the orders for medicines received at the Residential Centre from the Hospital Pharmacy of the San Pedro de Alcántara Hospital in Cáceres. Data on the consumption of medicines were also obtained from the management software "Farmadosis" AMCO+ (FARMADOSIS S.L., Palma de Mallorca, Spain), a management programme used by the Pharmacy Office for the weekly production of SPDA medication that connects the SPDA robot with healthcare professionals of the residential centre, as well as with the technology and information systems necessary to coordinate the pharmaceutical care with healthcare professionals in the nursing of this centre. The Pharmacy Office carries out the weekly emblisting of patients' medication for production with the SPDA system, which makes it possible to obtain metrics related to medication consumption.
All patients of this centre who agreed to participate were included in a consecutive selection process. For this purpose, the patients were given an informed consent form explaining the study in detail and that the data obtained would be anonymous and would only be used for the purpose of the study. In September 2018, the situation analysis phase was carried out prior to the implementation of the pharmaceutical care model in the residential centre. Meetings were held with healthcare staff, especially for reviewing the processes related to patient treatment. A working methodology adapted to what they were doing was proposed so that the implementation of personalised pharmaceutical care could be integrated into the usual working system. The healthcare staff of the residential centre, responsible for administering medicines to patients, were trained in SPD systems for medicines to ensure patient safety and without extensive modification of routine clinical practice.
We retrieved data on medications consumed by patients throughout 2018 and demographic data on all patients discharged from the centre. They were segmented by age, gender and type of health coverage. Training of healthcare staff was intensified for the implementation of this new model.
To this end, the technology and information systems in place were used to coordinate pharmaceutical care with the healthcare professionals at the residential centre.
During the period analysed, in 2018 and 2019, a total of 198 patients who met the inclusion criteria for the study were included each year.
The expenditure associated with the manual dosing of the residential center in manual devices was quantified and compared with the expenditure associated with the elaboration of the SPDA made by the Pilar Leal Carbajo Community Pharmacy Office.
The efficiency of personnel and material resources was analysed. At all times, the preparation of active ingredients of medicines was ensured through the SPDA. Qualified and duly certified staff came from the community Pharmacy Office of Pilar Leal Carbajo.
The time spent by healthcare staff at the residential centre who prepared the individual pill dispensers with medicines using the manual method in 2018 was analysed, along with the associated economic expenditure, which was compared with the preparation of medication using the SPDAs in the Pharmacy Office in 2019 and its associated economic expenditure.
Results
Patients included in the study who met the inclusion criteria were 198 patients in each study year.
Results Regarding Polymedication:
The proportion of polymedicated patients in the study population was 97.47% of the total number of patients with any medication. When analysing the polypharmacy of the population, it was observed that only five patients had less than three drugs.
A total of 276 active pharmaceutical ingredients consumed each year by the patients included in the study were recorded. Of these, 99 active pharmaceutical ingredients were found that could not be reblistered due to their pharmaceutical form in the SPDA and were administered manually by the healthcare staff at the healthcare centre. The remaining 177 active ingredients of medicines consumed each year by the patients included in the study could be reblistered in the SPDA robot prepared in the Pharmacy Office. Of these 177, 72 active ingredients of medicines were supplied by the community Pharmacy Office through medical prescriptions, and the remaining 105 active ingredients of medicines were supplied by the Extremadura Health Service to the Ciudad Jardín residential centre.
Results
Relating to the Expenditure Obtained for the Medication Administered by SPDA: When analysing the data on essential medicines provided by San Pedro de Alcántara Hospital to the residential centre, a decrease in the number of units of medicines requested from the hospital by means of fortnightly orders by the residential centre was observed in 2019 compared to 2018.
The number of units of medicines provided by San Pedro de Alcántara Hospital and spent in 2019 when using the SPDA is 330,271 U (with an associated expenditure of EUR 83,374.90) compared to 2018 which resulted in 381,740 U (with an associated expenditure of EUR 88,437.29). This reduction in expenditure using the system proposed in this study was EUR 5062.39 per year.
The overall consumption of medicines (essential and non-essential) was analysed using the integrated pharmaceutical care system with SPDA: it was observed that in the model presented, a total of 452,997 units of medicines were consumed in 2018 at the cost of EUR 178,745.74. This compares to 2019, when 402,638 units of medicines were consumed with a PVP ii cost of EUR 172,625.34.
This represents a saving of EUR 6120.40, but this model has not had a significant influence on the decrease in the number of units of non-emblistable medicines.
This study did not quantify the medication changes that occurred from one year to the next (2018 to 2019) in patients, so this decrease can be partly can be associated with this fact. What was reduced was the overstocking of medicines in the residential centre by ordering medicines without precision in quantity, avoiding future expiry of these medicines.
The proposed system contributed to the detection of two cases of therapeutic duplicity, i.e., the presence of two medicines of the same therapeutic class, one of which was duplicity of analgesics and one was duplicity of analgesics; the patient had been prescribed paracetamol one gram and the combination tramadol/paracetamol 37.5 mg/325 mg, another was duplicity of anti-ulcer drugs, the patient had omeprazole 40 mg and esomeprazole 20 mg. An alert record sheet was created for the nursing staff, who systematically checked the patient's visual check by them prior to the administration of medication, and reported the subjective perception of more cases where they self-corrected the duplicity error but did not record it prior to the implementation of the integrated pharmaceutical care model.
The study population comprised 46 men and 152 women. Nine patients were younger than 65 years, 40 patients between 65 and 80 years and 149 patients older than 80 years.
According to the classification established by the Ministry of Health, with regard to the TSI assigned, we see that 186 residents out of the 198 total have a TSI002, 93.93% of the total.
After analysing all the data according to the TSI of the patients in the residential centre, the only TSI to be considered is TSI001, TSI002 and TSI003, as no results were obtained for the rest as there were no patients in this range. From TSI001, a total of eight users were obtained for consideration. TSI002 yielded 186 users, of which 4 had a co-payment ceiling of 18.52 euros, leaving 182 residents with a co-payment ceiling of 8.23 euros for medicines.
Finally, from TSI003, there was only one user who paid 40% of the cost of his medication. The data on the medicines financed and the co-payments made by patients with TSI002 were included, as well as the co-payments that patients would make if the supply were managed entirely by the pharmacy that serves the geriatric centre, instead of being shared with the Extremadura Health Service.
In the Ciudad Jardín Residential Centre, there is a shared supply between the Extremadura Health Service and the Pharmacy Office of Alcuéscar. It was observed that the pharmaceutical co-payment expenditure was lower than that which each resident should have to pay for the prescribed treatments. This is usually the case in residential centres in Extremadura with less than 100 beds; it is the user who finances this expense and not the Extremadura Health Service, according to the agreement between the Pharmaceutical Association of Cáceres and the Extremadura Health Service [3].
On analysing the consumption generated by the patients of the Ciudad Jardín Residential Centre, a deviation of pharmaceutical products non-financed pharmaceutical products supplied from the Pharmacy Department of the San Pedro de Alcántara Hospital in Cáceres. This justifies a separate quantitative analysis of the amount at PVL ii and PVP ii according to Botplus as of 2018 [24] and the number of units served during the year. These medicines are financed for patients who are in facilities with more than 100 beds but not for patients residing in their own homes or in facilities with less than 100 beds, as currently established.
In other words, in order to dispense these medicines, the Extremadura Health Service buys them at the LMP, so an expenditure of EUR 18,888.11 is imputed. However, if these same patients were to reside in their own homes or in a residential centre with fewer than 100 beds, this cost would not be borne by the Extremadura Health Service but would be financed by the patient themself, regardless of the patient's other socio-economic variables.
This agreement creates a situation in which residents in centres with more than 100 beds would have a saving of EUR 28,119.46 and the Extremadura health service would have an expense of EUR 18,888.11. Individual medication preparation time was accounted for both with the SPDA (2019) and manual preparation of weekly pill dispensers (2018). Both processes were carried out with two pharmacists for SPDA preparation and three nurses for the manual preparation of pill dispensers, who were familiar with each of the preparation processes. Average preparation time and human resource expenditure were calculated. The analysed expenditure for the manual preparation of weekly medication for 240 residents is shown in Tables 1 and 2. The cost of producing SPDA blister packs of medicines was also analysed. The costs of consumables are borne by the Pharmacy Office, which was cost-efficient for the health system, as well as all costs of supplies, personnel and legislative requirements. The expenditure analysed for automated SPDA preparation of weekly medication for 240 patients of the residential centre is shown in Table 3. The SPDA was faster in the preparation of a patient's monthly medication than the manual procedure that has been used to the current time. On the basis of these working times, the total cost was estimated: a total cost of EUR 3348 per medication preparation process per month for 240 patients in 2018 (manual system) and a total expenditure in 2019 (with SPDA) of EUR 1437.6 for 240 patients.
At the time of the study, as the SES supplies the user without strict control, it receives EUR 5978.30 less and has an expenditure of EUR 18,888.11. In the study, the SES has a total loss of EUR 24,866.41 (EUR 5978.30 + EUR 18,888.11). In addition to showing a difference in access to medication between residents in centres with more than 100 or less than 100 beds or at home, regardless of the socio-economic situation of each patient.
Discussion
The implementation of this model of personalised pharmaceutical care integrated into the care of institutionalised patients in the population studied shows an economic impact on both the Extremadura Health Service and the Ciudad Jardín Residential Centre. As it is included in comprehensive pharmaceutical care, with appropriate information systems, the model presented allows for the detection of duplicity and other medication errors and solving them in real-time [11,24,25]. This means improving the quality of patient prescribing and, therefore, patient compliance, contributing to the optimal control of chronic diseases in terms of medication management [26]. The professionalisation of medication supply and the presence of a pharmacist to review and monitor treatments, duplications and interactions on a personalised basis for each patient has been shown to contribute to the rationalisation of healthcare expenditure.
Non-financed medicines were quantified at PVL ii because this is the purchase price of these products, according to Botplus, as of 2018 [27]. This potentially represents a direct saving for the Extremadura Health Service in pharmaceutical expenditure for these patients in one year of using this personalised pharmaceutical care model [18]. As a cost-saving measure for the Extremadura Health System, by ceasing to request fortnightly bulk orders, non-financed medicines and medical devices would change the financer by being to be charged to the patient in centres with more than 100 beds [17]. Until now, in facilities with less than 100 beds or residents at home, this expenditure is charged to the user.
The residential centre that was under this system would have more hours per month of qualified health personnel to reassign to other occupations that would improve the quality of the centre and its patients, saving time in storage and placement of medicines and checking expiry dates, as well as saving money on the purchase of medication trolleys and their maintenance.
The pharmacy that participates in the pharmaceutical care of this model, which was implemented in the Ciudad Jardín residential centre, assumed direct investment costs in technology and information systems to carry out the SPDA of medicines, consumable materials necessary for weekly production, and the relevant legal obligations. Likewise, the hiring of personnel to provide a quality service to the residential centres is financially compensated by the profit margin on the increase in sales of the medicines and medical devices discussed during this study.
It is therefore important to consider this model of integrated pharmaceutical care in residential centres as an alternative to the current model in order to guarantee the sustainability of social and healthcare services while at the same time providing quality, patient safety and equity.
A more extensive study on equity of access to medication for patients living in facilities with less than 100 beds and no and not institutionalized would be needed [28].
One of the limitations of the study is that it focuses on a single centre and takes as a reference the medication control system of the residents of a single centre. However, it would be useful to extrapolate this system to other centres, so we believe that future research can be based on our work in order to contribute more scientific quality to the community. To date, there is no article that discusses this process of cost analysis associ-ated with personalised dosing systems, comparing the two possibilities: manual versus automated preparation.
Conclusions
SPDAs are effective tools to optimise the pharmacological treatment of institutionalised elderly people and are safer and less costly than manual dosing of medicines. Their implementation in facilities has a positive economic impact on health services, as well as a way of incorporating integrated pharmaceutical care. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patients to publish this paper. | 2023-02-22T16:12:47.708Z | 2023-02-01T00:00:00.000 | {
"year": 2023,
"sha1": "46932a593ef536456a3ca5c630031206d0b0d0fc",
"oa_license": "CCBY",
"oa_url": null,
"oa_status": null,
"pdf_src": "PubMedCentral",
"pdf_hash": "b84f4bb3342a0f11bdac35dc101fbc6740a113fc",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
3355583 | pes2o/s2orc | v3-fos-license | Pauci Immune crescentic glomerulonephritis in a patient with T-cell lymphoma and argyria
Background Silver is a transition metal, toxic when ingested in significant amounts, causing argyria (skin deposition) and argyrosis (eye deposition). It is excreted mainly via the gastrointestinal tract with only small amounts eliminated by the kidneys, and rarely have cases of nephrotoxicity due to silver been reported. Here we present the case of a woman who used colloidal silver as an alternative remedy for a T cell lymphoma, who subsequently developed argyria and a pauci-immune crescentic glomerulonephritis with evidence of extensive glomerular basement membrane silver deposition. Case Presentation A 47 year old woman of Indo-Asian descent with a T-cell lymphoma who refused conventional chemotherapy for 18 months but self-medicated with a remedy containing colloidal silver, was admitted with acute dialysis-dependent kidney injury. A kidney biopsy demonstrated a pauci-immune crescentic glomerulonephritis with deposition of silver particles in the mesangium and along the glomerular basement membranes. The patient was treated with intravenous methylprednisolone and intravenous cyclophosphamide and recovered independent renal function. Conclusion Chronological evolution of the the pauci-immune glomerulonephritis suggests that a cellular immune-mediated process was induced, potentially mediated by lymphomatous T cells directed at the glomerular basement membrane, following silver deposition. Immunosuppressive therapy improved the situation and allowed cessation of haemodialysis, supporting the hypothesis of an immune-mediated process.
Background
Silver is a transition metal element with a range of industrial and ornamental uses. Silver also has pharmaceutical applications, primarily as a disinfectant and antimicrobial, and is known to be toxic when ingested in significant amounts. The recognised manifestations of this are irreversible deposition in the eye, causing argyrosis, and in the skin, mucous membranes and internal organs, causing argyria.
Ingested silver is thought to be excreted mainly via the gastrointestinal tract with only small amounts eliminated by the kidneys [1]. Nevertheless some cases of nephrotoxicity due to silver have been reported [2][3][4][5][6]. Here we present the case of a woman who used colloidal silver as a remedy for an un-treated T cell lymphoma and who subsequently developed argyria and a pauci-immune crescentic glomerulonephritis.
Case presentation
A 47-year-old woman of Indo-Asian decent was admitted to the hospital with a two week history of peripheral oedema, shortness of breath and lethargy, as well as an itchy pigmented rash on her arms, legs and groin of two months' duration. She reported a large swelling of the right side of her neck, which had been biopsied eighteen months earlier and diagnosed as a T cell lymphoma, for which she had refused conventional chemotherapy. She denied taking any regular prescribed medications, but began using alternative remedies since the appearance of the neck mass.
On examination the rash was raised and hyperpigmented with evidence of excoriation, and the patient had a grey tint to her skin. She had a blood pressure of 164/96 mmHg. The neck mass was firm and non-tender, measured 10cm x 4cm, and had two overlying scars from previous biopsies.
Laboratory results demonstrated a serum creatinine of 600 μmol/l, urea 21.2 mmol/l, haemoglobin 6.3 g/dl, neutrophils 21.9 x 10 9 /l, CRP 118 mg/l, LDH 675 iU/l, and serum bicarbonate 10 mmol/l. Urinalysis confirmed proteinuria and haematuria and urinary protein creatinine ration (UPCR) revealed nephrotic range proteinuria of 515 mg/mmol. Virological tests for HIV, HCV and Hepatitis B sAg were negative. Serological tests for anti-dsDNA, ANA, ANCA and anti-GBM antibodies were all negative, complement levels C3 and C4 were normal.
The patient was commenced on haemodialysis. A CT scan of the chest, abdomen and pelvis confirmed that the neck mass was an enlarged lymph node, with additional axillary and mediastinal lymphadenopathy.
Biopsies of the kidneys, neck mass and skin lesions were performed. The kidney biopsy revealed pauci-immune crescentic glomerulonephritis, and glomerular tufts with fine, dark, granular material scattered within the mesangium and along the basement membrane ( Fig. 1). Electron microscopy confirmed the presence of scattered electrondense granules along and within the glomerular basement membrane as well as in the mesangium (Fig. 2). This was strikingly similar to the skin biopsy showing scattered fine black granules within the basement membranes of the sweat gland epithelial cells consistent with argyria.
Electron probe microanalysis demonstrated that these granules contained predominantly silver and some selenium atoms (Fig. 3). The biopsy of the neck mass confirmed a peripheral T-cell lymphoma.
On further questioning the patient admitted using colloidal silver as another treatment for the neck swelling, made by electrolytic solubilisation of a pair of silver electrodes and consumption of the resulting solution (Fig. 4). The patients' serum silver concentration was strikingly elevated at 127.1 nmol/l (reference value <2.8 nmol/l).
The patient was treated with 3 pulses of intravenous methylprednisolone and intravenous cyclophosphamide, adjusted for age and renal function, every 2-3 weeks, followed by a tapering dose of oral prednisolone. After three weeks, partial recovery of renal function permitted withdrawal of haemodialysis. The patient declined conventional chemotherapy for her lymphoma, but continued
Conclusion
Irreversible silver exposure causing argyria (skin deposition) and argyrosis (eye deposition) is well-recognised. Current evidence for an effect of silver administration on the kidney is limited. Studies of workers occupationally exposed to silver have produced variable results: Rosenman et al. [7] observed a significantly lower creatinine clearance in precious metal manufacturing workers, but could not exclude a contribution of cadmium and solvent exposure, and renal biopsies were not performed. Pifer et al. [8] observed no impairment of creatinine clearance in silver reclamation workers.
A number of case reports of renal impairment following silver exposure have been published. However in several the route of exposure was through the application of topical silver sulfadiazine [2][3][4], the sulfadiazine component in itself being potentially nephrotoxic, whilst there were also other potentially confounding comorbidities such as diabetes, hypertension [2] and severe burns [3]. Watanabe et al. [5] reported an instance of nephrotic syndrome and membranous glomerulonephropathy in a patient who had had argyria for 8 years. More recently, Mayr et al. [6] published a case of renal impairment in a hypertensive patient who had been ingesting colloidal silver. His renal biopsy appearances were similar in part, with deposited silver particles along the glomerular basement membrane, but unlike our case glomerular damage was limited to hypertensive and ischaemic glomerulosclerosis.
Crescentic glomerulonephritis results from disruption of the glomerular basement membrane (GBM) which may be induced by immune complexes or by cellular mediators such as proteinases and reactive oxygen species. Examples of damage to the GBM initiating a pathological immune response by exposure of an immune neo-epitope and subsequent crescentic glomerulonephritis include membranous glomerulonephritis [9] and lithotripsy [10] in association with anti-GBM antibodies. ANCA positive pauci-immune glomerulonephritis has been previously described in a patient with intravascular large B cell lymphoma who presented with livedo reticularis and systemic symptoms of fever and lethargy however in this case renal excretory function was preserved [11]. Hamidou et al. described 2 cases of ANCA vasculitis in the context of chronic lymphocytic leukaemia and T cell lymphoma with both cutaneous vasculitis and a pulmonary renal syndrome [12]. However, our patient Fig. 3 Electron probe microanalysis of a granules and b background kidney. Silver was detected using energy dispersive analysis of X-rays (EDAX) on the transmission electron microscope. Electron probe microanalysis of background kidney shows a relative absence of silver and selenium atoms with much carbon. Analysis of the granules shows that they contain predominantly silver and some selenium atoms Fig. 4 Silver electrode for production of colloidal silver suspension. The device is shown switched on with electrodes placed in a beaker of plain water which has turned cloudy following 15 min of electrolysis had no autoantibodies, and no deposited immunoproteins. Our case is unique in that it represents an ANCA negative pauci-immune crescentic glomerulonephritis with evidence of intense silver deposition along the GBM in a patient with T cell lymphoma.
Although most cases of pauci-immune GN are associated with ANCA, between 5-10 % cases are ANCA negative. The pathogenesis in these cases may be related to other autoantibodies (such as anti-endothelial antibodies) or to direct leukocyte-induced GBM damage via soluble mediators, which in this case may have been due to a combination of T cell lymphoma and argyria. We demonstrate that immunosuppressive therapy improved the situation and allowed cessation of haemodialysis, supporting a hypothesis of an immune-mediated process most likely directed towards the glomerular basement membrane associated with silver deposition.
Ethics and consent to participate
All relevant ethics for the case report and its publication were obtained.
Consent to publish
Written informed consent was obtained from the patient for publication of this case report and any accompanying images. A copy of the written consent is available for review by the editor of this journal. | 2017-03-31T08:35:36.427Z | 2016-05-17T00:00:00.000 | {
"year": 2016,
"sha1": "eadb1ceff27882d2f6d0333c9faa894078045696",
"oa_license": "CCBY",
"oa_url": "https://bmcnephrol.biomedcentral.com/track/pdf/10.1186/s12882-016-0259-x",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "eadb1ceff27882d2f6d0333c9faa894078045696",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
221131929 | pes2o/s2orc | v3-fos-license | Race, Heart Rate, and Temperature Are Strongly Associated With COVID-19 at a Community-based Clinic in New Orleans
Federally qualified health centers are on the frontlines of the coronavirus disease 2019 (COVID-19) pandemic in the United States. It is essential to develop the workflows necessary to evaluate patients, perform appropriate diagnostics, make clinical recommendations, and provide public health messaging. This brief report presents findings from our COVID-19 response and compares the characteristics between the 345 patients screened between March 16 and April 10, 2020. One hundred seventeen patients tested positive for COVID-19, an overall rate of 33.9%; and Black race, increased heart rate, elevated temperature, and the use of antipyretic agents were associated with positive results.
T he novel coronavirus (severe acute respiratory syndrome coronavirus 2) pandemic continues to exert unprecedented challenges to the global health care system. 1 New Orleans is among the cities bearing the highest burden. 2, 3 Given the high rate of community transmission and expected predominance of a mildly symptomatic disease course, an ambulatory-based approach to coronavirus disease 2019 (COVID-19) management and mitigation is essential. 4,5 Evidence-based recommendations support patients receiving a clinical assessment, access to diagnostics, treatment guidance, health education on limiting transmissions, and triage for the utilization of higher levels of care. 6,7,8 At CrescentCare, a federally qualified health center in New Orleans, Louisiana, we developed a community-based response with a walk-in COVID-19 clinic. Here, we present the initial findings of our intervention and compare positive cases with negative cases.
The objective of this study was to assess the rate of COVID-19 positivity in a community-based health center, evaluate the clinical symptoms, and follow patient outcomes.
PATIENTS AND METHODS
The Advarra Institutional Review Board granted a full waiver of the Health Insurance Portability and Accountability Act authorization and deemed the study exempt. On March 16, 2020, our clinic developed a dedicated COVID-19 walk-in clinic. The clinic was open to all patients 17 years and older, new and existing, regardless of insurance coverage. It was intentionally not a drive-through testing site and centered on the patient/medical provider interaction. Clients entering the health center were screened for symptoms and triaged to the COVID-19 clinic. All 18,000 existing patients received text messages and e-mails directing them to the walk-in site if symptomatic. Radio service announcements, social media, and local press informed the community of our services.
Three tents were set up in the outdoor garage of our health center. Providers wore synthetic suits that were disinfected between patients, as well as N95 masks, gloves, and goggles. Only providers interacted within 6 ft of patients. Demographic information was obtained consistent with routine clinical procedures, a focused history including fever within the past 72 hours, cough, dyspnea, and pharyngitis as well as comorbidities were recorded. Testing was performed for patients with documented or subjective fever within the past 72 hours. They were swabbed for point-of-care influenza and streptococcal A, when appropriate. Coronavirus disease 2019 testing was performed using LabCorp's nucleic acid amplification nasopharyngeal swab transported in viral culture medium.
Patients were informed of their results by a provider and asked follow-up questions about resolution or worsening of symptoms or hospitalization. Symptomatic patients had additional telephone encounters to continue to assess symptoms and limit unnecessary utilization of hospital-based services. Social distancing was reinforced during each contact.
Data were presented as means AE SDs and frequencies (percentages). Differences in distribution between those who tested positive and those who tested negative were assessed using the t test (Wilcoxon rank sum test for oxygen saturation) for continuous variables and the Pearson chi-square test for categorical variables. Logistic regression methods were used to assess combinations of predictors. All analyses were performed using SAS version 9.4 (SAS Institute).
RESULTS
Our clinic tested 345 patients between March 16 and April 10, 2020. One hundred seventeen patients tested positive for COVID-19, a rate of 33.9% (Figure). Temperature reading at visit (mean, 99.4 F vs 98.6 F), heart rate (mean, 93.9 beats/min vs 86.8 beats/min), and the use of antipyretic agents were significantly associated with positive test results. Dyspnea and pharyngitis were not associated. Blacks had a significantly higher rate of test positivity (odds ratio, 2.1; 95% CI, 1.4 to 3.3) (Table).
Of the 117 positive cases, 9 required emergency department services, with 6 hospitalized. The 6 patients hospitalized had a median age of 48.5 years (range, 28-67 years) and consisted of 5 women and 1 man; 3 were black, 2 were Latinx, and 1 was white. Two of these patients died after hospitalization in an intensive care unit. All other patients have recovered. Anosmia or ageusia was noted in 41.4% of COVID-19 cases.
The median time of the receipt of test results was 5 days, and our clinic was able to reach all patients to communicate their results.
CONCLUSION
The high rate of positive results underscores the widespread community transmission in New Orleans. Louisiana has experienced a considerable impact from COVID-19, especially in the Black community, with disparities in both hospitalization rates and mortality. Our study reports similar disparities in the outpatient setting in which Blacks are at higher risk of community transmission. Community health centers play a vital role in ensuring access to care during this pandemic, and this model prioritizes provider contact and patient education.
Overall, our patients recovered from COVID-19, with very few requiring referrals for immediate care. Over the course of the pandemic, our program noted an initial decline in percentage of those testing positive for COVID-19, supporting the public health benefit of social distancing initiatives. However over recent weeks, positive cases seen in our clinic have mirrored regional trends, further supporting the role of community health clinics through this pandemic.
Our findings elucidate the symptoms that are most prevalent in an outpatient setting during a community-wide outbreak. Elevated temperature, increased heart rate, and the use of antipyretic agents were most associated with positive test results. Subjective symptoms were similar between positive and negative cases. Reports have noted the wide range of initial symptoms in the outpatient setting. A limitation of this study is that our federally qualified health center patient population is younger, with fewer comorbidities. In this study, fewer positive cases had a smoking history. This could explain our lower rate of hospitalization. Additionally, at the onset of testing, we encountered delays in the receipt of test results, limiting our ability to triage patients appropriately.
Community health centers play an essential role during this pandemic by evaluating symptoms, reinforcing public health messaging, and limiting unnecessary utilization of emergency services. | 2020-08-16T13:06:11.813Z | 2020-08-15T00:00:00.000 | {
"year": 2020,
"sha1": "1a921ccf247a7a026be84f0daa344447a0593f5e",
"oa_license": "CCBYNCND",
"oa_url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7428678",
"oa_status": "GREEN",
"pdf_src": "PubMedCentral",
"pdf_hash": "b95ef08c9e3684d45009de32fd8dbc4c734b55c6",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
16020449 | pes2o/s2orc | v3-fos-license | Effect of the EndoBarrier Gastrointestinal Liner on obesity and type 2 diabetes: protocol for systematic review and meta-analysis of clinical studies
Introduction Obese patients with type 2 diabetes undergoing bariatric surgery experience significant and lasting weight loss and improved glycaemic control. However, bariatric surgical procedures such as Roux-en-Y gastric bypass are irreversible and associated with considerable short-term and long-term risks. The EndoBarrier Gastrointestinal Liner or duodenal-jejunal bypass sleeve (DJBS) is a fully reversible procedure that has been developed to treat obesity and type 2 diabetes. We aim to perform a systematic review and meta-analysis of safety and efficacy of DJBS. Methods and analyses A systematic review with meta-analysis (as per the preferred reporting items for systematic reviews and meta-analyses) of randomised controlled trials of the device (vs no intervention, sham and/or low-calorie diet) will be performed. Primary endpoints include change in body weight and glycated haemoglobin and safety. Secondary endpoints constitute changes in other glycaemic parameters and blood lipids and the proportion of patients discontinuing antidiabetic medication. MEDLINE, EMBASE, The Cochrane Library and Science Citation Index will be sought electronically along with manual searches. The primary meta-analysis will use random effects models due to an expected intertrial heterogeneity. Fixed effect meta-analysis will be executed to assess the impact of small trials. Dichotomous data will be analysed using risk difference and continuous data using weighted mean differences, both with 95% CIs. Ethics and dissemination The study will describe the impact of DJBS on obesity and type 2 diabetes and possibly contribute to clinical decision-making. The results of this study will be disseminated by peer-reviewed publication and scientific presentations. Registration PROSPERO CRD42013004819
INTRODUCTION
Lack of physical exercise and excess nutrient intake constitute important factors leading to obesity and overweight. Worldwide, more than 1.4 billion adults (≥20 years old) are overweight with a body mass index (BMI) ≥25 kg/m 2 . Of these, approximately 500 million adults are obese (BMI ≥30 kg/m 2 ). WHO estimates that the number of obese persons has doubled since 1980. 1 Overweight and obesity are risk factors that increase the risk of cardiovascular disease, musculoskeletal disorders, cancer, type 2 diabetes and premature death. Dietary treatments are ineffective in the long-term treatment of overweight and obesity and the current antiobesity medications are few and largely ineffective. 2 In contrast, bariatric surgery has proven effective-also in the longer term-and leads to an improved glucose homoeostasis. Patients with type 2 diabetes undergoing bariatric surgery experience improved glycaemic control or remission of diabetes, reducing or even eliminating their need for medication. 3 Current clinical practice-the bariatic surgical procedure Roux-en-Y gastric bypass Interestingly, rerouting of nutrient flow through the gastrointestinal tract (bypassing the proximal small intestine) following the surgical bariatric procedure Roux-en-Y gastric bypass (RYGB) has been shown to dramatically improve glucose metabolism within a few days-prior to any weight loss occurrence-among obese patients with type
ARTICLE SUMMARY
Strength and limitations of this study ▪ Our group has great experience in conducting a systematic review with meta-analysis. ▪ This study may help guide clinical decisionmaking and procure better treatment of obesity and type 2 diabetes. ▪ Small studies with high heterogeneity and varying quality may be this study's limitation.
2 diabetes. Depending on the definition of remission, remission rates of 40% 4 to 80% 2 have been reported. The predominant hypotheses on the physiological background for the metabolic advantages after bariatric surgery include the changed release of gastrointestinal hormones (increased secretion of hormones with antidiabetic and/or antiobesity properties, eg, glucagon-like peptide-1 (GLP-1) and peptide YY (PYY) and reduced secretion of 'diabetogenic' hormones, for example, glucose-dependent insulinotropic peptide (GIP) combined with surgery-induced restriction of food intake. Despite the short-term and long-term benefits RYGB provides for obese patients with type 2 diabetes, the procedure-like most other bariatric surgical proceduresis invasive, irreversible and potentially lethal. In a meta-analysis from 2004, Buchwald et al 2 report a 30-day mortality after gastric banding, RYGB and biliopancreatic diversion of 0.1%, 0.5% and 1.1%, respectively. The most frequent short-term causes of mortality after RYGB are venous thromboembolism and cardiorespiratory disease. 5 Additionally, several short-term and long-term complications are associated with the procedures including anastomotic leaks, bleeding, infections, small-bowel obstruction, hernias, dumping syndrome and malabsorption of micronutrients and macronutrients. 5 6 Finding a less invasive bariatric procedure to treat obesity and type 2 diabetes would be of great interest not only for the patients but also for the society in general. The minimally invasive and fully reversible duodenal-jejunal bypass sleeve (DJBS) may represent an alternative to the most commonly used bariatric techniques. With this protocol, we intend to investigate the efficacy and safety of DJBS.
Description of the intervention
The EndoBarrier Gastrointestinal Liner (a polymer DJBS) consists of a nickel-titanium anchor and a 60 cm impermeable sleeve made of fluoropolymer (figure 1). The device, which is open at both ends, is endoscopically placed in the duodenum through an over-the-wire system. The anchor is fixed to the intestinal wall within the duodenal bulb by small barbs grasping the intestinal mucosa. 7 Ingested nutrients pass down to the stomach and onwards directly and mostly undigested into the sleeve. The pancreatic and bile juices pass naturally into the intestinal tract, flowing down between the sleeve and the intestinal wall. They mix together with the undigested nutrients at the distal end of the DJBS, that is, in the jejunum. 8 Placing the DJBS endoscopically makes the procedure minimally invasive. Furthermore, DJBS has the advantage of being fully reversible; the device can easily be removed using an endoscope. 9 The producer of the device (GI Dynamics Inc) recommends that treatment with DJBS is accompanied with dietetic counselling to optimise the effect and to prevent device malfunction. Currently, the device is approved for a maximal treatment period of 12 months. In 2010, DJBS received European Communauté Européenne (CE) marking and achieved conditional approval by the US Food and Drug Administration in August 2012.
How the intervention might work
The mechanisms behind the body weight lowering and antidiabetic effects of DJBS are unknown, but are thought to involve less absorption of nutrients and have been speculated to encompass changes in gut hormone secretion. Up to now, several human studies with a duration from 12 to 52 weeks report that implanted participants lose weight and achieve improvements in their diabetic state after treatment with the device. Tarnoff et al 10 reported in their 12 week open-label prospective randomised controlled trial an excess weight loss (EWL) of 22.1% and 5.3%, respectively, for implanted participants and participants treated with a low-calorie diet. Another randomised sham-controlled trial showed EWL of 11.9% and 2.7%, respectively, for the device group and the sham group. 8 Regarding changes in glycaemic parameters, Rodriquez et al 11 and Schouten et al 12 have reported improved glycaemic control (greater reduction in glycated haemoglobin (HbA1c)) when treated with DJBS compared to controls. De Jonge et al 13 report in their study of 17 obese participants with type 2 diabetes that DJBS changes the gut hormone secretion favouring postprandial release of GLP-1 and lowering the secretion of GIP within 1 week after implantation before any significant weight loss occurred. This emphasises that changes in gut hormones may constitute one of the mechanisms by which DJBS exerts antidiabetic antiobesity effects.
Why it is important to do this review As aforementioned, overweight and obesity represent major concerns for the individual and the society. The growing number of obese people has also led to a worrying increase in the incidence of people with type 2 diabetes. Nearly 350 million people suffer from this disease worldwide. 14 Bariatic surgery has proven to be effective as a method of reducing body weight and improving type 2 diabetes. However, the potentially serious complications during and following the invasive and irreversible surgical procedures are incontrovertible. Thus, there is currently a strong need for new and less invasive, safer and preferably reversible alternatives to bariatric surgical procedures. DJBS may provide a modality fulfilling these conditions. Current data on the effects of DJBS stem from rather small studies. Therefore, it seems of major importance to compile and analyse current evidence of the effect of DJBS on obesity and/or type 2 diabetes. Such evidence may help guide clinical decision-making and procure better treatment of obesity and type 2 diabetes.
OBJECTIVES
The primary objectives of the present protocol are to evaluate the effect of the DJBS on weight loss as assessed by change from baseline or the per cent of excess weight lost (%EWL), glycaemic control as assessed by HbA1c, and safety. Secondary objectives include evaluation of the proportion of patients with type 2 diabetes being able to reduce or discontinue antidiabetic medication and changes in glycaemic parameters other than HbA1c (fasting plasma glucose or fasting blood glucose) and total cholesterol.
METHODS AND ANALYSES
The systematic review will be performed according to the recommendations specified in the Cochrane Handbook for Intervention Reviews. 15 The reporting of the review will follow the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement. 16 Criteria for considering studies for review Types of studies The review will include randomised clinical trials, prospective non-randomised trials, case-control studies and case series investigating the effects of the DJBS, irrespective of blinding and publication status. Unpublished trials will be included if data and methodology are accessible in written form.
Types of participants
Adult overweight or obese patients (age 18 years or older) with or without type 2 diabetes treated with DJBS will be included. Preferably the definition of overweight, obesity and type 2 diabetes should follow the criteria from WHO, the European Association for the Study of Diabetes or the American Diabetes Association, 17 but if necessary, trials will be included using the author's definition of obesity and type 2 diabetes.
Types of interventions
The comparisons will assess implantation of DJBS versus no intervention, sham-endoscopy and/or low-calorie diet.
Types of outcome measures
The outcome measures will be assessed based on analysis of individual patient data from included trials or from published reports when available. Searching other resources Manual searches will include scanning of reference lists in relevant papers, specialist journals and conference proceedings. Additional trials will be sought through the WHO Trial Register 18 and through correspondence with experts. The website of the producer of the DJBS device (GI Dynamics Inc) will be sought for available material. 19 Data collection and analysis Two authors (UR and NH) will independently extract data and resolve disagreements through discussion before analysis. In case of unresolved matters, a third party will be involved. If necessary data are not included in the published trial reports, the authors of the included trial will be contacted for further information.
Selection of studies
The trials identified through electronic and manual searches will be listed. Included trials will be selected using the aforementioned criteria. Trials that are excluded will be listed with the reason for exclusion. All authors will participate in the selection of trials.
Data extraction and management
The following data will be extracted from the included trials: ▸ Patient characteristics: inclusion criteria, proportion of patients with type 2 diabetes, mean age, mean BMI, proportion of men/women, mean HbA1c and mean body weight ▸ Characteristics of interventions: type and duration of interventions ▸ Characteristics of trial: number of clinical sites, country of origin and funding
Assessment of reporting bias
We will compare trial protocols with subsequent publications when available and we will extract whether clinically relevant outcomes are reported.
Assessment of risk of bias in included trials
Owing to the expected inclusion of different types of studies, the following assessment of risk of bias will be used. For randomised studies, randomisation methods will be extracted as the primary measure of bias control. The randomisation methods will be assessed on the allocation sequence generation (which will be classified as adequate if based on computer-generated random numbers, a table of random numbers or similar), allocation concealment (which will be classified as adequate if randomisation was performed through a independent central unit, identically appearing treatments, serially numbered opaque sealed envelopes or similar) and incomplete data outcome (whether all patients were accounted for). With regard to blinding (detection and performance bias), data will be extracted in order to assess whether single or double blinding was performed. Blinding methods will be evaluated (eg, use of placebo). Persons who were blinded with regard to the intervention will be assessed (ie, patients, healthcare providers or other persons involved in the trial). For other types of studies, incomplete outcome data (attrition bias), for example, patients lost to follow-up, will be evaluated as a measure of attrition bias. Outcome reporting (reporting bias)-the extent to which clinically relevant outcome measures are reported-and differences between trial protocols and subsequent reports will be evaluated and reported as a marker of reporting bias. Other biases will include sample size calculations and the extent to which the planned sample size was achieved. All nonrandomised studies will be classified as high risk of bias.
Measures of treatment effect Dichotomous data will be analysed using risk differences and continuous data using weighted mean differences, both with 95% CIs. Relative risk will be calculated.
Assessment of heterogeneity
The intertrial heterogeneity will be expressed as I 2 values. The general interpretation of I 2 values is: ▸ 0-40%: might not be important ▸ 30-60%: may represent moderate heterogeneity ▸ 50-90%: may represent substantial heterogeneity ▸ 75-100%: considerable heterogeneity Intertrial heterogeneity, small study effects and risk of bias will be evaluated through regression analysis (Egger's test).
Dealing with missing data Intention-to-treat analyses including all patients randomised will be performed. In the case of patients with missing outcome data, the last observation carried forward will be used. Individual patient data will be sought from the original source or from the published trial reports where individual patient data are unavailable.
Data analysis STATA (Stata Corp, Texas, USA, V.12) will be used for analyses. The primary meta-analyses will be performed using random effects models due to an expected intertrial heterogeneity.
Subgroup analysis and investigation of heterogeneity Subgroup analyses will be performed to assess the impact of the patient, intervention, trial characteristics and intertrial heterogeneity. The test for subgroup differences will be calculated for all subgroups and the results presented as p value and I 2 value, respectively.
Sensitivity analysis
To assess the impact of small trials, fixed effect meta-analyses will be executed. Additional sensitivity analyses with exclusion of trials classified as having unclear adequate randomisation will also be performed.
Unit-of-analysis issues
In the analysis, each patient will be counted only once. If necessary, the same follow-up time point will be chosen to have as much data as possible to perform the analysis, even though the follow-up period may be longer for the individual trial. This will increase heterogeneity with regard to follow-up time, but may increase the possibility of reporting bias. Otherwise, the longest follow-up will be used.
ETHICS AND DISSEMINATION
This study will evaluate the impact of DJBS on weight loss, type 2 diabetes (HbA1c) and safety.
Furthermore, the effect on fasting plasma or blood glucose, reduction in antidiabetic medication and changes in blood lipids will be investigated. The study will hopefully shed light on the novel, minimally invasive and reversible technique of DJBS and thus provide knowledge about the use of it in the treatment of obesity and type 2 diabetes. The study will be disseminated by peer-review publication and conference presentation.
Contributors UR and NH have prepared this protocol in collaboration with TV, LLG and FKK. All authors have participated in the search strategy development. UR and NH extracted data and drafted a paper describing the systematic review. The remaining authors critically reviewed the manuscript.
Funding URand FKK are supported by The Danish Council for Independent Research Medical Sciences (grant number 11-107682). Otherwise, this study did not receive funding from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
Open Access This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 3.0) license, which permits others to distribute, remix, adapt, build upon this work noncommercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http:// creativecommons.org/licenses/by-nc/3.0/ | 2016-05-12T22:15:10.714Z | 2013-09-01T00:00:00.000 | {
"year": 2013,
"sha1": "9f37af384d9e44db0b227e8c133ac8eca580916f",
"oa_license": "CCBYNC",
"oa_url": "https://bmjopen.bmj.com/content/bmjopen/3/9/e003417.full.pdf",
"oa_status": "GOLD",
"pdf_src": "Highwire",
"pdf_hash": "48eaf62238ee040173bff5851db92cfe4023cafc",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
245769846 | pes2o/s2orc | v3-fos-license | Skip Vectors for RDF Data: Extraction Based on the Complexity of Feature Patterns
The Resource Description Framework (RDF) is a framework for describing metadata, such as attributes and relationships of resources on the Web. Machine learning tasks for RDF graphs adopt three methods: (i) support vector machines (SVMs) with RDF graph kernels, (ii) RDF graph embeddings, and (iii) relational graph convolutional networks. In this paper, we propose a novel feature vector (called a Skip vector) that represents some features of each resource in an RDF graph by extracting various combinations of neighboring edges and nodes. In order to make the Skip vector low-dimensional, we select important features for classification tasks based on the information gain ratio of each feature. The classification tasks can be performed by applying the low-dimensional Skip vector of each resource to conventional machine learning algorithms, such as SVMs, the k-nearest neighbors method, neural networks, random forests, and AdaBoost. In our evaluation experiments with RDF data, such as Wikidata, DBpedia, and YAGO, we compare our method with RDF graph kernels in an SVM. We also compare our method with the two approaches: RDF graph embeddings such as RDF2vec and relational graph convolutional networks on the AIFB, MUTAG, BGS, and AM benchmarks.
Introduction
The Resource Description Framework (RDF) is a framework for describing metadata, such as attributes and relationships of resources on the Web. The Semantic Web aims to increase the machine-readability of data on the Web by the semantic description of resources in the RDF. In fact, RDF datasets from a wide variety of fields are available on the Web as Linked Open Data (LOD) [18]. In order to make effective use of such RDF data, retrieval, inference, and learning for RDF graphs have been widely investigated in the field of the Semantic Web [3,5,7]. In particular, machine learning [2,16] and data mining for RDF data have been actively studied in recent years.
RDF data has two significant properties that correspond to the description of semantic structures in the Semantic Web. The first is that it has a graph structure to facilitate schema-less data integration. Due to this property, conventional machine learning algorithms, such as neural networks, cannot directly apply RDF data because the input data are limited to vectors. Second, RDF data contains different types of relations and values, such as metadata, ontological data, and attribute data, which are represented together in a graph structure. As a semi-supervised learning approach, the graph convolutional network (GCN) [11] deals with graph data that separates the graph structure from the feature vector of each node. However, RDF data is described in a graph structure without separating the feature vector of each node. Machine learning tasks for such RDF data adopt three methods: (i) support vector machines (SVMs) with RDF graph kernels, (ii) RDF graph embeddings, and (iii) relational graph convolutional networks (R-GCNs).
In this paper, we present a method for extracting features from various combinations of neighboring edges and nodes in an RDF graph. Based on the complexities of features and their patterns, we propose feature vectors (called Skip vectors) of target resources that represent attributes and To prevent the dimensionality of Skip vectors from becoming too large, we select important features for classifying resources based on the information gain ratio of each feature. For classification tasks in RDF data, the Skip vectors of target resources can be applied to conventional machine learning algorithms, such as SVMs, the k-nearest neighbors (KNN) method, neural networks, random forests (RFs), and AdaBoost (ADA). In the evaluation experiments with RDF data such as Wikidata, DBpedia, and YAGO, we compare our method with RDF graph kernels in SVMs. On the AIFB, MUTAG, BGS, and AM benchmarks, we compare our method with two approaches: RDF graph embeddings such as RDF2Vec and relational graph convolutional networks. The remainder of this paper is organized as follows. In Section 2, we introduce the basic concepts of RDF graphs, the feature extraction of RDF resources, and the information gain ratio. In Section 3, we theoretically analyze the complexities of features and feature patterns for each resource in RDF graphs, and present a method for generating the Skip vectors as feature vectors based on feature patterns. In Section 4, we provide the evaluation experiments of our Skip vectors with conventional machine learning algorithms. In Section 5, we discuss related works. Finally, in Section 6, we conclude this paper and discuss future work.
RDF Graph
Let U be a set of Uniform Resource Identifier (URI) references, L be a set of literals, and B be a set of empty nodes. An RDF triple is a tuple (s, p, o) of subject s, predicate p, and object o with s ∈ U ∪ B, p ∈ U , and o ∈ U ∪ B ∪ L. In other words, an RDF triple is an element of (U ∪ B) × U × (U ∪ B ∪ L). An RDF graph G is defined as a finite set of RDF triples {(s 1 , p 1 , o 1 ), · · · , (s n , p n , o n )}.
For example, an RDF graph for describing fruits and animals is shown in Figure 1.
Feature Extraction for RDF Resources
As a feature extraction for graph and tree structures, kernel functions calculate the distance between data by counting the common substructures in two graphs [6,9,4,19,20]. Specifically, for RDF graphs, the Walk, Path, and Subtree kernels [12], which are based on intersection graphs and trees, have been devised by Lösch et al. Moreover, the PRO kernel [1] and Skip kernel [2], which are extensions of these kernel functions, have been proposed by Arai et al.
A simple path that exists in the neighborhood of a resource in an RDF graph G is called a Walk.
Definition 2.1 (Walk). A sequence of d triples whose object and subject match is called a Walk of depth d. It is simply a sequence of 2d + 1 elements starting from the subject s, which can also be written as In particular, a Walk of depth 0 starting at resource s is s .
Let X be the space of data points, and F be the space of their features. The mapping ex : X → F from a data point to a feature is called a feature extraction. Features in an RDF graph G are extracted by using a Walk starting from the target resource. Let W alk(s) be the set of Walks starting from resource s on the RDF graph G, and let EX be the set of feature extractions. The feature set F walk (s) of Walks for resource s is expressed as follows: The partial map v : N → { * } that transforms any position i of Walk e 1 , e 2 , ..., e n into a variable * , as in v(i) = * , is called a skip function. We denote the set of skip functions by V ar. A skip function v ∈ V ar is extended to a function on elements e i and element sequences e 1 , ..., e n as follows: v( e 1 , ..., e n ) = v(e 1 ), ..., v(e n ) A PRO is a feature in which there is no object o 1 , o 2 , ..., o d−1 on the path of a Walk of depth d.
Definition 2.2 (PRO). The feature set F pro of PROs for resource s is defined as follows.
A Skip is a feature in which arbitrary predicates and objects are made variables and missing from a PRO.
Definition 2.3 (Skip).
For any pro ∈ F pro (s), the feature set F skip (s) of Skips for resource s is defined as follows.
where |pro| is the depth of pro.
Information Gain Ratio
Let G be an RDF graph. A set of resources as training data is denoted by R G ⊂ U . Let C be a set of class labels. The set of resources with class label c ∈ C is denoted as R c G ⊆ R G . Let F be the set of all features in the RDF graph G. The set of resources that possess feature f ∈ F is denoted by R f G , and the set of resources that do not possess feature f is denoted by R −f G .
For an RDF graph G, the information entropies Info(R G ) for R G and Info f (R G ) for the resources partitioned by a feature f are defined as follows: The split information entropy SplitInfo f (R G ), which is the information entropy with feature f as a class label; the information gain IG RG (f ), which represents the decrease in information entropy if data is split by feature f ; and the information gain ratio IGR RG (f ), which is the normalized information gain, are defined as follows.
Feature Vectors from an RDF Graph
In this section, we define the representation vector of each resource that is obtained from its feature set based on a pattern of features in an RDF graph.
Complexity of Features
In Section 2.2, Walks were defined as the simple paths of each resource on an RDF graph. In addition, PROs and Skips are defined as feature extractions that take into account the nature of RDF data. As a more simplified variant of Walks, PROs are features consisting of the predicate sequence p 1 , ..., p n and the terminal object o without the intermediate nodes (objects), while Skips are features that extract the substructures of PROs by skipping any predicates and objects. First, we theoretically analyze each feature extracted on the RDF graph by counting the combinations of predicates and objects. The following theorem shows the complexities of features starting from a subject resource (node) in Subtrees, Walks, Paths, PROs, and Skips. Proof. (i) If the depth d = 1, the number of Subtrees with depth 1 is mn k=1 mn C k = 2 mn − 1, where the number of predicate-object pairs is m×n, and all predicate-object combinations mn C 1 ∼ mn C mn are features of Subtrees. The number of predicate-object combinations of depth 2 that can be reached from a predicate-object pair of depth 1 is 2 mn , and by adding the case of only a predicate-object pair of depth 1, the number of this part is 2 mn + 1. Since the combinations of the mn parts are all Subtrees, the number of Subtrees up to depth 2 is (2 mn + 1) mn − 1, except when there are not all predicates and objects. The number of predicate-object combinations of depth 3 that can be reached from predicate-object pairs of depth 1 and 2 is 2 mn , and considering the case with only predicate-objects of depth 1 and 2, and the case with only predicate-objects of depth 1, the number of this part is 2 mn + 1 + 1 = 2 mn + 2. Since the combinations of the (mn) 2 parts are all Subtrees, the number of Subtrees up to depth 3 is (2 mn + 2) mn 2 − 1, except when there are not all predicates and objects. Thus, the number of Subtrees up to depth of d is (ii) Because Walks and Paths are sequences of predicate-object pairs, the number of Walks and Paths is m × n at depth 1, (m × n) 2 at depth 2, and (m × n) 3 at depth 3, and the number of features at depth d is (m × n) d . Thus, the number of Walks and Paths up to depth of d is mn + (mn) 2 + · · · + (mn) n . Therefore, the number of Walks and Paths is in O(2 (mn) d ).
(iii) Because PROs are features that skip all objects from the Walk except for the endpoint, the number of PROs is m × n at depth 1, m 2 × n at depth 2, and m 3 × n at depth 3, and the number of features at depth d is m d × n. Thus, the number of PROs up to depth of d is mn + m 2 n + · · · + m d n. Hence, the number of PROs is in O(m d n).
(iv) Skips are features that skip any predicate and object from PROs. Because Skips of depth 1 consist of predicates (e.g., s, p 1 , * ), objects (e.g., s, * , o 1 ), and PROs of depth 1 (e.g., s, p 1 , o 1 ), the number of Skips is m+n+mn. The number of Skips to depth 2 is m+n+mn+mm+mn+m 2 n. This is because Skips to depth 2 consist of predicates of depth 2 (e.g., s, * , * , p 2 , * ), objects of depth 2 (e.g., s, * , * , * , o 2 ), predicate and object of depth 2 pairs (e.g., s, * , * , p 2 , o 2 ), predicates of depth 1 and depth 2 pairs (e.g., s, p 1 , * , p 2 , * ), predicate of depth 1 and object of depth 2 pairs (e.g., s, p 1 , * , * , o 2 ), and PROs of depth 2 (e.g., s, p 1 , * , p 2 , o 2 ). Therefore, the number of Skips to depth 2 is (m+n+mn+mm+mn+m 2 n)+(m+n+mn). Let the number of Skips up to the depth d be S d ; then S d can be expressed as In practice, it is difficult to compute all the Subtrees that produce very large combinations in O(2 (mn) d ). In RDF data, the number of predicates m, the number of objects n, and the depth d are at very different scales, where n is on the order of 10,000, m is roughly 10 to 100, and d is approximately 2 to 5, respectively. Since the number of predicates m is much smaller than the number of objects n (i.e., m ≪ n), m d is computationally small. Therefore, O(m d n) for PROs and Skips is properly less complex than O((mn) d ) for Walks and Paths. In addition to the complexity, Skips are highly expressive as an extension of PROs. As a result, the Skips can achieve high accuracy in classifying resources, while suppressing combinatorial explosions such as Subtrees.
Complexity of Feature Patterns
We carry out the following method to limit the exponential combinations with respect to the depth of features from each resource.
• The features of each resource are categorized into patterns that are divided into predicate and object components (called feature patterns).
• The search cost of the RDF graph is reduced by limiting the feature depth and the number of feature patterns.
We classify patterns of the feature sets Subtrees, Walks, Paths, PROs, and Skips by focusing on the types of predicates and objects. The elements in each feature set are rewritten as predicate type p, object type o, and variable * , which are called feature patterns. For example, two features * , p 1 , * , p 2 , o 1 and * , p 3 , * , p 4 , o 2 belong to the same feature pattern * ,p, * ,p,o . With this feature pattern, each feature can be grouped according to its structure and variables, and the types of predicates and objects.
The following theorem shows the complexities of feature patterns (called Subtree, Walk, Path, PRO, and Skip patterns).
Theorem 3.2 (Complexity of feature patterns). Let the total number of predicates and objects in an RDF graph G be m and n, respectively. A feature pattern from depth 1 to d for a resource satisfies the following properties.
(i) The number of Subtree patterns is in O(2 n d ).
(ii) The number of Walk and Path patterns is in O(d).
Proof. (i) The number of Subtree patterns at depth 1 is n, because each Subtree pattern has at most n Walks. There are n Subtree patterns of depth 2 reaching from a Walk of depth 1, only a Walk of depth 1, and not also a Walk. Then, the number for this part is n + 1 + 1 = n + 2. There are n of these parts, but if we do not count the Subtree patterns once they are chosen, the combination of these parts is (n + 2) · (n + 1) · · · 2 · 1. Because this is all the Subtree patterns, the number of Subtrees up to depth 2 is (n + 2)!. There are n Subtree patterns of depth 3 reaching from a Walk of depth 2, only a Walk of depth 2, only a Walk of depth 1, and not also a Walk. Then, the number for this part is n + 1 + 1 = 1 = n + 3. There are n 2 of these parts, but if we do not count the Subtree patterns once they are chosen, the combinations for these parts is ((n + 3) n )!. Because this is all the Subtree patterns, the number of Subtrees up to depth 3 is ((n + 3) n )!. Thus, the number of Subtree patterns up to depth of d is (ii) The number of Walk and Path patterns increases by one for each additional depth: po at depth 1, popo at depth 2, and popopo at depth 3. Thus, the number of Walk and Path patterns up to depth d is (iv) Skip patterns are constructed by arbitrarily dropping predicates and objects from PRO patterns. The number of skip patterns increases by a factor of two as the depth increases, from three p,o,po at depth 1 to six * p, * o, * po,pp,ppo,p * o at depth 2. Therefore, the number of Skip patterns at depth d can be expressed as 3 · 2 d−1 . Thus, the number of Skip patterns up to depth d is This means that if the depth d is limited to a small value (e.g., d = 2), then the number of Skip patterns is limited to a few (e.g., 3 · 2 d − 3 = 3 · 4 − 3 = 9), which indicates that Skip patterns are computable in practice. Table 1 summarizes the Skip patterns of depth d ≤ 2. Each Skip pattern is expressed by a feature pattern, where o and p indicate the object and predicate types, and the variable * means to skip one predicate or object. For classification tasks of resources s ∈ R G , we define simple feature sets based on the nine Skip patterns SP = {p, o, po, * p, * o, * po, pp, ppo, p * o}. Namely, sp ∈ {p, o, po} has a depth of 1 and sp ∈ { * p, * o, * po, pp, ppo, p * o} has a depth of 2. The depth d of each Skip pattern sp ∈ SP indicates the number of triples that are extracted from each resource s in an RDF graph G.
Feature Sets Based on Skip Patterns
For the example in Figure 1, the feature sets of objects for the resource "apple" are represented as follows: For the example in Figure 1, the feature sets of predicates and objects for the resource "apple" are represented as follows: F po (apple) = { * , color, green , * , color, red , * , shape, sphere , * , taste, sweet , * , rdf:type, fruits } F * po (apple) = { * , * , * , rdfs:subClassOf, foods } Figure 1, the feature sets of predicates for the resource "apple" are represented as follows: Let sp ∈ SP be a Skip pattern. The feature set of sp for all resources s ∈ R G is represented as follows:
Skip Vectors
A Skip vector is a vector representation of the feature set F sp (s) of sp ∈ SP for resource s. . . , f n }, the Skip vector V sp (s) of sp ∈ SP for resource s ∈ R G is defined as follows: where each x sp i is defined by: The concatenation of two Skip vectors V sp 1 (s) and V sp 2 (s) is denoted as V sp 1 (s) V sp 2 (s). The concatenation of Skip vectors V sp (s) by all sp ∈ SP is denoted as V all (s) = sp∈SP V sp (s). We abbreviate V sp (s) and V all (s) as V sp and V all if the resource s is not explicitly specified.
Feature Selection by Information Gain Ratio
Depending on the Skip pattern, the dimensionality of Skip vectors may be too large, making the machine learning computational requirement unrealistically high. Therefore, we use the information gain ratio to reduce the dimensionality of Skip vectors by selecting features that are useful for classification tasks.
Let F sp (R G ) = {f 1 , . . . , f k } be the feature set of a Skip pattern sp ∈ SP. We define F ǫn sp (R G ) = {f ′ 1 , . . . , f ′ n }( F sp (R G )) as the set of features selected from the top n(≤ k) of all features in the order of increasing information gain ratio IGR RG (f 1 ), . . . , IGR RG (f k ). The Skip vector for resource s ∈ R G created from the feature set F ǫn sp (R G ) is denoted as V ǫn sp (s). Furthermore, V ǫn all (s) = sp∈SP V ǫn sp (s) denotes the concatenation of the Skip vectors V ǫn sp (s) created from the top n feature sets F ǫn sp (R G ) for all Skip patterns sp ∈ SP. We exploit a discount factor λ to reduce the effect of features further from the resource on the similarity between resources. The following summarizes the procedure for generating the Skip vectors from an RDF graph whose features are selected by the information gain ratio.
(i) First, the feature set F sp (R G ) for all resources is generated by exploring an RDF graph.
(ii) Next, the information gain ratio of each feature f ∈ F sp (R G ) is calculated from the resource set R c G for each class label c ∈ C. (iii) Finally, the feature set F ǫn sp (R G ) is selectively extracted through the information gain ratio of each feature in F sp (R G ), and then the Skip vector V ǫn sp (s) of F ǫn sp (s) for each resource s ∈ R G is generated.
Experiments
We evaluate the Skip vectors generated from RDF data by applying them to conventional machine learning algorithms. In the evaluation experiments, we compare our method with existing RDF graph kernel methods, as well as RDF graph embeddings and relational graph convolutional networks.
The process of generating the Skip vectors from an RDF graph was implemented in Java 11. We efficiently searched an RDF graph in the RDF datastore FROST 1.1.1. 1 We implemented neural networks in Keras; SVMs, RFs, KNN, ADA in scikit-learn [15]; and GCN in TensorFlow 2.1.0.
Datasets
We summarize eight binary classification problems for the RDF datasets Wikidata, DBpedia, and YAGO as follows. Table 2 shows the number of features |F sp (R G )| for each Skip pattern sp ∈ SP (before feature selection). For all datasets, we perform 10-fold cross-validation with 90% of target resources as training data and 10% as test data.
Experimental Setting
We generate the Skip vectors V ǫn all , where the number of features for each Skip pattern sp ∈ SP is limited to n, by selecting features using the information gain ratio. We apply the conventional machine learning algorithms SVM, KNN, neural networks (NN), RF, and ADA. For each of the algorithms, the number of features n is set to n SVM , n KNN = 100, n NN = 150, n RF = 350, and n ADA = 450. For SVM, KNN, and NN, the discount factor λ varies from 0.1 to 1.0 in increments of 0.1. The features occurring once in training data are considered to be noise data and are excluded.
For SVM, we use a regularization parameter of 1.0 and the hinge loss function. For KNN, we select the value k=5. For NN, we use the ReLU activation function on the input, at most four hidden layers, and the softmax activation function on the output layer. The hidden layer size is reduced from the previous layer size to 1 5 , and the output layer size is two. We train the neural network model for 200 epochs by minimizing the loss function categorical cross entropy with the Adam optimizer and a mini-batch size of 32. We stop training if the loss function does not decrease in 10 epochs. For RF, we use the Gini impurity as the splitting criterion, and set the number of decision trees to {5, 10, 20, 40, 60, 80, 100, 150, 200, 300}. For ADA, the maximum depth of decision trees, as weak classifiers, is set from 1 to 10.
Results
In Table 3, we provide the results of classification accuracies for Skip vectors in the conventional machine learning algorithms. We compare our method against the classification accuracies for the RDF graph kernels Skip, Hop [2], PRO [1], Walk, Path, Full Subtree, and Partial Subtree [12] in SVM, reported in a previous work [2]. The discount factor λ is set to 0.0001, 0.001, 0.01, and 0.1 for the Partial Subtree kernel, and from 0.1 to 1.0 in increments of 0.1 for the other kernels. The performance of Skip vectors, even with the conventional machine learning algorithms, is superior to the existing RDF graph kernels. The best accuracies of our method are equal to or better than the best of all graph kernel methods on 7 out of 8 datasets (i.e., Gender1, SeaLake, NetIncome, ScientistArtist, BoxOffice, Gender2, PopulationDensity). In particular, the best accuracies of Gender1, NetIncome, and PopulationDensity are improved by 0.5%-4%.
Datasets
We summarize multiclass classification problems for the four dataset benchmarks [17]. AM Classification of artifacts from the Amsterdam Museum (5988321 triples, 11 classes, 1000 resources) By following the experimental setting in R-GCN [17], the RDF triples that were used to create the class labels for target resources are deleted in each dataset. For all datasets, we split the target resources into 80% and 20% for training and testing, respectively.
Experimental Setting
We use the Skip vectors V ǫn all with the number of features for each Skip pattern n = 1000. The features occurring once in the training data are considered as noise data and are excluded. For SVM, KNN, NN, and GCN, we choose the discount factor λ ∈ {0.1, 0.2, 0.3, 0.4 ,0.5, 0.6, 0.7, 0.8, 0.9, 1.0} based on the validation set performance. 6 For SVM, we use a regularization parameter of 1.0 and the linear kernel. For KNN, we select the value k=5. For NN, we use the ReLU activation function on the input, at most four hidden layers, and the softmax activation function on the output layer. The hidden layer size is reduced from the previous layer size to 1 5 , and the output layer size is two. We train the neural network model for 200 epochs by minimizing the loss function categorical cross entropy with the Adam optimizer and a mini-batch size of 32. We stop training if the loss function does not decrease in 10 epochs. For RF, we use the Gini impurity as the splitting criterion, and choose the number of decision trees to {5, 10, 20, 40, 60, 80, 100, 150, 200, 300} based on validation set performance. 6 For ADA, the maximum depth of decision trees, as weak classifiers, is chosen from 1 to 10 based on validation set performance. 6 For GCN, we use the ReLU activation function on the input and intermediate layers, and the softmax activation function on the output layer, where the intermediate layer size is 16, the optimizer is Adam, the L2 regularization weight is 5e-4, the dropout rate is 0.5, and the learning rate is 0.01. For 100 epochs, we train the GCN model, where the feature matrix is constructed by the Skip vectors, and the adjacency matrix is derived by converting an RDF graph to an undirected graph.
Results
We compare our Skip vectors with RDF2Vec and R-GCN. The classification accuracies for applying the Skip vectors to SVM, KNN, NN, RF, ADA, and GCN are displayed in Table 4. We have not shown the results on the datasets BGS and AM for SkipVec-GCN due to out of memory. The experimental results for Feat [14], WL (Weisfeiler-Lehman kernels) [19], RDF2Vec [16], and R-GCN [17] are shown in the bottom part of Table 4, reported in a previous work [17].
The average accuracies for SkipVec-SVM and SkipVec-RF on all datasets outperform the previous methods. In particular, the best accuracies for SkipVec-GCN on the AIFB dataset, SkipVec-SVM, SkipVec-RF, and SkipVec-ADA on the datasets BGS, and SkipVec-RF on the dataset AM outperform the previous methods on the AIFB, BGS, and AM datasets. These results indicate that the Skip vectors successfully represent the features of each resource in RDF graphs and achieve high accuracy in combination with conventional machine learning algorithms and the basic GCN.
Graph Kernels
There are several works on graph kernels [6,9,4,19,20,8,10] that apply tree or graph structure data without the input of vectors to machine learning tasks. The graph kernels are defined by functions that calculate the distance between data by counting the common substructures for two graphs or nodes, enabling machine learning on graph data. For example, an SVM can be trained to classify two graphs or nodes by calculating the similarity between data in the graph kernel instead of the inner product of vectors.
RDF Graph Kernels
Lösch et al. [12] have proposed the graph kernels Walk, Path, Full Subtree, and Partial Subtree for specializing in RDF graphs. Based on intersection graphs, the Walk kernel and Path kernel count the number of common walks and paths for two graphs, respectively. In addition, based on intersection trees, the Full Subtree kernel and Partial Subtree kernel count the number of common full and partial subtrees for two graphs, respectively.
By extending these RDF graph kernels, Arai et al. [1,2] have proposed the PRO kernel, which eliminates all intermediate nodes (objects) from Walks, and the Skip kernel, which arbitrarily eliminates edges (predicates) and nodes (objects) from PROs. The Skip kernel improved classification accuracy compared with the other RDF graph kernels.
However, the use of the RDF graph kernels is limited to be incorporated in machine learning algorithms such as SVM. On the other hand, the Skip vectors can be used as the input of feature vectors for applying most conventional machine learning algorithms.
RDF Graph Embeddings
Related to learning with RDF data, an approach to RDF graph embeddings has been studied. RDF2Vec [16] embeds target resources in an RDF graph into a low-dimensional vector space, inspired by the word embeddings of word2vec [13]. Given a sequence of words, Word2vec trains the continuous bag-of-words (CBOW) model to predict the target word from surrounding words (context words), and the skip-gram model to predict surrounding words (context words) from the target word. Instead of word sequences in word2vec, RDF2Vec learns the vector representations of resources in an RDF graph by applying graph structures such as Walks and Subtrees to the CBOW and skip-gram models.
The Skip vectors do not learn RDF graph embeddings as proposed in RDF2Vec, but generate feature vectors by extracting and selecting substructures from an RDF graph.
Relational Graph Convolutional Networks
GCN [11] performs semi-supervised learning on a graph dataset with the feature vectors for all nodes. For example, there is a dataset that consists of a graph network representing the relationships among papers and the feature vectors of papers created by the bag-of-words from the documents of each paper. Furthermore, R-GCN [17] has been extended as a variant of GCN to relational graph data, such as RDF graphs.
Unlike learning the R-GCN model, the Skip vectors aim at generating feature vectors from an RDF graph that can generally be used in combination with other learning models and methods, such as SkipVec-RF, SkipVec-ADA, and SkipVec-GCN, as described in Section 4.2.
Conclusion
In this paper, we proposed a method for generating the Skip vectors on various feature patterns (Skip patterns) by extracting predicates and objects from target resources in an RDF graph. In particular, we have theoretically proved the complexity of Skip patterns to limit the exponential combinations of features in RDF graphs. In addition, the dimensionality of Skip vectors is reduced by selecting features using the information gain ratio. In the experiments for classification tasks using conventional machine learning algorithms, we achieved better performance in comparison with the existing RDF graph kernels, RDF graph embeddings, and relational graph convolution networks. Therefore, for classification tasks in RDF data, Skip vectors successfully represent the features of target resources and can be combined with SVM, KNN, NN, RF, ADA, and the basic GCN.
In future work, we plan to apply Skip vectors to link prediction in knowledge graphs and logical reasoning with ontology embeddings. | 2022-01-07T02:15:36.474Z | 2022-01-06T00:00:00.000 | {
"year": 2022,
"sha1": "036dcd3838db8515cc118322b6e6efc0d33248dc",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "036dcd3838db8515cc118322b6e6efc0d33248dc",
"s2fieldsofstudy": [
"Computer Science"
],
"extfieldsofstudy": [
"Computer Science"
]
} |
213878626 | pes2o/s2orc | v3-fos-license | Optimization of Flow Field and Nox Concentration Field of Denitrification System in Coal-Fired Power Plant
A unit of a coal-fired power plant in North China was selected to study. The flow field distribution and NOx concentration distribution at the inlet and outlet of the denitrification system are tested. Then, the flow field adjustment and optimization of the unit are carried out based on the above test results, and the verification test of optimized flow field and NOx concentration distribution are carried out. After the flue optimization, the uniformity of NOx concentration distribution at SCR reactor inlet and outlet was improved, and the flue rate at SCR reactor inlet need further optimization through combustion adjustment.
Introduction
With the increasing attention to environmental protection in China, people pay more attention to the pollutants emission from coal-fired power plants [1][2]. In order to further reduce the emission of atmospheric pollutants from coal-fired power plants, the "provincial plan of action for energy conservation and emission reduction of coal-fired power plants (2014-2020)" issued in September 2014 requires that the emission limit of nitrogen oxides (NOx) for key areas and newly built coal-fired generating units was 50 mg/m 3 [3]. The implementation of "Environmental protection tax law of China" on January 1, 2018 stipulated that the emission concentration of air pollutant below the standard 50% cut tax in half [4]. That is to say, when the emission concentration of NOx reaches 25mg/m 3 , the environmental protection tax will be reduced by half. Therefore, it is of great economic and practical significance to explore and optimize the denitrification capacity of the existing denitrification system to reach the 50% of the existing standard.
This paper chooses a unit of a coal-fired power plant in North China to study. The denitrification system of the unit is tested, and the flow field distribution, NOx concentration distribution at the inlet and outlet of the denitrification system and its influence on NOx control are evaluated. Then, the flow field adjustment and optimization of the unit are guided, and the test of optimized flow field and NOx concentration distribution of NOx are carried out. Through the optimization and adjustment of flue, this paper provides a basis for improving the denitrification capacity of the unit meeting the 50% of the existing NOx emission standard.
Flue gas denitrification system
The flue gas denitrification system (Selective catalytic reduction, SCR) of selected power plant is described as follows: flue gas is extracted from the outlet flue of economizer, and enters the SCR reactor through horizontal flue, vertical flue, top of reactor and rectifying grid. There are five groups of diversion plates in the original denitrification system. The first group of diversion plates is located at the inlet of the horizontal flue enlargement section, the second group is located at the outlet of the horizontal flue enlargement section, the third group is located in the lower elbow of the vertical flue, the fourth group is located in the upper elbow of the vertical flue, and the fifth group is located in the upper elbow of the vertical flue.
Test scheme
The NOx concentration, temperature and flue gas velocity at inlet and outlet of SCR reactor before and after modification were measured. The flue gas NO X was measured by flue gas analyzer (NOVA 4000), and the flue gas temperature was measured by thermocouple and electronic thermometer. The flue gas dynamic pressure was measured by micro-pressure gauge and Pitot tube according to grid method, and the flue gas velocity is calculated via values of flue gas dynamic pressure and temperature [5].
Pre-modification testing
The test results of flue gas velocity and NOx concentration at the inlet and outlet of SCR reactor are shown in Table 1.
The average flue gas velocity at the inlet of SCR reactor is 11.8 m/s for both sides (A and B), and the flue gas velocity at the far side of SCR reactor is lower than that near the furnace side. The distribution relative standard deviations (RSD) of gas velocity on A and B sides of SCR reactor inlet were 19.99% and 14.21% respectively. The required maximum RSD of flue gas velocity at SCR inlet section was 15% [6], and the distribution uniformity of flow velocity was not satisfied in a side. The average NOx concentration at the A side and B side of SCR reactor inlet were 129.9 mg/m 3 (standard state, dry base, 6% O 2 ) and 106.2 mg/m 3 (standard state, dry base, 6% O 2 ), respectively. The distribution RSD of NOx concentration on A and B sides were 58.84% and 66.24% respectively, and the distribution deviation was large which mainly reflect the NO X concentration at the near furnace side of both sides are larger than those at the far furnace side.
The average NOx concentration at the A side and B side of SCR reactor outlet were 48.7 mg/m 3 (standard state, dry base, 6% O 2 ) and 17.8 mg/m 3 (standard state, dry base, 6% O 2 ), respectively, showing a larger concentration difference between two sides. The distribution RSD of NOx concentration on A and B sides of SCR reactor outlet were 23.36% and 47.33%, respectively, and the deviations of NOx concentration distribution on both sides were different. Table 1. The NOx concentration and flow rate at the inlet and outlet of SCR reactor before modification.
Reform scheme
The pre-modification experimental results showed that the distribution of NOx concentration flow fields in the both sides of SCR reactor inlet were not uniform, so the diversion and equalization of flue should be adjusted first; the distribution of NOx concentration at the outlets of A and B sides were not uniform also, and the catalytic performance of catalysts should be checked while adjusting the flow fields of SCR reactor. Combining with the structure of the reactor, the optimization scheme of the mixed flow field is as follows: (1) The length of the horizontal straight section of the second diversion plate group was reduced to 350 mm; the number of diversion plates in 3m was increased to 11, and the distance between diversion plates was 250 mm; the number of diversion plates in 9.2 m was increased to 23, and the distance between them was 383.3 mm.
(2) The length of the horizontal straight section of the fourth diversion plate group was rudeced to 500 mm.
(3) The number of the fifth group of diversion plates was added to 7 along the height direction, and the diversion plate spacing was 400 mm.
(4) The first group of mixers was installed before the expansion joints of the horizontal flue, and the second group of mixers was installed at the vertical flue.
Revamping effect of SCR reactor
After modification, the NOx concentration and flow rate at the inlet and outlet of SCR reactor are also detected, as shown in Table 2. The distribution RSD of NOx concentration on A and B sides of SCR reactor inlet were 2.54 % and 4.06 % respectively, which decreased from the distribution RSD of 65.8 % and 73.88 % of premodification test. The uniformity of NOx concentration distribution at SCR reactor inlet was improved.
The distribution RSD of flow velocity on A and B sides of SCR reactor inlet were 16.76% and 22.13 % respectively, which present different improvement results compared with the distribution RSD of pre-modification test. The uniformity of flow field at the inlet of a side was improved and B side was deteriorated, but the change of both sides was not significant, which need further optimization through combustion adjustment.
The distribution RSD of NOx concentration on A and B sides of SCR reactor outlet were 7.42 % and 9.14 % respectively, which decreased from the distribution RSD of 21.05% and 29.63% of premodification test. The uniformity of NOx concentration distribution at SCR reactor outet was improved.
Conclusion
The distribution RSD of NOx concentration on A and B sides of SCR reactor inlet decreased from 65.8 % to 2.54 % and from 73.88 % to 4.06 % respectively after optimization.
The distribution RSD of NOx concentration on A and B sides of SCR reactor outlet decreased from 21.05 % to 7.42 % and from 29.63 % to 9.14 % respectively after optimization.
The uniformity of flow field at the inlet of a side was improved and B side was deteriorated, but the change of both sides was not significant after optimization, which need further optimization through combustion adjustment. | 2019-12-12T10:05:49.172Z | 2019-12-10T00:00:00.000 | {
"year": 2019,
"sha1": "3a97cf17974c8de333763d919b68d7ec15086dec",
"oa_license": null,
"oa_url": "https://doi.org/10.1088/1757-899x/677/3/032037",
"oa_status": "GOLD",
"pdf_src": "IOP",
"pdf_hash": "61b49c42e62978804b7f3803fba80ed85d2e4a04",
"s2fieldsofstudy": [
"Physics",
"Engineering"
],
"extfieldsofstudy": [
"Physics",
"Environmental Science"
]
} |
229929954 | pes2o/s2orc | v3-fos-license | Surgical Strategy for Sacral Tumor Resection
Purpose This study aimed to present our experiences with a precise surgical strategy for sacrectomy. Materials and Methods This study comprised a retrospective review of 16 patients (6 males and 10 females) who underwent sacrectomy from 2011 to 2019. The average age was 42.4 years old, and the mean follow-up period was 40.8 months. Clinical data, including age, sex, history, pathology, radiographs, surgical approaches, onset of recurrence, and prognosis, were analyzed. Results The main preoperative symptom was non-specific local pain. Nine patients (56%) complained of bladder and bowel symptoms. All patients required spinopelvic reconstruction after sacrectomy. Three patients, one high, one middle, and one hemi-sacrectomy, underwent spinopelvic reconstruction. The pathology findings of tumors varied (chordoma, n=7; nerve sheath tumor, n=4; giant cell tumor, n=3, etc.). Adjuvant radiotherapy was performed for 5 patients, chemotherapy for three, and combined chemoradiotherapy for another three. Six patients (38%) reported postoperative motor weakness, and newly postoperative bladder and bowel symptoms occurred in 5 patients. Three patients (12%) experienced recurrence and expired. Conclusion In surgical resection of sacral tumors, the surgical approach depends on the size, location, extension, and pathology of the tumors. The recommended treatment option for sacral tumors is to remove as much of the tumor as possible. The level of root sacrifice is a predicting factor for postoperative neurologic functional impairment and the potential for morbidity. Pre-operative angiography and embolization are recommended to prevent excessive bleeding during surgery. Spinopelvic reconstruction must be considered following a total or high sacrectomy or sacroiliac joint removal.
INTRODUCTION
Sacral tumors are relatively uncommon. The majority of sacral tumors are metastatic tumors, although primary sacral tumors, such as primary bone tumors, neurogenic or congenital tumors, arising from the sacrum do occur. [1][2][3] During diagnosis, sacral tumors are often missed at an early stage because of their indefinite clinical and unclear radiological characteristics. Due to their characteristics and growth patterns, these locally aggressive tumors often have huge masses and lead to complications. 4 Computed tomographic (CT) scanning and magnetic resonance imaging (MRI) can aid in diagnosis in terms of the location, size, characteristics of these tumors, as well as in preoperative planning.
Most sacral tumors, except for metastatic tumors, are resistant to radiotherapy (RTx) and chemotherapy (CTx). Therefore, wide excision is typically the treatment of choice, and several surgical approaches and techniques have been described. Depending on the size and extent of the tumors, a single-stage posterior or a combined antero-posterior approach can be used.
The aim of this study was to review the management of sacral tumors, including symptoms, tumor features, perioperative management, surgical treatments, and complications.
MATERIALS AND METHODS
We retrospectively reviewed the records of 77 patients with sacral tumors who were treated between October 2011 and September 2019. Patients who underwent only tumor removal or surgical biopsy, as well as those with an intradural tumor, dermal or epidermal lesions, or recurrence after a previous surgery at other hospitals, were excluded. A total of 16 patients (6 males and 10 females) were analyzed. The mean age was 42.4 years (range, 16-79 years), and the mean follow-up period was 40.8 months (range, 12-79 months).
All of the patients underwent radiographs of the lumbosacral spine, including both sacroiliac (SI) joints, as well as CT scans and MRIs. The tumors were then classified by growth patterns according to guidelines for sacral tumors proposed by Wei, et al. 3 (Table 1).
The clinical data of the patients, including age, sex, history, pathology, radiographs, treatment, recurrence, and prognosis, were analyzed. We obtained approval for this study from our Institutional Review Board (number: 3-2019-0128).
Surgical procedures
The surgical approaches differed between cases depending on the pathology, location, and extent of the tumor.
There is a notable difference between the upper and lower sacrum regarding spinopelvic continuity. 4 Additionally, presacral tumor protrusion of the tumor into the pelvic cavity and involvement of the SI joint and vessels surrounding and feeding the tumor are key factors in determining the surgical approach, specifically between a single posterior or a combined antero-posterior approach.
After deciding on the surgical approach, sacrectomy was considered. According to the system of Fourney, et al., 1 the sacrum is divided into three regions of the upper, middle, and lower sacrum by S1-S2 and S2-S3 junctions. Based on the tumor extension, en bloc resection of primary sacral tumors was classified into five types (Table 2). 1,5,6 Combined antero-posterior approach A combined antero-posterior approach is appropriate for type I tumors with presacral masses or type II tumors complicated by anterior masses greater than 5 cm. In type I, an en bloc resection with an anterior lumbosacral discectomy is useful. In lateral tumor invasion with involvement of the SI joint, en bloc resection and anterior SI joint removal are appropriate. Additionally, if the peritumoral surrounding vessels are complicated, an anterior approach is helpful for vessel ligation.
Surgery was performed in two stages. The tumor was approached anteriorly through an anterior midline transperitoneal or retroperitoneal approach. 7 Recently, an anterior approach using laparoscope and retroperitoneoscope has been introduced. 8,9 In most cases, it was difficult to accurately identify masses in the anterior; therefore, the tumor margin was ex-plored and surgical vessel ligation was performed. Lateral dissection of the sacral ala allowed for identification of the lumbar trunk (L4-5) of the lumbosacral plexus. The SI joint was then identified lateral to these nerve roots, and bilateral partial ventral SI osteotomy was performed. The lumbosacral disc was exposed and removed along with the anterior aspect of the annulus fibrosis.
In stage II, via a posterior incision extending from L2 to beyond the coccyx, the posterior iliac crests, greater sciatic foramina, and sciatic nerves were exposed bilaterally, as well as the L3-5 spinous processes, facet joints, and transverse processes. An L5 laminectomy exposed the dural sac and cauda equine below this level. The sacral nerve roots were then divided, and the dural sac was amputated according to one of two methods: 1) amputation before closure, after which the dural sac was closed with a double layer of sutures, or 2) the dural sac was tied up several times first, after which the amputation was performed (Fig. 1). The remaining posterior L5-S1 intervertebral disc was then excised, and the posterosuperior iliac spines were removed, facilitating cutting the bilateral osteotomes lateral to the ala of the sacrum and parallel to the SI joints, thus completing the osteotomy cuts made in these planes during stage I. Partial mobilization of the sacrum facilitated identification of the sacrospinous and sacrotuberous ligaments, which were then transected. The sacral nerve roots were divided as they exited the sacrum, protecting the sciatic nerves from injury. The entire sacrum along with the neoplasm was then removed en bloc. 1,10-12
Single posterior approach
A single posterior approach is suitable for type II or III tumors with tumor extension to the anterior of less than 5 cm.
Reconstruction
In patients who underwent total sacrectomy, spinopelvic reconstruction was performed to facilitate early mobilization and better ambulation because of the spinopelvic discontinuity and instability. 13,14 Fixation methods included various combinations of spinopelvic fixation (SPF), iliac screw fixation (ISF), and pelvic ring reconstruction (PR). Two vertical L-shaped rods were positioned bilaterally in a manner allowing fixation to the L3-5 pedicles on each side according to the Galveston technique. 15 Two to three cross-connecting rods were used to secure the vertical rods to each other. Distally, the vertical rods were directed laterally into the ilium between the two cortices. Both autologous and allogenic bone grafts were placed to promote fusion of the transverse processes and lamina from L4 distally to the medioposterior aspect of the transected ilium bilaterally. An allograft strut was used to close the space between the two ilia, and a bone fusion promoter and bone chips were added across the graft area to facilitate fusion of the entire defect. In cases where the dead space was wide, a gluteus muscle and skin flap were then utilized. In ordinary cases, a conventional muscle and skin closure was performed after placement of closed drains. [16][17][18] Recently, 3D-printed implant reconstruction (3DIR) has been attempted. Therein, customized implants, which are fitted to a patient's anatomy, minimize dead space and eliminate the need for additional reconstruction. 19
Patient characteristics
The chief complaint of the patients was non-specific local pain. In terms of pre-operative motor deficits, 3 patients (19%) had leg weakness. Nine patients (57%) complained of bladder and bowel symptoms, such as voiding difficulty, urinary frequency or urgency, constipation, and residual sense after urination or defecation (Table 3).
Pre-operative management
Before surgery, 13 patients underwent CT-guided biopsy for surgical and adjuvant therapy planning.
Among possible complications, excessive blood loss is the most fatal and dangerous. Preoperative angiography was performed to identify highly vascular tumors. Subsequently, the decision between angiographic embolization or an anteriorly approached surgical ligation, or both, was made. In this study, 8 patients underwent preoperative angiography. For six of those patients, angiographic embolization was then performed using polyvinyl alcohol (150-250 µm or 250-300 µm) or coils via multiple feeding arteries from the internal iliac artery (Fig. 2).
Among six male patients, three were adults in their 20s who deposited their sperm in a sperm bank due to the risk of sexual dysfunction. The other patient in his 30s tried to deposit his sperm in a sperm bank, but was unable to do so, because ejaculation was impossible due to pain and it was thought that there would be no problems with sexual function after surgery.
Surgical management
Twelve patients underwent a single-stage posterior approach, and 3 patients underwent a two-stage antero-posterior approach. One patient underwent a two-stage anterolateral approach due to a chondrosarcoma located in an eccentric hemisacrum. The average blood loss was 5381 mL, and the average operation time was 8.24 hours.
In total, 5 patients underwent surgical iliac vessel ligation, all of which were performed after angiography to confirm a highly vascular tumor. In three cases, pre-operative embolization was performed first, but was not successful.
All patients who underwent a total sacrectomy and one patient who underwent a hemi-sacrectomy required SPF be-cause the tumor extended S1 superiorly and the SI joint laterally. Because SI joint removal was required when the tumor was removed, they also required ISF and PR. A total of 2 patients underwent 3DIR with SPF and ISF, according to the method of Kim, et al. 19 Two patients, one each with a high or middle sacrectomy, also underwent SPF. The patients' tumors were large and extended to S1 but did not involve the root. Therefore, the upper part was removed only, and high or middle sacrectomy was done. In addition, because the SI joint was also involved, ISF was done in addition to SPF. For the patient who received a high sacrectomy, PR was also performed to prevent complications stemming from a pelvic gap after tumor removal. The procedure was performed by orthopedic surgeons using a universal locking system plate (Zimmer Biomet, Warsaw, IN, USA) and fresh-frozen femoral head allograft struts (Fig. 3).
A vertical skin incision was performed on most patients who underwent a posterior-only approach. However, for 5 patients with a large tumor extending up to S1 or laterally to the SI joint, a total or high sacrectomy was performed using an inverted Tor Y-shaped skin incision. The inverted Y incision allowed for easier tumor removal due to a wider operation field, compared to the inverted T incision, although the ensuing large void and severe skin defect required a muscle flap using the gluteus maximus to be created by plastic surgeons (Table 4).
Of the patients with chordoma, five received adjuvant RTx.
Fig. 2. (A)
Angiography for highly vascular tumors. There were multiple feeding arteries from the internal iliac artery (white arrow). (B) Angiographic embolization was performed using polyvinyl alcohol via a feeding artery from the internal iliac artery (white dotted circle). After angiographic embolization, there was no vascular flow from feeding arteries.
A B
The patients with chondrosarcoma and osteosarcoma and one with giant cell tumor (GCT) received adjuvant RTx and CTx. One patient with neurofibroma received adjuvant CTx due to lung metastasis, not the sacral tumor. An additional patient with GCT received neoadjuvant CTx to decrease their tumor size. Another patient with Ewing's sarcoma received palliative CTx. Three patients (19%) experienced post-operative recurrence (37 and 45 months for chordoma and 13 months for chondrosarcoma), and all of those patients expired (Table 5).
DISCUSSION
Sacral tumors are account for approximately 1-7% of all spinal tumors. 20 Chordoma is the most common among the primary malignant bone tumors, and GCT is one of the most frequently seen benign lesions arising from the sacrum. Although GCT is a benign tumor, it is very vulnerable to local recurrence. 21,22 Neurogenic tumors can also occur in the sacrum. Sacral neurogenic tumors, which comprise schwannoma and neurofi- bromas, arise from the sacral nerve, grow along the bony neural foramen, and extend inside the sacral canal. 4,23 In our study, chordoma was the most prevalent, followed by nerve sheath tumors.
En bloc resection is associated with decreased local recurrence and increased survival rates compared to intralesional resection. However, injury to the sacral nerve roots may occur intra-operatively, leading to post-operative neurological dysfunction. 1,3,5,6,10 To do sacrectomy, dural sac amputation must be performed. In order to prevent postoperative cerebrospinal fluid leakage, careful watertight ties with non-absorbable silk suture material are required. 24 As mentioned above, angiography is needed to identify highly vascular tumors. Pre-operative embolization can reduce intra-operative blood loss and time, increase tumor resectability, and improve visualization of the operative field. Even partial embolization may reduce intraoperative bleeding. 25,26 In addition, patients who may wish to procreate in the future should be advised that sexual dysfunc-tion may occur after surgery, allowing them to store their sperm in advance of the procedure.
Identifying tumor characteristics in advance through preoperative CT-guided biopsy can also determine the surgical methods used and the appropriate treatment after surgery. However, during sacrectomy in a chordoma patient and biopsy in a resected tumor removed for intraoperative frozen, highpressure exudate was observed coming from inside the tumor (Fig. 4). If a pre-operative CT-guided biopsy had been performed, we could not have ruled out the possibility that tumor cells had come out along the wound tract and invaded the surrounding area. Therefore, we have to ensure extensive tumor removal including the biopsy tract and skin. In addition to this method, if small chordoma is suspected based on preoperative MRI, a total tumor removal without pre-operative biopsy is recommended to avoid tumor seeding.
The surgical methods to be used are determined in accordance with the size, location, and extension of the tumor. De- cisions are based on the level of sacrifice of the sacral root, which can have a direct effect on motor deficits, bladder and bowel symptoms, and sexual dysfunction that can occur after surgery. 1,4,5,10,11,17 Therefore, it is optimal to save the root if possible. In this study, a single-stage posterior approach was used for a GCT patient with a large, hypervascular tumor covering the lower part of the S1 body and around the S2 root. During the stripping process to save the root, an uncontrolled iliac vessel injury occurred, causing the patient to undergo vessel ligation, resulting in increased operation time and bleeding. If a combined antero-posterior approach been planned for the surgery, the patient would likely have had a better result: the patient has had three surgeries, fortunately leading to a very good outcome overall. Indeed, Lee, et al. 27 reported, in a case of presacral giant schwannoma, that an anterior approach could achieve total resection of presacral tumor without sacrificing sacral bone and sacral nerve root.
For tumors extending laterally to the SI joint, SI joint removal must be included when the tumor is removed. Six patients (38%) underwent SPF due to spinopelvic discontinuity and instability. SPF was performed in the 3 patients who underwent a total sacrectomy, followed by ISF and PR. An additional 3 patients underwent SPF and ISF, two of whom also received 3DIR.
In cases with a large tumor, long incision, and notable muscle dissection, the resulting void may be large after tumor removal, which can lead to wound dehiscence or necrosis. In our study, an inverted T or Y-shaped skin incision was performed in 7 patients. A gluteus muscle and/or skin flap was also utilized in these patients to prevent wound problems.
Adjuvant CTx or RTx should be considered as post-operative treatments according to the tumor pathology. In this study, pa-tient prognosis was primarily affected by pathology and age. Three patients expired after recurrence. For the chondrosarcoma patient, the sacral lesion was well removed, but the tumor had heavily invaded the abdominal cavity such that only a neartotal removal was possible. The remnant tumor was then subjected to adjuvant CTx and RTx, but the patient did not survive. Five of the 7 chordoma patients were under 60 years old, and the two others were over 70 years old. Unfortunately, despite the total sacrectomy and adjuvant CTx and RTx, both of the older patients experienced recurrence and expired.
The recommended treatment for sacral tumors is to remove as much of the tumor as possible. Surgeons should be careful of excessive bleeding during tumor removal, and therefore, preoperative angiography and embolization are recommended. In addition, the type of sacrectomy should be determined according to the location and extension of the tumor, and to reduce post-operative complications, all possible efforts should be made to save up to the S2 root. In a systematic review paper, Zoccali, et al. 28 reported that patients who underwent a sacrectomy maintained functionally normal ambulation in 56.2% of cases when both S2 roots were spared, in 94.1% when both S3 roots were spared, and in 100% with more distal resections. Normal bladder and bowel function were not present when both S2 were cut. When one S2 root was spared, normal bladder function was present in 25% of cases; in 39.9% when both S2 roots were spared, in 72.7% when one S3 root was spared, and in 83.3% when both S3 roots were spared. Abnormal bowel function was present in 12.5% of cases when both S1 roots and one S2 root were spared, in 50.0% of cases when both S2 roots were spared, and in 70% of cases when one S3 root was spared. If both S3 roots were spared, bowel function was normal in 94% of cases. When even one S4 root was spared, normal bladder and bowel function were present in 100% of cases. Unilateral sacral nerve root resection preserved normal bladder function in 75% of cases and normal bowel function in 82.6% of cases. Motor function depended on S1 root involvement. In our study, Patients 14 and 15 had tumor invasion around the sacral root, and the pathology was GCT. Thus, if en-bloc resection was not performed, they were judged to be at high risk of recurrence, and the tumors were removed accordingly.
A limitation of our study was the relatively small number of sacrectomy cases. This number alone may not be enough to establish an optimal surgical strategy. In addition, there were differences in the treatment and follow-up periods depending on disease entities. Also, with the presence of several different operators for each individual, the resulting surgical strategy was not unified either. However, we believe that our data will contribute to and help in establishing theories as more data are added by other researchers.
If a tumor is large, a combined antero-posterior approach is recommended over a single-step approach. Spinopelvic reconstruction must be considered following a total or high sacrectomy or SI joint removal, and for cases with a large incision or void following tumor removal, a muscle and/or skin flap should be considered. We have summarized our treatment algorithm in Fig. 5. | 2020-12-31T09:12:41.712Z | 2020-12-23T00:00:00.000 | {
"year": 2020,
"sha1": "4e30f106266c8869b0381dc88cc1a961c0b3b420",
"oa_license": "CCBYNC",
"oa_url": "https://doi.org/10.3349/ymj.2021.62.1.59",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "2b6cf9bef5814362fe62016209abedb81ff1d227",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
3194870 | pes2o/s2orc | v3-fos-license | Characterization of a spontaneous disease of white leghorn chickens resembling progressive systemic sclerosis (scleroderma)
University of California, Davis (UCD) line 200 White Leghorn Chickens spontaneously develop a syndrome that has many analogous features to human progressive systemic sclerosis. This syndrome is characterized by progressive involution of comb, dermal fibrosis, and distal polyarthritis. These three features occur within 6 wk after hatching, and are accompanied by a 40% mortality as a result of vaso-occlusive disease, with development of secondary infection of peripheral gangrenous lesions. Birds that survive greater than 2 mo after hatching progressively develop fibrosis of the esophagous and mononuclear infiltration of heart and kidney, with prominent occlusion of small and medium sized blood vessels. In addition, line 200 chickens develop rheumatoid factors, antinuclear antibodies, and antibodies to collagen, but do not have antibodies to thymocytes, DNA, or extractable nuclear antigens. Moreover, antinuclear antibodies when studied using HEp-2 cells as substrate demonstrate predominantly a speckled pattern. This syndrome of line 200 chickens is not detectable in F1 crosses to several UCD inbred lines. F1 X parental line BC1 backcrosses have an approximately 50% incidence of disease, suggesting that this syndrome is inherited as autosomal recessive. However, only 4% of F2 generation birds show abnormal symptoms, suggesting the presence of modifying genes. There is no appearance of IgG deposition, as determined by immunofluorescence, in either skin, blood vessels, esophagus, or heart. However, approximately 20% of chickens have a glomerulonephritis; this feature appears to be a terminal event and does not appear clinically significant. Although this syndrome of line 200 chickens has several features that are in sharp distinction to human scleroderma, the presence of common immunologic and pathologic denominators suggest that this spontaneous disease may be an appropriate model to develop a better understanding of autoimmune connective tissue diseases.
atrophy at an early age, with adult birds exhibiting variable degrees of missing combs. This abnormality was termed "self-dubbing" ("dubbing" being defined as the cutting or removing the comb artificially) and was seen originally in only a few males of this outbred flock. Because of its potential practical usefullness, the line was propagated by Dr. Bernier. Upon his retirement, Dr. Bernier offered this interesting bird model to one of us (Hans Abplanalp). At this point, only three males and five females were alive to propagate this line of self-dubbing chickens. Thanks to special efforts of Dr. Bernier's successor, Dr. J. Renden, we were able to establish an initial flock of 15 males and 25 hens at the Department of Avian Science at the University of California, Davis, Calif. This material is now referred to as UCD line 200. All experimental work is derived from offspring from these original chickens.
Animals. All line 200 and control line (UCD lines 003, 011, 140, 159, and 446) chickens were bred aqd maintained by the Department of Avian Sciences, University of California, Davis. (10). Hens were artificially inseminated and eggs collected for 2 wk. All eggs were incubated for 21-22 d in Jamesway 252 incubators (Butler Mfg. Co., Jamesway Div., Fort Atkinson, Wis.). On the day of hatch, chicks were banded and moved to floor pens or to battery brooders at the Hopkins poultry plant, University of California, Davis. Chickens were vaccinated against Marek's disease on day 1, Newcastle disease at 4-6 wk and again at 10-12 wk, and fowl pox at 10-12 wk. Birds were transferred to individual cages in windowless houses cooled by evaporation: males at 14 wk and females at 15 wk of age. The characteristics of the control chickens used in this study have previously been noted (11).
Genetics. Abnormal Pathology. 10 UCD line 200 and control line 446 were killed at 6 wk and 6 mo (the median survival) of age. The comb, skin (from the back of the neck), esophagus, small intestine, heart, lung, and kidney were removed and preserved in formalin. Sections 5-~m thick were prepared and stained with hematoxylin and eosin (HE) and Masson's trichrome (12). Direct immunofluorescence was performed on fresh-frozen tissue as described (10).
Hematology. At 6 mo of age, packed cell volumes (PCV), hemoglobin concentrations, and blood cell counts were determined in 20 line 200 and line 159 birds of each sex. PCV were performed in mierohematocrit capillary tubes (13). Hemoglobin concentrations were assayed by the cyanmethemoglobin method (13). Blood cell counts were made by a modification of the Rees-Ecker method utilizing the following staining solution: 3.9 g sodium citrate, 0.2 ml neutral formalin, 0.5 g brilliant cresyl blue, and 100 ml distilled water (10). Fresh blood samples were diluted 1:200 in this solution in a blood-diluting pipet and the cells microscopically counted in a Neubauer hemocytometer.
Coombs's Tests. At 6 wk and 6 mo of age, heparinized blood samples were obtained from the wing vein of 20 line 200 and line 003 controls. The erythrocytes were washed three times in phosphate-buffered saline (PBS), removing the huffy coat after each wash, and the cells suspended to a final concentration of 2% in PBS. Rabbit anti-chicken y-globulin, rabbit antichicken 7S Ig (Fe specific), and rabbit anti-chicken IgM (/1 specific), were all heat-inactivated in a 56°C water bath for 30 min and absorbed with an equal volume of packed normal chicken erythrocytes. A direct Coombs's assay was performed in microtiter plates by addition of 25 #1 cell suspension to 25 #1 serially diluted antisera. Known positive samples (UCD line 140) and negative controls were incubated in each test (10,11). Additionally, direct erythrocyte agglutination tests were performed in microtiter plates by adding 25 v,l 2% erythrocyte suspensions from normal chickens to serially diluted test chicken sera (10).
Co~oprecipitates. Blood was drawn into prewarmed syringes from the jugular vein of 13 line 200, 19 line 3, and 14 line 159 chickens at 6 mo of age and transferred to siliconized glass tubes kept at 38-40°C. To test for the presence of serum proteins that precipitate in the cold, the method of Weisman and Zvaifler (14) was employed. The blood was allowed to clot for 3-4 h at this temperature, and the serum was harvested by centrifugation at 500 g for 10 min at room temperature. Then, 1 ml of serum was dispensed into a 12-× 75-mm tube and placed at 4°C for 72 h. Samples were handled in as sterile manner as possible throughout the procedure. Known positive chicken sera (UCD line 140) were included as previously described (10). Rheumatoid Factor. At 6 mo of age, 17 line 200, and 11 line 446 birds were bled. Rheumatoid factor was quantitated by the Rose-Waaler test (10). Sheep erythrocytes cells (SRBC) were coated with chicken 7S Ig by incubating a subagglutinating dose of the 7S Ig fraction of chicken anti-SRBC sera for 30 min at 37°C. The coated cell suspension was washed three times and resuspended in PBS to a final concentration of 5%. Test chicken sera were serially resuspended in PBS to a final concentration of 5% and serially diluted in microtiter plates. Agglutination was examined after a 2-h incubation at room temperature. Known rheumatoid factor-positive samples (UCD line 140) were included.
Anti-Thymocyte Antibodies. 15 6-mo-old line 200 and line 003 controls were assayed for the presence of anti-thymocyte antibodies. An indirect immunofluorescent test was performed on thymus cells obtained from 2-4-wk-old normal chicks (10,11). Thymuses were removed, and the cells teased into cold RPMI-1640 media (Pacific Biologicals) with forceps, and then pressed through a stainless-steel wire screen. The cell suspension was washed three times in cold RPMI-1640 and adjusted to a concentration of 10 X 106/ml after the final wash. 1 ml of the cell suspension was delivered to 12-X 75 mm tubes and the cells pelleted by centrifugation. To the pellet, 0.1 ml undiluted test chicken sera was incubated at 37°C for 60 rain and at 4°C for an additional 60 min. The cells were then washed three times in cold RPMI-1640 and pelleted. After removal of the supernate, 0.1 ml of diluted fluorescein isothiocyanate-(FITC) conjugated anti-chicken ),-globulin was added to the cells and incubated for 30 min at room temperature. The FITC antisera were used at a 1 : 15 dilution to eliminate nonspecific labeling of thymocytes. Known positive and negative samples were included in all assays.
Antinuclear Antibodies. At 6 mo of age, the titer of antinuclear antibodies in 40 line 200 and 28 line 011 controls was quantitated using HEp-2 cells as substrate (15). Known positive samples were included from UCD line 140 (10). Antibodies to DNA and extractable nuclear antigens (ENA) were performed as described (16,17). Anti-Collagen Antibodies. Type II chicken collagen was purified as previously described from 2-wk-old control line 446 chicks (19). This material was coupled to SRBC and the presence of antibodies was determined by enzyme-linked immunosorbent assay (ELISA) and by hemagglutination in 46 6-too-old line 200 chickens and 29 birds from control lines 003, 159, and 446 (19)(20)(21).
Results
Natural HistoTy of Disease Expression. Line 200 chicks, when compared with other UCD lines, are normal until ~ 1-2 wk of age. At that time, a series of abnormalities appear that are readily detectable by casual observation. First, swelling and erythema of the comb and wattle occurs. (Fig. 1). Subsequently, over the next several weeks, a polyarthritis involving only peripheral joints develops (Fig. 2). The severity and frequency of comb involvement peaks at 2-4 wk of age. Concurrently, the skin on the neck and back begins to swell, becomes severely indurated, loses feathers, and becomes taut (Fig. 3). These lesions, however, are not found in all birds. Indeed, the most prevalent finding is the self-dubbing comb abnormality initially observed by Bernier. This involvement of the comb is found in "'90% of birds at 4 wk after hatching (Fig. 4). In contrast, polyarthritis is found in only 20% of birds at 3 weeks and in 35% at 4 wk after hatching. Finally, dermal lesions are significantly more delayed than either comb or joint pathology and are found in 20% of birds at 5 wk and 45% at 7 wk after hatching (Fig. 4). There are no differences in disease expression between males and females, and thus data from both sexes are combined. The mortality of line 200 chickens is -20% at 1 mo of age, 40% at 4 mo of age, and 55% at 10 mo of age (Fig. 5). This compares with a <10% mortality at 1 yr in line 159 controls (Fig. 5). Birds that develop the acute episode of swelling of comb, polyarthritis, and dermal alterations account for the early accelerated mortality. The cause of death in such chickens appears to be secondary cutaneous infections of the comb, wattle, neck, and feet. Indeed, in virtually all such birds, a gangrenous necrosis of comb and digits develops after severe vascular occlusion. The vascular lesions that predispose to autoamputation of the comb are readily visible. For example, in line 200 chickens, multiple microvascular infarcts are seen. Nail beds reveal severe vasodilation, even in the absence of arthritis.
The mortality of older birds is a result of a wasting-like syndrome, characterized by listlessness, fatigue, anorexia, and loose droppings. At autopsy, such birds exhibit signs of pneumonitis but also have severe small vessel occlusive disease with involvement of esophagus, heart, and/or kidney (see below: pathology). (Table I). Finally an F~ generation produced by crossing F1 individuals with each genetic subline produced only 4% abnormals.
Pathology. The pathology of line 200 chickens has thus far been limited to extensive examination of 6-wk-and 6-mo-old chickens. 6-wk-old chickens have abnormalities of skin and comb, but do not appear to have significant internal manifestations, i.e., ther visceral organs examined appear similar to controls (Table II). The only major exception to this observation is the presence of pneumonia in autopsied birds.
The pathology of skin and comb of 6-wk-old line 200 chickens is noteworthy for three major features: First, an intense mononuclear cell infiltrate (Fig. 6A) present diffusely throughout all layers of skin and subcutaneous tissue is apparent. Second, there is an marked proliferation of small vessels (Fig. 6 B). This proliferation of small vessels is accompanied by an intense collagen deposition by a rich fibroblastic-like network and appears concurrent with the mononuelear cell infiltrate. However the proliferation of small vessels continues after the mononuclear cell infiltrate begins to wane. Finally, skin and comb develop severe occlusive disease of small arteries with vessels developing thickening of the muscle wall, narrowing of the lumen, and intimal proliferation (Fig. 6C). The overall result is replacement of subcutaneous fat and muscle with collagen ( Fig. 7).
At 6 mo of age, the majority of line 200 chickens have the self-dubbed appearance but <10% have clinically visible joint or skin involvement. In contrast, they have extensive involvement of esophagus, heart, and kidney (Table II). The esophagus from control chickens has an intact epithelium with loose underlying connective tissue and striated muscle layer (Fig. 8A). T h e esophagus of chickens does not show a gradual change from skeletal muscle to smooth muscle and all levels are comparable.
In contrast, 60% of 6-mo-old line 200 birds have a distinctly different esophagus than that seen in controls with extensive thickening of the underlying connective tissue by densely packed collagen. This collagen extends down into and through the muscle layer (Fig. 8 B). In addition, mononuclear focal infiltrates in mucus glands are present in the subepithelial layer of the esophagus of abnormal birds (Fig. 8 B); these glands, in the chicken, are analogous to h u m a n salivary glands. Finally, in addition to the intense fibrosis, there is also significant small vessel proliferation in subepithelial layers of tissue. These vessels appear thick-walled, much like those in skin. The kidneys of line 200 chickens reveal a variety of abnormalities. Because serial studies were not done, it is unclear as to whether or not these changes represent primary or secondary features of disease. However, in virtually all (greater >90%) 6mo old line 200 chickens, significant swelling and thickening of muscular layers of renal blood vessels are present. Some vessels have surrounding round cell infiltrates. Further, ~20% of birds have a significant glomerulonephritis and, by immunofluorescence, have IgG deposits within glomeruli. IgG deposits were not found in skin, blood vessels, or the other organs described.
The thoracic cavity of line 200 chickens is most noteworthy for the appearance of pericardial effusions in 40% of 6-mo-old animals. In addition, sections of heart show moderate to severe prominance of blood vessels and cellular proliferation on the epicardial surfaces. In contrast, the lung, at 6 mo of age, appears to be relatively normal. In only 1 of 10 birds was there a significant pulmonary abnormality. In this animal, there were thickened blood vessel walls present similar to those described earlier. In addition, there was an intense cellular proliferation obliterating the aveolar space. The culture of this lung was sterile.
Hematology. At 6 mo of age the hemoglobin, hematocrit, and leukocyte counts of line 200 chickens were similar to line 159 controls (Table III). However, the differential leukocyte count revealed a moderate (9% male; 14.5% female) eosinophilia in line 200 chickens compared with <2.5% in control line 159 chickens.
Autoantibodies. Both the direct and indirect Coombs's tests performed on line 200 chickens were negative. Similarily, there were no cryoprecipitates present in any of the 5-mo old line 200 chickens; samples from abnormal line 140 birds produced cryoprecipitates. At 4 mo of age, rheumatoid factor was found in 60% of line 200 chickens examined with a mean (log2) titer of 5.3 + 1.8; no detectable rheumatoid factor was found in 17 control line 446 birds (Table IV). There were no naturally occurring anti-thymocyte antibodies detected by the immunofluorescent methods described herein. Similarily, line 200 chickens were'found to have no detectable antibodies to native DNA or ENA. Known reference standards from UCD line 140 chickens gave expected results and served as quality controls as previously described. In contrast, 26 of 40 6-mo-old line 200 chickens revealed a positive ANA test with titers of 1:20-1:160; 0 of 28 line 011 controls gave a positive ANA (Table V). Pattern reading using the HEp-2 cell line revealed the majority of positive ANA (20 of 26) to be speckled (Fig. 9). Finally, by hemagglutination, 45 or 46 line 200 chickens were Fro. 7. Sections are of line 200 chicken skin (top). The epidermis is atropic. Notice the dense collagen bundles that have replaced the usual delicate collagen of the superficial dermis. The remaining dermis is infiltrated with chronic inflammatory cells and adipose tissue has been replaced by fibroblasts and. lymphocytes. Compare with sections of normal skin (bottom). Notice the delicate superficial dermis; the subcutaneous tissue contains fat and striated muscle; HE staining; X 77.
S P O N T A N E O U S S C L E R O D E R M A -L I K E DISEASE OF W H I T E L E G H O R N S
Fla. 8 A Flo. 8. Line 446, 6-too-old esophagus (A). Note the delicate nature of the underlying collagen; this collagen does not extend into the muscle layer. In contrast, in a line 200 6-mo-old bird (B), note the lymphocytic infiltrates around esophageal mucus glands. In addition, there is an extremely thickened and dense collagen which extends down through the muscle layer; HE staining; × 100.
Fie. 8 B noted, at 6 mo of age, to have antibodies to type 2 collagen with a mean titer of 3.1 _ 1.4 (Table VI). In contrast, there were no detectable anti-collagen antibodies present in 29 control chickens; ELISA gave similar results.
Quantitative Serum Immunoglobulins. At (Table VII). Biochemical Markers. There were no significant differences between line 200 and control chickens at 6 wk or 6 mo with regard to sodium, potassium, chloride, COs, BUN, creatinine, SCOT, SGPT, alkaline phosphatase, uric acid, CPK, or aldolase.
Discussion
There are several well-catalogued mutations of chickens that have provided useful models for connective tissue and immunologic research. These models include congenital ichthyosis, muscular dystrophy, thyroiditis, and spontaneous acquired dysgammaglobulinemia (22)(23)(24)(25). In addition, in 1945, an unusual autosomal recessive disease termed "dactylolysis" was reported by Shoffner in a short communication (26). This report described a mutation in White Leghorn chickens characterized by accelerated mortality, involution of comb, and peripheral arthritis. Unfortunately, although the author suggested that the syndrome was similar to human scleroderma, no further attention was given to these birds and the relationship between the disease of Shoffner's flock and the birds studied herein is unknown. Finally, although there are a large number of inherited diseases of collagen of dog, cat, mink, mice, and cattle, these syndromes have no similarity to those found in line 200 chickens nor any relationship to scleroderma in humans (26, 27). In humans, familial clustering of scleroderma is generally very rare. Of the few published reports describing scleroderma in more than one family member, siblings were usually involved, although there have been reports involving father and son and a mother and four daughters (28,29). Because there are few well-documented familial cases of scleroderma, it has not been possible to characterize scleroderma as a heritable disorder. Moreover, a variety of immunogenetic and epidemiologic studies have failed to conclusively support an association between major histocompatibility antigens and scleroderma (30,31). Nonetheless, it is of interest to note that close relatives of patients with scleroderma have been found to have low titers of ANA and an abnormally high frequency of chromosome breakage (31). These two factors certainly lend support to the notion of a familial predisposition to PSS.
The syndrome of line 200 chickens has a number of similarities as well as differences with human scleroderma. First, similar to the human syndrome, affected chickens develop fibrosis of skin and esophagus. Moreover, intense mononuclear cell infiltrates accompanied by collagen deposition and severe vascular occlusion appear early. In addition, as described in human scleroderma, there is marked proliferation of small vessels at the site of involvement. Nonetheless, there are a number of significant histologic differences. Unlike the histocytic infiltrate and intimal damage of small vessels of human scleroderma, affected chickens develop severe muscular hypertrophy of small vessels. This hypertrophy is a major basis for the occlusive vascular disease. Secondly, although an intense deposition of collagen occurs, the onset is significantly more acute than in humans. Further, a significant percentage of older chickens develops glomerulonephritis, a feature uncommon in human scleroderma. This glomerulonephritis, however, may be a secondary feature of infection and is probably clinically insignificant because chickens do not develop azotemia. The urine of chickens drains into the cloaca, making urinalysis difficult. The immunologic features of this syndrome in line 200 chickens include the presence of ANA, rheumatoid factors, and antibodies to type 2 collagen, but absence of positive Coombs's test, cryoprecipitates, or antibodies to thymocytes, DNA, or ENA. These immunologic features are again similar to large series of patients with scleroderma (1,15). ANA, in particular, give a speckled pattern in ~50% of chickens.
This syndrome, however, in sharp distinction to human scleroderma, appears to be significantly more fulminant in chickens. The dramatic mortality of young chickens is a result of secondary infection of peripheral gangrenous lesions in regions vascularized by occluded vessels. Nonetheless, as emphasized earlier, these chickens have been maintained in a closed flock for nearly forty years. Moreover, as they were selected on the basis of the comb lesion, it is not surprising that this manifestation is so evident. Birds that survive the acute episode and develop the self-dubbed appearance later develop a progressive disease involving internal organs; fortunately, sufficient enough fertility is maintained to permit propagation of the line.
The use of line 200 chickens as a model of human scleroderma must be contrasted with other proposed animal models. The first proposed model of PSS was homologous disease of rats (5,6). Experimental homologous disease is induced by injection of lymphoid cells from donors of an inbred strain into tolerant rats. Among the prominent features of homologous disease are lesions of skin, joints, and heart with impressive similarities to scleroderma. Moreover, of particular importance to scleroderma, there is immunologic reactivity directed against host skin. Indeed, affected rats reject grafts of their own skin at the same time that donor grafts are accepted• This autograft rejection can be seen to be highly specific for host skin. Chronic skin lesions of homologous disease show extreme epidermal atrophy with markedly increased thickening and collagenization of the dermis• An inflammatory reaction, if present, is usually mild in such lesions. Similarly chronic graft-vs.-host disease (GVHD) after bone marrow transplantation has been found to have scleroderma-like features in humans (7,8,32). Spielvogel et al. (7) have noted that patients with chronic GVHD have cutaneous abnormalities including degeneration and loss of cellularity of surface epithelium, hyperkerotosis, hyperpigmentation, and an infiltrate of the mid and deep dermis by monocytes, histiocytes, and fibroblasts (7). Moreover immune complexes are found in vessel walls and at the dermal-epidermal junction• The latter features have suggested that chronic GVHD may be more similar to either vasculitis or systemic lupus erythematosus. Nonetheless, other clinical features of both scleroderma as well as SjiSgren's syndrome have also been reported in chronic GVHD. In contrast to line 200 chickens, however these models do not develop significant titers of ANA, antibodies to collagen, or rheumatoid factors. Moreover, and most important, they do not permit establishment of a distinct animal colony with a spontaneous disease• We do not, of course, propose that this syndrome of line 200 chickens is the identical disease process known as human progressive systemic sclerosis. At best it is a new disease of chickens, in a highly inbred flock of birds, characterized by severe vasoocclusive disease, fibrosis, and the presence of a variety of autoantibodies and hypergammaglobulinemia. A significant volume of work remains to be done to place these observations in perspective and to define the associations of specific pathologic and immune abnormalitites with the clinical expression of disease and the genetic basis of the syndrome• Summary University of California, Davis (UCD) line 200 White Leghorn Chickens spontaneously develop a syndrome that has many analogous features to human progressive systemic sclerosis• This syndrome is characterized by progressive revolution of comb, dermal fibrosis, and distal polyarthritis. These three features occur within 6 wk after hatching, and are accompanied by a 40% mortality as a result of vaso-occlusive disease, with development of secondary infection of peripheral gangrenous lesions. Birds that survive >2 mo after hatching progressively develop fibrosis of the esophagous and mononuclear infiltration of heart and kidney, with prominent occlusion of small and medium sized blood vessels• In addition, line 200 chickens develop rheumatoid factors, antinuclear antibodies, and antibodies to collagen, but do not have antibodies to thymocytes, DNA, or extractable nuclear antigens. Moreover, antinuclear antibodies when studied using HEp-2 cells as substrate demonstrate predomi-nantly a speckled pattern. This syndrome of line 200 chickens is not detectable in Ft crosses to several UCD inbred lines. F1 X parental line BCi backcrosses have an ~50% incidence of disease, suggesting that this syndrome is inherited as autosomal recessive. However, only 4% of F2 generation birds show abnormal symptoms, suggesting the presence of modifying genes. There is no appearance of IgG deposition, as determined by immunofiuorescence, in either skin, blood vessels, esophagus, or heart. However, -20% of chickens have a glomerulonephritis; this feature appears to be a terminal event and does not appear clinically significant. Although this syndrome of line 200 chickens has several features that are in sharp distinction to human scleroderma, the presence of common immunologic and pathologic denominators suggest that this spontaneous disease may be an appropriate model to develop a better understanding of autoimmune connective tissue diseases. | 2014-10-01T00:00:00.000Z | 1981-06-01T00:00:00.000 | {
"year": 1981,
"sha1": "9f1b525b22f7a78a86620d8388d848b2e59e2064",
"oa_license": "CCBYNCSA",
"oa_url": "http://jem.rupress.org/content/153/6/1640.full.pdf",
"oa_status": "BRONZE",
"pdf_src": "PubMedCentral",
"pdf_hash": "9f1b525b22f7a78a86620d8388d848b2e59e2064",
"s2fieldsofstudy": [
"Biology",
"Medicine"
],
"extfieldsofstudy": [
"Biology",
"Medicine"
]
} |
72070387 | pes2o/s2orc | v3-fos-license | Hidden foreign body in an unexplained asthma
Foreign body aspiration (FBA) is a common condition with potential to cause significant morbidity and mortality. In children, highest incidence is within the first 3 years of life. Without witnessed aspiration or acute symptoms such as choking, coughing or respiratory distress, it may go unnoticed as clinical presentation and radiological investigations generally have low diagnostic value. Delay in diagnosis increases the risk of developing severe complications. Unexplained recurring or persistent lung pathologies in this age group should raise suspicion of FBA, and early bronchoscopy is warranted for confirmation of diagnosis. We report a case of FBA in a 2-year-old boy whom had been symptomatic for 3 months and was treated as bronchial asthma prior to diagnosis of foreign body in the left bronchus.
Introduction:
Foreign body aspiration is a common and potentially fatal accident.According to the National Safety Council, it is the fourth leading cause of death due to unintentional injury in the United States in 2004, and the mortality rate was estimated to be 1.5 per 100,000 population 1 .Classical history consists of choking episode followed by coughing, wheezing, stridor, or acute respiratory distress 2 .Especially in children, the acute event may go unnoticed, and foreign body aspiration may be mimicked as other illness such as bronchial asthma, bronchitis, or pneumonia.The delay of making an accurate diagnosis may lead to dangerous consequences for the patient 4 .
Case report:
A 2-year-old boy presented with history of persistent non-productive cough for 3 months, followed by fever in past 1 week associated with occasional noisy breathing.He was brought to emergency department due to progressive dyspnea.Prior to this, he has been seen by several primary care practitioners and the diagnosis of bronchial asthma has been made.However the treatment given failed to control his symptoms.There was no history of witnessed foreign body aspiration.He was otherwise well and active at home.Examination revealed a febrile boy but not in respiratory distress.He was not cyanosed and there was no stridor.Wheeze was heard on auscultation at the middle and lower zone of the left lung.
Breath sound was equal bilaterally.Baseline blood investigation was normal, including the white blood cell count.Chest X-ray showed a radiopaque spring-like foreign body situated in the left main bronchus.No other significant finding was noted in the x-ray.
Rigid bronchoscopy was done under general anesthesia, revealed a small metal spring in the left main bronchus.It was removed completely without difficulty by using the rigid optical forceps.The left main bronchus noted to be mildly inflamed without any laceration, slough or pus.No other foreign body was found.
He had an uneventful recovery, and was discharged home a day after the bronchoscopy.
Discussion:
In children, FBA most commonly occurs within the first 3 years of life 3,[5][6][7] .This can be attributed to several factors: tendencies of putting things in their mouth for exploration; incomplete molar reducing effective chewing; and incomplete swallowing reflex 3 .Male to female ratio is 2:1, reflecting the more adventurous games and impulsive nature of the boys 3,5,6 .Majority of the aspirated material is organic, seeds and nuts being the most common, comprising between 50 to 78% of all foreign body found in the reviewed studies 3,5 7 .Size, shape, and surface of the aspirated foreign body, and anatomical condition of the patient determine the lodgement location.Preferential of the foreign bodies to be located in right bronchus can be explained by the wider and more in-line right bronchus with the trachea as compared to the angulated left bronchus.
Positive history of aspiration noticed by a witness is the most important clue for FBA, which unfortunately is not always available 2 .Most common presentation of FBA is persistent cough [5][6][7] as in this case.Other sign and symptom includes dyspnea, wheezing, stridor, or cyanosis.Fever is suggestive of contaminated or chemically irritating foreign body, or patient has developed infective complications such as pneumonia or lung abscess 2 .Clinical triad of cough, localized wheezing and localized decreased breath sound has been reported to be present in 15-40% of patients.About 5% of patients with positive FBA were asymptomatic on presentation 3,6 .
The value of chest X-ray (CXR) in diagnosing FBA remains controversial.Radiopaque foreign body is only present in 2-15% of cases, and CXR was reported as normal in 16-61%
Conclusion:
In summary, FBA is a common accident in children with potential to cause significant morbidity and mortality.High index of suspicion is needed in cases with atypical presentation or unexplained persistent bronchial asthma.
Figure 1 :
Figure 1: Erect PA chest x-ray showing radiopaque material in the region of left main bronchus.
Figure 2 :
Figure 2: A metal spring removed from the left main bronchus.
6he technique used depends on the type of foreign body and the ability of the individual operator6. | 2018-10-22T21:20:51.981Z | 2012-11-24T00:00:00.000 | {
"year": 2012,
"sha1": "36580f7e46b42db41c8de827085eaf01f22309ba",
"oa_license": "CCBY",
"oa_url": "https://www.banglajol.info/index.php/BJO/article/download/12017/8767",
"oa_status": "HYBRID",
"pdf_src": "Anansi",
"pdf_hash": "36580f7e46b42db41c8de827085eaf01f22309ba",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
259993706 | pes2o/s2orc | v3-fos-license | Homoharringtonine-Based Induction Therapy Reduces the Recurrence Rate of Pediatric Acute Myeloid Leukemia After Allogeneic Hematopoietic Stem Cell Transplantation
Allogeneic hematopoietic stem cell transplantation (HSCT) is an effective treatment for acute myeloid leukemia (AML). Pediatric patients with AML who relapse after HSCT have an extremely poor prognosis. We performed a retrospective study of pediatric patients diagnosed with AML from August 2015 to October 2019 who were treated with HSCT. Kaplan–Meier analyses were used to evaluate overall survival (OS), event-free survival (EFS), and cumulative recurrence rate (CRR). Cox regression analysis was used to determine the association between the baseline characteristics and relapse. A total of 37 pediatric patients met the inclusion criteria. Twenty-eight (75.7%) patients survived, and 9 (24.3%) patients died. The OS rates of AML patients treated with HSCT were 89.2% ± 5.1%, 75.7% ± 7.1%, and 75.7% ± 7.1% at 1, 3, and 5 years, respectively, and the CRRs were 11.4% ± 5.4%, 24.7% ± 7.7%, and 33.1% ± 10.4% at 1, 3, and 5 years after HSCT, respectively; four of nine children who relapsed after transplantation died. Induction with etoposide rather than homoharringtonine and fungal infections could be high-risk factors for recurrence after transplantation. The association between homoharringtonine-based induction therapy and a low recurrence rate persisted after adjusting for age, sex, risk stratification, fusion genes, and fungal infections. This study clarifies the clinical features and poor prognosis of post-transplant relapse in pediatric AML and indicates the urgent need for effective therapy for patients who relapse after HSCT.
Introduction
Leukemia is the most common type of cancer in children (aged 0-14 years). The incidence of leukemia is 42.33 per million in China 1 . Although AML accounts for approximately only one in four of all acute leukemias in children, it is the second most common cancer type in children aged 5 years and older and in adolescents (aged 15-19 years) 1 . AML is characterized by greater than 20% myeloid blasts and is the most aggressive cancer with a variable prognosis depending upon the molecular subtype. In recent years, treatment outcomes for AML have improved due to precision risk stratification and optimized chemotherapy regimens, with overall survival (OS) reaching 70% 2,3 . The current standard regimen for the treatment of AML is induction chemotherapy with anthracyclines and cytarabine, followed by a choice of chemotherapy consolidation for patients with low to moderate risk or HSCT for patients with high risk, which depends on the patient's risk stratification 4 . Approximately 80% to 90% of patients with AML achieve complete remission (CR) after induction therapy; however, AML has a high recurrence rate of approximately 40% 5 . After aggressive treatment with strong chemotherapy and HSCT, the prognosis for patients with recurrent AML remains very poor 6 . HSCT is undoubtedly an important treatment for AML, especially for patients with high risk and refractory/relapsed disease 7 . However, the high incidence of transplantation complications, the prognostic impact, the risk stratification of transplantation patients, and the timing of transplantation interventions in the overall management of AML remain controversial in different research centers. This study retrospectively analyzed the clinical outcomes of 37 pediatric patients with AML treated with HSCT after induction of remission with the Chinese Children's Leukemia Group (CCLG)-AML2015 regimen at the Hematoma Center of Beijing Children's Hospital, Capital Medical University, and analyzed possible factors affecting prognosis.
Patients
The study included 37 patients who underwent HSCT for AML at the Hematology Center of Beijing Children's Hospital, Capital Medical University. Patients were selected if they met all of the following criteria: (1) have a clear diagnosis and typing of morphology-immunology-cytogeneticsmolecular biology (MICM) and risk stratification, (2) highrisk AML treated with the CCLG-AML2015 regimen before HSCT, (3) first allogeneic HSCT, and (4) initial diagnosis between 2015.8 and 2019.10. Patients who met any of the following criteria were not included in this study: (1) diagnosis of acute promyelocytic leukemia (APL) and (2) temporary or palliative treatment without following the standard treatment protocol. The following data were recorded from the database of the hospital: demographics, diagnosis of MICM, induction regimen, disease status before HSCT, transplant characteristics, and complications.
Treatment Protocols
AML-CCLG2015 regime. Induction therapy included two courses. Patients were randomly randomized 1:1 into one of the following two groups: the VP-16 group (DAE + IAE) and the HHT group (DAH + IAH). Consolidation therapies included MA, HA, CLASP, and HAs. However, not every consolidation regimen was used in patients enrolled prior to HSCT, and physicians suggested transplantation for patients after they achieved remission depending on their condition. The specific medications used in the regimen are detailed in Table 1. The CCLG-2015 protocol was a prospective and randomized clinical trial (registered trial number: ChiCTR-IPR-15006816). We used block randomization without stratification to minimize investigator and participant bias. The block size was set to 8, and a random sequence was generated according to a random number table. Allocation was concealed with random envelopes. This clinical trial was approved by the Institutional Ethics Committee of Beijing Children's Hospital.
HSCT
(1) Indications for HSCT. If there is an appropriate donor, HSCT should be considered for patients with high-risk AML after course III of consolidation treatment.
(2) Donor and stem cell source. Donor sources include unrelated donors and allogeneic and haploidentical donors. Stem cell sources include bone marrow and peripheral blood stem cells.
Prevention of graft versus host disease. In this study, mycophenolate mofetil (MMF) was used in combination with cyclosporine (CsA) and short-course methotrexate (MTX) for the prevention of graft versus host disease (GVHD). CsA was given at 2.5 mg/kg.d intravenously from D7, and after stabilization, it was changed to 4-6 mg/ kg.d orally. The dose of CsA was adjusted according to the serum drug concentration to control the serum concentration at approximately 200 ng/mL, and the dose was gradually reduced from D60. For related haploidentical grafts, MMF was given orally from D1, and MTX was given 15 mg/m 2 intravenously (D1, 3,6,9). Some patients transplanted with unrelated grafts were given antithymocyte globulin (ATG) to prevent acute GVHD (aGVHD), and ATG was given 2.5 mg/m 2 d, D-5 to -2. The dose and duration of administration were adjusted according to the presence or absence of GVHD.
Implantation standard assay. The chimerism rate was detected by polymerase chain reaction (PCR) for short tandem repeats (STR). Sex chromosome transformation and STR transferred to donor type or blood type transformation were used as evidence of whether the donor graft had implanted. The date of granulocyte implantation was the first day when neutrophils were ≥0.5 × 10 9 /L for three consecutive days. The date of platelet implantation was the first day when platelets were ≥20 × 10 9 /L for seven consecutive days and no platelet transfusion was given.
Donor lymphocyte infusion.
Measurable residual disease (MRD) was detected by flow cytometry, 1 month after transplantation. MRD was tested every 3 months thereafter if MRD turned negative and monthly if MRD remained positive. Donor lymphocyte infusion (DLI) was given monthly to AML patients with MRD positivity or MRD elevation greater than 1 log level 6 months after HSCT or with recurrence of extramedullary lesions, with CD3+ cell count starting at 1 × 10 6 /kg.
Definition
High risk at the time of initial diagnosis was defined as having one adverse genetic and molecular abnormality, myeloid sarcoma or transformed AML. CR was defined as the presence of fewer than 5% blasts in the bone marrow, no leukemic blasts in the peripheral blood or extramedullary sites, and recovery of blood counts. Relapse was defined as loss of CR (morphological relapse) or recurrence of leukemic blasts in the extramedullary sites (extramedullary relapse), and MRD turned to >10 -2 with CR also belonging to relapse (molecular biology relapse).
Statistical Methods
The study population was analyzed by disease outcomes (relapse or nonrelapse). Patients' demographic and clinical characteristics are described as the median (interquartile range [IQR]: Q1-Q3) or as percentages, as appropriate. Differences between groups were evaluated using the Mann-Whitney U test for continuous variables and the chi-square test for categorical variables. Kaplan-Meier methods were used to evaluate the incidence rate of outcomes. We selected the variables with P < 0.20 in the analysis of the single factors for Kaplan-Meier survival curve estimation. Variables with P < 0.1 and well-established predictors were selected as confounding variables in the multivariable analyses. Cox regression analysis was used to determine the association between the baseline characteristics and relapse. In this model, we adjusted for age and sex. Mononuclear cell (MNC) counts were transformed into a classification variable based on the tertile and incorporated into survival analysis. The primary objective of this study was to evaluate cumulative recurrence rate (CRR), which was defined as the interval from the date of HSCT to the first relapse post-transplant or the end of the study (death or last follow-up). OS was defined as the interval from the date of AML diagnosis to death or the last follow-up. Event-free survival (EFS) was defined as the interval from the date of HSCT to the first event or last follow-up, and death and relapse were treated as events. The probabilities of CRR, OS, and EFS were estimated by the Kaplan-Meier method. Analyses were performed with SPSS 26.0 (IBM, Chicago, IL, USA) and R version 3.4.0 (R Core Team, Vienna, Austria). All the tests were two-sided, and a P value of <0.05 was considered statistically significant.
Baseline Characteristics of Patients
Among the 37 enrolled patients, the median age was 6.00 (IQR: 3.00-10.50) years, and 26% were male. Baseline characteristics among the nonrelapse and relapse groups are presented in Table 2. In general, the baseline characteristics were well balanced between the two groups; however, patients with relapse had not been induced with the DAH regimen (P = 0.005), had higher MNC counts (P = 0.040), had paternal haploSCT (P = 0.143), had cytomegalovirus (CMV) infection (P = 0.143), had fungal infection (P = 0.025), and had a shorter follow-up time after HSCT (P = 0.009).
GVHD
There were 31 patients (83.8%) who had acute GVHD, 22 of whom had I-II GVHD. Twenty-five patients (67.6%) had chronic GVHD, one patient had interstitial lung disease resulting in decreased lung function, and one patient died of chronic intestinal GVHD.
Discussion
Allogeneic HSCT in AML patients in remission is the strongest anti-leukemic therapy available because of the cytoreductive conditioning and the immunologic anti-leukemic graft-versus-leukemia effect. Allogeneic HSCT is more suitable for the subtype of AML where conventional chemotherapy is unlikely to achieve long-lasting remission, especially in refractory/recurrent AML 8 . Previous studies 9 have shown that allogeneic HSCT treatment reduces the relapse rate by 10% in children with newly diagnosed AML, but post-transplant relapse is the leading cause of post-transplant mortality, and once relapsed, treatment options are limited with extremely poor efficacy.
In this study, we focused on analyzing and summarizing the recurrence and death after transplantation in pediatric AML patients and identifying adverse factors of post-transplantation recurrence by analyzing the primary disease characteristics, transplantation characteristics, and post-transplantation complications. Our study showed that the 5-year OS rates of AML patients treated with HSCT were 75.7% ± 7.1% at 5 years, and the 5-year CRR rates were 33.1% ± 10.4% after HSCT, while a study 10 of transplanted AML patients in our hospital during 2005 to 2015 showed that the 5-year OS was 61. 1% ± 8.7%, and the 5-year CRR was 16. 0% ± 6.0%. Data from different time periods showed a significant improvement in the OS rate and an increase in the CRR in transplant patients. The improvement in survival may be due to improvements in HSCT conditioning regimes, diversity in donor selection, infection control and means of pathogen surveillance, advances in comprehensive supportive care, and maturation of transplant teams. The increased recurrence rate and higher survival rates seem to be contradictory phenomena, which we consider to be related to advances in MRD detection technology, allowing physicians to detect recurrence earlier and give patients pre-emptive treatment, such as DLI.
As treatment-related toxicity and mortality continue to improve, relapse has become the most important cause of treatment failure, and new approaches are needed to make the next large leap forward in transplantation for AML. This study showed that post-transplant leukemia relapse was the leading cause of death, with 4 of 9 children who relapsed after transplantation experiencing death. It is worth noting that up to 44.4% of deaths were attributed to leukemia, which highlights how difficult it is to manage AML once patients have developed post-transplant relapse. DLI can enhance the graft-versus-leukemia effect and is an important method to reduce relapse after HSCT 11 . In this study, 14 patients (37.8%) were treated with DLI for positive MRD conversion or logarithmic grade elevation, and only 5 of them progressed to relapse. A retrospective study 12 showed that 46 patients with high-risk AML given DLI treatment 120 days after HSCT had a higher 7-year OS than matched controls (34 patients) (67% vs. 31%, P < 0. 001). The success of therapeutic DLI is strongly correlated with the leukemia load at the time of relapse and the remission status at the time of DLI application 13 . Therefore, it is important to monitor for MRD regularly after transplantation for early detection of relapse.
Regrettably, there is a lack of clear guidelines on the indications, timing, and dose of prophylactic DLI. AML therapy is approaching an era where the promise of targeting small molecule inhibitors and biologics agents has been realized since 2017 14 . Therefore, new directions have emerged for post-transplant maintenance therapy to prevent recurrence. Clinical trials using post-transplant FLT3 inhibitors, such as sorafenib and midostaurin, have shown feasibility, safety, and encouraging post-transplant outcomes 15 Venetoclax has been shown to be effective in combination with hypomethylating agents or low-dose cytarabine in children with AML 16 and should be taken seriously as maintenance therapy after HSCT. In addition, one study 17 showed that donor memory-like natural killer (NK) cells robustly expand and persist with potent anti-leukemic activity in the absence of exogenous cytokines and presented a novel immunotherapy platform for AML that relapsed after allogeneic HSCT in combination with DLI 17 . The role of other cellbased therapies, including chimeric antigen receptor T-cell immunotherapy (CAR-T), in AML is currently being investigated 18 . In this study, univariate analysis showed that fungal infections after transplantation might be a high-risk factor for recurrence after transplantation. Although multifactor analysis was not statistically significant, it was still evident on the survival curve that patients without fungal infections had a lower recurrence rate. Interestingly, the patients included in this study were treated with two regimens of DHA and DAE for induction therapy, and the Kaplan-Meier curve showed that patients treated with the induction regimen of DAE experienced a significantly higher risk of relapse after HSCT than patients induced with the DAH regimen. The association between homoharringtonine-based induction therapy and low recurrence rate persisted after adjusting for age, sex, risk stratification, fusion genes and fungal infections, providing more favorable evidence for considering homoharringtoninebased induction therapy as a favorable factor in reducing post-transplant recurrence. Although homoharringtonine is a commonly used chemotherapeutic agent that has been used as standard therapy for AML in China 19 , the relationship between homoharringtonine and relapse after transplatation and the reason homoharringtonine reducing relapse have not been reported in previous studies. But it is a great interest, and a larger/prospective study should been considered. Previous studies 10,20 have shown that myeloid sarcoma and the number of remissions are important poor prognostic factors. Myeloid sarcoma leads to lower survival rates and shorter survival times. Those with a high number of remissions have a worse prognosis than patients in first remission. A retrospective study 20 showed that the 5-year OS and 5-year disease-free survival (DFS) of 51 patients treated with HSCT because of myeloid sarcomas were 47% and 36%, respectively. All patients were transplanted after their first remission in this study, and no adverse association between myeloid sarcoma and relapse was found, which was considered to be related to the small sample size. However, of the five children who received a second HSCT treatment, one patient who had a recurrence of myeloid sarcoma after the first HSCT had a second recurrence of myeloid sarcoma after the second transplantation and eventually died, while the remaining four children with nonmyeloid sarcoma recurrence had a good prognosis, suggesting to some extent there is a poor prognosis for myeloid sarcoma recurrence.
GVHD is an important transplant-related comorbidity. In this study, due to the short observational follow-up period, the impact of chronic GVHD later on survival is not clear. Attention should be given to chronic GVHD, and its impact on the quality of survival should not be ignored.
This study was influenced by the small sample size due to the prevalence of AML in children, and future multicenter studies should include larger sample sizes and longer followup to reduce bias. HSCT is an effective treatment for AML, especially refractory/relapsed AML. Post-transplant relapse is a major cause of death, and secondary transplantation after relapse is an effective means of salvage for such patients. It is believed that with the in-depth understanding of the pathogenesis of AML, the improvement of chemotherapy regimens and supportive therapy, and the advancement of transplantation technology, we can further reduce transplant infection-related mortality, explore post-transplant immunotherapy to reduce AML recurrence rates, and further improve overall AML survival.
(ethical approval number: 2015-43). Informed consent was obtained from all patients or legal guardians in accordance with the principles of the Declaration of Helsinki.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. | 2023-07-21T06:17:50.686Z | 2023-01-01T00:00:00.000 | {
"year": 2023,
"sha1": "d723fe3165c138c6615a146420a646757e72f18d",
"oa_license": "CCBYNC",
"oa_url": null,
"oa_status": null,
"pdf_src": "PubMedCentral",
"pdf_hash": "723761c677361736bf15c8b69a4c7f722975463e",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
248563540 | pes2o/s2orc | v3-fos-license | Genome-Wide Characterization of the Methyl CpG Binding Domain-Containing Proteins in Watermelon and Functional Analysis of Their Roles in Disease Resistance Through Ectopic Overexpression in Arabidopsis thaliana
Methyl-CPG-Binding Domain (MBD) proteins play important roles in plant growth, development, and stress responses. The present study characterized the MBD families in watermelon and other cucurbit plants regarding the gene numbers and structures, phylogenetic and syntenic relationships, evolution events, and conserved domain organization of the MBD proteins. The watermelon ClMBD proteins were found to be localized in nucleus, and ClMBD2 and ClMBD3 interacted with ClIDM2 and ClIDM3. ClMBD2 bound to DNA harboring methylated CG sites but not to DNA with methylated CHG and CHH sites in vitro. The ClMBD genes exhibited distinct expression patterns in watermelon plants after SA and MeJA treatment and after infection by fungal pathogens Fusarium oxysporum f.sp. niveum and Didymella bryoniae. Overexpression of ClMBD2, ClMBD3, or ClMBD5 in Arabidopsis resulted in attenuated resistance against Botrytis cinerea, accompanied by down-regulated expression of AtPDF1.2 and increased accumulation of H2O2 upon B. cinerea infection. Overexpression of ClMBD1 and ClMBD2 led to down-regulated expression of AtPR1 and decreased resistance while overexpression of ClMBD5 resulted in up-regulated expression of AtPR1 and increased resistance against Pseudomonas syringae pv. tomato DC3000. Transcriptome analysis revealed that overexpression of ClMBD2 in Arabidopsis up-regulated the expression of a small set of genes that negatively regulate Arabidopsis immunity. These data suggest the importance of some ClMBD genes in plant immunity and provide the possibility to improve plant immunity through modification of specific ClMBD genes.
INTRODUCTION
As sessile organisms, plants have to face invasive attacks from diverse pathogenic microorganisms in the environment. To defense these pathogenic invasions, plants have evolved a complicated but fine-tuned innate immune system (Jones and Dangl, 2006;Yuan et al., 2021;Ngou et al., 2022). The first layer of the innate immunity, called PAMP-triggered immunity (PTI), is triggered by the recognition of microbial patterns via cell surfacelocalized pattern-recognition receptors, while the second layer, called effector-triggered immunity (ETI), is activated by the direct or indirect interaction between predominantly intracellularly localized nucleotide-binding leucine-rich repeat receptors and pathogen effectors. Activation of PTI and/or ETI is fine-tuned by a complicated molecular and genetic network, in which epigenetic regulation including DNA methylation/demethylation play critical roles (Huang and Jin, 2022).
Methylation of DNA, a conserved epigenetic mark, is one of the main mechanisms that play critical roles in epigenetic regulation of various biological processes including plant growth, development, and response to environmental cues (He et al., 2011;Arıkan et al., 2018;Zhang et al., 2018). Among DNA methylation, cytosine methylation (5-mC) is the most common epigenetic phenomenon that regulates the fate of gene expression (Grafi et al., 2007;Zhang et al., 2018). In higher plants, 5-mC occurs in CG dinucleotide regions, and CHG/CHH (H represents A, T, or C) trinucleotide regions (Gruenbaum et al., 1981). DNA methylation is a dynamic process that are achieved by different enzymes (Moore et al., 2013;Zhang et al., 2018) and is involved in many molecular processes, including genome stability, gene regulation, transposon silencing, and chromosome interactions (Zhang et al., 2006;Cokus et al., 2008;Lang et al., 2017). Extensive studies have indicated that DNA methylation plays important roles in plant growth and development, such as vegetable growth, pattern formation, flowering time, seed development, and fruit ripening (Gehring et al., 2009;He et al., 2011;Ibarra et al., 2012;Arıkan et al., 2018;Zhang et al., 2018) as well as in abiotic stress responses (Rambani et al., 2015;Xu et al., 2015;Yong-Villalobos et al., 2015;Hewezi et al., 2017). Importantly, DNA methylation, as one of the epigenetic regulation mechanisms, plays crucial roles in plant immunity (Huang and Jin, 2022); for example, Arabidopsis mutants with DNA hypomethylation are more resistant to Pseudomonas syringae pv. tomato (Pst) DC3000 and exhibit an elevated salicylic acid (SA)-dependent response (Dowen et al., 2012;Yu et al., 2013;Cambiagno et al., 2021).
The functions of MBD genes in plant growth, development, and response to abiotic stress have been explored. Mutation in AtMBD8 or knockdown of AtMBD11 led to a delay in flowering time, while the atmbd9 mutant showed a significantly earlier flowering time (Berg et al., 2003;Peng et al., 2006;Stangeland et al., 2009). Overexpression of Salix viminalis L. SvMBD5 led to an early flowering phenotype in transgenic Arabidopsis (Cheng et al., 2020). These observations indicate that the MBD proteins play critical roles in regulation of flowering in plants. The AtMBD11 knockdown mutant also displayed a variety of phenotypic effects, e.g., aerial rosettes, serrated leaves, abnormal position of flowers, and fertility problems (Berg et al., 2003), while the atmbd9 mutants produced more shoot branches (Peng et al., 2006). Overexpression of OsMBD707 leads to larger tiller angles and reduced photoperiod sensitivity in rice (Qu et al., 2021). The atmbd4 mutant exhibited altered root architecture and up-regulated expression of many phosphate transporters and transcription factors, indicating that AtMBD4 negatively regulates the phosphate starvation response (Parida et al., 2019). Some of the wheat TaMBD genes and most of the petunia PhMBD genes were highly induced by abiotic stress and hormones (Hu et al., 2011;Shi et al., 2016Shi et al., , 2022. However, the possible involvement of the MBD genes in plant immunity remains elusive. Watermelon (Citrullus lanatus L.) is one of important horticultural crops, providing favorite fresh fruits worldwide. Fusarium wilt, caused by Fusarium oxysporum f.sp. niveum (Fon), and gummy stem blight, caused by Didymella bryoniae (Db), are two of the most devasting fungal diseases that lead to significant yield losses in watermelon industry (Michielse and Rep, 2009;Keinath, 2011). However, knowledge on the molecular mechanism of resistance in watermelon against Fon and Db is currently limited, which significantly impedes the breeding for watermelon cultivars with improved resistance against these two fungal diseases. The present study aimed to identify the watermelon ClMBD family by characterization and expression analyses and explore the putative mechanism of the ClMBD family in disease resistance. The transcript levels of the ClMBD genes were changed after treatment with SA and methyl jasmonate (MeJA) and infection by Fon and Db. Functional analyses revealed that ClMBD2, ClMBD3, and ClMBD5 negatively regulate resistance against Botrytis cinerea and that ClMBD1 and ClMBD2 negatively while ClMBD5 positively regulate resistance against Pst DC3000 in Arabidopsis.
Plant Materials and Growth Conditions
Watermelon (Citrullus lanatus) cv. Zaojia was used for all experiments. Nicotiana benthamiana plants expressing a known nucleus-localized marker protein RFP-H2B (Chakrabarty et al., 2007) were used for subcellular localization and bimolecular fluorescence complementation (BiFC) assays. Plants were grown in a soil mix (clay: soil = 3:1) in a growth room under fluorescent light (200 µE m 2 s −1 ) at 22-24 • C with 70% relative humidity (RH) and a 14 h light/10 h dark cycle. Arabidopsis seeds were surface sterilized in 75% ethanol for 5 min and 4% sodium hypochlorite for 10 min, rinsed with sterile water for three times, sowed on 1/2 MS plates and vernalized for 2 days at 4 • C. Arabidopsis seedlings were grown on 1/2 MS plates at 22 • C with 75% RH with a 16 h light/8 h dark cycle for 7 days and then transplanted to a soil mix (clay: soil = 1:1) in a growth room at 22 • C with 75% humidity under a 16 h light/8 h dark cycle for normal growth or under a 8 h light/16 h dark cycle for disease assays.
Hormone Treatment and Pathogen Inoculation for Gene Expression Analysis in Watermelon
For analysis of tissue-specific expression, leaf, stem and root samples of 4-week-old watermelon plants were collected and stored at −80 • C till use. For SA and MeJA treatment, 4-weekold watermelon plants were treated by foliar spraying with 1 mM SA, 100 µM MeJA or an equal volume of solution containing only 0.1% ethanol and 0.02% Tween-20 as controls, and leaf samples were collected at different time points after treatment.
For analysis of gene expression in response to Fon infection, pathogen inoculation was performed according to a previously reported method (Song et al., 2015). Briefly, mycelial plugs from 6-day-old culture of Fon race 1 strain ZJ1 were transferred into 200 mL mung bean liquid broth (mung bean 20 g/L, boiled for 20 min, pH7.0) and incubated with shaking (250 rpm) at 26 • C for 2 days. Spores were collected and spore suspension was adjusted to 1 × 10 7 spores/mL for inoculation. Three-week-old watermelon plants were uprooted, washed in tap water, the main roots were cut up of one-third, and then dipped for 15 min in Fon spore suspension or in mung bean liquid broth as mockinoculated controls. The inoculated plants were replanted in soil and allowed to grow in the same growth room as described above. Root samples were collected, frozen in liquid nitrogen and stored at −80 • C until use.
Db strain DBTL4 was grown at 26 • C on PNA (potato 200 g/L, NH 4 H 2 PO 4 2 g/L, agar 15 g/L, pH7.0) for 6-7 days in dark and then treated with a 12 h UV light/12 h dark cycle for 5 days to induce spore production. After induction, the mycelial plugs were picked into distilled water, spores were collected and the spore suspension was adjusted to 2 × 10 6 spores/mL. Fiveweek-old watermelon plants were foliar sprayed with Db spore suspension containing 0.05% Tween-20 or with an equal volume of 0.05% Tween-20 solution as mock controls. The inoculated plants were placed in a 22 • C chamber with 100% RH for 48 h. Leaf samples were collected, frozen in liquid nitrogen, and stored at −80 • C until use.
Identification of Watermelon ClMBD Genes and Proteins
Arabidopsis AtMBD protein sequences were obtained from TAIR 1 and were used as queries to search for putative MBD genes and proteins in watermelon, melon, cucumber, pumpkin, and zucchini genomes at Cucurbit Genomics Databases. 2 The obtained nucleotide and protein sequences were examined by domain analysis programs PFAM 3 (PF01429) and SMART 4 with the default cutoff parameters. The isoelectric points and molecular weights were predicted on the ExPASy Proteomics Server. 5 Sequence alignment was carried out by the ClustalX program. Phylogenetic trees were constructed using the neighbor-joining method of the MEGA7 program with the p-distance and complete deletion option parameters.
Synteny Analysis of the ClMBD Genes
The reliability of the obtained trees was tested using a bootstrapping method with 1,000 replicates. The MCScanX algorithm with default parameters (Wang et al., 2012) was used to scan orthologous regions containing the watermelon ClMBD genes. The corresponding plot was created by Dual Synteny Plot for MCscanX in TBtools software (Chen et al., 2020). The chromosomal localization of ClMBDs in the C. lanatus genome was obtained by TBtools software (Chen et al., 2020) according to the annotation data of the C. lanatus genome. The genomic and annotation data of melon, cucumber, zucchini, and pumpkin were downloaded from the Cucurbit Genomics Database (see text footnote 2), and those of Arabidopsis were downloaded from TAIR (see text footnote 1). The synteny relationship of the orthologous MBD genes obtained between watermelon and other selected species was visualized by the Advance Circos package of TBtools (Chen et al., 2020). DnaSP software was used to calculate the non-synonymous (Ka)/synonymous (Ks) values of the duplicated ClMBD gene pairs (Librado and Rozas, 2009).
Cloning of the ClMBD Genes
Total RNA was extracted using RNA Isolater reagent (Vazyme, Nanjing, China) according to the manufacturer's instructions. RNA was treated with RNase-free DNase and then reversetranscribed into cDNA using the HiScript QRT SuperMix kit (Vazyme, Nanjing, China). The obtained cDNAs were used for cloning, semi-RT-PCR and qRT-PCR. The coding sequences (CDs) of ClMBDs were amplified using gene-specific primers (Supplementary Table 1) and cloned into pCAMBIA1300s vector, yielding pCAMBIA1300s-ClMBDs-GFP. After confirmation by sequencing, these pCAMBIA1300s-ClMBDs-GFP plasmids were used as templates to amplify the target genes for further experiments.
Yeast Two-Hybrid Assays
Putative interactions between ClMBDs and ClIDM2 or ClIDM3 were examined using the yeast two-hybrid (Y2H) System according to the manufacturer's instructions (Clontech, Mountain View, CA, United States). The CDs of ClMBDs were amplified using gene-specific primers (Supplementary Table 1) from pCAMBIA1300s-ClMBDs-GFP and cloned into pGBKT7 vector, yielding pGBKT7-ClMBDs. ClIDM2 and ClIDM3 were obtained by homologous searching using Arabidopsis AtIDM2 and AtIDM3 as queries and the CDs of ClIDM2 and ClIDM3 were amplified with gene-specific primers (Supplementary Table 1) and cloned into pGADT7 vector, generating pGADT7-ClIDM2 and pGADT7-ClIDM3. The resultant pGBKT7-ClMBD plasmids were transformed with or without pGADT7-ClIDM2 or pGADT7-ClIDM3 into yeast strain Y2HGold by the LiAc/SS carrier DNA/PEG method and confirmed by colony PCR. The transformed yeasts were cultivated on DDO (SD/-Leu/-Trp) medium (Clontech, Mountain View, CA, United States) at 30 • C for 3 days, followed by screening on QDO medium containing 40 µg/mL X-α-Gal (Clontech, Mountain View, CA, United States) and 125 ng/mL Aureobasidin A (Clontech, Mountain View, CA, United States). Interactions between ClMBDs and ClIDM2/3 were evaluated according to the growth performance of the transformed yeast cells on QDO and the production of blue pigments after the addition of X-α-Gal. Co-transformation of pGBKT7-53 or pGBKT7-Lam and pGADT7-T were used as positive and negative controls, respectively.
Bimolecular Fluorescence Complementation Assays
The CDs of ClMBD2 and ClMBD3 were amplified using genespecific primers (Supplementary Table 1) and inserted into p2YN vector, yielding p2YN-ClMBD2 and p2YN-ClMBD3. Similarly, the CDs of ClIDM2 and ClIDM3 were inserted into p2YC vector, yielding p2YC-ClIDM2 and p2YC-ClIDM3. Agrobacteria harboring different indicated pairs of plasmids were infiltrated into leaves of N. benthamiana plants expressing a red nuclear marker protein RFP-H2B (Chakrabarty et al., 2007). At 48 h after agroinfiltration, YFP and RFP signals were detected and photographed under a Zeiss LSM780 confocal laser scanning microscope (Zeiss, Oberkochen, Germany).
Electrophoretic Mobility Shift Assays
The CDs of the ClMBD genes were amplified using genespecific primers (Supplementary Table 1) and inserted into pGEX-4T-3 vector, generating pGEX-4T-3-GST-ClMBDs, followed by transforming into Escherichia coli strain BL21 (DE3), a widely used non-T7 expression strain that is suitable for transformation and protein expression (New England BioLabs, Beverly, MA, United States). To induce the expression of GST-ClMBD proteins, isopropyl-D-thiogalactoside was added to the bacterial cultures to a final concentration of 1 mM and incubated at 18 • C for 20 h. The recombinant GST-ClMBD fusion proteins were purified using glutathione resin columns (Genscript, Shanghai, China) according to the manufacturer's protocol. The following double-stranded DNA probes were synthesized and used in EMSA assay: 5mCG (GCTCGTAGCTAACGAGCTCGACTCGTTGACATAGGCCAT GGCGTAGACTC) (methylated nucleotides underlined) and its complementary strand with m5C at symmetrical positions, 5mCHG (GCTCTGAGCTAACAGGCTCAGC TCTGTGACATAGGCCATGGCTGAGACTC) (methylated nucleotides underlined) and its complementary strand with m5C at symmetrical positions, 5mCHH (GCTCTTAGCTAACA AGCTCAACTCTATGACATAGGCCATGGCTTAGACTC) (methylated nucleotides underlined) and its complementary strand (GAGTCTAAGCCATGGCCTA TGTCATAGAGGTGAGCTTGTTAGCTAAGAGC) (Ito et al., 2003). Equal volumes of single-stranded DNAs were mixed in annealing buffer (10 mM Tris-HCl, 1 mM EDTA, 100 mM NaCl, pH7.5) and incubated at 85 • C for 5 min to form double-stranded DNAs. EMSA was performed as previously described (Yuan et al., 2019) using LightShift Chemiluminescent EMSA Kit (Thermo Fisher Scientific, Waltham, MA, United States). In brief, binding reactions (10 µL) contained 1 µL 10 × binding buffer, 2 µg GST-ClMBD protein or GST protein (as a negative control) and 1 µL biotin-labeled 5mCG, 5mCHG, or 5mCHH probe. In the competitive reactions, unlabeled 5mCG probe was added in excess of 500 times. The binding reactions were incubated at 28 • C for 20 min and separated on 6% native PAGE gels. After electrophoresis, the gels were transferred onto Amersham Hybond-N + nylon membrane (GE Healthcare, Buckinghamshire, United Kingdom), and signals from the biotin-labeled probes were detected using a Chemiluminescent Biotin-labeled Nucleic Acid Detection Kit (Beyotime Biotechnology, Haimen, China) according to the manufacturer's recommendations.
Generation and Characterization of ClMBDs-Overexpressing Transgenic Arabidopsis Lines
Arabidopsis transformation was performed using the floral dip method as previously described (Clough and Bent, 1998). In brief, flowers of 5-week-old Arabidopsis plants were dipped in a suspension of agrobacteria carrying pCAMBIA1300s-ClMBD-GFP plasmids for 1 min. The infected plants were placed in dark for 12 h under 100% RH, returned to the growth room with normal conditions and allowed to grow until the silique maturation. T0 seeds were surface sterilized and then plated on 1/2 MS plates containing 50 µg/mL hygromycin. After treatment at 4 • C for 2 days, the plates were transferred to 22 • C under a 16 h light/8 h dark cycle, seedlings showing hygromycin resistance, regarded as positive transgenic plants, were transferred in the mixed nutrient soil and allowed for growth for 5-6 weeks to collect seeds. Putative single-copy transgenic lines and homozygous lines were obtained by screening for a 3:1 segregation ratio of hygromycin-resistant (Hgr R ) character and 100% Hgr R phenotype in T2 and T3 generations on 1/2 MS medium supplemented with 50 µg/mL hygromycin, respectively. The transcript levels of the ClMBD genes in the transgenic Arabidopsis lines were analyzed by semi-PCR and qRT-PCR. Two homozygous transgenic Arabidopsis lines with single-copy for each of the ClMBD genes (T3 generation) and similar expression levels of the transgenes were chosen for further experiments.
Disease Assays on Transgenic Arabidopsis Plants and Measurement of in planta Pathogen Growth
Disease assays with B. cinerea were performed as previously described (Wang et al., 2009). Briefly, spores were collected from 8∼10-day-old culture of B. cinerea strain BO5.10 grown on 2 × V 8 plates and resuspended in 4% maltose and 1% peptone buffer to a final concentration of 2 × 10 5 spores/mL. Four-weekold Arabidopsis plants were inoculated by foliar spraying with the spore suspension containing 0.05% Tween-20 or with an equal volume of 0.05% Tween-20 solution as mock controls. The inoculated plants were placed in a 22 • C chamber with 100% RH for 48 h, and disease development was continuously observed. Measurement of in planta fungal growth was performed by analyzing the transcript level of B. cinerea BcActin gene and comparing with the transcript level of an Arabidopsis Actin gene as an internal control according to a previously reported protocol (Wang et al., 2009).
Disease assays with Pst DC3000 were carried out as previously described . Pst DC3000 was grown on King's B (KB) broth and bacteria were collected and resuspended in 10 mM MgCl 2 solution to OD 600 = 0.002. The bacterial inoculation was performed by hand infiltration using 1-mL syringes without needle into rosette leaves of 4-weekold Arabidopsis plants and the inoculated plants were kept in sealed containers 22 • C for 72 h. For quantification of in planta bacterial growth, leaf discs from inoculated leaves were collected and homogenized in 10 mM MgCl 2 . After a series of gradient dilutions, the homogenate was plated on KB plates supplemented with 25 µg/mL rifampicin and bacterial colonies were counted at 3 days after incubation at 28 • C.
In situ Detection of H 2 O 2 Accumulation
Detection of H 2 O 2 was performed using the DAB staining method (Thordal-Christensen et al., 1997). Leaf samples were collected from Arabidopsis plants with or without infection of B. cinerea at 24 h post inoculation (hpi) and dipped into DAB solution (1 mg/mL) in 10 mM Na 2 HPO 4 (pH7.0). After incubation for 5 h in dark with shaking (80 rpm) at room temperature, the DAB-treated leaves were transferred into acetic acid/glycerol/ethanol (1:1:1, vol/vol/vol) and boiled for 5 min, followed by several washes with the same solution. The DABstained leaves were photographed using a digital camera.
RNA-Seq Analyses
Leaf samples were collected from 4-week-old Col-0 and ClMBD2-OE2 Arabidopsis plants, frozen in liquid nitrogen and stored at −80 • C. RNA-seq was performed by BioMarker Technologies (Beijing, China) on Hiseq 2500 platform (Illumina). Raw data were filtered to get clean data, sequence comparison with the GCF_000001735.4_TAIR10.1. FPKM (Fragments Per Kilobase of transcript Per Million Fragments Mapped) was used to analyze the level of gene expression (Florea et al., 2013). The expression changes of differentially expressed genes (DEGs) ≥ 1.5-fold and P-value < 0.05. Gene Ontology (GO) enrichment analysis of DEGs was implemented by the GOseq R packages based Wallenius non-central hyper-geometric distribution (Young et al., 2010). KOBAS (Mao et al., 2005) software were used to test the statistical enrichment of differential expression genes in KEGG pathways (Kanehisa et al., 2008).
Semiquantitative RT-PCR and qRT-PCR Analyses
Extraction and treatment of total RNA were performed as mentioned above. Semiquantitative RT-PCR reactions contained 0.5 µL Phanta Max Super-Fidelity DNA Polymerase (Vazyme, Nanjing, China), 0.5 µL dNTP Mix, 12.5 µL 2 × Phanta Max Buffer, 0.1 µg cDNA, 7.5 pmol of each of gene-specific primers (Supplementary Table 1), and 8.5 µL ddH 2 O in a final volume of 25 µL. Arabidopsis AtActin was used as the control. Each qPCR reaction contained 10 µL 2 × AceQ qPCR SYBR Green Master Mix (Vazyme, Nanjing, China), 0.1 mg cDNA and 7.5 pmol of each of gene-specific primers (Supplementary Table 1) in a final volume of 20 mL, and had two independent biological replicates. The qPCR was performed in a CFX96 real-time PCR detection system (Bio-Rad, Hercules, CA, United States). Watermelon ClGAPDH or Arabidopsis AtActin were used as internal controls to normalize the data. Relative gene expression level was calculated using 2 − CT method as described.
Statistical Analysis
All experiments were independently repeated three times and the obtained data were subjected to statistical analysis according to the Student's t-test. The probability values of p < 0.05 were considered as significant difference between the treatments and corresponding controls.
Identification and Characterization of the Watermelon ClMBD Family
To identify putative ClMBD genes in watermelon, BLASTp searches were performed against the watermelon genome database using the Arabidopsis AtMBDs as queries and 10 non-redundant sequences that are putative ClMBD genes were identified ( Table 1). For convenience, unique identities to each of the identified ClMBD genes were assigned as ClMBD1-10 according to their chromosomal locations ( Table 1). The CDs of ClMBD1-10 were confirmed by cloning of the full-length cDNAs using primers designed according to their predicted cDNA sequences. The sizes of the ClMBD open reading frames (ORF) ranged from 798 bp (ClMBD2) to 6,636 bp (ClMBD8) and the sizes of the encoded proteins varied from 265 amino acids (ClMBD2) to 2,211 amino acids (ClMBD8), with molecular weight of 23.34∼244.99 kDa and pI of 4.80∼9.49 (Table 1). Similarly, the MBD families in other cucurbit plants were also characterized and 9, 10, 15, and 16 MBD genes in melon, cucumber, zucchini, and pumpkin, respectively, were identified (Supplementary Table 2).
Structure of ClMBD Genes and Organization of Conserved Domains in ClMBD Proteins
The 10 ClMBD genes are unevenly distributed on eight chromosomes in the watermelon genome and chromosomes 2, 4, and 8 do not host any ClMBD gene (Table 1 and Supplementary Figure 1). Chromosomes 9 and 11 harbor two ClMBD genes while each of the other chromosomes 1, 3, 5, 6, 7, and 10 carry one ClMBD gene (Table 1 and Supplementary Figure 1). Phylogenetic tree analysis revealed that the watermelon ClMBD proteins were divided into two clades: Clade I contained six ClMBD proteins (ClMBD1, 3, 4, 5, 6, and 9) while Clade II contained four ClMBD proteins (ClMBD2, 7, 8, and 10) ( Figure 1A). Phylogenetic tree analysis of the MBD proteins from cucurbit plants showed that the MBD proteins from watermelon, melon, cucumber, pumpkin, and zucchini have a high level of similarity in the amino acid sequences (Supplementary Figure 2).
The structure of the ClMBD genes in the CDs is highly divergent in terms of the exon and intron numbers, with two (ClMBD9) to 11 (ClMBD8) exons and one (ClMBD9) to 10 (ClMBD8) introns ( Figure 1B). A similar diverse exonintron structure was also observed in the MBD genes in other cucurbit plants such as melon, cucumber, zucchini, and pumpkin (Supplementary Figure 3). The divergent gene structure may imply that the MBD genes in cucurbit plants possess divergent functions during their evolution.
The ClMBD proteins contain a characteristic conserved MBD domain (Figure 1C), ranging from 60 to 125 aa in size. The MBD domains in ClMBDs show 13∼43% of sequence identity and harbor some conserved amino acids, e.g., 15W/F, 35Y/F, 38P, and 54L/V (Supplementary Figure 4). Notably, ClMBD1 and ClMBD3 have two MBD domains while the other ClMBDs contain a single MBD domain ( Figure 1C). Except for ClMBD7 whose MBD domain locates at the C-terminal, the MBD domains are generally located as the N-terminals in the ClMBD proteins ( Figure 1C). ClMBD1, 2, 3, 4, 6, and 9 harbor the sole conserved MBD domains, while ClMBD5, 7, 8, and 10 contain other conserved domains in addition to the MBD domain ( Figure 1C). For example, ClMBD5 has a Glycoside Hydrolase Family 17 domain; ClMBD10 has a zf-CW domain; ClMBD7 has a SPARK domain, a PKINase domain, a zf-CW domain; ClMBD8 has a Bromo Domain, a WHIM1 domains and 2 PHD domains ( Figure 1C). Similar features in the presence of conserved amino acids in MBD domains (Supplementary Figure 4) and of the additional conserved domains in MBD proteins from melon, cucumber, zucchini, and pumpkin (Supplementary Figure 5) were also detected. The divergence of conserved domains between watermelon and other cucurbit plants may result in the diversity of functions and complexity of the biochemical and molecular mechanisms of the MBD proteins in plants.
Four colinear gene pairs of watermelon ClMBDs distributed on each of Chr06, Chro9, and Chr11, while one colinear gene pair existed on each of Chr03, Chr07, and Chr10 (Figure 2 and Supplementary Table 4). These genes may originate from the same ancestors. Overall, there are more collinear gene pairs between watermelon and other cucurbit plants, indicating that these species were associated with the phylogenetic relationship and that the ClMBD gene family may be considered as marker genes in plant evolutionary.
ClMBDs Are Nucleus-Localized Proteins
To explore the subcellular localization of the ClMBD proteins, agrobacteria carrying ClMBD1-10-GFP or GFP was infiltrated into leaves of N. benthamiana plants expressing a red nuclear marker RFP-H2B protein (Chakrabarty et al., 2007). The ClMBD1-10-GFP protein was solely localized to the nucleus, which was co-localized with the known nucleus marker RFP-H2B protein ( Figure 3A). By contrast, GFP alone distributed ubiquitously throughout the cell without specific compartmental localization ( Figure 3A). These results indicate that ClMBD1-ClMBD10 are nucleus-localized proteins.
ClMBD2 Specifically Binds to Methylated CG DNA
It is well known that MBD proteins have the capability to bind methylated DNA (Ito et al., 2003;Zemach and Grafi, 2003;Grafi et al., 2007). To explore the biochemical activity of the watermelon ClMBDs, recombinant GST-tagged ClMBD1-7, 9, 10 proteins were purified (Supplementary Figure 7) and their binding activity to methylated CG DNA was examined by EMSA. Two complementary single-stranded DNA probes with 5 methylated CG sites (5mCG) were synthesized and the doublestranded 5mCG DNA was generated ( Figure 4A). In repeated EMSA, only ClMBD2 bound to labeled double-stranded 5mCG DNA, and the binding of ClMBD2 to labeled double-stranded 5mCG DNAs was specific as this binding was completely suppressed by the excessive unlabeled double-stranded 5mCG DNA in the competition binding assay (Figure 4A). The remaining ClMBDs did not show binding activity to the labeled double-stranded 5mCG DNA ( Figure 4A). However, ClMBD2 did not bind to DNA harboring mCHG or mCHH sites ( Figure 4B). These results indicate that ClMBD2 specifically binds to mCG DNA, but not to mCHG DNA or mCHH DNA.
ClMBDs Have Similar Expression Patterns in Root, Stem, and Leaf Tissues
The expression patterns of the ClMBD gene in root, stem and leaf tissues of 4-week-old watermelon plants were analyzed and the qRT-PCR results showed that the ClMBD genes have similar expression patterns: highest expression in leaves, moderate in stems, and lowest in root (Supplementary Figure 8).
ClMBDs Are Responsive to Defense Hormones Salicylic Acid and Methyl Jasmonate
To explore the possible involvement of the ClMBD genes in disease resistance, expression changes of the ClMBD genes were analyzed in watermelon plants after treatment with SA and MeJA. After foliar spraying with 1 mM SA, the expression of ClNPR1 and ClPR1, the marker genes of SA signaling pathway, was significantly up-regulated. Particularly, the expression of ClPR1 significantly up-regulated at 6 h after treatment and peaked at 12 h, showing a > 59-fold increase, as compared with that in mock control ( Figure 5A). After SA treatment, the expression of most of the ClMBD genes were up-regulated with distinct patterns: ClMBD2, 3, 6, 7, 8, 9, and 10 were up-regulated at 6 h; ClMBD1, 4, 5, 7, and 10 were up-regulated at 12 h; while ClMBD1, 6, and 8 were markedly up-regulated at 24 h, as compared with those in the mock controls ( Figure 5A). Notably, ClMBD2, 3, 9, and 10 showed similar expression patterns after SA treatment, implying that these ClMBD genes may have similar functions. These data indicate that the ClMBD genes can respond to SA and thus may be involved in disease resistance in watermelon. After foliar spraying with 100 µM MeJA, the expression of ClJAZ1 and ClPDF1.2, the marker genes of the JA signaling pathway , was significantly up-regulated and peaked at 24 h ( Figure 5B). After MeJA treatment, ClMBD2, 6, 7, 8, and 10 were highly up-regulated, while ClMBD5 and ClMBD9 were significantly down-regulated, as compared with those in the mock controls ( Figure 5B). ClMBD4 was up-regulated at 12 h but down-regulated at 24 h, as compared with those in the mock controls ( Figure 5B). Notably, ClMBD8/ClMBD10 and ClMBD1/ClMBD4 exhibited similar expression patterns in response to exogenous MeJA, indicating these two pairs of the ClMBD genes may have similar biological functions. These data indicate that the ClMBD genes differentially respond to MeJA and thus may play different roles in disease resistance in watermelon.
ClMBDs Differentially Respond to Fungal Pathogens
To explore the possible involvement of ClMBDs in watermelon disease resistance, the expression changes of the ClMBD genes in watermelon plants after infection with different fungal pathogens were analyzed. Fon is the most important soilborne vascular pathogen that causes Fusarium wilt, one of the most devastating fungal diseases in watermelon (Michielse and Rep, 2009). Fon infects watermelon plants through root system and proliferation within the xylem vessels (Michielse and Rep, 2009). Therefore, the expression changes of the ClMBD genes in root tissues of watermelon plants after Fon infection were analyzed. The qRT-PCR results showed that the expression level of ClPR1 started to increase at 3 days post inoculation (dpi), peaked at 6 days, leading to 65.5-fold higher over that in mock-inoculated plants, and then decreased ( Figure 5C). The expression of ClMBD1 and ClMBD10 in root tissues of the Fon-infected watermelon plants were up-regulated, as compared with those in the mockinoculated plants, at 3 dpi ( Figure 5C). As compared with those in the mock-inoculated plants, the expression of ClMBD2, 6, and Agrobacteria carrying pCAMBIA1300-ClMBDs-GFP or pCAMBIA1300-GFP were infiltrated into leaves of N. benthamiana plants expressing a known nucleus-localized marker protein RFP-H2B. At 48 h after agroinfiltration, GFP signals were visualized under a confocal laser scanning microscope in dark field for green fluorescence (left), red fluorescence (middle left), white field for cell morphology (middle right) and in combination (right), respectively. (B) Interactions of ClMBD2 and ClMBD3 with ClIDM2 and ClIDM3. Agrobacteria carrying indicated pairs of p2YC and p2YN plasmids were infiltrated into leaves of N. benthamiana plants expressing a red nuclear marker RFP-H2B protein, and YFP and RFP signals were observed at 48 h after infiltration. Images were taken in dark field for green fluorescence (left) and red fluorescence (middle right), white field for cell morphology (middle left) and in combination (right), respectively. Experiments in (A,B) were repeated for three times with similar results. 9 was up-regulated at 6 dpi, while the expression of the ClMBD genes, except for ClMBD3, 5, and 10, was up-regulated at 9 dpi ( Figure 5C). Overall, the expression changes of the ClMBD genes exhibited similar significant up-regulation patterns in root tissue at 3, 6, or 9 dpi; for example, the expression level of ClMBD7 in root tissue was markedly up-regulated with a > 639-fold increase FIGURE 4 | Binding activity of ClMBD2 to methylated CG DNA. (A) Binding of ClMBD proteins to the methylated CG DNA. Biotin-labeled mCG DNA (for binding assays) or biotin-labeled mCG DNA in combination with excessive unlabeled mCG DNA (for competitive assays) were incubated with GST-ClMBD or GST (as a negative control). (B) Binding activity of ClMBD2 to methylated CG DNA, methylated CHG DNA and methylated CHH DNA. Biotin-labeled mCG DNA, biotin-labeled mCHG DNA and biotin-labeled mCHH DNA (for binding assays) or biotin-labeled mCG DNA in combination with excessive unlabeled mCG DNA (for competitive assays) were incubated with GST-ClMBD or GST (as a negative control). The mCG, mCHG, and mCHH DNA sequences are shown and the methylated sites are indicated in red color. Experiments in (A,B) were repeated for three times with similar results. than that in mock-inoculated plants at the 9 dpi ( Figure 5C). Notably, the expression changes of ClMBD8, 9, and 10, and of ClMBD1, 2, 4, 5, 6, and 7 showed similar patterns in response to Fon, implying that these two groups of the ClMBD genes may be involved in resistance with similar functions. The results suggest that the ClMBD genes are responsive to Fon infection during the relative late stage of the pathogenesis and thus may play roles in the process of regulating watermelon resistance to vascular Fusarium wilt disease.
Db is another devasting fungal pathogen that infects leaf and stem tissues and causes gummy stem blight, which is a very common fungal disease on cucurbitaceous crops including watermelon (Keinath, 2011). The responsiveness of the ClMBD genes to Db infection was also analyzed in leaf tissues of watermelon plants after foliar spraying with a fungal spore suspension. After Db inoculation, the expression of ClPR1 in leaf tissues started to increase at 12 hpi, gradually increased and peaked at 48 h, leading to a 55-fold increase over that in mockinoculated plants (Figure 5D). The expression of the ClMBD genes in leaf tissues exhibited distinct patterns in response to Db infection. The expression levels of ClMBD3, 4, 7, and 9 in Dbinfected leaves were significantly up-regulated at 12, 24, or 48 hpi, as compared with those in mock-inoculated leaves ( Figure 5D). By contrast, the expression levels of ClMBD2, 6, 8, and 10 in Dbinfected leaves were highly down-regulated at 12, 24, or 48 hpi, as compared with those in mock-inoculated leaves ( Figure 5D). The expression of ClMBD1 and ClMBD5 in leaf tissues was not significantly affected by Db infection during a 48 h period of the experiments (Figure 5D). Notably, the expression changes of ClMBD3, 5, and 7 exhibited similar patterns in response to Db, indicating similar involvement for these three ClMBD genes in Db resistance. These results indicate that the expression of Three-week-old plants were inoculated by dipping the roots in spore suspension (1 × 10 7 spores/mL) of F. oxysporum f.sp. niveum or in mung bean liquid broth as mock-inoculated controls, and root samples were collected at indicated time points after inoculation. (D) Expression changes of the ClMBD genes to Didymella bryoniae. Five-week-old watermelon plants were inoculated by foliar spraying with D. bryoniae spore suspension (2 × 10 6 spores/mL) or similar volume of solution as mock controls, and leaf samples were collected at indicated time points after inoculation. qRT-PCR was performed using the watermelon ClGAPDH gene as an internal control. Relative expression was calculated using the 2 − CT method. Experiments were repeated for three times and the data presented are the means ± SE from three independent experiments. * or ** above the columns indicate significant difference at p < 0.05 or p < 0.01 levels (Student's t-test), respectively, between treatment/inoculation and mock controls at the same time point.
the ClMBD genes in leaf tissues exhibited distinct patterns in response to Db infection and therefore may play different roles in the process of regulating watermelon resistance against Db.
Generation and Characterization of ClMBD-Overexpressing Arabidopsis Lines
To investigate the functions of the ClMBD genes, transgenic Arabidopsis lines with overexpression of an individual ClMBD gene were generated. The ClMBD genes were transcribed normally in their own transgenic Arabidopsis lines (Supplementary Figures 9A,B). The ClMBD-overexpressing Arabidopsis plants showed no significant defect in growth and development, including plant height and size, in comparison to WT plants, when grown in a greenhouse (Supplementary Figure 10).
ClMBD2, ClMBD3, and ClMBD5
Negatively Regulate Arabidopsis Immunity Against Botrytis cinerea To explore the possible functions of the ClMBD genes in plant immunity, disease resistance phenotype of the ClMBDoverexpressing Arabidopsis lines and the wild-type (WT) Col-0 plants after infection with B. cinerea, a necrotrophic fungus causing grey mold disease, was assessed. In repeated detached leaf punch inoculation assays, B. cinerea-caused necrotic lesions on leaves detached from the ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants were significantly larger than those on leaves of WT plants, resulting in increases of approximately 88.9, 55.6, and 66.7% in lesion length, respectively, as compared with those on WT leaves (Supplementary Figure 11). By contrast, B. cinerea-caused necrotic lesions on leaves detached from the ClMBD1-OE, ClMBD4-OE, ClMBD6-OE ClMBD7-OE, ClMBD8-OE, ClMBD9-OE, and ClMBD10-OE plants were comparable to those on leaves of WT plants (Supplementary Figure 11). To confirm these results, the ClMBD-overexpressing plants were inoculated by foliar spraying with B. cinerea spore suspension and disease severity and fungal growth were compared with those in WT plants. After infection, typical B. cinerea-caused disease symptom was seen at 3 dpi. Much severe diseases were observed on leaves of the ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants, especially the B. cinerea-infected ClMBD2-OE plants decayed and died at 3 dpi ( Figure 6A). By contrast, disease severity on leaves of the B. cinerea-infected ClMBD1-OE, ClMBD4-OE, ClMBD6-OE, ClMBD7-OE, ClMBD8-OE, ClMBD9-OE, and ClMBD10-OE plants were similar to that in WT plants (Supplementary Figure 12A). Accordingly, the ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants supported more in planta fungal growth, leading to increases of 57.5-851.3% over that in WT plants (Figure 6B), while the ClMBD1-OE, ClMBD4-OE, ClMBD6-OE, ClMBD7-OE, ClMBD8-OE, ClMBD9-OE, and ClMBD10-OE plants supported similar amounts of in planta fungal growth (Supplementary Figure 12B). These data from detached leaf punch inoculation and whole plant inoculation assays indicate that overexpression of ClMBD2, 3, and 5 attenuates the resistance of transgenic Arabidopsis plants against B. cinerea, while overexpression of each of the remaining ClMBD genes does not affect the resistance against B. cinerea.
To get insights in the possible mechanism of the attenuated B. cinerea resistance, the expression of a marker defense gene AtPDF1.2 and accumulation of reactive oxygen species (ROS) were analyzed and compared between the ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants and WT plants after infection by B. cinerea. In mock-inoculated plants, the expression level of AtPDF1.2 in the ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants was similar to that in WT plants (Figure 6C). At 24 hpi with B. cinerea, the expression level of AtPDF1.2 in the ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants and WT plants were markedly up-regulated, as compared with those in the mock-inoculated plants; however, the expression level of AtPDF1.2 in the ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants were significantly suppressed, resulting in a decrease of 73-81, 80-84, and 59-63%, respectively, as compared with that in Col-0 plants ( Figure 6C). Similarly, no difference in accumulation of H 2 O 2 , as revealed by in situ DAB staining, was observed among the WT, ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants without B. cinerea challenge (Figure 6D). At 24 hpi with B. cinerea, accumulation of H 2 O 2 increased markedly in B. cinerea-infected leaves of the WT, ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants, as compared with those in mockinoculated controls ( Figure 6D). However, more staining for H 2 O 2 in B. cinerea-infected leaves of the ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants was detected, as compared to that in WT plants ( Figure 6D). These data indicate that overexpression of ClMBD2, 3, and 5 in transgenic Arabidopsis plants attenuates the B. cinerea-induced expression of defense genes but promotes the B. cinerea-induced ROS accumulation.
ClMBD1 and ClMBD2 Negatively but ClMBD5 Positively Regulate Arabidopsis Immunity Against Pseudomonas syringae pv. tomato DC3000 The possible involvement of the ClMBD genes in resistance against Pst DC3000, a hemibiotrophic pathogen that causes bacterial spot disease, was also investigated. At 3 dpi, typical Pst DC3000-provoked symptom with chlorotic lesions was seen in WT plants and the ClMBD-overexpressing plants ( Figure 7A and Supplementary Figure 13A). The ClMBD1-OE and ClMBD2-OE plants displayed much severe disease with extensive chlorotic lesion while the ClMBD5-OE plants showed less severe disease ( Figure 7A). Accordingly, the bacterial growth in the ClMBD1-OE and ClMBD2-OE plants was 0.43-0.69 order of magnitude higher while the growth in the ClMBD5-OE plants was ∼1.0 order of magnitude lower, as compared to that in WT at 3 dpi ( Figure 7B). Disease severity and bacterial growth in the ClMBD3-OE, ClMBD4-OE, ClMBD6-OE, ClMBD7-OE, ClMBD8-OE, ClMBD9-OE, and ClMBD10-OE plants were indistinguishable to those in WT plants (Supplementary Figure 13B). These data indicate that overexpression of ClMBD1 and ClMBD2 leads to attenuated D) were repeated for three times with similar results, and results from one representative experiment are shown. Data presented in (B,C) are the means ± SE from three independent experiments and * or ** above the columns indicate significant differences at p < 0.05 or p < 0.01 levels (Student's t-test), respectively, between the ClMBD2/3/5-OE plants and WT plants at the same time point. FIGURE 7 | ClMBD1 and ClMBD2 negatively but ClMBD5 positively regulate resistance of the transgenic Arabidopsis plants against Pseudomonas syringae pv. tomato DC3000. (A) Typical P. syringae pv. tomato DC3000-caused disease on WT, ClMBD1-OE, ClMBD2-OE, and ClMBD5-OE plants. Four-week-old plants were inoculated by injecting with P. syringae pv. tomato DC3000 bacterial suspension (OD 600 = 0.0002) and photographed at 72 hpi. (B) In planta bacterial growth in inoculated leaves. Leaf samples were collected at 0 and 3 dpi and bacterial growth in CFU/cm 2 leaf area are shown. (C) Relative expression of AtPR1 in the mockand P. syringae pv. tomato DC3000-inoculated plants. qRT-RCR was performed using AtActin as an internal control to analyze the expression level of AtPR1 Experiments in (A,B) were repeated for three times with similar results, and results from one representative experiment are shown. Data presented in (C) are the means ± SE from three independent experiments and * or ** above the columns indicate significant differences at p < 0.05 or p < 0.01 levels (Student's t-test), respectively, between the ClMBD1/2/5-OE plants and WT plants at the same time point.
resistance while overexpression of ClMBD5 results in increased resistance against Pst DC3000.
To gain insights in the possible mechanism of the altered Pst DC3000 resistance, the expression of a marker defense gene AtPR1 was analyzed and compared between the ClMBD1-OE, ClMBD2-OE, and ClMBD5-OE plants and WT plants after infection by Pst DC3000. In mock-inoculated plants, the expression level of AtPR1 in the ClMBD1-OE, ClMBD2-OE, and ClMBD5-OE plants was not significantly affected, as compared with that in WT plants (Figure 7C). At 24 hpi with Pst DC3000, the expression level of AtPR1 in the ClMBD1-OE, ClMBD2-OE, and ClMBD5-OE plants and WT plants were markedly up-regulated, as compared with those in the mock-inoculated plants ( Figure 7C). However, the Pst DC3000-induced expression of AtPR1 in the ClMBD1-OE and ClMBD2-OE plants was significantly suppressed, resulting in a decrease of 51-55 and 39-47%, respectively, while the Pst DC3000-induced expression of AtPR1 in the ClMBD5-OE plants was markedly increased by ∼21-folds ( Figure 7C). These data indicate that overexpression of ClMBD1 and ClMBD2 in transgenic Arabidopsis plants attenuates while overexpression of ClMBD5 strengthens the Pst DC3000-induced expression of defense genes.
Identification of Differentially Expressed Genes in ClMBD2-OE Plants
Considering that overexpression of ClMBD2 led to attenuated resistance against B. cinerea and Pst DC3000 (Figures 6, 7), transcriptome profiling of the ClMBD2-OE2 and WT plants grown under normal growth conditions was performed to gain insights into the possible molecular mechanisms of ClMBD2 in regulating resistance against the two pathogens. With criteria of expression change > 1.5-folds and P < 0.05, a total of 70 genes (52 up-regulated and 18 down-regulated) were found to be differentially expressed genes (DEGs) in the ClMBD2-OE2 plants as compared with WT plants (Supplementary Tables 5, 6). The DEGs in the ClMBD2-OE2 plants grown under normal condition were categorized into functional groups based on Gene Ontology (GO). DEGs that were up-regulated (Supplementary Table 7) and 22 categories (Supplementary Table 8), respectively. Some overrepresented categories include genes involved in DNA binding, molecular transducer activity, and transcriptional factor activity in molecular function category, and response to stimulus, immune system process, signaling, and biological regulation in biological processes category (Figures 8A,B), implying that overexpression of ClMBD2 in transgenic Arabidopsis plants may affect the immune signaling and response. Among the DEGs (Supplementary Tables 5, 6), some genes have been previously reported to be involved in Arabidopsis immunity, including AtWRKY18 (Xu et al., 2006), AtWRKY30 (Zou et al., 2019), AtWRKY54 (Chen et al., 2021), AtANAC019 (Zheng et al., 2012), AtMLO6 (Acevedo- Garcia et al., 2017;Kuhn et al., 2017), and AtNATA1 (Lou et al., 2016). The expression patterns of 10 selected genes were further validated by qRT-PCR in ClMBD2-OE and WT plants before and after the infection of B. cinerea and Pst DC3000. In the ClMBD2-OE plants without pathogen infection, the expression levels of AtWRKY18, AtWRKY30, AtANAC019, AtARCK1, AtMLO6, and AtERF54 were significantly up-regulated while the expression levels of AtMAF5, AtBEE1, AtbZIP34, and AT5G52190 were markedly down-regulated ( Figure 8C). After infection of B. cinerea, the expression of AtbZIP34 was down-regulated, while the expression of other genes was up-regulated in ClMBD2-OE and WT plants ( Figure 8C). The expression levels of AtWRKY18 and AtbZIP34 were significantly increased, while the expression levels of AtANAC019 and AtBEE1 were significantly suppressed in ClMBD2-OE plants after infection of B. cinerea, as compared with those in WT plants ( Figure 8C). After infection of Pst DC3000, the expression of AtWRKY18, AtWRKY30, AtANAC019, AtARCK1, AtERF54, AtMLO6, and AtMAF5 was up-regulated, but the expression of AtBEE1, AT5G52190, and AtbZIP34 was down-regulated in ClMBD2-OE and WT plants ( Figure 8C). The expression levels of AtWRKY18, AtERF54, AtMAF5, AT5G52190, and AtBEE1 were significantly decreased in ClMBD2-OE plants after infection of Pst DC3000, as compared with those in the WT plants ( Figure 8C). These data consistently conformed the results from RNA-seq analysis and indicate that ClMBD2 regulates a small set of defense and signaling genes that are involved in Arabidopsis immunity.
DISCUSSION
It has been documented that the MBD proteins play important roles in plant growth, development, and abiotic stress response; however, the involvement of the MBD proteins in plant immunity has not been established. The present study characterized the MBD families in watermelon and other cucurbit plants, examined the subcellular localization and binding activity of ClMBDs to 5-mC DNAs, analyzed the expression patterns of ClMBDs in response to defense hormones and pathogens, and explored the functions of ClMBDs in disease resistance. The functional analysis in transgenic Arabidopsis revealed that ClMBD1, 2, 3, and 5 play roles in immunity against B. cinerea and Pst DC3000, providing novel insights into the function of the MBD genes in plant immunity and a possibility to improve plant disease resistance through genetic manipulation of specific MBD genes.
The present study identified 10 watermelon ClMBD genes and 9, 10, 15, and 16 MBD genes in melon, cucumber, zucchini, and pumpkin, respectively ( Table 1 and Supplementary Table 2). The numbers of ClMBDs in watermelon and MBD genes in other cucurbit plants are comparable to those of Arabidopsis (13), rice (17), maize (14), poplar (14), potato (15), tomato (18), and petunia (11) (Grafi et al., 2007;Parida et al., 2018;Shi et al., 2022). The presence of gene pairs in duplicated genomic regions of the watermelon genome and syntenic collinearity gene pairs between watermelon and other cucurbit plant species (Figures 1, 2 and Supplementary Tables 3, 4) suggests that gene duplication events occurred during the evolution of the ClMBD family. In addition to the typical MBD domain, other conserved domains such as zf-CW domain, SPARK domain, PKINase domain, and Bromo domain were also identified in some of the ClMBD proteins (Figure 1). Similar conserved domains are also present in Arabidopsis AtMBDs, tomato SlMBDs, and potato StMBDs (Grafi et al., 2007;Parida et al., 2018). It is thus likely that some watermelon ClMBD proteins may exert their functions in affecting transcription of target genes through different biochemical mechanisms including protein-protein interactions. Subcellular localization observations revealed that the ClMBD proteins were localized in nucleus when transiently expressed in N. benthamiana (Figure 3). This is consistent with the previous observations that most of the Arabidopsis AtMBD proteins displayed clear localization within the nucleus in onion cells (Berg et al., 2003;Ito et al., 2003) and that ZmMBD101 localized to nucleoplasmic foci (Questa et al., 2016). Another, ClMBD2 and ClMBD3 interacted with ClIDM2 and ClIDM3, and the interactions occurred in nucleus in BiFC assays (Figure 3). This feature is similar to the interaction of the Arabidopsis AtMBD7 with AtIDM2 and AtIDM3 (Lang et al., 2015), and further confirmed the nuclear localization of the ClMBD2 and ClMBD3. ClMBD2 is phylogenetically related to AtMBD5 and AtMBD6 and also shows an evolutionary syntenic relationship with AtMBD5 (At3G46580) and AtMBD6 (At5G59380) (Supplementary Table 4), implying that ClMBD2 may have a similar biochemical activity to AtMBD5 and AtMBD6. In the present study, ClMBD2 showed the ability to bind to mCG DNA (Figure 4), similar to AtMBD5 and AtMBD6, maize ZmMBD101, and tomato SlMBD5, which have the binding ability to mCG DNA (Ito et al., 2003;Scebba et al., 2003;Grafi et al., 2007;Questa et al., 2016). However, ClMBD2 did not bind to mCHH and mCHG DNA (Figure 4), different from AtMBD5 and AtMBD6, which also have the ability Data presented in (C) are the means ± SE from three independent experiments and * or ** above the columns indicate significant differences at p < 0.05 or p < 0.01 levels (Student's t-test), respectively, between the ClMBD2-OE plants and WT plants. to bind to mCHH DNA (Ito et al., 2003;Scebba et al., 2003;Grafi et al., 2007). Surprisingly, the binding activity of the other ClMBD proteins to mCG DNA was not detected in the present study (Figure 4), implying that the ClMBD proteins may have different biochemical activities in recognizing methylated or unmethylated DNA and thus confer specific biological functions.
It was previously observed that the expression of some tomato SlMBD, wheat TaMBD and petunia PhMBD genes were affected by abscisic acid and abiotic stress, e.g., drought, salt, and cold stress (Li et al., 2008;Hu et al., 2011;Parida et al., 2018;Shi et al., 2022). The expression of most of the watermelon ClMBD genes was up-regulated after SA or MeJA treatment, except that ClMBD5 and ClMBD9 were down-regulated by MeJA and that the expression of ClMBD1 and ClMBD3 was not affected by MeJA (Figure 5). In response to Fon, the expression of almost all of the ClMBD genes in root tissues was up-regulated ( Figure 5). By contrast, the expression of ClMBD3, 4, 7, and 9 was up-regulated, while the expression of ClMBD2, 6, 8, and 10 was down-regulated in response to Db (Figure 5). Notably, the expression changes of the ClMBD genes exhibited differential but inconsistent patterns in leaf and root tissues of watermelon plants in response to treatment of SA and MeJA and to infection of Fon and Db. However, the expression changes induced by the two defense hormones and the two fungal pathogens imply the involvement of the watermelon ClMBD genes in disease resistance, probably through affecting transcription of a set of genes including those involved in defense response.
The responsiveness of the watermelon ClMBD genes to exogenous SA and JA, two hormones that mediated defense response against (hemi)biotrophic pathogens such as Pst DC3000 and necrotrophic fungi like B. cinerea (Glazebrook, 2005;Grant and Jones, 2009) led to evaluate the disease resistance phenotype of the ClMBD-OE Arabidopsis lines against Pst DC3000 and B. cinerea. In the present study, ectopic overexpression of ClMBD2, 3, and 5 in transgenic Arabidopsis led to increased susceptibility to B. cinerea (Figure 7 and Supplementary Figure 12), suggesting that ClMBD2, 3, and 5 are negative regulators of defense response against B. cinerea. This is further supported by the suppression of pathogen-induced expression of defense gene AtPDF1.2, an indicator gene of defense response against necrotrophic fungal pathogens, and overaccumulation of ROS in ClMBD2-OE, ClMBD3-OE, and ClMBD5-OE plants (Figure 6). This is consistent with the general concept that excessive ROS accumulation during early stage often benefits the infection by the necrotrophic fungi like B. cinerea, but is different from the phenomenon that early ROS accumulation is critical to the activation of immune response against (hemi)biotrophic pathogens (Mengiste, 2012). On the other hand, overexpression of ClMBD1 and ClMBD2 in transgenic Arabidopsis attenuated while overexpression of ClMBD5 strengthened resistance to Pst DC3000 (Figure 7), indicating that ClMBD1 and ClMBD2 are negative regulators while ClMBD5 is a positive regulator of immunity against this bacterial pathogen. This is consistent with the suppression of expression of AtPR1 in ClMBD1-OE and ClMBD2-OE plants but elevation of expression of AtPR1 in ClMBD5-OE plants after infection of Pst DC3000 (Figure 7). This is also indirectly supported by the down-regulated expression of ClMBD2 in leaf tissues of watermelon plants after infection of Db, a heminecrotrophic fungal pathogen (Figure 5). Notably, overexpression of ClMBD2 in transgenic Arabidopsis resulted in attenuated immunity against both B. cinerea and Pst DC3000; however, overexpression of ClMBD5 led to opposite functions in immunity against these two pathogens, e.g., attenuated immunity against B. cinerea but strengthened immunity against Pst DC3000 (Figures 6, 7). It is generally accepted that immune response against (hemi)biotrophic pathogens such as Pst DC3000 is modulated through the SA signaling while the defense response against necrotrophic pathogens like B. cinerea is regulated by the JA/ET signaling (Glazebrook, 2005;Grant and Jones, 2009). Both antagonistic interaction and synergistic cross-talks between the SA and JA/ET signaling pathways occur and allow plants to mount appropriate immune responses against different invading pathogens (Glazebrook, 2005;Koornneef and Pieterse, 2008;Verhage et al., 2010). It is therefore likely that ClMBD2 and ClMBD5 function in immunity through regulating different mechanisms.
Transcriptome profiling identified a limited number of DEGs in the ClMBD2-OE2 plants grown under normal growth conditions (Supplementary Tables 5, 6). The fact that genes involved in DNA binding, transcriptional factor activity, response to stimulus, and immune system process were overrepresented in DEGs in ClMBD2-OE2 plants (Figures 8A,B) further confirms the function of ClMBD2 in immunity of the transgenic Arabidopsis plants against B. cinerea and Pst DC3000. Generally, the MBD proteins recognize the methylated CG sites and recruit chromatin remodelers and histone deacetylases to repress transcription of target genes (Lang et al., 2015). Considering the attenuated immunity against B. cinerea and Pst DC3000 (Figures 6, 7), it is speculated that overexpression of ClMBD2 should lead to down-regulation of a set of genes that are involved in Arabidopsis immunity. Surprisingly, only 18 genes were identified as down-regulated genes (expression change > 1.5-folds and P-value < 0.05) in the ClMBD2-OE2 plants (Supplementary Table 6). Among these down-regulated genes, AtWRKY54 was previously reported to function as a positive regulator of SARD1 and CBP60g expression in immunity against P. syringae pv. maculicola (Chen et al., 2021;Figure 8). No other gene with known function in Arabidopsis immunity was identified in the down-regulated genes in ClMBD2-OE2 plants (Supplementary Table 6). This might be due to the fact that samples from healthy ClMBD2-OE2 plants without pathogen infection were used for RNA-seq analysis. Indeed, the expression of defense genes such as AtPR1 and AtPDF1.2 was suppressed significantly in the ClMBD2-OE plants upon infection with Pst DC3000 and B. cinerea, respectively (Figures 6, 7). If it is the case that ClMBD2, like its closely related AtMBD5 and AtMBD6, acts to repress transcription of target genes, this may imply that the function of ClMBD2 in suppression of transcription of defense genes in transgenic Arabidopsis plants occurs upon pathogen infection. By contrast, some genes that negatively regulate Arabidopsis immunity were found to be up-regulated in the ClMBD2-OE2 plants (Figure 8 and Supplementary Table 5). For example, AtWRKY18 negatively regulates resistance against Pst DC3000 but positively modulates resistance to B. cinerea (Xu et al., 2006), AtANAC019 negatively regulates immune response through repressing AtICS1 and thus inhibiting SA accumulation (Zheng et al., 2012), AtMLO6 is a susceptible gene for powdery mildew disease (Acevedo-Garcia et al., 2017;Kuhn et al., 2017), and AtNATA1 negatively regulates immunity against Pst DC3000 by acetylating putrescine and decreasing ROS accumulation (Lou et al., 2016). It seems that overexpression of ClMBD2 activates an unknown pathway that up-regulates the expression of a subset of genes with negative functions in Arabidopsis immunity.
In summary, the present study characterized the watermelon ClMBD family and the MBD families in other cucurbit plants in terms of gene structures, conserved domain organization, phylogenetic and syntenic relationships, evolution events, subcellular localization, biochemical activity, and expression patterns in response to defense hormones and pathogen infection. The present study also provided the information on the possible involvement of each of the watermelon ClMBD genes in disease resistance when they were ectopically expressed in Arabidopsis. Functional analyses in transgenic Arabidopsis revealed that CMBD2, 3, and 5 negatively regulate Arabidopsis resistance against B. cinerea, and that ClMBD1 and ClMBD2 negatively while ClMBD5 positively regulate Arabidopsis resistance against Pst DC3000. Transcriptome analysis showed that overexpression of ClMBD2 in transgenic Arabidopsis affected the expression of a small set of genes that are involved in Arabidopsis immunity. Further analyzing the DNA methylation levels and characterizing the genomewide binding sites in the ClMBD2-OE and ClMBD5-OE transgenic Arabidopsis plants will definitely provide detailed molecular mechanisms by which ClMBD2 and ClMBD5 regulate immunity against B. cinerea and Pst DC3000. Due to the divergence of gene functions in immunity between Arabidopsis and watermelon, the functional analysis in the present study performed by ectopic overexpression in Arabidopsis may not reflect the intrinsic functions of the ClMBD gene in watermelon disease resistance. Therefore, further investigations in watermelon through overexpression and CRISPR/Cas9-based knockout approaches will be critical to elucidate the functions and molecular mechanisms of the ClMBD genes, especially the ClMBD1, 2, 3, and 5 in disease resistance against Fon, Db, and other pathogens.
DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm. nih.gov/, PRJNA803007. | 2022-05-09T13:26:14.648Z | 2022-05-09T00:00:00.000 | {
"year": 2022,
"sha1": "03d8ed3dd0bda56c27c55e5da2e72bc387e3fa71",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Frontier",
"pdf_hash": "03d8ed3dd0bda56c27c55e5da2e72bc387e3fa71",
"s2fieldsofstudy": [
"Biology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
5893114 | pes2o/s2orc | v3-fos-license | Characterization of the novel mutant A78T-HERG from a long QT syndrome type 2 patient: Instability of the mutant protein and stabilization by heat shock factor 1
Background The human ether-a-go-go-related gene (HERG) encodes the α-subunit of rapidly activating delayed-rectifier potassium channels. Mutations in this gene cause long QT syndrome type 2 (LQT2). In most cases, mutations reduce the stability of the channel protein, which can be restored by heat shock (HS). Methods We identified the novel mutant A78T-HERG in a patient with LQT2. The purpose of the current study was to characterize this mutant protein and test whether HS and heat shock factors (HSFs) could stabilize the mutant protein. A78T-HERG and wild-type HERG (WT-HERG) were expressed in HEK293 cells and analyzed by immunoblotting, immunoprecipitation, immunofluorescence, and whole-cell patch clamping. Results When expressed in HEK293 cells, WT-HERG gave rise to immature and mature forms of the protein at 135 and 155 kDa, respectively. A78T-HERG gave rise only to the immature form, which was heavily ubiquitinated. The proteasome inhibitor MG132 increased the expression of immature A78T-HERG and increased both the immature and mature forms of WT-HERG. WT-HERG, but not A78T-HERG, was expressed on the plasma membrane. In whole-cell patch clamping experiments, depolarizing pulses evoked E4031-sensitive HERG channel currents in cells transfected with WT-HERG, but not in cells transfected with A78T-HERG. The A78V mutant, but not A78G mutant, remained in the immature form similarly to A78T. Maturation of the A78T-HERG protein was facilitated by HS, expression of HSF-1, or exposure to geranyl geranyl acetone. Conclusions A78T-HERG was characterized by protein instability and reduced expression on the plasma membrane. The stability of the mutant was partially restored by HSF-1, indicating that HSF-1 is a target for the treatment for LQT2 caused by the A78T mutation in HERG.
Introduction
Long QT syndrome (LQTS) is characterized by QT intervals longer than 450 ms that lead to a torsades de pointes ventricular tachycardia and sudden death. LQTS is attributed to mutations in genes that encode ion channels or their regulatory proteins responsible for the proper repolarization of cardiomyocytes [1][2][3], Contents lists available at ScienceDirect Abbreviations: LQTS, long QT syndrome; HERG, human-ether-a-go-go-related gene; HS, heat shock; HSP, heat shock protein; HSF, heat shock factor; GGA, geranyl geranyl acetone n Correspondence to: Division of Regenerative Medicine and Therapeutics, Tottori and is classified into 13 types based on the mutated genes. LQTS types 1, 2, and 3 account for 90% of LQTS cases [4].
Human-ether-a-go-go-related gene (HERG) encodes the αsubunit of rapidly-activating delayed-rectifier potassium channels that generate I Kr . This outward potassium current is elicited during the plateau phase of action potentials and is required for repolarization [4]. After being translated in the endoplasmic reticulum (ER), the HERG protein is transported to the Golgi apparatus where it undergoes glycosylation [5]. This modification transforms the protein from the premature form into the mature form. Eventually, HERG is transported to the plasma membrane [6]. LQTS type 2 (LQT2) is caused by mutations in HERG [7]. In most LQT2 cases, the mutations destabilize the HERG protein and impair its maturation and intracellular transport [8]. Sorted by the ER quality control system, the mutant HERG protein is reverse-transported from the ER to the cytoplasm where it is degraded by the ubiquitin-proteasome system, resulting in reduced HERG channel currents and impaired repolarization of ventricular action potentials [9]. It has been reported that heat shock (HS) assists both in the folding of newly synthesized proteins and the refolding of denatured proteins [10]. Molecular chaperones such as heat shock proteins (HSPs), including Hsp90 and Hsp70, induced by HS play an important role in the maturation of HERG [11,12]. In mammals, the heat shock factor (HSF) family consists of 4 subtypes and increases in response to HS to activate the transcription of molecular chaperones.
In the present study, we characterized a novel mutant, A78T-HERG, found in a patient with LQT2. The A78T-HERG protein showed reduced stability and failed to transfer to the plasma membrane. Its stability was partially restored by HSF-1, indicating that HSF-1 is a target for the treatment for LQT2 caused by the A78T mutation in HERG.
Cell culture and transfection
HEK293 cells were cultured in DMEM (Sigma, St. Louis, MO) supplemented with 10% fetal bovine serum (JRH Biosciences, Inc., St. Louis, MO) and penicillin/streptomycin/geneticin at 37°C, in a 5% CO 2 atmosphere. Missense mutations were introduced into pcDNA3/HERG-FLAG by site-directed mutagenesis. pcDNA3 expression plasmids for HSF-1, 2, and 4 were provided by A. Nakai. Cells were transfected using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. The total amount of cDNA was adjusted using vector cDNA.
Immunoblotting and immunoprecipitation
Cells were harvested 48 h after transfection and lysed by sonication in lysis buffer (PBS supplemented with 1% polyoxyethylene octylphenyl ether, 0.5% sodium deoxycholate, 0.1% SDS, 1.5 mM aprotinin, 21 mM leupeptine, 15 mM pepstain, and 1 mM phenylmethylsulfonylfluoride). After removal of insoluble materials by centrifugation, protein concentrations were determined by the bicinchoninic acid protein assay using a commercially available kit (Pierce Biotechnology, Rockford, IL). Proteins were separated by SDS-PAGE and electrotransferred to PVDF membranes. The membranes were blocked with 5% non-fat dry milk in PBS containing 0.1% Tween and probed with the primary antibody. Antibodies against the following proteins were used: Fisher Scientific, Waltham, MA). The blots were developed using an ECL system (Amersham, Biosciences, Amersham, UK). Immunoprecipitation was carried out at 4°C overnight in PBS supplemented with 1% Triton X-100, 0.5% SDS, 0.25% sodium deoxycholate, 1 mM EDTA, and protease inhibitors. Immune complexes were collected using protein G agarose (Pharmacia, Uppsala, Sweden) and bound proteins were analyzed by SDS-PAGE followed by immunoblotting. Band densities were quantified using NIH image J software (Bethesda, MD) and normalized to the densities of β-actin.
Immunofluorescence
HEK293 cells on gelatin-coated coverslips were transfected with pcDNA3/HERG-FLAG and pAcGFP-Mem constructs (Clontech Laboratories, Inc., Mountain View, CA). Twenty-four hours after transfection, the cells were fixed in 4% paraformaldehyde/PBS and then permeabilized with 0.5% Triton X-100. Next, the cells were incubated for 1 h at room temperature with the primary antibody (FLAG, 1:1000). After blocking in 3% albumin, bound antibodies were visualized with Alexa Fluor 546-conjugated mouse secondary antibody (1:2000), and images were obtained using a Bio-Rad MRC 1024 confocal microscope (Bio-Rad, Hercules, CA).
Statistical analysis
All data are presented as the mean 7SEM. For statistical analysis, Student's t-test and repeated measures analysis of variance (two-way ANOVA) were used, with Po 0.05 considered statistically significant.
Characterization of the LQT2 patient
A 7-year-old boy was diagnosed with LQT2. The QT interval was 535 ms and the QT interval corrected for heart rate was 496 ms (Fig. 1A). There were no abnormal findings on the chest X-ray film, ultrasonic cardiogram, or treadmill exercise test. Direct sequencing of HERG (KCNH2) alleles identified the mutation G232A (A78T) in one allele (Fig. 1B). A78 is in the N-terminal intracellular domain of the HERG channels (Fig. 1C). No other mutations were detected. 220 230 240
Impaired maturation of A78T-HERG protein
When expressed in HEK293 cells, WT-HERG-FLAG gave rise to immature and mature proteins of 135 kDa and 155 kDa, respectively. In contrast, A78T-HERG gave rise to an immature protein only ( Fig. 2A). There was a significant reduction in the mature protein in cells transfected with A78T-HERG compared to transfection with WT-HERG (Fig. 2B). Cotransfection of the A78-HERG and WT-HERG plasmids (0.5 μg each) reduced the mature band of 155 kDa, while transfection with the same amount of WT-HERG and vector plasmid (0.5 μg) yielded the mature band, indicating a dominant-negative effect of A78T-HERG ( Supplementary Fig. 1A).
Ubiquitination of A78T-HERG proteins and effects of proteasome inhibitor
Representative ubiquitination of the A78T-HERG and WT-HERG proteins is shown in Fig. 3A. Ubiquitination of A78T-HERG proteins was much more abundant than that of WT-HERG proteins. Quantification of ubiquitination, obtained from 4 different experiments, is shown in Fig. 3B. The A78T-HERG protein was significantly ubiquitinated compared to the WT-HERG protein.
Treatment with the proteasome inhibitor MG132 increased the mature form of WT-HERG proteins but not that of A78T-HERG, whereas the immature forms of both WT-HERG and A78T-HERG were increased (Fig. 4A and B).
Cell-surface expression of A78T-HERG and its channel activity
Representative immunofluorescence images of cells expressing WT-HERG or A78T-HERG are shown in Fig. 5A. The signal from the WT-HERG protein, but not that of the A78T-HERG protein, colocalized with that from AcGFP-Mem. Representative original traces of HERG channel currents in HEK293 cells expressing WT-HERG or A78T-HERG are shown in Fig. 5B. The E4031-sensitive outward currents and tail currents were largely time-dependent and observed in cells expressing WT-HERG, whereas there were only small E4031-sensitive outward currents with no detectable tail current in cells expressing A78T-HERG. The current-voltage relationship of HERG channel currents in cells expressing WT-HERG or A78T-HERG (Fig. 5B, lower panel; n ¼10) shows that there was a significant reduction in peak outward HERG currents in cells expressing A78T-HERG. As shown in Supplementary Fig. 1B, cells transfected with both the WT-HERG and A78T-HERG plasmids gave rise to reduced E4031-sensitive currents, indicating a dominant negative effect of the mutant.
Effects of amino acid substitutions on HERG expression
The type and position of a mutation in a gene influences protein structure and stability [13]. To clarify whether the key factor for instability of the A78T mutant is the change in molecular size or hydrophobicity (or both) at position 78, we investigated the effect of substituting the amino acid alanine (A) at position 78 with valine (V), which is similar in molecular radius to a threonine residue (larger than an alanine residue), but is a non-polar amino acid similarly to alanine. For A78T, A78V failed to give rise to the mature form of HERG. Next, in order to evaluate the importance of the methyl group, we constructed another mutant, A78G-HERG, in which the alanine at position 78 was substituted with glycine (G), which has no methyl group and is smaller than alanine. The mature form of the HERG protein was detected when A78G-HERG was expressed, suggesting a tolerance for deletion of the methyl group, i.e., a decrease in size and hydrophobicity (Fig. 6).
Effects of HS, HSF-1, and a HSF-1 inducer on the maturation of A78T-HERG protein
Heat shock increases HSF levels and facilitates the maturation of WT-HERG proteins [11]. HS at 42°C for 1 h increased the expression of the mature 155 kDa protein of A78T-HERG (Fig. 7A), which was confirmed in four different experiments (B). HSP expression is regulated by HSF-1, 2, and 4. HSF-1 increased the mature form of A78T-HERG, whereas neither HSF-2 nor 4 influenced the maturation of A78T-HERG (Fig. 8A, left panel) Fig. 8B (left panel). A significant increase in the E4031-sensitive mutant channel current was observed in cells treated with HS. The current-voltage relationships of A78T-HERG currents in transfected cells with or without HS treatment, obtained from ten experiments, are also shown in Fig. 8B (right panel). The E4031sensitive peak outward A78T-HERG currents significantly increased in cells treated with HS. This effect of HS was mimicked by co-expression with HSF-1.
Discussion
The novel HERG mutation A78T, found in a patient with LQT2, was located at the N-terminus of HERG proteins. The A78T-HERG protein failed to mature and was expressed as a highly ubiquitinated 135-kDa immature protein. Since ubiquitination of the protein to be degraded by the proteasome is known to occur in the core-glycosylated immature form at the ER [11,12], this finding indicates the impaired stability of A78T-HERG proteins to be degraded via the ubiquitin proteasome system. Protein maturation is facilitated by the accumulation of immature proteins (mass effect) as well as by posttranslational modifications including glycosylation. In the present study, MG132 increased the expression of the 135-kDa immature form of A78T-HERG, but not that of the 155-kDa mature form, suggesting that the accumulation of immature A78T-HERG did not facilitate its maturation. We previously demonstrated that MG132 increased the expression of the mature form of Kv1.5, a voltage-gated potassium channel, indicating that accumulated immature wild-type Kv1.5 protein could be converted to its mature form as a result of the mass effect [14]. Since a substantial portion of Kv1.5 is degraded via the ubiquitinproteasome system, proteasomal inhibition results in the accumulation of a large amount of immature Kv1.5, facilitating its maturation as a mass effect. In contrast, during the synthesis of WT-HERG, most of the core-glycosylated immature protein is converted to the fully-glycosylated mature form [6]. Thus, accumulation of the immature form of HERG protein in a setting of proteasomal inhibition is lower than that of Kv1.5. The failure of MG132 to increase the expression of mature A78T-HERG suggests that accumulation of immature A78T-HERG is not sufficient to exert a mass effect, while accumulation of immature A78T-HERG may not increase its mature form because of severe dysfunction of its maturation process.
Since the type and position of a mutation in the gene influence protein structure and stability [13], the A78T mutation may destabilize the HERG protein because of an alteration in the molecular radius or hydrophobicity of the amino acid at position 78. A78G-HERG yielded the mature protein, whereas A78V-HERG, similarly to A78T-HERG, did not. The former finding excludes the possibility that deletion of the methyl group by substitution for alanine destabilizes the A78T-HERG protein. The latter finding suggests that the larger molecular radius but not hydrophilicity of the threonine residue destabilizes A78T-HERG.
Heat shock facilitates the folding of various proteins as mediated by the HSP family and can stabilize mutant proteins. HS has been reported to facilitate maturation of WT-HERG as well as mutant HERG by activating the HSF family [15]. HSF-1, 2, and 4 were reported to be expressed in humans. HSF-1 is a master regulator of Hsp70 and Hsp90. HSF-2 plays roles in the development of the brain and reproductive organs and acts as a major regulator of the proteostasis capacity against febrile-range thermal stress to prevent the accumulation of misfolded proteins [16]. HSF-4, which possesses transcriptional repressor properties in vivo, inhibits basal transcription of Hsp27 and Hsp90 in cultured cells [17]. In the present study, we found that HSF-1, but not HSF-2 or 4, increased the expression of mature A78T-HERG protein. These results suggest that HSF-1 stabilizes A78T-HERG protein, promotes its transport to the plasma membrane, and increases the membrane current I Kr .
The ubiquitin proteasome system in cardiomyocytes, including HL-1 cells and neonatal rat cardiomyocytes, involves three activities: chymotrypsin-like, trypsin-like, and caspase-like activities. The degradation of several ion channel proteins showed similar time courses between cardiomyocytes and HEK293 cells, indicating no difference in ubiquitin-proteasome system activities in these cells. Thus, mutant HERG proteins such as A78T-HERG may be degraded through the ubiquitin-proteasome system in cardiomyocytes as well [18]. Accordingly, the clinical implications of our results are as follows. To date, many studies have shown that maturation of mutant HERG proteins can be improved by low temperature or HS and by application of E-4031 or a gastroprokinetic agent such as cisapride [19][20][21]. GGA is a non-toxic acyclic isoprenoid compound with a retinoid skeleton that increases Hsp70 expression by activating HSF-1 [22]. In the present study, GGA at the clinical concentration of 200 nM facilitated maturation of the A78T-HERG protein, suggesting that GGA is of therapeutic value for patients with LQT2 caused by the A78T mutation in HERG.
Conclusions
The A78T-HERG protein showed reduced stability and failed to transfer to the plasma membrane. Its stability was partially restored by HSF-1, indicating that HSF-1 is a potential target for the treatment for LQT2 caused by the A78T mutation in HERG. | 2018-04-03T03:29:18.380Z | 2015-11-25T00:00:00.000 | {
"year": 2015,
"sha1": "7e287747e25be822aab3b3849feea3612c4616d5",
"oa_license": "CCBYNCND",
"oa_url": "https://doi.org/10.1016/j.joa.2015.10.005",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "7e287747e25be822aab3b3849feea3612c4616d5",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Biology",
"Medicine"
]
} |
14158806 | pes2o/s2orc | v3-fos-license | Heterogeneous Mechanisms of Secondary Resistance and Clonal Selection in Sarcoma during Treatment with Nutlin
Nutlin inhibits TP53-MDM2 interaction and is under investigation in soft-tissue sarcomas (STS) and other malignancies. Molecular mechanisms of secondary resistance to nutlin in STS are unknown. We performed whole-transcriptome sequencing (RNA-seq) on three pretreatment and secondary resistant STS cell lines selected based on their high primary sensitivity to nutlin. Our data identified a subset of cancer gene mutations and ploidy variations that were positively selected following treatment, including TP53 mutations in 2 out of 3 resistant cell lines. Further, secondary resistance to nutlin was associated with deregulation of apoptosis-related genes and marked productive autophagy, the inhibition of which resulted in significant restoration of nutlin-induced cell death. Collectively, our findings argue that secondary resistance to nutlin in STS involved heterogeneous mechanisms resulting from clonal evolution and several biological pathways. Alternative dosing regimens and combination with other targeted agents are needed to achieve successful development of nutlin in the clinical setting.
Introduction
The tumor suppressor TP53 plays a crucial role in protection from malignant tumor development. It is a transcription factor which is activated following stress and regulates multiple downstream genes involved in cell cycle control, apoptosis, DNA repair, and senescence [1]. In non-stressed cells, the level of T53 is controlled tightly by MDM2 (murine double minute 2). MDM2 regulates p53 through a negative-feedback loop. When the nuclear TP53 level is elevated, it activates the transcription of the MDM2 gene. In turn, MDM2 binds to TP53 and blocks its transactivation domain. MDM2 also serves as a TP53 ubiquitin ligase that targets TP53 for ubiquitin-dependent degradation in the proteasome [2].
Treatment of cancer cells expressing wild-type TP53 with TP53-MDM2 interaction antagonists should result in the concurrent transcriptional activation of TP53 downstream genes, cell cycle arrest, and apoptosis.
Cell viability
Nutlin effects on cell viability were investigated using the MTT assay [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide] (Sigma-Aldrich Chimie, Saint-Quentin-Fallavier, France) as an indicator of metabolically-active cells [16]. Known number (2000 or 3000) of STS cells was transferred into 96-well plates incubated for 24 h before addition of test compound. Cells were then exposed for 72 h at 37°C to an increasing concentration range of RG7388, for 48 h to doxorubicine or gemcitabine, MTT at the final concentration of 0.5mg/ml was added and, following incubation for 3h, Formazan crystals were dissolved in DMSO. Absorbance of the colored solution was measured on a microplate-photometre (Bio-Tek Instruments, Colmar, France) using a test wavelength of 570 nm and a reference wavelength of 630 nm. The concentration of substance required for 50% growth inhibition (IC50) was estimated with the Graphpad Prism software (GraphPad software INC, San Diego, USA).
Cell cycle analysis
Cell cycle distribution of sensitive and resistant cell lines was studied by DNA content by fluorescence-activated cells sorting and analyzed using the Cell Quest Pro software (BD Biosciences, San Jose, USA). After 48h of treatment with two different concentrations of RG7388, cells were centrifuged at 1500g for 5 min and washed twice with PBS. Cells were then fixed with 70% ethanol at 4°C overnight. Ethanol was removed and cells washed twice with PBS. 300μl of a propidium iodide, ribonuclease-containing solution were added to cells and then analyzed by FACS. Data were analyzed with FlowJo v.7.6.3. software and results expressed in terms of percentage of cells in a given phase of cycle.
Apoptosis
The mitochondrial membrane depolarization (ΔCm) was measured by tetramethyl-rhodamine ethyl staining (TMRME). 5000 cells were seeded in 96-well plates, incubated 24h before adding an increasing concentration of RG7388 over 48h. Cells were loaded with TMRM (200nM) for 30 min at 37°C and then trypsinized and harvested in saline extracellular solution. Cells were then analyzed on a FAC scan flow cytometer (BD biosciences, San Jose, USA).
For apoptosis assessment, 1.105 cells were seeded in 6-wells plates and after 24h cells were treated by Nutlin for 72hr and exposed to FITC-Annexin and propidium iodide (PI) according to the manufacturer's protocol (BD Biosciences, Erembodegem, Belgium). It allows distinguishing annexin V positive cells in early apoptosis, versus annexin V and PI positives cells in late apoptosis. Cells were analyzed in cytomerty using FL1 Annexin-V, whereas FL2 was used for Propidium iodide.
Confocal microscopy
Cells were seeded on coverslips and treated with RG-7388 for 72 hours. Slides were then washed twice with PBS, fixed in formaldehyde 4% and incubated with anti-LC3IIB monoclonal antibody (Sigma Aldrich, Saint Louis, USA) overnight, and then with a goat anti-rabbit Alexa fluor 488 antibody (Invitrogen, Paisley, United Kingdom). Slides were then counterstained by 4,6-diamidino-2-phenylindole (Hoechst).
RnaSeq protocols for Illumina HiSeq2000 NGS
RnaSeq was sequenced by the Centro nacional de análisis genómico (Barcelona, Spain) using standard Illumina (Illumina Inc., 9885 Towne Centre Drive, San Diego, CA, USA) protocols. Briefly, the library from total RNA was prepared using the TruSeq™ Stranded mRNA Sample Preparation kit (Illumina Inc.) according to manufacturer's protocol. 1μg of total RNA was used for poly-A based mRNA enrichment selection using oligo-dT magnetic beads followed by fragmentation by divalent cations at an elevated temperature resulting into fragments of 80-450nt, with the major peak at 160nt. First strand cDNA synthesis by random hexamers and reverse transcriptase was followed by the second strand cDNA synthesis, performed in the presence of dUTP instead of dTTP. Blunt-ended double stranded cDNA was 3´adenylated and the 3´-"T" nucleotide at the Illumina indexed adapters was used for the adapters' ligation. The ligation product was amplified with 15 cycles of PCR.
Each library was sequenced using TruSeq SBS Kit v3-HS, in paired end mode with the read length 2x101bp. We generated 135 million minimally-paired end reads for each sample run in one sequencing lane on HiSeq2000 (Illumina, Inc) following the manufacturer's protocol. Images analysis, base calling and quality scoring of the run were processed using the manufacturer's software Real Time Analysis (RTA 1.13.48) and followed by generation of FASTQ sequence files by CASAVA.
NGS RnaSeq sequence alignment and quality control pipeline
Raw RnaSeq sequences where quality controlled using a set of published tools in order to produce curated reads. Briefly, the package sickle [18] was used to trim raw reads of the 5' and 3' low quality bases (phred cut off 20, max trim size 30nc, package SeqPrep (St John GitHub, https://github.com/jstjohn/SeqPrep) was used to remove sequencing adaptors from the raw reads. The application of these standard packages revealed a proportion of RNA fragments of smaller size, whose R1 and R2 paired end reads were overlapping. To keep exploiting these fragments without biasing the count of reads in reads overlapping regions, we developed a home-made script (awk language) that merged the overlapping R1 and R2 and split it into new non-overlapping read sequences.
Curated read sequences were then aligned using TopHat2 [19] on both genome and transcriptome with fragment size 175 +/-70. Following, we applied strict quality control of aligned reads, by filtering the reads whose alignment score was lower than 20 (samtools) and removing the reads with identical align starting positions that were possibly caused by PCR amplification during the creation of the RnaSeq libraries (picard MarkDuplicates at http://broadinstitute. github.io/picard/).
The statistics of the number of raw, curated and aligned sequences and curated aligned sequences was that created with a home-made tool (bash, awk)
NGS allelic variant detection and annotation pipeline
Positions in the transcriptome showing at least one alternative allele were detected via samtools mpileup [20] taking care to filter out bases with less than 20 as phred quality score. Variant Calling format information was created via bcftools view (v 0.1.19) of the samtools suite.
Positions detected as variant in several samples (sensitive and resistance cell lines) were merged using a homemade script (bash awk).
Variant positions were annotated using the Annovar tool( [21]. Briefly, Annovar determines the type of genomic location of the variant for example exonic, intergenic, and its class (e.g. missense, nonsense), its presence in population genetics databases dbSNP [22] and 1000g database [23] and somatic variants (COSMIC, http://cancer.sanger.ac.uk/cancergenome/projects/ cosmic). For variants in coding positions the impact of the aminoacid change in the protein domains is reported via the use of Sift [24] and PolyPhen2 [25] databases.
We obtained on average 135M and 120M Paired End reads respectively after sequencing and post-alignment quality controls. As a starting point of our analysis we produced a table, based on post-alignment controlled reads, collecting all genetic variants (either polymorphic or not, single nucleotides or Indels) found in sequences in at least one of the sensitive or resistant samples, of any of the three cell lines. The table resumes the variants that present at least 1 alternative allele counts in any sample, each base having sequencing quality (phred score) of at least 20, without additional filters. The table, containing 285,999 variants (herein called Merged Variant Table) allows checking the alternative and reference allele counts at a position, in any of the sequenced samples.
We extracted from the Merged Variant table the variants detected in sensitive and resistant cells. We then included in the analysis only variants whose genotype was estimated as a nonhomozygous reference in both samples, with more than 10 reads at the variant position, localized in coding regions and of non-synonymous type. We included all variants whose positions were reported in the COSMIC database (http://cancer.sanger.ac.uk/cancergenome/projects/cosmic) and whose alternative allele frequency (AF) in the Caucasian population (CEU), as reported in 1000g, was lower than 5%. If a variant was neither reported in 1000g CEU nor dbSNP, it was included only if either SIFT or Polyphen2 HDIV functional interpretations were reported as "deleterious" for the protein. If a variant was not reported in 1000g CEU but reported in dbSNP, it was included only if both SIFT or Polyphen2 HDIV functional interpretations were reported as "deleterious" for the protein. Also, if a variation was located in a region of the genome that is annotated as Segmental Duplication [26], then it was not reported. In order to distinguish variants that were shared between sensitive and resistant cells, we tested whether the Alternative Allele Ratios (AAR) were statistically different between the two conditions (Fisher test P< = 10-3, and difference between AAR < = 0.1). Variants whose differences in AAR were not significant and superior to 0.30 in both conditions were associated to the ancestral primary tumor clone, while the other variants identified the differences between the clones sensitive and secondary resistant to RG7388.
Detection of genomic ploidy and Differential Gene Expression/IPA
RnaSeq was used to estimate the ploidy of the tumor samples, where the term ploidy defines the total number of parental alleles in large genomic regions (see S1 Methods). The RnaSeq Differential Gene Expression between two NGS samples was calculated using the methodology described in S1 Methods and implemented in R language. The selection criteria for detection of Differentially Expressed Genes were based on a Fold Change greater or equal to 2.5 and an adjusted p-value of less than or equal to 1% (Benjamini Hochberg adjustment). Data were analyzed through the use of QIAGEN's Ingenuity 1 Pathway Analysis (IPA 1 , QIAGEN Redwood City,www.qiagen.com/ingenuity) in order to predict the interactions between the genes and the potential effects on cellular pathways and networks.
Results
RG7388 activates the TP53 pathway, induces significant proliferation inhibition, cell-cycle arrest and apoptosis in MDM2-amplified and nonamplified STS As predicted by the mechanistic model of TP53 regulation, the nutlin compound RG7388 inhibited significantly the proliferation of 5 out of 11 STS cell lines with no TP53 mutations as assessed by Sanger sequencing (IC50: 2-50 nM), but not of the 6 out of 11 cell lines with TP53 mutations (Fig 1A). The most sensitive cell lines were the DDLPS cell lines IB111 and IB115 characterized by an amplification of the MDM2 gene and the extraskeletal osteosarcoma cell line IB128 characterized with no alteration of the MDM2 gene copy numbers. In agreement with the mechanism of action of nutlins, treatment of wild-type TP53 (Sanger sequencing) STS cell lines with RG7388 showed an accumulation of the TP53 protein and its targets, P21 and MDM2, as revealed by Western blotting (Fig 1B and 1C).
One of the main cellular functions of activated TP53 is blocking cell cycle progression in the G1 and G2 phase. Treatment of exponentially proliferating STS cell lines with RG7388 for 48 hours led to a dose-dependent cell cycle block in G1 and G2/M phase and a depletion of the S phase compartment (Fig 2A and 2B). One of the other main functions of activated TP53 is induction of apoptosis. Exposure of exponentially proliferating STS cell lines to RG7388 RO5503781 for 72 hours led to the induction of apoptosis in a dose-dependent manner as revealed by an increase in the percentage of TMRM-staining cells ( Fig 2C). The 3 STS cell lines that elicited the most significant apoptotic responses were IB111, IB115 and IB128.
Repeated exposures of STS cell lines sensitive to RG7388 lead to the emergence of sub-cell lines strongly resistant to the RG7388 anti-tumor activity Among the entire panel of STS cell lines tested, RG7388 displayed the highest cell growth inhibition effects in IB111, IB115 and IB128. In order to investigate mechanisms of secondary resistance to RG7388, IB111, IB115 and IB128 cells were treated with RG7388 for 3 days. The cells were then rinsed to remove RG7388, and the remaining cells were expanded in normal medium (minus RG7388). We repeated the process four times, and we obtained populations that survived 1-4 rounds of RG7388 treatment (P1 for one round-P4 for four rounds). We compared the extent to which IB111, IB115 and IB128 underwent apoptosis and cell cycle inhibition when treated for 3 days with RG7388. Results indicated that the selected populations became progressively more resistant to cell growth inhibition ( Fig 3A) to apoptosis ( Fig 3B) and cell cycle inhibition (data not shown). For instance, whereas 1 μM of RG7388 induced significant apoptosis in parental IB115 cells (80.7% apoptosis) after 3 days of treatment with RG7388, this effect was significantly reduced in IB115P2 (36.7% apoptosis) and IB111P4 (8.8% apoptosis) cells. Beside, we confirmed resistance to apoptosis at 1μM of RG7388 in resistant cell lines IB115P4 (69.4% apoptosis) and IB111P4 (28.4% apoptosis) versus parental cell lines IB115 (98.7% apoptosis) and IB111 (76.7% apoptosis) by FITC annexin-V and propidium iodide assay ( Fig 3C). STS sub-lines with secondary resistance to RG7388 display sensitivity to cytotoxic drugs similar to the parental cell lines In order to assess specificity of secondary resistance to RG7388, we compare the sensitivity profiles of IB111, IB115, IB128 and their RG7388-resistant counterparts IB111P4, IB115P4 and IB128P4 to doxorubicin and gemcitabine, two cytotoxic drugs commonly used to treat STS patients with advanced disease. Doxorubicin and gemcitabine significantly inhibited the proliferation with the same extent in the parental cell lines and the RG7388-resistant sublines (S1 Fig). This indicated that secondary resistance to RG7388 in IB111P4, IB115P4, IB128P4 did not result from a multidrug resistance phenotype and may involve specific mechanisms related to the mechanisms of action of RG7388.
Deep sequencing identifies ploidy variations involved in secondary resistance of STS cells to RG7388
Several studies have shown correlations between gains and losses of chromosomes and cancerspecific drug resistance [27][28][29]. We wanted to investigate whether regional ploidy variations were associated with secondary resistance to RG7388 in STS cells. Results are summarized in Table 1 and S2 Fig. IB111 had the greater number of allelic imbalances, followed by IB115 and IB128. Some of them were shared by both the parental and secondary resistant cell lines. However, all the secondary resistant cell lines were characterized by at least one specific ploidy variation not found in the parental ones. For instance, the secondary resistant IB111 cell line was also characterized by an unbalanced tetraploidy of the 12q13.2-qter region in which MDM2 is located. The secondary resistant IB128 cell line was also characterized by an unbalanced pentaploidy of the short arm of chromosome 17 including the TP53 gene.
Secondary resistance of STS cells to RG7388 is characterized by clonal selection of cancer gene mutations
To identify changes in the mutation profiles of secondary resistant cells, we compared the abundance of somatic mutations found in the parental and secondary resistant cell lines. For each cell line, we examined a conservative list of mutations and used Fisher's exact test to identify those specifically associated with secondary resistance (see material and methods). In all Table 1. Ploidy variations between secondary resistant and parental soft-tissue sarcoma cell lines.
only sensitive clone only resistant clone -: absence of allelic imbalance cell lines the majority of variants were common to the parental and resistant cells (Fig 4). The number of non-synonymous mutations with significant changes in normalized abundance between parental and resistant cell lines ranged from 10 to 23 for each case. These include mutations in well-known cancer genes, genes linked to drug resistance and drug metabolism, and genes not previously associated with carcinogenesis or therapy resistance ( Gene expression analysis identifies defects in apoptosome activity and autophagy induction as a mechanism of secondary resistance to RG7388 We also analyzed the sequencing data to identify differences in gene expression between secondary resistant and parental cell lines. We found 196, 105 and 370 genes with 2.5X or more fold changes (adjusted p-value < 0.01) between the parental and secondary resistant cell lines in IB111, IB115 and IB128 respectively (S2 Table). We then applied these genes to Ingenuity Pathway Analysis software (IPA 1 ,QIAGEN Redwood City, www.qiagen.com/ingenuity). The results showed that these genes were mainly enriched in proliferative, growth and movement networks ( Table 3). We focus particularly on the genes differentially expressed between the resistant and parental IB111 cells in order to identify some genetic and transcriptomic alterations that may compensate for the lack of TP53 mutations observed in the IB111 resistant cells. Strikingly, in the resistant IB111 cells, we found a significant alteration of genes involved in the regulation of apoptosis and more particularly, a down-regulation of pro-apoptotic activators such as BMF, BIM and PUMA (S3 Fig). Among them, the most strongly down-regulated was BMF with a fold change of 17. BMF also has a role as a regulator of autophagy. Since autophagy has been associated with resistance to radiation, chemotherapy, and targeted agents [30][31][32][33][34][35][36][37][38][39], we decided to investigate whether secondary resistance to RG7388 was associated with autophagy induction in STS cell lines. When autophagy is not activated, LC3 is localized homogeneously in the cytoplasm, while upon initiation of autophagy, it associates with the membrane of autophagosomes. Since an increase in LC3-II levels or GFP-LC3 vesicles can occur not only due to increased autophagosome synthesis but also due to impaired autophagosome-lysosome fusion, we assessed LC3-II levels also in the presence of chloroquine, a blocker of LC3-II degradation. Analysis of LC3-II levels by western blotting and by fluorescence microscopy allowed us to detect autophagy induction in the resistant IB111 cells but not in the IB115 and the IB128 resistant cells (Fig 5A and 5B). We therefore wondered whether blocking the induction of autophagy with chloroquine could restore sensitivity to RG7388 in IB111 resistant cells. The RG7388 IC50's were significantly reduced (divided by 6) in the secondary resistant IB111 cells when they were treated with chloroquine whereas RG7388 IC50 was almost unchanged in parental IB111 cells treated with chloroquine in the same conditions. (Fig 5C).
Discussion
The new paradigm of targeted therapies has dramatically impacted oncology practice with the discovery and development of 'personalized' anti-cancer drug medicines that produce remarkable clinical responses in a subset of patients with advanced disease. While there is rapidly growing enthusiasm for this new paradigm, there is also increasing realization that such targeted therapies suffer from the same major limitation as traditional chemotherapy-clinical benefits are of limited duration due to secondary drug resistance.
Establishing specific molecular mechanisms of resistance to classical cytotoxic drugs has been challenging due to the nonspecific nature of their antitumor mechanisms of action. On the other hand, identification of mechanisms of acquired resistance to targeted therapies is crucial since the discovery of such mechanisms can prompt the development of strategies specifically designed to overcome them. The design of new drugs to treat lung cancer patients who become resistant to EGFR-directed therapy due to the emergence of the T790M secondary mutations represents a perfect example of the importance of research efforts in the field of secondary resistance [40].
In Phase I studies, RG7388 has been associated with disease control in a significant subset of sarcoma patients. However all of them had disease progression within a median of 6 months after treatment onset suggesting the occurrence of mechanisms of secondary resistance involved in tumor progression [14]. We report here the first comprehensive analysis of genetic mechanisms involved in secondary resistance to nutlin using next-generation sequencing of isogenic pairs of nutlin-sensitive and nutlin-resistant STS cell lines. The aberrations involved in secondary resistance we described here can be considered specific to RG7388 given that no cross-resistance was noted with other drugs commonly used for the management of STS patients such as doxorubicin or gemcitabine.
By performing deep sequencing of secondary resistant and parental sensitive cell lines, we showed here that secondary resistance to nutlin was associated with several mutations whose Among them, we identified TP53 mutations in the secondary resistant cells that were not present in the parental counterpart. Previous reports investigating mechanisms of secondary resistance to nutlin in neuroblastoma and osteosarcoma cell lines suggested that exposure to nutlin induces the emergence of TP53-mutated clones [41,42]. Whether these mutations appeared de novo in the secondary resistant cells or were already present in a minority of clones was not known. Using deep sequencing, our results suggest that nutlin favors the occurrence of TP53 mutations in initially TP53 wild-type cells. Besides TP53 mutations, our data identified a subset of other gene mutations that were positively selected in secondary resistant cells, many of which have been previously involved in the control of cell growth or apoptosis. Some of them may represent 'passenger' mutations or false-positives, but some are likely to contribute to resistance to RG7388. We highlight two examples here. In IB128 resistant cells, we observed a significant increase in the abundance of an activating mutation in N-RAS in the IB128 resistant cells. Activation of the MAPK/ERK kinase (MEK)1/2 kinase pathway has been shown to impair TP53-dependent apoptosis in U20S cells treated with nutlin [43]. Recent data showed that combining the MDM2-p53 protein-protein interaction inhibitor SAR405838 and the MEK inhibitor pimasertib resulted in synergic anti-tumor activity in several K-RAS, N-RAS and B-RAF mutant tumor models [44]. The possibility that combining a MEK inhibitor with nutlin could overcome secondary resistance deserves further investigation. In IB111 resistant cells, we observed an increase in abundance of a missense mutation of the ECT2 gene, the expression of which has been shown to be normally down-regulated by Nutlin in a TP53-dependent manner [45]. Inactivation of ECT2 was shown to be sufficient to prevent cell death induced by ionizing radiation underlying its potential important role in resistance to cancer therapy resistance [46].
We have also found that secondary resistance to RG7388 was associated with few acquired ploidy variations. This observation fits with the results of a previous study showing that instable aneuploidy is associated with acquired drug resistance and that nutlin treatment of U2OS osteosarcoma cells and HCT116 colon cancer cells can result in the emergence of tetraploid cells [47]. Here we have extended these previous observations, using deep sequencing, demonstrating that the majority of allelic imbalances were shared by the parental and resistant cells. Altered intrinsic tumor sensitivity to RG7388 was associated with only a few more aberrations. Some of them targeted genomic regions including genes that are directly linked to nutlin activity such as MDM2 (12q13-15) for IB111 resistant cells or TP53 (17p) for the IB128 resistant cells. This implies that some ancestral cells that carried this complement of ploidy aberrations emerged under RG7388 selection pressure. These additional ploidy aberrations demarcate the split between fully clonal versus subclonal ploidy variations.
It was recently observed that tumor cells vary in terms of their apoptotic activity potential, and that alteration of the apoptotic machinery is a cause of chemoresistance and oncogenic transformation [48][49] By comparing gene expression of parental and secondary resistant cells, we observed in IB111 resistant cells a strong down-regulation of genes encoding for BH3-only factors such as BIM, BMF and PUMA. Interestingly, down-regulation of these genes has been associated with resistance to targeted therapies such as imatinib in bcr-abl+ leukemia cells [50] or vemurafenib in human melanoma cells [51]. Besides being a pro-apoptotic molecule, BMF is also an important regulator of autophagy [52]. Since autophagy has been associated with resistance to radiation, chemotherapy, and targeted agents [30][31][32][33][34][35][36][37][38][39], we decided to investigate whether secondary resistance to RG7388 was associated with autophagy induction in STS cell lines. Interestingly, we found that IB111 resistant cells showed pronounced autophagy whereas no autophagy was observed in IB115 and IB128 resistant cells. Moreover, we have demonstrated that autophagy inhibition significantly enhanced RG7388-induced cell death in IB111 resistant cells. We report here that down-regulation of genes encoding pro-apoptotic molecules and autophagy induction are important mechanisms involved in secondary resistance to nutlin. These observations indicate that in the presence of nutlin autophagy can promote cell adaptation and survival.
In conclusion, our results demonstrate that secondary resistance of STS cells to nutlin involves multiple and complex mechanisms and that the interplay between these mechanisms warrants extensive investigation. As such events do not appear to impact sensitivity to cytotoxic agents commonly given to STS patients, therapeutic strategies preventing their emergence should be designed. One of them could be administering nutlin for a restricted period, alternating with other drugs and then restarting, irrespective of disease progression status in the period between treatments. Table. (DOCX) S2 Table. (DOCX) | 2018-04-03T02:05:31.982Z | 2015-10-01T00:00:00.000 | {
"year": 2015,
"sha1": "fdd294c58ea9d524cca9b078cb931319c005f650",
"oa_license": "CCBY",
"oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0137794&type=printable",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "fdd294c58ea9d524cca9b078cb931319c005f650",
"s2fieldsofstudy": [
"Biology",
"Medicine"
],
"extfieldsofstudy": [
"Biology",
"Medicine"
]
} |
235458055 | pes2o/s2orc | v3-fos-license | Wigner transform and quasicrystals
Quasicrystals are tempered distributions $\mu$ which satisfy symmetric conditions on $\mu$ and $\widehat \mu$. This suggests that techniques from time-frequency analysis could possibly be useful tools in the study of such structures. In this paper we explore this direction considering quasicrystals type conditions on time-frequency representations instead of separately on the distribution and its Fourier transform. More precisely we prove that a tempered distribution $\mu$ on ${\mathbb R}^d$ whose Wigner transform, $W(\mu)$, is supported on a product of two uniformly discrete sets in ${\mathbb R}^d$ is a quasicrystal. This result is partially extended to a generalization of the Wigner transform, called matrix-Wigner transform which is defined in terms of the Wigner transform and a linear map $T$ on ${\mathbb R}^{2d}$.
Introduction
By a Fourier quasicrystal we mean a tempered distribution µ ∈ S ′ (R d ) of the form µ = λ∈Λ a λ δ λ for which µ = s∈S b s δ s , where δ ξ is the mass point at ξ, Λ and S are discrete subsets of R d . Λ and S are called respectively the support and the spectrum of µ.
The basic examples of Fourier quasicrystals follow from the Poisson summation formula. N. Lev and A. Olevskii [7] proved that if a measure µ on R d (which is assumed to be positive in the case d > 1) is a Fourier quasicrystal and both the support and the spectrum of µ are uniformly discrete (see Section 2) then there are a lattice L on R d , vectors θ j ∈ R d and trigonometric polynomials P j (1 ≤ j ≤ N) such that We refer to [5,8] for examples of quasicrystals with other structures. See also [3,10].
The Wigner transform W (µ) of a tempered distribution µ gives a description of the time-frequency content of µ. Hence, it is reasonable to study which information about the structure of µ follows from the knowledge that W (µ) is a measure supported on a uniformly discrete set.
The answer to this question does not follow directly from [7]. Let us assume that µ ∈ S ′ (R d ) is an even distribution whose Wigner transform W (µ) is a measure on R 2d supported on a uniformly discrete set. The relation W (f )(− ω 2 , x 2 ) = 2 d W (f )(x, ω) when f is an even square integrable function (see [4, 4.3,4.2]) can be appropriately extended to even tempered distributions, implying that also the Fourier transform W (µ) is a measure supported on a uniformly discrete set. However, since the Wigner distribution is almost never non negative (see [4,Theorem 4.4.1]) we cannot apply [7,Theorem 2] to obtain a precise description of W (µ).
Furthermore, from the fact that W (µ) is supported on a uniformly discrete set, we cannot even deduce that the support of µ is discrete. This is due to the interaction between the Wigner distribution and the metaplectic operators. More precisely, to every symplectic matrix A ∈ Sp(2, R) one can associate a unitary operator T A acting on L 2 (R) (denoted µ(A) in [2]) such that W T A f, T A g (z) = W (f, g)(A * z) ∀z = (x, ω) ∈ R 2 .
Our objective will be to obtain information about µ ∈ S ′ (R d ) from the fact that the Wigner transform W (µ) is a measure supported on the product of two uniformly discrete subsets of R d . The main result of the paper is as follows.
are uniformly discrete sets. Then µ and µ are measures with supports contained in A and B respectively.
The proof is contained in section 3. To facilitate the reading, we have preferred to include the complete proof in dimension d = 1 and then indicate the necessary modifications to obtain the result in arbitrary dimension.
The usefulness of time-frequency representations in the study of quasicrystals is not limited to the Wigner transform, actually in section 4 we consider a generalization of the Wigner transform, called matrix-Wigner transform which is defined in terms of the Wigner transform and a linear map T on R 2d and contains for particular choices of T most of the classic time-frequency representations. First we obtain some results that relate the support of two distributions µ, ν (or µ, ν) with that of the cross matrix-Wigner transform W T (µ, ν), thus generalizing the information contained in Theorem 1 relative to the supports. Then, we focus on the one-dimensional case and obtain a version of Theorem 1 for the cross matrix-Wigner transform. We remark moreover that a particular choice of T connects our framework to that of Lev and Olevskii [7], which essentially corresponds to the case of the Rihaczek representation.
We suppose that the link between quasicrystals and time-frequency analysis presented in this paper could lead to further developments, both in view of recent results e.g. as in [9], as well as in the direction of using specific time-frequency representations to enlighten particular features of quasicrystals structures.
Notation
We use brackets µ, g to denote the extension to S ′ (R d ) × S(R d ) of the inner product f, g = R d f (t)g(t)dt on L 2 (R d ). We will write g, µ instead of µ, g .
The cross-Wigner distribution of f, g ∈ L 2 (R d ) is It is a quadratic representation of the signal f both in time and frequency and it is covariant, which means that Here, T α and M β are the translation and modulation operators, defined by The cross Wigner distribution can be extended as a continuous map from for any φ ∈ S(R 2d ), where F 2 denotes the partial Fourier transform and T s is the symmetric coordinate change defined by This extension satisfies Moyal's formula, that is, for any functions φ, ψ ∈ S(R d ) one has (see for instance [4, 4.3.2]) A set A ⊂ R d is said to be uniformly discrete (u.d. from now on) if there is δ > 0 such that |r − s| ≥ δ whenever s, r ∈ A, s = r.
The proof of Theorem 1
The next result is well-known and will play a role in the proof of Theorem 1. We include a proof for the convenience of the reader. As usual, for a multiindex α ∈ N d 0 we denote its length by |α| = α 1 + . . . + α d . B ε stands for the ball with radius ε centered at the origin.
Then there are N ∈ N 0 and complex numbers {b α Proof. Let N ∈ N 0 and C > 0 satisfy for every f ∈ S(R d ). Since µ has u.d. support we find complex numbers {b α r : |α| ≤ N, r ∈ A} such that for every ϕ ∈ D(R d ). Let us check that the right hand side on (3) defines a tempered distribution. We take 0 < ε < inf{|r − r ′ | : r, r ′ ∈ A}. For |α| ≤ N we take ϕ α ∈ D(B ε ) such that ϕ where the constantC does not depend on α or r ∈ A. Therefore, the right hand side in (3) defines a tempered distribution. Finally, the density of D(R d ) on S(R d ) gives the conclusion. ∈ A for any r 1 , r 2 ∈ supp µ.
Proof. The inclusions supp µ ⊂ A and supp µ ⊂ B follow from standard properties of the Wigner transform. Let us now fix r 1 , r 2 ∈ supp µ and consider ν := T −r 1 µ, so that 0, r 0 := r 2 − r 1 ∈ supp ν. From the covariance property of the Wigner transform we obtain a representation r for some N ∈ N and b α r ∈ C. We aim to prove that r 0 2 ∈ A 1 , which means r 1 +r 2 2 ∈ A as desired. Proceeding by contradiction, let us assume that r 0 2 / ∈ A 1 . We choose 0 < ε < dist r 0 2 , A 1 . Since 0 ∈ supp ν we can find g ∈ D(B ε ) real-valued and satisfying ν, g = 0. For every α ∈ N N 0 with |α| ≤ N let We conclude b α r 0 = 0 for every |α| ≤ N, which is a contradiction since r 0 ∈ supp ν.
Remark. Under the hypothesis of Lemma 3 the set supp µ + supp µ 2 is u.d.. We note however that there are u.d. sets A such that A+A 2 has accumulation points. As an example in dimension d = 1 we can consider The next elementary result will be used in the proof of Theorem 1. We omit the proof.
Proof of Theorem 1 in dimension d = 1: Let us write D for the support of µ, which is contained in A. According to Lemma 2 we can put with a j r ∈ C. We now assume N ≥ 1 and show that a N r = 0 for all r ∈ D. For any real-valued functions A simple calculation gives where for every 0 ≤ j ≤ N, an application of Lemma 4 permits to conclude that the double series in the right hand side of (4) is absolutely convergent. We take By Lemma 3, r+s 2 ∈ A for any r, s ∈ D, therefore { r+s 2 : r, s ∈ D} has no accumulation points. Fix r 0 ∈ D and choose Here D(r 0 ) := {(r, s) : r, s ∈ D; r + s = 2r 0 }. Hence, if φ 2 has also compact support, where the constant C depends on the (compact) support of φ 1 ⊗ φ 2 .
Next, we fix ψ ∈ D(−ε, ε) such that ψ (n) (0) = 0 for n = 0, . . . , 2N − 1 and ψ (2N ) (0) = 1, and for each t ≥ 1, let us consider ψ t (x) := ψ(tx) and φ t 1 (x) = ψ t (x − r 0 ). Hence, as the supports of the φ t 1 's shrink as t increases, we have, for every t ≥ 1, where C depends on the support of φ 1 1 ⊗φ 2 . Taking limits as t goes to infinity we conclude that defines a tempered distribution, which coincides with the series being convergent in S ′ (R). Hence, by density, equation (6) holds for every φ 2 ∈ S(R). Now we consider φ 2 ∈ S(R) such that supp φ 2 is a so small compact set that it does not contain other points of (D − r 0 ) other than possibly 0, a fact which is possible since D ⊂ A has no accumulation points. Then equation (6) reduces to a N r 0 a N r 0 = 0, i.e. a N r 0 = 0. Proceeding by recurrence, we finally get that a j r = 0 for all r ∈ D whenever j ≥ 1. This proves that µ is a measure, as desired.
The conclusion for µ now follows from If T is obtained as the Wigner transform of a tempered distribution then a more restrictive condition on the coefficients is satisfied.
Proof. According to Theorem 1 we can apply [7, Theorems 1,3] to conclude that there are a > 0 and N ∈ N such that µ = N j=1 µ j , where each µ j is a finite linear combination of time-frequency shifts of n∈Z δ na . The conclusion follows from the properties of the cross-Wigner distribution (see for instance [4, 4.3.2(c)] in the L 2 setting) and the fact that, for Λ = aZ, where Λ * = 1 a Z is the dual lattice. Our next goal is to adapt the previous arguments to the case of arbitrary dimension d. We first need a technical lemma. For any γ ∈ N d 0 we denote . Lemma 6. Let N ≥ 1 and let us assume that the family of complex numbers for every γ ∈ N d 0 with |γ| = N. Then a γ = 0 whenever |γ| = N.
Proof. We proceed by induction on d. For d = 1 the lemma is obvious since condition (7) means |a γ | 2 = 0 for γ = N. Let us now assume that the lemma holds in dimension d − 1 (d ≥ 2) for every N ≥ 1 and let a γ : γ ∈ N d 0 ⊂ C be given such that condition (7) is satisfied. Now we proceed by induction on k to prove that a γ = 0 whenever |γ| = N and at least one component of γ equals k. Let us first assume that k = 0, so γ has at most d − 1 non-null components. Without loss of generality, we can assume γ d = 0. For every Our hypothesis on the validity of the lemma in dimension d − 1 permits to conclude that a γ = 0. Let us now assume that a γ = 0 whenever |γ| = N and at least one component of γ is less than k (k ≥ 1) and let us fix γ ∈ N d 0 such that |γ| = N and at least one component equals k. We can assume and we put δ = (δ ′ , k). Then, for every δ ′ ∈ N d−1 We observe that |δ| = N and condition (α, β) ∈ F d δ implies α d + β d = 2k, hence we can assume α d = β d = k. Otherwise some of the coefficients α d , β d is less than k, from where it follows a α a β = 0. Our hypothesis on the validity of the lemma in dimension d − 1 (applied to N − k instead of N) permits to conclude that b δ ′ = 0 for any δ ′ ∈ N d−1 0 such that |δ ′ | = N − k. In particular, a γ = b γ ′ = 0. Here γ ′ = (γ 1 , . . . , γ d−1 ). The proof is finished.
Proof of Theorem 1 for arbitrary d: As in the case d = 1 we only need to check the statement concerning µ. Let us write S µ for the support of µ, which is contained in A. According to Lemma 2 we put µ = r∈Sµ |α|≤N a α r δ (α) r , with a α r ∈ C. Our aim is to show that a α r = 0 for all r ∈ S µ and |α| ≥ 1. Recall that { r+s 2 : r, s ∈ S µ } has no accumulation points. Fix r 0 ∈ S µ and γ ∈ N d 0 with |γ| = N and choose a smooth function φ 1 supported on a sufficiently small neighborhood of r 0 and satisfying φ (2γ) 1 (r 0 ) = 0 while φ (α) 1 (r 0 ) = 0 for each α = 2γ. Then, for every φ 2 ∈ S(R d ) (real valued), we have where D(r 0 ) := {(r, s) : r, s ∈ S µ , r + s = 2r 0 }. Proceeding as in the case d = 1 we obtain An application of Lemma 6 gives a γ r = 0 whenever |γ| = N and r ∈ S µ . Now a recurrence argument shows that a γ r = 0 for every r ∈ S µ and γ ∈ N d 0 with |γ| ≥ 1. The conclusion follows.
The matrix-Wigner transform
A natural generalization of the hypothesis of Theorem 1 is the case where different input functions or distributions µ and ν are considered for the Wigner transform. This situation is more involved but still some results can be obtained. Furthermore we shall consider a generalization of the Wigner transform, called matrix-Wigner transform, see [1], which, using a composition with linear maps, will yield a unifying framework connecting our results to those of [7], [8]. We need some preliminaries.
We begin by recalling the following notations. For a set E ∈ R 2d = R d x × R d ω we indicate the projections on the x and ω-coordinates as: When it is clear from the context we shall however omit the subscripts x and ω in R d x and R d ω . We recall without proof the following well-known property which will be used later.
We need now to discuss some properties concerning supports.
Proof. Let I be an interval in R d (i.e. the cartesian product of d open intervals of R). Preliminarly we observe that a distribution Ψ ∈ S ′ (R 2d ) vanishes on the strip I × R d if and only if the restriction of F 2 Ψ to the same strip also vanishes. Indeed if the distribution Ψ vanishes on I × R d , then for every which means that also F 2 Ψ vanishes on I × R d because F 2 : S(R 2d ) −→ S(R 2d ) is a bijection which preserves the inclusion of the supports in I × R d . The converse is analogous.
Let us suppose now that Π 1 supp Ψ is a u.d. set and x 0 / ∈ Π 1 supp Ψ, then there exists an open interval I ⊂ R d containing x 0 such that Ψ vanishes on I × R d . From the first part of this proof we know that also F 2 Ψ vanishes on I × R d and therefore x 0 / ∈ Π 1 supp F 2 Ψ. This proves the inclusion which in particular shows that also Π 1 supp F 2 Ψ is u.d. The opposite inclusion Π 1 supp Ψ ⊆ Π 1 supp F 2 Ψ, is proved by the same argument and we have therefore Finally the case where Π 1 supp F 2 Ψ is a u.d. set can be proved in analogous way.
Remark. We observe that the hypothesis of u.d.ness of either Π 1 supp Ψ or Π 1 supp F 2 Ψ in the previous proposition is essential. Consider for example Then 0 belongs to Π 1 supp F 2 Ψ but not to Π 1 supp Ψ.
In order to treat the matrix-Wigner transform, to be defined later, we need to consider some properties of linear maps in connection with supports. Proposition 9. Let T : R n −→ R n be a linear bijective map. Then for every Ψ ∈ S ′ (R n ) we have supp (Ψ • T ) = T −1 (supp Ψ).
Proof. As T is a diffeomorphism, by definition of the composition with a distribution, we have for φ ∈ S(R n ): As Ψ ∈ S ′ (R n ) ⊂ D ′ (R n ) we have that x ∈ supp Ψ implies that for every neighborhood U of the origin there exists a function φ ∈ C ∞ c (x + U) such that Ψ, φ = 0.
We have then Remark. Ψ • T is the pull-back T * Ψ of Ψ, here for analogy with the case where Ψ is a function we shall write for short T (Ψ).
In particular for n = 2d and µ, ν ∈ S ′ (R d ), we have that x 1 ∈ supp µ, where T s is the symmetric coordinate change operator defined in (2).
More generally let T : (x, y) ∈ R 2d −→ (u, v) = T (x, y) ∈ R 2d be an invertible linear transformation, with abuse of notation we shall still indicate with T the matrix associated with the transformation and the change of coordinate operator given by with natural extension to distributions. We have then the following: Proposition 10. Suppose that µ, ν ∈ S ′ (R 2d ), and T : R 2d −→ R 2d is a linear invertible transformation. Let us write the inverse matrix of T as: with A, B, C, D submatrices of dimension d × d. Then the following hold: Proof. We prove (i), the others are analogous. By contradiction suppose that supp µ is not u.d.. Then for every ǫ > 0 there exist x, y ∈ supp µ such that 0 < x − y < ǫ. Let z ∈ supp ν, then (x, z) and (y, z) belong to supp (µ ⊗ ν).
As mentioned above, following [1], we introduce next a Matrix-Wigner transform which is a natural generalization of the Wigner transform W (µ, ν) = F 2 (T s (µ ⊗ ν)) where the change of coordinates T s have been replaced by a general bijective linear map T . This sesquilinear transform turns out to be a quite comprehensive tool including most of the basic time-frequency representations, we refer to [1] for details and properties.
Definition 11. Let T and A, B, C, D be as before, then the Matrix-Wigner transform of µ, ν ∈ S ′ (R d ) is defined as: As usual we shall write W T (µ) for W T (µ, µ).
In connection with our previous discussion we have the following property.
In particular for the classical Wigner transform we have where Id is the identity, therefore, as all subdeterminants are non zero, we have that Π 1 supp W (µ, ν) u.d. implies that both supp µ and supp ν are u.d.
Our next aim is to obtain an analogous of Proposition 12 for the case of the Fourier transforms of µ and ν. Before giving the statement we need the following remark on block matrices.
Proposition 13. If Π 2 supp W T (µ, ν) is a u.d. set and T is an invertible matrix satisfying as in Proposition 12, then
Now, from (13) we have and so, by (15) and (16) the thesis follows by an application of Proposition 12 to Π 1 supp W R ( µ, ν).
From Propositions 12 and 13 we immediately have the following corollary.
The basic connection between our setting and the hypothesis assumed in [7] is the following remark.
The previous results, Propositions 12, 13 and Corollary 14, where we can obtain u.d.ness of the supports of signals from that of the their matrix-Wigner transform, include many classical time-frequency transforms such as the STFT and the Rihaczek transforms. As an example we consider the Ambiguity function.
Example 15. The Ambiguity function is defined as for µ, ν ∈ S(R d ), which is generalized to . Then Π 1 supp A(µ, ν) u.d. implies supp µ and supp ν u.d. In fact, it suffices to apply Proposition 12.
Our next aim is to give a version of Theorem 1 for the case of W T (µ, ν) for a 2 × 2 matrix T , giving then a general picture of the situation for the case µ, ν ∈ S ′ (R).
Let T : R 2 → R 2 be an invertible linear transformation with inverse We assume from now on that ab = 0.
We show next that for every fixed j, l, k, m the double sum over r, s in (17) is absolutely convergent. We recall the following facts. There exists α > 0 such that, setting for short t = 1 + |t|, The functions φ are in S(R), therefore, for every β > 0 and suitable constant C > 0, we have Furthermore we remark that linear bijections on R 2 are bi-continuous and therefore preserve u.d. sets, so that T −1 (S µ × S ν ) is u.d. Supposing then that T has matrix and indicating by C generic (possibly different) suitable constants, we have: where β can be chosen arbitrary large and T −1 (S µ × S ν ) is a u.d. set, and therefore the last sum is convergent, i.e. (17) is absolutely convergent. Let us take now 0 < ε < δ(A) := inf{|r − r ′ | : r, r ′ ∈ A, r = r ′ }.
For r ∈ S we denote by s(r) the unique s ∈ S ν such that (r, s) ∈ D. Observe that there are constants α, β such that α = 0 and cr + ds = αr + β whenever (r, s) ∈ D. From (18) and the fact that S is a u.d. set it follows that r∈S a N r b M s(r) δ αr+β defines a tempered distribution. Condition (19) means that the Fourier transform of that distribution vanishes. Consequently a N r b M s(r) = 0 ∀r ∈ S.
Since s(r 0 ) = s 0 and b M s 0 = 0 we conclude a N r 0 = 0. Since r 0 ∈ S µ is arbitrary we conclude a N r = 0 for every r ∈ S µ . Proceeding by recurrence on the order of µ we finally get that a j r = 0 for all r ∈ S µ and 0 ≤ j ≤ N, from where it follows µ = 0. This contradiction proves that M = 0 and ν is a measure. The same argument but swapping the role of the distributions proves that also µ is a measure. Hence, where, as ab = 0, by direct calculation or from (8), (9), we haveãb = 0, and we can then apply Theorem 16 to conclude.
Corollary 18. Let µ, ν ∈ S ′ (R) \ {0} satisfy W T (µ, ν) = (r,s)∈A×B c r,s δ (r,s) where A, B are u.d. sets. Then, there are a, b > 0 such that µ is a finite linear combination of time-frequency shifts of n∈Z δ na and ν is a finite linear combination of time-frequency shifts of n∈Z δ nb . | 2021-06-18T01:15:54.659Z | 2021-06-17T00:00:00.000 | {
"year": 2021,
"sha1": "4b16da39c91e2f29a28aa98de612c27ed1ad43ed",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "4b16da39c91e2f29a28aa98de612c27ed1ad43ed",
"s2fieldsofstudy": [
"Mathematics"
],
"extfieldsofstudy": [
"Mathematics"
]
} |
134681647 | pes2o/s2orc | v3-fos-license | Eco-Etnoturism : An Option for Sustainable Environmental Management
The purpose was to develop a theoretical knowledge to the scientific knowledge of the sustainable Eco-Ethnotourism activity tending to minimize the problems of environmental contamination of the study area and the preservation of the Wayuu culture, based on the empirical dynamics of the community. This study was conducted in the Integrated Management District (DMI) delta of the Ranchería-Riohacha River. The results are made up of a project management and planning tool that helps to strengthen conservation and sustainable management actions in the area. The methodology was based on the qualitative approach of [1], [2], using the ethnographic method of [3] and [4], the observation technique, the survey and the semi-structured interview, the environmental and cultural characteristics of the study area, to make a participative diagnosis, definition of the sustainability indicators of the eco-ethnotourism activity in a participative way with the inhabitants, the potential ecoethnotouristic index was established,. It was concluded that there is illegal occupation of the river bank, constructions without planning, pressure of resources, especially of mangroves; there is inter-institutional disarticulation and community, there is no continuity of the plans and programs of the public institutions in charge of the management and preservation of this ecosystem. Keyword Delta of the Ranchería River, environmental sustainability, environmental management, ethnotourism, eco-tourism.
II. PERSPECTIVE OF THE SITUATION
Colombia is one of the Latin American countries that has a great tourist interest due to its variety of landscapes, large area of coastal areas, cultures, festivities, traditions and folklore, among other attractions.This country biodiverse in its Caribbean region strategically, has one of the departments that in recent years has shown an accelerated tourist growth; this is how in La Guajira, there has been a boom in the development of tourism activities in natural settings, with the largest attraction being the indigenous reservation territory; where these activities are developed without any planning, without control of the negative impacts that are caused to the environment, natural resources and ethnic cultures.
These statements are also corroborated by [7], when they argue that "mangroves are also subject to anthropogenic tensioners, such as deforestation, grazing, the accumulation of solid, organic and inorganic waste, as well as wastewater" (2014, p 40). ), aspects evidenced in the observations and the field work carried out in the DMI of the Ranchería river delta.
III. TOPICAL CONCEPTIONS OF RESEARCH A. Ethnotourism
In the DMI of the Ranchería delta, the interrelation of the environment with ethnicity and tourism is clearly evident, there being a natural conjugation between the ethnic, the Eco and their relationship with the same environment, it should be noted that when interacting with the population in a direct perspective, this interrelation between nature and community can be perceived in the same space.In ethnoturism, tourists seek to satisfy that need to know those ancestral cultures, within their own paradigm in the search for historical vestiges [8].Similarly, the development of ethnoturism in a specific territorial space is recognized as an indigenous settlement that according to Ojeda is defined as "the ancestral territory that is constituted in the vital space where the spiritual, social, cultural, economic and political life of the Be Wayuu "(2012, p. 15).
The community settled in the delta of the Ranchería River, is directly related to the biotic environment (we are a single world) that has a richness in florist resources and is the means of subsistence for these communities, nevertheless, it is evident, a relevant ecosystem importance, ecotourism and sociocultural, which is closely related to the socioeconomic aspect of the communities in this area; in view of this repercussion found in the research links, the opportunity for the generation of income in the settlers can be missed, through the hunting of some minor species, fishing and the use of the forest resource.B. Ecotourism [9] Describe ecotourism as a favorite specialty for tourists, seeking the enjoyment of the natural to appreciate and in some cases study the natural attractions and / or cultural manifestation, this tourism modality seeks to promote the conservation of ecosystems and the preservation of the cultures of the tourist destination, with a low environmental and cultural impact, which promotes an active and socioeconomically beneficial involvement.
Ecotourism represents a strategy of benefits for the communities belonging to the population area under study, being, not only an allowed economic activity, but a conservation strategy for ecosystems and / or protected areas with ecotourism vocation, [10]; therefore, the ecotourism exercise implies a cooperation between all the actors involved.
With this conceptual manifestation, a sustainable tourism is sought that allows an enjoyment of the natural goods, knowledge and preservation of the cultural traditions, generating satisfaction and social welfare in the tourist destination [11]; in the ecosystem related to the case of the delta DMI of the Ranchería River, there is a relict of mangroves, which, when viewed from a vulnerability perspective due to its high environmental impact, requires immediate preservation strategies, since the life of this ecosystem is favourable an active and socioeconomically beneficial relationship for local populations, as stated in [12].
C. The sustainability of eco-ethnotourism
In recent times, sustainable development is a topic of interest and study in all fields; economic, political, environmental and social; to what [13] it defines as a methodological theoretical foundation that gives rise to studies, economic projects, programs and government plans to promote development.
In this sense, the environmental sustainability of ecosystems is a process of continuous improvement, flexible collective construction and active participation of the various actors involved to enjoy the environmental goods of the present, conserving resources for the enjoyment of future generations reason that eco-ethnotourism is considered an economic activity closely related to the sustainable development of peoples and the conservation and preservation of natural resources, biodiversity, the quality of environmental services, cultural and ethnic identity.
D. Integrated management district (DMI)
According to [14], an Integrated Management District (DMI) is the geographic space, in which landscapes and ecosystems maintain their composition and function, although their structure has been modified, whose associated natural and cultural values are within reach of the human population to use for their sustainable use, preservation, restoration, knowledge and enjoyment.Similarly, in [15] the District of Integrated Management of Renewable Natural Resources is described as a space of the biosphere, which due to environmental or socioeconomic factors, is delimited so that within the criteria of sustainable development it is ordered, planned and regulate the use and management of renewable natural resources and the economic activities that take place there.
IV. METHODOLOGY This research has a qualitative approach that, according to [4], is selected when: "It seeks to understand the perspective of the participants (individuals or small groups of people being investigated) about the phenomena that surround them, to deepen their experiences, perspectives, opinions and meanings, that is, the way in which the participants they perceive their reality subjectively.It is also advisable to select the qualitative approach when the subject of the study has been little explored, or has not been done research in a specific social group".
In the same way, the qualitative approach tries to make an approximation of social situations to explore, describe and understand them; based on the knowledge of the people involved in them, this means that individuals interact with other members of their social context, understanding their meaning and knowledge of their reality [1], Thus, taking into account the above, the research sought to analyze facts within the focal community of the Ranchería river delta.
The data collection techniques that were used, focused on investigating the environmental and cultural situation in the communities settled in the area of influence of the DMI of the Ranchería river delta, which focused on the environment -sociocultural factors and which they frame the eco-ethnotourism system of the area.
According to [16], when the research touches the social sphere for obtaining information, the main techniques for gathering qualitative data are observation, the survey, the interview, the focus groups, the collection of documents, materials and the histories of life, which were considered and carried out in the present investigation.Likewise, since the research is focused on an area of social influence, a review of several ethnographic studies was carried out to find patterns, and thus describe the social world as it is continually being constructed, emerging as objective, ordered, intelligible and familiar reality.
The ethnographic perspective, in this type of approach, is relevant in the observations that were made, its study, provided reconstructions in the cultural context, the values, and beliefs of the participants of the Wayuu ethnic group; the researcher was placed in a situation and disposition that allowed to observe the behaviors of the community within the environmental context and tried by all possible means to obtain structures of meanings that testify the social behaviors, in relation to environmental factors that can affect the environment of the community and eco-ethnotourism environmental management of the DMI delta del río Ranchería.
The techniques used to obtain the information were: direct observation of the Villa Fátima neighborhood and the indigenous communities of Pasito and Cangrejito, the visits to the different ranches, the population of the community and specific sites of the delta, surveys conducted at random with 200 families in the community, interviews applied to Wayuu leaders and external experts, talks to target groups of young people and adults, documentary review, life stories and workshops held in concert with the community.
To obtain the information, the survey is divided into three important components that extract specific characteristics of the population, such as: 1) The family component, the information of the nucleus, its location and the level of study; 2) The socioeconomic component, commercial activity for the generation of income; 3) The environmental component, based on the use of ecosystem resources, the characteristics of the dwelling, the behavior of public services and the management of solid and liquid waste.
E. Data collection
The interview technique was applied to collect information from community leaders and local experts, who were chosen for their knowledge of the culture and environmental dynamics within the community of the Ranchería river delta and external experts, researchers, representatives of public and private entities that are related to the environment; undoubtedly, all the actors with which a margin of direct intervention was established, possessed continuous action on issues of culture and / or tourism and were of a significant nature for the contribution and collection of the different points of view related to the Topics raised from eco-ethnotourism as an option for sustainable environmental management.
At the same time, we observed the need to conduct a semi-structured interview to determine in advance what is the relevant information that is required; semi-structured interviews are based on a guide of issues or questions and the interviewer is free to introduce additional questions to clarify concepts or obtain more information on desired topics [16].One of the characteristics of the semi-structured interview is that open questions are asked in it, giving an opportunity to receive more nuances of the answer; it allows to interweave subjects, but it requires a great attention on the part of the investigator to be able to channel the subjects and an attitude of active listening.
Three interview formats were applied, each according to the characteristics of the interviewees; one to local experts, external experts and tourism entrepreneurs.
Another technique, which was implemented, was based on carrying out workshops, such as the Problem Tree, where the causes and effects of the problem presented in the Delta of Ranchería area were identified [17]; the fundamental basis for the development of the participatory diagnosis, where children, youth and adults were clearly outlined the main environmental problems that the inhabitants of the Ranchería delta have in their communities.
The workshops planned, designed, prepared, organized and carried out in the different meetings, were structured as spaces to share and build knowledge scenarios, taking into account the participation of those who with their contributions generated an exchange of experiences between the community and the project of research developed in the study area.
F. Population and sample
The universe population is La Guajira, which has immersed the enigmatic of its landscapes and Wayuu culture where you can take advantage of all the ecological and the ethnographic potential to develop a sustainable activity; the population under study is constituted by the delta of the Ranchería river with three populations equidistant from the Villa Fátima neighborhood and the indigenous communities El Pasito and Cangrejito.The population sample, are the 200 randomly surveyed families that allowed to collect reliable and direct information to develop an analysis on the population under study.
For the determination of the population, different instruments were used for the data collection, among the most transcendental, the observation, the recording, the anecdotal record, the field diary, the notebook of annotations, the photographic record, the chronogram of activities and the work log.
V. RESULTS
In the research, important information was obtained on all the issues raised.Based on this, the analytical and interpretative management was executed in order to develop the research report based on the data obtained, under the following correlated aspects: It was possible to gather great information provided by the inhabitants of the delta of the Ranchería River during the visits, meetings and interviews conducted; under the bower they could be evidenced by direct contact with the actors involved, a number of significant aspects and great value of the culture, so they were urged towards a compilation of all the information accumulated through the techniques and instruments used throughout the investigation.
The collection of information led to the realization of some concrete and effective activities among which are: The visits, which were made by initial agreement as a measure of cultural respect with some leaders, in order to explain the importance of research.Previous exploratory visits were made, where the environmental, natural and population characteristics found during the first hearing were identified at the start of the experiential investigation.After establishing the first contacts with the traditional authority and obtaining the endorsement of the leaders, the territorial entry to the study area was continued in order to initiate the investigation, establishing the dates and times for the meetings and the respective workshops with the community.
After having the endorsement of the traditional authority and the way leaders, exploratory visits were made where the different situations on the environmental and social aspects were observed, such as: poor vegetation, unpaved streets in the neighborhood of Villa Fátima, grooves of water accumulated after the rains that become decomposed by organic matter, flooding in patios and free spaces taken as a soccer field, housing in various materials, wood, blocks, bricks, mud or clay, garbage, especially plastic bags in the surroundings, empty patios and mountains, bridleways, and mountains of rubble.
A. Expected impact
The impacts generated by the tourist activity are given by the interaction between the tourist, the ecosystem and the local culture, when the load capacity is exceeded and becomes a nuisance for both the visitor and the target population.In the last four decades, tourism has grown rapidly and the number of tourist destinations to visit has increased [18], increasing with them the impacts on the environment, economic and social-cultural aspects.
The reflection made on what was found is evident in the research process framed in eco-ethnotourism in the integrated delta management district of the Ranchería River, La Guajira: an option of sustainable environmental management, has characteristics that led to a learning and appropriation of knowledge for the description and analysis of the information obtained according to the nature of the research, where a process was conceived and conceived with the which was determined the type of research to be used, having a clear focus that led to the design of the investigation, throwing today a series of concrete results as a result of the whole investigative process.
In this reflection, it can be seen that based on the results obtained, a clear and evident corollary was generated, which gave rise to a participatory diagnosis, which served as a study material to define the sustainability indicators and thus determine the eco-ethnotouristic potential of the delta of the ranchería river.
B. Environmental impact
Conventional tourism can generate negative impacts such as the loss or destruction of ecosystems due to the massive presence of visitors and the need of tourism entrepreneurs to expand areas for hotels, and infrastructure, the need to use more resources such as water for the daily hotel service or entertainment infrastructure such as swimming pools, the consumption of food and beverages that generates a large amount of solid and liquid waste; also the means of transport used.
The generation of social and environmental impact, whether positive or negative will depend on the quality of planning that is implemented, which should include the multiple dimensions included in tourism development.The ideal is that the impact is positive, the beneficial effects of tourism for culture and the environment arise when considering that tourism is based on the use and conservation of natural and cultural heritage [19]; meaning for the tourist destination improvements in economic, sociocultural and environmental conditions.In the last four decades, tourism has grown rapidly and tourist destinations have diversified [18], increasing the impact on the environment.
With specialized tourism, such as ecotourism, tourists are attracted by the enjoyment of nature as birds, rivers, lush vegetation; where the environmental impact tends to be positive because the tourist delights in the natural beauty of the place; however, with the increase in the number of tourists, ecosystems tend to be vulnerable due to the continuous presence of visitors and the use of resources, that is, their greater margin of affectation is evident when the carrying capacity of the population is exceeded area (tourist destination).The way to create tourism by the producer is carried out quite often through personal development ("Encounter of two traditions") [20].
C. Socioeconomic impact
The socioeconomic impact is evident from the development of tourism, it begins to generate a pool of offers in services aimed at the satisfaction and enjoyment of tourists, in order to boost investment and promote job creation.In the delta DMI of the Ranchería River, the eco-ethnotourism activity would generate an economic impact, both in the tour operators and in the inhabitants of the area; this economic scenario would represent a margin of profitability for the way people, since they increase the sales margin of their handicrafts, offer their services such as attention in the rancherías, ranchería evenings where they make cultural samples of their traditions, myths and legends, with its lodging, guiding, hiking and traditional gastronomy are provided.
It is important to point out that with the development of the eco-tourism activity significant income would be generated for the continuous improvement of the local economy, which in turn would contribute to the increase of district revenues.
D. Sociocultural impact
To talk about social-cultural impacts, it is important to review a series of variables such as the number of days the visitor visits, their route through the protected area, the ecological fragility of the system and the type and behavior of the users; in order to be able to determine, the sociocultural impact generated by tourism in the areas determined as a tourist destination, [21].Table 1 shows the impacts of eco-ethnotourism.VI.FINAL THOUGHTS After carrying out the present investigation, it was possible to affirm by way of conclusion that there is an illegal occupation of the Ranchería river bank in the delta, which constitutes a socio-environmental and security problem caused by displacement, as well as problems of drug addiction, to which the district administration must implement strategies for the recovery of the eco-tourism path.
Likewise, it was observed that there is an inter-institutional disarticulation and community, each entity has its projects, its programs, there is no continuity of the plans and programs of the public institutions in charge of the management, nor the preservation of species together with the ancestral community settled in the delta del Rio Ranchería.
Finally, it was observed that the communities settled in the delta of the Ranchería River: Villa Fátima, Cangrejito and El Pasito have an organization and articulation of programs, government plans to carry out environmental and cultural processes conducive to the improvement of the quality of life of the settlers
TABLE I .
Impacts of Eco-ethnotourism | 2019-02-17T22:12:54.467Z | 2018-12-31T00:00:00.000 | {
"year": 2018,
"sha1": "0448a72cb76e51bd4ab6173ba125bd9d00390578",
"oa_license": "CCBY",
"oa_url": "http://www.enggjournals.com/ijet/docs/IJET18-10-06-202.pdf",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "0448a72cb76e51bd4ab6173ba125bd9d00390578",
"s2fieldsofstudy": [
"Economics"
],
"extfieldsofstudy": [
"Business"
]
} |
258505960 | pes2o/s2orc | v3-fos-license | Application of single-cell RNA sequencing on human testicular samples: a comprehensive review
So far there has been no comprehensive review using systematic literature search strategies to show the application of single-cell RNA sequencing (scRNA-seq) in the human testis of the whole life cycle (from embryos to aging males). Here, we summarized the application of scRNA-seq analyses on various human testicular biological samples. A systematic search was conducted in PubMed and Gene Expression Omnibus (GEO), focusing on English researches published after 2009. Articles related to GEO data-series were also retrieved in PubMed or BioRxiv. 81 full-length studies were finally included in the review. ScRNA-seq has been widely used on different human testicular samples with various library strategies, and new cell subtypes such as State 0 spermatogonial stem cells (SSC) and stage_a/b/c Sertoli cells (SC) were identified. For the development of normal testes, scRNA-seq-based evidence showed dynamic transcriptional changes of both germ cells and somatic cells from embryos to adults. And dysregulated metabolic signaling or hedgehog signaling were revealed by scRNA-seq in aged SC or Leydig cells (LC), respectively. For infertile males, scRNA-seq studies revealed profound changes of testes, such as the increased proportion of immature SC/LC of Klinefelter syndrome, the somatic immaturity and altered germline autophagy of patients with non-obstructive azoospermia, and the repressed differentiation of SSC in trans-females receiving testosterone inhibition therapy. Besides, the re-analyzing of public scRNA-seq data made further discoveries such as the potential vulnerability of testicular SARS-CoV-2 infection, and both evolutionary conservatism and divergence among species. ScRNA-seq analyses would unveil mechanisms of testes' development and changes so as to help developing novel treatments for male infertility.
Background
Human testis is the core organ of the male reproductive system [1], and acts as the cornerstone of male fertility since it is the place of spermatogenesis [2]. On the one hand, the testis is lionized by developmental biologists because it's the start of sex determination [3] and the kindergarten of male germline cells [4]. Meanwhile, human testis is also the priority of andrologists because testicular failure is the leading cause of male infertility [4]. Therefore, scientists have delved into both normal [5,6] and abnormal testes [7][8][9], and detected mechanisms behind them especially those related to male infertility [10,11]. Although numerous approaches have been used to learn about human testis including cell experiments [12], animal experiments [13] or even new methods like testicular organoid [14], most of the above-mentioned methodologies are indirect. It is no doubt that the most ideal model to study human testes is the testis itself. Therefore, human testicular biological sample-based studies could be important sources of knowledge of changes in both normal and impaired testes.
Ivyspring International Publisher
As early as the 1910s, human sample-based studies had started to give direct evidence of testicular diseases [15]. Yet until the 1970s, most of those studies were restricted to anatomy and morphology/histology [16][17][18][19] perhaps because at that time nonspecific histological staining was the most popular way to study human testes. Then with the development of some more specific methods such as immunohistochemistry staining (IHC), researches on the human testis reached protein levels [20][21][22]. A huge breakthrough in this field was the application of microarray and high-throughput RNA-seq on human testicular samples [23][24][25][26], which gave scientists a transcriptional view of both the normal testis's development and the pathogenesis of spermatogenic dysfunctions [27,28]. Nevertheless, both microarray and traditional RNA-seq are based on bulk tissue, for which the testicular sample is considered as a whole, so we could only obtain the mainstream of transcriptional changes in the sample but ignore the heterogeneity of different cells. To solve this problem, Tang et al. introduced single-cell RNA sequencing (scRNA-seq) [29], which could detect the transcriptome at a single-cell resolution. The technology was created on blastomere and oocytes and soon applied to human testicular cells [30] and then to unsorted testicular tissues [31]. So far, there have been plenty of studies using human testicular scRNA-seq data and several reviews partially or roughly touched upon this topic (human testicular scRNA-seq analysis) [32][33][34][35][36][37][38][39]. Nevertheless, some of these reviews didn't fully focus on "human testis" and/or "scRNA-seq" (for example, some included much more about female gonad or animal testis [32,34,40] and some focused much on techniques other than scRNA-seq [33,37,38]), which led to limited paragraphs and blurry outline on this topic. Other reviews were almost restricted to germ cells [35,36] or after-birth gonads [39], which gave an incomplete depiction of this topic. Most importantly, none of these articles was with systematic literature searching strategies, which might result in a biased interpretation. Using systematic literature searching, this comprehensive review plans to fully summarize the new findings offered by scRNA-seq analyses of human testicular biological samples, concerning both testicular cells and unsorted testicular tissues, both prenatal and postnatal male gonads, both underage and adult testes, both young and aged testes, both normal spermatogenic and impaired spermatogenic patients, and both original sequencing data and reusage of published data. We hope this review will give scientists detailed and all-round information on testicular development, spermatogenesis as well as testicular pathogenesis in the single-cell resolution, and humbly suggest the future directions in this field.
Search methods
The searches were conducted on 29 Aug 2022. First, a systematic search in PubMed database was conducted. Based on the fact that this review mainly focused on two aspects, including "scRNA" and "human testicular samples", we used a matched searching strategy to link these two aspects. In detail, for the former aspect, the terms "single-cell" (term A) or "scRNA" (term B) were selected. And each of these two terms was combined with one of the following testis-related (the latter aspect-related) search terms, including "testis" (term 1), "spermatogenesis" (term 2), "Klinefelter Syndrome" (term 3), "azoospermia" (term 4), "Y chromosome microdeletion" (term 5), "cryptozoospermia" (term 6), "oligozoospermia" (term 7), "asthenospermia" (term 8), "teratospermia" (term 9), "orchitis" (term 10), "cryptorchidism" (term 11) and "male and gonad" (term 12). In order to minimize the chance of omitting related studies especially preprint studies, we do an extra search on GEO dataset (https://www.ncbi.nlm.nih.gov/gds) using four combination of advanced search terms including "((testis) AND scRNA) AND Homo sapiens [Organism]", "((testis) AND single-cell) AND Homo sapiens[Organism]", "(((male) AND gonad) AND scRNA) AND Homo sapiens[Organism]" and "(((male) AND gonad) AND single-cell) AND Homo sapiens [Organism]", which also referred to the data related to both focused aspects. For filtering, the "series" of Entry type and "Expression profiling by high throughput sequencing" of Study type were also chosen. For each identified GEO series, its related articles were obtained from GEO or bioRxiv (https://www.biorxiv.org/) if there was no published article at the searching time.
Selection criteria
The inclusion criteria were (1) studies with single-cell RNA sequencing applied on human testicular biological samples (including testicular tissues, cells and prenatal male gonads) and/or (2) studies with re-analyses of published/public scRNAseq data of human testicular biological samples (not including pure literature review of published articles).The exclusion criteria were (1) articles not in English; (2) articles published before 2009 (the year when scRNA-seq technology was created [41]); (3) non-research articles (such as pure review/case report/guideline/protocol/technical note/correction/comment/chapter of book/letter) without scRNA-seq analysis on human testis; (4) full-length articles not available; (5) tumor studies without scRNA-seq data analyzing or not relevant to human testis; (6) studies based on sequencing methods other than scRNA-seq (e.g. single-cell small RNA-seq or spatial transcriptomics sequencing); (7) pure animal or botanic studies; (8) human-related studies but with no scRNA-seq analysis on human testicular samples.
Application of scRNA-seq on human testicular samples: an overview
The workflow of literature searching and selections was in Figure 1. The searches returned about 1300 records (609 after de-duplication). By further filtering, 81 full-length articles were enrolled into the final analysis. There were 33 articles containing original scRNA-seq data (at least from one donor) on human testicular samples, while the rest 48 articles were based on the re-analysis of previously published data (no matter whether the data was previously published by the same team). Since the year of 2015, Guo et al. for the first time do scRNA-seq on human fetal testicular cells [30], followed by adult testicular cells sequenced by two studies in 2017 [42,43]. Interestingly, although those two adult studies were from different teams, both of them focused on spermatogonia of one accord. From 2018, numerous studies have done scRNA-seq on human testicular samples regarding both prenatal and postnatal, both normal and abnormal testes. Table 1 summarized articles that have originally published scRNA-seq data on human testicular samples. We could conclude that most of the studies were conducted by scientists from the USA and China. As for library techniques, most of the studies used the 10×Genomics platform (https://www.10xgenomics.com/) to carry out their works. While other studies chose techniques/ plateforms such as BD Rhapsody [44], (modified) smart-seq2 [45][46][47], STRT-seq [48], Microwell-seq [49], Drop-seq [50], inDrop [51], Slingleron GEXSCOPE ® [52,53] and some earlier platform such as Fluidigm C1 [54] and Tang method [29,41]. In some studies [55,56], two or more methods simultaneously appeared for cell capturing. So far, no comparative study has focused on which technique is more suitable for scRNA-seq of human testicular samples. Nevertheless, some reviews and comparative studies based on scRNA-seq of other tissues have gradually revealed that the technological differences of scRNA-seq, including both the wet-lab parts (e.g., cell capturing/sequencing/barcoding) [44,[57][58][59] and the dry-lab parts (e.g., data-processing/batch-effect correction/dimension reduction/quality control) [60][61][62][63][64] might have profound impact on the final results. Especially, unavoidable technological batch effects will be generated using data generated from different sequencing platforms [65]. Although the data will go through batch effects correction before downstream analysis [60], we still recommend that researchers should choose one suitable techniques and consistent data-analyzing methods to do their testicular scRNA-seq works. If you need to do integrated analyses and have to combine your own scRNA-seq data with public scRNA-seq data, we suggest that you should choose the public data that used the same cell-capturing techniques as your own data, to minimize the non-biological batch effects.
In terms of sample types, one-third of the studies choose prenatal male gonads (the youngest sample was from the gonad from a 4-week male embryo). For adult testes (the oldest sample was from a 76-year-old male for now), most of the studies included samples with complete/full/normal spermatogenesis. But the sources of these normal spermatogenic samples varied. Some collected deceased healthy male testicular samples [31] to serve as the normal groups, which were ideal controls. For many institutes, obstructive azoospermia (OA) samples with full spermatogenesis confirmed by histopathological diagnosis, including congenital bilateral absence of the vas deferens [66], post-vasectomy [67], and physical obstruction of the vas deferens [66] were considered as workarounds for normal/control groups. It is noteworthy that in the original manuscripts of Hermann et al. [68], Tan et al. [69] and Sohni et al [67], some of the adult testicular biopsies were obtained from patients receiving vasectomy reversal, and the samples in Xia et al.'s work [70] were from testicular sperm extraction (TESE) surgeries of patients with obstructive infertility (they were also post-vasectomy patients after confirming with the first author). Interestingly, none of these patients was clearly claimed as OA patients, although they all had obvious obstructive etiology. Frankly speaking, these post-vasectomy patients were extremely unlikely to have sperm in ejaculate (only if some rare situations happened such as duplication of vas deferens or natural recanalization [71]). But in order to be strict and consistent, in the current review, all patients with obstructive etiology (no matter they were diagnosed with obstructive azoospermia or not) were also unitedly abbreviated as obstruction (OB). Besides, testes removed from patients receiving transsexual operation might also be an option for sample collection [43,72]. Additionally, normal testicular samples at age 17 were collected by both Zhao's and Guo's studies, but were assigned to different groups (adult for Guo's and underage for Zhao's studies) [31,55], maybe due to the different definition of adult (biological adult versus legal adult). Till now, the number of scRNA-seq testicular samples in each study has been limited, maybe due to the high price of scRNA-seq commercial service and limited sample donors. To partially solve this dilemma, some studies chose to do combined analyses of their own scRNA-seq data and public scRNA data or bulk RNA-seq data. And most of the studies in Table 1 did extra experiments (such as histological experiments, animal or cell experiments) to validate the findings from scRNA-seq analyses. Additionally, some studies were conducted on both testes and ovaries [22,73], or on testes from both human and animals [68], in order to do comparisons between genders or among species. Garcia-Alonso et al did integrated analyses by the combination of testicular scRNA-seq with spatial transcriptomics [74], which is a novel direction. Nevertheless, so far there has been no guideline for scRNA-seq studies on the human testis. By reviewing all related studies, we summarized a standard guideline(workflow) of testicular scRNA-seq-based studies (Figure 2). We believe that with the wider application of scRNA-seq techniques, there will be more scRNA-seq-based studies of testicular samples published, with more comprehensive analyzing methods used.
Cell identifications by scRNA-seq on human testicular samples
The fundament of scRNA-seq analysis is cell identification. Apart from a few studies based on isolated cells by fluorescence activated cell sorting (FACS) or magnetic activated cell sorting (MACS), the majority of studies in Table 1 were based on testicular tissues. For adult testes, the representatively identified cell types in scRNA-seq analyses were summarized in Table 2. ScRNA-seq-identified testicular cells could be roughly divided into three classifications, including spermatogenic cells (germ cells), non-immune somatic cells and immune cells [66]. The widely used and some newly identified markers for cell identification were also summarized ( Table 2). Importantly, because of the different study purposes and different sample selections, the types and subtypes of testicular cells in different studies were slightly variant. For germ cells, spermatogonia (SPG), spermatocyte (SPC) and spermatid could be broadly identified in a full-spermatogenic sample if the focus of the study is not spermatogenic cells [55]. Nevertheless, a more detailed classification of these three germ cells could be done based on stage-specific markers. Here we recommended a moderate classification of germ cells used by Nie et al., in which SPG could be further divided into spermatogonial stem cell (SSC) (also called undifferentiated SPG) and differentiating SPG (diff_SPG), SPC into early primary SPC and late primary SPC, spermatid into round and elongating/elongated spermatid [75]. One can adjust the subdivision of each germ cell type according to your research focus. For example, if you were not interested in the development of SPC, you could easily keep SPC as an unclassified type but classifying SPG and spermatid into more detailed subtypes [56]. If the whole spermatogenesis is your attention, you might choose a more aggressive strategy for germ cell classification [66,76]. [79], but five of them were first reported by Chitiashvili et al. [178] (these two studies were from the same research team). So, in the item of Guo's study, we only summarized the information of the rest two cases (w12 and M5) which were reported for the first time. Apart from germ cells, it is well known that somatic cells also played important roles in spermatogenesis. Such somatic cells could be further divided into non-immune somatic cells and testicular immune cells. For non-immune somatic cells, Sertoli cells (SC) were one of the most important cell types in spermatogenesis, as they could not only form blood-testis barrier but also directly support germ cell survival [55]. And Leydig cells (LC) could affect spermatogenesis by producing testosterone [75]. Although the roles of testicular immune cells in spermatogenesis have not been fully revealed, more evidence indicated that they could interact with germ cells and take part in spermatogenesis [77].
The rules of cell identification in germ cells also worked when annotating somatic cells. For example, although the rest studies all keep testicular macrophages as a whole, Mahyari et al. chose to further divided them into M1 and M2 macrophages [78]. Nevertheless, the above-mentioned all classifications were "well-known" classification of adult testicular cells. One important function of scRNA-seq is to identify novel subtypes of cells. For example, traditionally SC were considered as one type of cells, but Zhao et al. and Guo et al. found three subtypes of SC in the human testis [55,72]. Similar findings also occurred in LC [78]. Owing to the lack of unified understanding, sometimes similar novel subtypes of certain cells were independently reported by different teams with different nomenclatures. For instance, an important subtype of prenatal gonadal cells which could bifurcate into the developmental path of both SC and LC were reported by two different teams with different names [79,80]. Meanwhile, samples with different pathological status might deeply influence the identified cell types and numbers. A case in point is testicular immune cells. In the microenvironment of testes with spermatogenic dysfunction, an increasing number of immune cells were observed [66]. Additionally, the main type of immune cells in full spermatogenic testes was macrophages, while a large number of mast cells were identified in testes from non-obstructive azoospermia (NOA) patients, especially in Sertoli cell-only syndrome (SCOS) [55,56].
ScRNA-seq demonstrated the development and changes of testicular cells from all ages
Testicular development from prenatal to postnatal period
Overview
Before the year 2015 in which the first prenatal testicular sample was scRNA-seqed, the understanding of the development of different types of testicular cells in prenatal testes was limited to the changes of cell number, morphology and distribution [81]. So far, with over ten articles published their own scRNA-seq data on human prenatal testicular samples, we already have an outline of both germline and somatic development at single-cell transcriptional level.
Germline transition
For germline development pre-and postnatally, Guo et al pointed out that prenatal germs cells generally have two main forms of transcriptional profiling, including a form of primordial germ cells (PGC) (expressing PGC markers such as TFAP2C, KIT, NANOG, POUF51 and SOX17) and form of the most naïve SSC (expressing earliest SSC markers such as PIWIL4, EGR4, MSL3 and TSPAN33) [79]. PGC form mainly started at the embryotic time and then gradually stopped meiosis and inhibited pluripotency and at transcriptional level became cells with high similarity to the early status of adult SSC [79]. The latter form (named State f0) [79] strode across human birth and kept a quite similar transcriptional status until the adult time (named State 0 in adult) [31]. During the germline transition from PGC to (f)State 0, pathways related to gonad and stem cell development were downregulated while signaling related to transcription/homeobox were enhanced [79]. Similarly, Wang et al. found a transition from mitotic (mainly existing before 15 weeks) to meiotic arrest (mainly existing after 15 weeks) male fetal germ cells (FGC) [82], which was in line with Guo's [79] and Li's results [73], indicating that the downward status of meiosis (proliferative-quiescent-arrested) was the symbol of male fetal germ cell development. Garcia-Alonso et al. called the latter form of FGC prespermatogonia and also showed the activation of EGR4 in this stage [74]. Interestingly, Wang et al. also reported an extra subtype of FGC called SPARC + POU5F1 + FGC, which might keep the ability to migrate locally within the fetal testis [82].
Somatic development
As for prenatal somatic cells, most of the researches focused on SC and LC due to the isogenesis of these two cells. The source of testicular somatic cells was embryotic coelomic epithelium (CE) [83]. By re-clustering of scRNA-seqed CE cells, Cheng et al. found two subclusters of CE, of which one was NR0B1 + STAR + NR5A1 + but GATA4 − LHX9 − (annotated as adrenogenic CE), while the other was GATA4 + LHX9 + HOXA9 + HOXD9 + but NR5A1 − (known as posterior/gonadogenic CE) [74,80]. The former subcluster will develop into the adrenal gland while the latter goes to the testis. Noteworthily, gonadogenic CE will step into a sort of bipotent somatic progenitors (symbolized by the expression of NR5A1), which had been recognized by several studies [79,80]. In Wang et al.'s study, a subpopulation of KRT19 + somatic cells were identified mainly in 6-to 8-week testes, which also resembled these somatic progenitors [82]. Starting from 7-8 weeks, these somatic progenitors bifurcated into two different developmental paths, one of which is embryonic Sertoli progenitors (marked by SRY, similar to the "early supporting gonadal cells" in Garcia-Alonso et al.'s work [74]) and then became fetal SC [79]. The rest path, on the contrary, led to embryonic/fetal interstitial progenitors (marked by ARX and TCF21, also called Leydig precursor cells by Li et al. [73] and called somatic progenitor by Wang et al. [84]), and finally part of the fetal interstitial progenitors became fetal Leydig cells (marked by HSD3B2 and CYP17A1, also called differentiated LC by Li et al. [73]) [79,80]. During the development of prenatal interstitial/ Leydig lineage, the enriched molecular pathways altered from "extracellular matrix"/ "cell adhesion" to "steroid biosynthesis". Interestingly, at the perinatal stage, fetal SC successfully passed the birth (with genes related to translation and respiratory chain enhanced, while genes related to the endoplasmic reticulum and steroid biosynthesis reduced) and became postnatal SC including stage_a and b (discussed in more detail below) [55,79], while fetal Leydig cells disappeared in the postnatal time [79]. In fact, it was CYP17A1-negative fetal interstitial cells that might finally pass the birth and finally become the common progenitors of adult LC and PTM [79]. Regarding EC, although some studies proved the early appearance of EC in prenatal male gonad [73,84], so far there have been few studies that focused on EC development. Like the previous murine studies [85,86], scRNA-seq data on human the fetal testis also revealed two distinct types of macrophages in the fetal testis, of which one is named SIGLEC15 + fetal testicular macrophages (osteoclastlike) and the other is TREM2 + fetal testicular macrophages (microglia-like) [74]. Further study showed that SIGLEC15 + macrophages were usually located in the interstitium and could interact with EC/mesenchymal cells and play roles in mesonephric endothelial cell migration, while TREM2 + macrophages (located in the testis cord) interacted with fetal SC and FGC and worked as an immune regulator [74]. Unfortunately, the transition from fetal testicular macrophages to postnatal testicular macrophages remained unexplored. Further scRNA-seq-based studies might have the opportunity to reveal the relationship between fetal and infant testicular macrophages.
After-birth testicular maturation from underage to adult
Overview
Among all scRNA-seq studies on testicular samples, several studies did original scRNA-seq on underage testicular samples [31,55,67,72,79,87]. And in some other studies [78,88], previously published underage testicular scRNA-seq data were re-analyzed. Those studies pointed out that huge heterogeneity existed in both spermatogenic cells and somatic cells during the maturation of testes after birth.
Spermatogenic maturation
For analyzing the after-birth development of spermatogenic cells, pseudotime trajectory analysis was widely used. In fact, pseudotime trajectory analysis on spermatogenic cells could be divided into three distinct types. First, if we only focused on the normal testis of an adult man, spermatogenesis itself, starting from SSCs and ending at mature spermatids, was a continuous process running along a time axis of maturation [89]. Hence, the complete spermatogenesis process could be analyzed using pseudotime trajectory in scRNA-seq data. In 2018, Guo et al. did this task and showed there was only one no-branch pseudotime trajectory in spermatogenesis of healthy male [31], which started with SSCs, passed diff_SPGs, early and late SPCs, then round and elongated spermatids and ended up with testicular sperms (also known as mature spermatids [90]). The second type of pseudotime trajectory analysis on spermatogenic cells is to extract and re-cluster a certain type of germ cells and do pseudotime trajectory analysis on this type of germ cells. For instance, Sohni et al. did pseudotime trajectory analysis on SPGs and found a minor group of SPGs called "transitional cells" which was the linkage between SSCs and diff_SPGs during SPGs differentiation [67]. Besides, Guo et al. did pseudotime analysis on SPGs and reported a five-stage development of SPG, starting from state 0 (markered by PIWIL4, similar to the above-mentioned state f0 in the fetal testis), then state 1 (ID4, GFRA1 expressed), followed by state 2-3 (marked by KIT and MKI67) and finally reaching state 4 (marked by STRA8) [31,42]. By comparing these 5 stages with the results from Wang et al. [76] and Sohni et al [67]., we could conclude that state 0 and 1 were two types of SSC (0 was more naïve and fetal-like, and resembled the "SSC-1" in Sohni's work, while 1 resembled "SSC-2" in Sohni's work), and state 2-3 were diff_SPG while state 4 were diffed_SPG. Third, pseudotime analysis could also be used to study the maturation of germ cells from underage to adult. In fact, the discovery of state 0 was based on pseudotime analysis on the combination of infant and adult spermatogenic cells, from which they found this special state of SPGs (which was close to the infant germ cell cluster) in adult [31]. Sohni et al. further re-clustered infant germ cell cluster into three subclusters, including primordial germ cells (PGCs)-like cells (PGCLs) (marked by POU5F1 and NANOG), Pre-SPG1 (marked by DOCK8 and SERINC2) and Pre-SPG2 (marked by COL1A2 and TIMP2) [67,91]. And further pseutotime trajectory showed that the developmental order of fetal/infant/adult naïve germ cells was PGCL→Pre-SPG1→Pre-SPG2→adult SSC subcluster (SSC-1) [67]. During puberty, the spermatogenic status changed obviously with age, and pseutotime trajectory analysis of a series of testicular scRNA-seq data from 1-to 25-years old males showed the dynamic changes of germ cells at different ages [72]. In detail, during the infant and child period (1 and 7 years old), there were only SSC in testes (also known as undifferentiated SPG or SPG state 0 and 1, which all referred to the naïve status of SPG) [72], and this finding was further validated by scRNA-seq data on testes from 2-and 5-years old males [55]. At year 11, both diff_SPG and SPC appeared, while spermatid appeared at 13 years old samples [72]. Zhao et al. further showed that diff_SPG had already existed in the 8-year-old testis but SPC was still absent at that time [55]. The first maturity of spermatogenesis was observed at 14 years old, by which time the proportion of each type of spermatogenic cells was almost the same as adult testes [72]. In terms of molecular signaling, Activin pathway was downregulated in SSC but was obviously upregulated during spermatogonia differentiation [72]. In a word, scRNA-seq together with pseutotime trajectory analysis is very powerful in analyzing the after-birth maturation of spermatogenic cells, especially in the identification of new cell subclusters and finding new stages of cells.
Somatic maturation
For somatic cells, the development of Sertoli cells has been fully analyzed by scRNA-seq analysis. By integratedly analyzing testicular scRNA-seq data of five young adult with normal spermatogenesis (age from 23 to 31 years), three children (age 2, 5 and 8 years) as well as two adolescents (age 11 and 17 years), Zhao et al. revealed that the normal maturation of testicular Sertoli cells consisted of three stages (termed as stage_a, b and c) [55]. In detail, stage_a (marked by JUM) was the majority of Sertoli cells at infant time (including the neonatal testis) and gradually decreased with the age increasing. While stage_c (marked by DEFB119) appeared at the time of puberty (after 11 years old) and became the dominant type of Sertoli cells in adult people. Besides, they also found that the Wnt signaling pathway should be the key of regulating Sertoli cell maturation [55]. Such findings are outstanding because it focused on the normal development of postnatal somatic cells and revealed not only the phenotype development of Sertoli cells but their related molecular mechanisms as well. Similarly, Guo et al. also found two immature stages of SC, named immature 1# (marked by PDPN) and 2# (with higher expression of TOMM7 and ATP5E), which finally merged into mature SC after 11 years old [72]. By comparing the differentially expressed genes (DEGs) of stage_a/b SC (from Zhao's study) versus those of immature 1#/2# SC (from Guo's study), we concluded that stage_a was similar to immature 1# while stage_b was closer to immature 2#. A huge difference between these two studies is that Guo's study showed 1# and 2# as two independent paths of immature SC, which will merge into mature SC after 11 years old. Zhao's study, however, showed a potential sequence of stage_a, b and c, (from a to be to c) by both analyzing the proportion of these three cells at different ages and analyzing the functions of these three cells. In detail, they found that before the appearance of mature SC, the ratio of SC a to SC b decreased with age, suggesting there might be a transition from a to b [55]. But was this change of ratio caused by "transition from a to b" or by "proliferation of b/death of a"? Enrichment analysis and cycle-specific gene analysis further showed that stage a SC were the most proliferative and had more features as progenitors, which supported the hypothesis that stage a, b and c were three consecutive development stage of SC [55]. Since Zhao's study was more comprehensive in terms of SC, hereinafter we chose the nomenclature of SC from Zhao's study [55].
Moreover, Mahyari et al. found that in adult human testes, the maturation of Leydig cells also consisted of three types including progenitor LCs (PLCs), immature LCs (ILCs) and mature LCs (MLCs) [78]. And pseudotime analysis confirmed that the maturation of LCs started from PLCs, followed by ILCs and ended at MLCs [78]. However, in underage testes, the LCs seemed to be a unique type which was totally separated from the above-mentioned three adult LC types [78]. In line with that finding, using pseudotime analysis on both infant and adult LCs, another study also found that infant LCs and adult LCs appeared at two different ends of the trajectory, indicating that the progress of LCs from underage to adult might be a discontinuous alteration [67]. Further, Guo et al. for the first time showed that the so-called "infant or child LCs" might actually be the common progenitors of LC and PTM [72], which will finally develop into both LC and PTM during puberty [72]. As for PTMs, although the developmental trend was similar to LCs (because they were homologous), pseudotime analysis of scRNA-seq data identified a special subtype of adult PTMs which was clustered together with neonatal PTMs and showed a neonatal-like gene expression characteristic [67]. For the molecular signaling involved in the maturation of LC and PTM, Guo et al. found that the common progenitors expressed high level of genes associated to "transcription" signaling, and the enriched pathways went to cytoskeleton/ cell adhesion signaling (if the progenitor differentiated into PTM) or secretion signaling (when it differentiated into LC) [72]. So far, the development and maturation of other types of somatic cells, especially testicular ECs and immune cells, have not been fully revealed. Future scRNA-seq studies should pay special attention to the differences of these rare somatic cells between underage and adult human testes.
Based on the published studies, we can conclude that during the maturation of after-birth testes, huge differences existed in both somatic cells and germ cells between underage and adult testis in terms of cell subtypes, expressed markers and regulating mechanisms.
Testicular senescence from adult to aged males
Overview
Unlike the deep studies on differences between underage and adult spermatogenesis, few scRNA-seq researches focused on the changes in aging testes. Until last May, Nie et al. published the world's first testicular scRNA-seq study based on the comparison between young and old adult men [75]. In this study, by integrated analysis of 4 young adults (age 17-22 years) and 8 older adults (age 62-76 years), this team found huge age-related changes in both spermatogenic cells and somatic cells.
Germ cell senescence
In terms of germ cells, they found that spermatogenic cells were affected by age in an inconsistent way (some old people still have complete spermatogenesis, while others showed impaired spermatogenesis), and in old males with obviously impaired spermatogenesis, the decreased number of SSC might be one of the causes of age-related spermatogenic dysfunction [75]. Mechanistically, using Cellchat analysis, a powerful function of scRNA-seq data to detect cell-cell communications [92], Nie et al found that Activin and KIT signaling pathways weakened between SCs and SSC/diff_ SPGs, which might lead to damaging effects on germ cells [75].
Somatic cell senescence
For somatic cells, scRNA-seq analyses revealed pan-somatic cell changes, including inflammation changes of SCs, loss and testosterone decline of LCs, decreased contraction and integrity of PTMs-based tubular walls and the senility of both ECs and Macrophages [75]. Mechanistically, Nie found that for SCs, metabolic signaling was significantly altered in aged males, while in senescent LCs, key components of the Hedgehog pathway were decreased [75].
Young infertile testes showing onsets of senescent process
For some infertile males who were still young, Alfano et al. found that their testes showed early onsets of senescent process [88]. Such abnormal senescence was observed in almost all somatic cells of idiopathic germ cell aplasia, and was characterized by the upregulation of proteins of innate immunity, the overexpression of UBA52/NACA and decreased pathways of amino acids metabolism [88]. This phenomenon suggested that the senescence of human testes was not only associated to the exact age of male, but affected by their pathological status as well.
Transitions of SC/LC/PTM in the context of the whole life cycle
So far, there has been limited scRNA-seq study that focused on the all-age process of the origin, development, maturation and senescence of a certain type of testicular cells by analyzing prenatal, infant, children, adolescent, adult and aged testicular samples. Due to the profound studies on the transitions of SC/LC/PTM and the low inconsistency among these studies, we tried to summarize the development and changes of these three main somatic cells in the context of the whole life cycle (from embryotic to aged) in Figure 3, based on the current evidence retrieved from human testicular scRNA-seq studies [55,67,[72][73][74][75][78][79][80]. Future studies should try to integrate the transcriptional phenotypes of testicular cells from all periods of the life cycle, and do a coherent study on the other types of testicular cells, especially testicular immune cells, to which less attention was paid.
Obstructive azoospermia
Obstructive azoospermia was clinically diagnosed by semen tests (no sperm in ejaculates) and the existence of excurrent duct obstruction [93]. According to traditional opinion, OA patients were with rather conserved spermatogenesis [94]. There-fore, in traditional studies, due to the difficulty in obtaining completely healthy and fertile male testicular samples [95,96], researchers would like to choose OA samples with confirmed normal histopathologic features as the controls [97][98][99][100]. In testicular scRNA-seq studies, this workaround was also widely used [55,66,76]. Although we could conclude that these OA control samples were with complete or full spermatogenesis because of the histopathologic confirmation, could these OA testes strictly be called as "normal" testes? In the past, due to the limitation of research methods, this question was not fully answered. By the combination of healthy and OA testicular scRNA-seq data, Chen et al. revealed an obvious loss of late spermatids and an increase of early SPCs in OA testes compared with testes from the healthy man [56]. Mechanistically, the changes were found in both germ cells and somatic cells. First, SPCs of OA testes might suffer from the weakening meiotic process and enhanced apoptosis, which eventually leading to the loss of spermatids [56]. As for somatic cells, PTMs might be the matchmaker between obstruction-caused pressure and the damage of spermatogenesis [56]. This study partially solved the unsettled issue and suggested that although the spermatogenesis of OA testes was functionally and pathologically normal, it was not technically normal. Meanwhile, this study only used scRNA-seq data of post-epididymitis and postvasectomy OA testes. Is there a difference in testicular function among OA patients with various causes (e.g., congenital bilateral absence of the vas deferens and post-vasectomy)? Whether the changes of OA testes found by scRNA-seq are obstructive time-dependent? These questions remain to be answered.
Ejaculation dysfunction
Ejaculation dysfunction contains a wide range of situations in which the normal ejaculation was disordered or even failed [101]. Infertility caused by ejaculatory dysfunction or retrograde ejaculation could also be accompanied by testes with complete/full spermatogenesis. Using testicular scRNA-seq data of a patient with retrograde ejaculation (post-colectomy), Mahyari et al. pointed out that the characteristics of cell score residuals (a statistic reflecting testicular cells' pattern) of the retrograde ejaculation sample and adult control samples were alike, which indicating that the patient with retrograde ejaculation had undisturbed spermatogenic status and could be considered as another type of control samples [78]. Nevertheless, the above-mentioned scRNA-seq study did not especially focus on ejaculation dysfunction, so the evaluation of testicular changes of ejaculation dysfunction was not that comprehensive. Further scRNA-seq studies should take several situations of ejaculation dysfunction, including "total anejaculation", "situational anejaculation" [102] and "retrograde ejaculation", into consideration, in order to profoundly reveal the potential alterations caused by ejaculation dysfunction.
ScRNA-seq revealed cellular and molecular disorders in patients with spermatogenic dysfunctions
Non-obstructive azoospermia Spermatogenic dysfunction was characterized by the impairment of intratesticular spermatogenesis, and NOA is the most severe condition of it [103]. NOA was clinically diagnosed with no sperm in ejaculates along with evidence reflecting/causing the impairment of testicular spermatogenesis (e.g., elevated FSH or history of chemotherapy) [104]. Maybe due to the relatively easy availability of NOA testicular samples (a large number of NOA patients need testicular sperm retrieval surgeries [105,106]), so far there have been seven studies that did their own scRNA-seq on at least one NOA testicular sample ( Table 3). It should be noted that in this part ( Table 3) we only discussed NOA samples without evidence of chromosome abnormality (NOA patients with chromosome abnormality will be discussed separately). Among these studies, over 70% (five) were conducted by Chinese research teams. Interestingly, due to the heterogeneity of NOA, the pathological status of scRNA-seq NOA samples used in each study varied across studies. Besides, maybe due to the decrease or lack of germ cells in the NOA testis, most of these studies focused on changes in somatic cells, especially SCs and LCs. Notably, most of the scRNA-seq studies did analyses on NOA samples in the following two ways: 1) identifying DEGs (in a certain type of cells) between control and NOA testes, and 2) finding new/major subtypes of a certain cell in NOA. For instance, Zhao et al. found that in iNOA patients, Sertoli cells were basically in stage a and stage b (mentioned above), and they revealed iNOA SCs showed different gene expressions pattern compared with normal testes [55], which were independently confirmed by Wang et al. [76] and Chen et al. [56]. In terms of somatic cells, two main somatic cells of the human testis, SCs and LCs, showed similar status in NOA patients, both were immature but with enhanced proliferative/divisive ability [55,78], which were in line with the fact that immature somatic cells have proliferative ability [107,108]. Alfano et al. further found that in SCOS (the severe pathology of NOA), the transcriptional and phenotypic features of LC were more like pre-pubertal LC, indicating the repressed maturation of LC in NOA [88]. As for germ cells in the NOA sample (if it had germ cells), Cst3 mediated autophagy might play an important role in the development of NOA by harming SSC maintenance (the ability of SSC to keep its population by self-renewal [109]), and knockdown of this gene might cause the impairment of SSC which might further lead to spermatogenic dysfunctions [110]. Unfortunately, maybe due to the reduction or absence of germ cells in the NOA testis, most of these studies merely focused on somatic cells, which made the changes of residual spermatogenic cells in NOA patients largely unknown. Future studies could especially select samples with spermatogenic arrest or hypospermatogenesis to do scRNA-seq, which will give a chance to solve this problem.
Cryptozoospermia
In the area of Oligo-astheno-teratozoospermia (OAT), so far there have been few scRNA-seq data reported. This phenomenon is easy to understand because the scRNA-seq data was based on testicular tissue which is usually dissected in sperm retrieval surgeries such as TESE or microdissection TESE (mTESE) [111], and for OAT patients, many of them could use ejaculated sperm to do in vitro fertilization (IVF). Only severe OAT patients might need to undergo testicular sperm retrieval surgeries [112]. Cryptozoospermia, which was characterized by the appearance of isolated sperms in semen under the microscopic examination after centrifuging [106], could be considered as the transitional situation between severe oligozoospermia and azoospermia. The previous study pointed out that for cryptozoospermia patients, surgically retrieved testicular sperm was better than ejaculated sperm when doing intracytoplasmic sperm injection (ICSI) [113], which could also provide more convenience for obtaining testicular samples to do scRNA-seq. Using three cryptozoospermia testicular samples versus three controls (OA samples), Persio et al. profoundly revealed the extensive changes of testicular microenvironment in cryptozoospermia patients [66]. Deeper changes occurred in spermatogenic cells, including the reduction of germ cells such as pachytene spermatocytes and beyond, the changes of transcriptional profiles of all-type spermatogenic cells, an increase of PIWIL4 positive spermatogonia (representing the most naïve germ cells [114], just like the above-mentioned state 0 SPG or SSC-1), and a decrease of Adark SPGs (reserved SSCs [115]) and UTF1 positive SPGs (a SSC/undifferentiated SPG marker [116]) [66]. More importantly, an enhanced and prolonged expression of EGR4 was noticed in SPGs (also resembled the SSC-1 cells in Sohni's study [67]), which could act as a transcriptional factor and inhibit the expression of UTF1 in part of SPGs from cryptozoospermia patients [66]. Besides, scRNA-seq data also revealed perturbed somatic cells in cryptozoospermia, including an obvious increase of PTMs (including fibrotic PTMs) and macrophages, a deep transcriptional change of PTMs, an elevated ratio of MUSTN1 postitive (a pan-musculoskeletal cell marker [117]) blood vessels and the gathering of CD3 T cells around blood vessels [66]. Besides, two crosstalk networks between SPGs and adjacent microenvironments including "SPGs-FGFR1/3-FGF2pachytene/diplotene SPCs" and "SPGs-ACKR2-CCL2/3/4/5/3L1/14-EC/immune cells/perivascular cells" were detected in cryptozoospermia rather than normal patients [66].
Trans-females after gender-affirming hormone therapy
The testis removed from patients receiving sexual reassignment surgery was another important tissue source for scRNA-seq studies [43,72]. Of course, if a physically healthy patient did not receive any medical treatment before transsexual surgery, the removed testis was definitely an ideal sample of the "normal" group. Unfortunately, many transgender individuals received gender-affirming hormone therapy [118], which might profoundly impair the spermatogenesis of their testes [119]. Guo et al. drew the single-cell transcriptional profiles of two trans-females who received long-term (over a year) testosterone antagonist and estradiol treatment before transsexual surgeries [72], and found huge disorders in both germ cells and somatic cells. In terms of germ cells, although they still existed in both donors, the proportions of advanced spermatogenic cells including spermatids, SPC and diff_SPG were extremely low or even absent, and such changes stopped at the level of undifferentiated SPG (or SSC, or state 0 & 1 SPG as mentioned above) [72], indicating that testosterone inhibition might have less impact on SSC itself at the transcriptional level but could stop its differentiation. We found that such status was similar to the testis of prepuberty boys, for whom diff_SPG and more developed germ cells did not appear before 8 years old [55]. As for somatic cells, SC was found to be immature with higher expression of AMH and HES1 in these two patients, and the transcriptional profiles of SC were more like pubertal males (stage_b SC was predominant during this period [55]) rather than adults [72].
Klinefelter syndrome
Klinefelter syndrome (KS) makes up the most common part of male chromosome disorders [120], which is also widely studied by scRNA-seq studies [55,78,121]. Importantly, usually the studies on KS patients were considered as a part of works on NOA patients because of the popular azoospermic phenotype [122]. Nevertheless, due to the special karyotype of KS (one or more extra X [123]), the changes and relative mechanisms in KS testes were distinguishable from the above-mentioned iNOA patients [124]. Interestingly, for scRNA-seq data of KS testicular samples with the presence of germ cells, there seemed to be no huge changes at the transcriptional level of KS germ cells compared with normal germ cells [121]. But such a conclusion was drawn indirectly based on the co-clustering of KS germ cells with normal germ cells, rather than differential expression analysis (because there was only one KS sample sequenced with only 39 germ cells identified in Laurentino's study [121]). Unlike the other type of male infertility (in which the scientist could deeply analyze germ cell changes), Laurentino's study was almost the "best" work of KS germ cells because most of the scRNA-seq samples on KS from other teams lacked spermatogenic cells and showed a SCOS or even tubular atrophy pattern [55,78]. Although previous study pointed out that spermatogenic cells were not entirely lost among all KS patients [125], we found it was hard to accurately get one sample with germ cells to do scRNA-seq. According to our experience, for KS patients, if the sample was coming from a random biopsy, we can't guarantee there were germ cells in it. And if the sample was coming from TESE or mTESE surgery (which we could judge the spermatogenic status by the form of tubules during surgeries), although it was possible for us to find a sample that contains germ cells or even spermatozoa, most of such "good" samples should be used for finding sperms and ICSI (since such good samples were quite rare and precious), so it was unlikely to have an extra "good" sample for scientists to do further studies including scRNA-seq. By combining scRNA-seq with bulk RNA-seq data and IHC staining, Winge et al. found that such germ cell loss in KS patients might start as early as in the fetal testis [126]. And another study found that KIF2C might be a germ-cell-related gene to regulate spermatogenic development in KS testes [127]. Frankly, based on the current evidence from scRNA-seq data, it's too early to answer the question that what was the transcriptional change in the residual germ cells of KS patients.
As for KS somatic cells, scRNA-seq data has offered us abundant findings which we didn't know before. SCs were found to be the most affected somatic cells in KS testes [126]. In detail, KS SCs were almost stage_b SCs and showed altered energy metabolism characteristics such as enhanced expression level of glycolysis and oxidative phosphorylation-related genes but rather low triglyceride metabolism level [55]. Huge transcriptional changes were observed in KS SCs including the upregulation of immune-related genes such as B2M [128] and MIF [129], the decreasing of sex hormone -regulating gene GNRH1 and the increasing expression of X-linked genes [55,78]. Notably, Mahyari et al. found a special subpopulation of KS SCs which showed lacking expression of XIST, which could partially explain the X-linked genes' enhancement in SCs [78]. The involvement of immune activation in KS testes found by Zhao et al. [55] were also validated in peripheral blood of KS patients [130], indicating the profound impact of immune-related pathogenic factors in the pathogenesis of KS. Similarly, the proportion of immature LCs was also increased in KS LC population and more LCs were found to undergo division/development (with higher expression level of RNA processing/splicing-regulating genes such as CCNL1, DDX17, PNISR and FUS) [78]. Interestingly, when comparing KS-NOA somatic cells with iNOA somatic cells, both shared and heterogenous features were found. For example, the expression of β-catenin protein level was both significantly enhanced in KS-NOA and iNOA SCs, indicating a shared pattern of activated Wnt/β-catenin pathways in both type of NOA compared with normal adults [55]. While SERPINE1, a tissue plasminogen blocker, was highly expressed in KS immature LCs but not in iNOA LCs, indicating the heterogenous pathogenesis of KS-NOA compared to NOA with other etiology [78].
Y chromosome microdeletion
The testicular evidence got from single-cell studies on patients with Y chromosome microdeletion was limited and less than the evidence from bulk studies [131,132], especially for AZFc microdeletion (the most common form of Y chromosome microdeletion in male infertility [133]). So far, there has been only one scRNA-seq testicular sample with azoospermia factor a (AZFa) deletion reported [55]. Unfortunately, this sample was with SCOS pathology, so we can't gather enough data on germ cells of patients with Y chromosome microdeletion. Pseudotime trajectory analysis in the AZFa testis found its SCs were in the early part of stage_c (a form of mature SCs), together with upregulation of MIF and DEFB119 (two important factors in immunoregulation), suggesting the involvement of immune regulating pathways in the etiology of AZFa SCs [55]. Considering the limited research on this field, we suggest future scRNA-seq studies should focus on the pathogenesis of azoospermia factor b (AZFb) and azoospermia factor c (AZFc) testes (especially those AZFc testes with germ cells), so as to profoundly reveal the special changes in patients with Y chromosome microdeletions.
Re-analysis of published scRNA-seq data on human testicular samples met diversified scientific requests
Overview With more and more scRNA-seq data on human testicular samples published, so far there have been plenty of studies that chose to re-analyze those published data in various aspects. Table 4 summarized the studies that re-analyzed the previously reported scRNA-seq data on human testicular samples. A special case is Chen et al.'s spatial transcriptomic work on testes [134]. In that Slide-seq-based study, the authors only employed published scRNA-seq data as a reference for cell-type assignment, and as an input for pseudotime value assignment, which (strictly speaking) were not the real "re-analyses" of scRNA-seq data itself. Therefore, we chose to exclude that study from Table 4. Among the reviewed 48 studies in Table 4, the majority of them combined the human testicular scRNA-seq data with other types of scRNA-seq data or bulk RNA-seq/microarray data in order to get integrated results, and most of them chose to experiments such as IHC/immunofluorescence (IF) staining on testicular sections to validate their findings. Hence, in Table 4 we also offered the important public data (other than human testicular scRNA-seq data) that the study used and listed the experiments/clinical studies or sequencing works conducted on their own. Although those studies had various designs and different findings, we still found universalities among these studies and categorized them into several categories. The detailed classifications as well as representative studies were discussed hereinafter, and many of the studies were with multi-categories. [192], and bulk RNA expression level of these three genes from HPA [192] and GTEx [193] 1.Ranking the organs and cells which were easier to be harmed by SARS-CoV-2 and finding the candidates with high risk including lung AT2 cells and macrophages, cardiomyocytes, stromal cells of the adrenal gland, the testis, the ovary and the thyroid 2.Finding that an acid condition could inhibit the activity of SARS-CoV-2 pseudovirus, which indicated the defensive role of the stomach against SARS-CoV-2 infection [196] 2.Gene information from Consensus Coding Sequence (CCDS) database [197], and gene variants from databases including 1000 Genomes Project [198], ESP6500siv2 [199] and gnomAD [200] 1.Drawing the maps of molecular signatures (marker genes, DEGs among subclusters, dynamic expression changes of functional genes such as cytoskeleton-associated genes, etc.) during female meiotic prophase I stages 2.Revealing both conserved and sex-distinctive transcriptional features as well as DNA methylational regulation during meiotic prophase I stages between female and male 3.Revealing a momentary increase of X-linked expression during female pachytene (which is opposite to meiotic sex chromosome inactivation of males) and revealing that it was due to a lower turnover or higher stability of X-linked genes rather than enhanced X-linked transcription Identifying four male-specific and five female-specific stages during the development of chicken germ cells, and revealing the paths as well as their related transcriptional patterns of chicken germ cells' stages transition (one path in male and two paths including meiosis and apoptosis in female) 3.Revealing both conserved and species-specific transcriptional patterns between human and chicken germ cell development *Country information was based on the institutional information of the first author **To be more brief, the references of some databases employed by multi-studies were listed here, including GEO (Gene Expression Omnibus) [244], ArrayExpress [245], HPA(Human Protein Atlas [192,246], GTEx (Genotype-Tissue Expression) [193], FANTOM5 (Function Annotation of The Mammalian Genome) [247], TCGA (The Cancer Genome Atlas) [248], GEPIA2 (Gene Expression Profiling Interactive Analysis) [249] and MGI (Mouse Genome Informatics) [250] databases.
Detecting the transcriptional pattern of interested genes in human testes As a transcriptome-level technology, the fundamental usage of scRNA-seq was to detect genes' transcriptional patterns such as the mRNA expression level of a certain gene in different types of cells. A typical study was carried out by Yang et al. [135], in which they first found a new variant of KASH5 by Whole Genome Sequencing (WGS) in an NOA patient, but did not know the detailed KASH5 expression pattern in the human testis. So, they chose scRNA-seq data (the data used here were previously reported by the same team) to see the expression of KASH5 and found it to be enriched in SPCs, especially leptotene to pachytene SPCs, and finally validated it by IF staining. This process of "finding target gene" and then "detecting it in human testicular scRNAs-seq data" (sometimes with a third step as "validating genes using experiments") was quite common for studies with interested genes [136][137][138][139][140][141]. Meanwhile, the interested genes were not restricted to one gene or several genes, the researchers could detect a gene family (such as Interleukins) or a series of function-related genes (such as inflammatory cytokine receptors) in scRNA-seq data [142], which could give researchers a view on macroscale. Moreover, an updated usage of this process was to not only show the type of cells that a certain gene was expressed in, but also show its transcriptional changes during spermatogenesis [143] or among different pathological groups [127,144]. A good example came from He et al.'s work [127], in which they identified four interested genes (KIF2C, MRPS2, RPS15 and TSFM), and then based on human scRNA-seq data of both KS testes and normal testes, they showed that KIF2C were downregulated while RPS15 were upregulated in KS testes. Additionally, pseudotime trajectories showed a growing tendency of KIF2C and a downtrend of RPS15 with the maturation of spermatogenic cells [127]. We need to point out that no matter what method the study used and no matter how complicated the study procedure was, the nature and essence of this type of re-analyzing were to show the expression of given genes in human testes, no matter how these interested genes were found out (by other technologies/experiments, by bulk RNA-seq/ microarray data, by scRNA-seq data of other organs/species, by literature reviews or even by the authors' own idea).
Indirectly revealing the vulnerability of testes to SARS-CoV-2 infection
Due to the Corona Virus Disease 2019 (COVID-19) pandemic and the high proportion of SARS-CoV-2-related articles in Table 4 (10 out of 48), we especially discussed the usage of human testicular scRNA-seq data here. As for the usage of testicular scRNA-seq data, most of these studies were actually in the form of "detecting the transcriptional pattern of interested genes in human testes" [145][146][147][148][149][150][151][152][153][154], since several important proteins, such as ACE2 [155], TMPRSS2 [156], Furin [157] and CD147 [158], were deeply involved in the entry of SARS-CoV-2. So far, the opinions on this topic have been controversial. In some studies, ACE2 was reported to be expressed in both germ cells (especially SSCs&SPGs) and somatic cells, including SCs, LCs as well as myoid cells [145,147,149,154]. And one study showed a higher positive rate of ACE2 in testes from infertile patients (OA and NOA) [148]. Those results indicated the potential vulnerability of human testes to SARS-CoV-2 infection, especially in infertile men. And some studies even called the human testis a "high-risk organ" for SARS-CoV-2 infection [152]. But based on the scRNA-seq-derived evidence that there was no co-expression of ACE2 and TMPRSS2 in the human testis, other two studies gave the opposite view that human testes might not be able to be long-term affected by SARS-CoV-2 infection [146,151]. From our point of view, those ten scRNA-seq-based studies only gave indirect evidence based on previously published data. By combing original and published scRNA-seq data on human testes, Liu et al. partially overcame the shortcomings of those studies and for the first time showed an expression of ACE2 in primordial germ cells of prenatal testes and a downward trend of ACE2 with the age increasing [159]. More importantly, although ACE2 had the highest expression in SCs, it significantly shrank in SCs of NOA patients [159]. But the results of Liu's study were still based on scRNA-seq of patients without histories of COVID-19, which were also indirect. So far, there has been no report of scRNA-seq data on human testes from COVID-19 recovered patients, maybe due to the huge difficulty of sample obtaining. But with the pandemic continuing, there must be patients who simultaneously suffered from COVID-19 and male infertility and needed testicular biopsy or TESE surgery. Therefore, in the future, researchers can use scRNA-seq to compare testicular samples from COVID-19-recovered patients with those from non-COVID-19 patients, which will offer direct evidence of the impact of SARS-CoV-2 on human testes.
Finding novel genes involved in normal testicular function and male infertility
Testicular scRNA-seq data could not only be used to draw expression patterns of certain genes in the testis, but could directly be used as the sources of such novel genes. The difference between this type of usage and the type "Detecting the transcriptional pattern of interested genes in human testes" is that this type of study directly used testicular scRNA-seq data as a tool to find novel genes, while the other one only used scRNA-seq data as a tool of validation or display. For example, by comparing the DEGs in SPG between OA and NOA scRNA-seq data, Yang et al. reported a novel SPG-specific gene ELAVL2 and found it to be significantly downregulated in SPGs of NOA patients [160]. Further experiments proved this gene as an important factor in regulating SSC maintenance and apoptosis [160]. Like that study, scientists could use testicular scRNA-seq data as a tool to reveal functional genes that may be related to the development of testicular diseases [161][162][163][164]. We recommend that this type of usage should be "cell-oriented", which means the researchers should focus on a certain type of testicular cells and try to dig up novel genes or mechanisms of this type of cells, rather than just comparing the differences between disease and control groups without considering cell heterogeneity (in other words don't use scRNA-seq data to do tasks that bulk RNA-seq can do).
Comparing conserved and species-or sex-specific transcriptional features
Another important type of re-analyses of human testicular scRNA-seq data is to do species or sex comparisons. For species comparison, so far, the spermatogenesis of mouse [165], mouse lemur [165], chicken [166], cynomolgus macaque [167], sheep [168] and the somatic development of pigs [169] has been fully compared with that of human. The similarities of these studies were that they reported both conserved (especially the conservatism of X chromosome inactivation) and species-specific features among different animals and human. A good example would be Tian et al.'s work, in which they compared the spermatogenesis of four different species (human, mouse, sheep and cynomolgus macaque) [168]. In that study, they found the spermatogenesis between sheep and human were partially conserved at the transcriptional level (e.g., similar regulators were expressed in the same stages of spermatogenesis in human and sheep) and further showed the high conservatism of X chromosome inactivation during meiosis among four species [168], which was in line with Lau et al.'s report [167]. On the other hand, Rengaraj et al. reported a large proportion of species-specific genes between human and chickens (24 out of 56 orthologs) during the development of male fetal germ cells [166]. Additionally, the species comparison (usage type 4) could be combined with detecting certain genes' expression patterns (usage type 1). For example, INPP4B was found to be highly expressed in postmeiotic spermatogenic cells in the testis of both human and mice [170], which remained evolutionary conservatism, while SFTPC was expressed in late spermatogenesis of human but were absent in murine spermatogenesis [171], which indicated evolutionary divergence. And the tasks of those two studies were using testicular scRNA-seq data to show the expression of certain genes, but such tasks were done cross-species by combining two types of usage. In this area, the spermatogenesis among mammalian species or some other common animals (such as chicken) has been largely compared, but two subjects remain to be discovered. First, are there any highly conserved genes in spermatogenesis between humans and rare species or other non-mammalian animals (which are evolutionarily far)? Second, what's the conserved mechanism of testicular somatic cells' development among species? We believe scRNA-seq analysis will give the answers in the future. Besides, it's well known that ovary is the female counterpart of the testis, thus scientists tried to compare gonads of two genders. The conservatism of germ cells for both genders was observed during meiotic prophase I, such as the high expression of the same marker genes at each stage of primary SPC/oocytes (TEX19 in leptotene; SPO11 in zygotene; BRDT in pachytene and H1FOO in diplotene) [172]. Wang et al. even identified shared subtypes of germ cells (SPARC + fetal germ cells) in prenatal testes and ovaries [82]. But huge heterogeneity of germ cells was also reported in the expression of X-linked genes, including the high expression of SMS in diplotene oocytes but not SPC and the lack of XIST in adult testes [73,172]. Both homogeneity and diversity between genders were also reported in somatic cells during scRNA-seq analyses. For example, shared transcriptional expressions of NR5A1/SOX9 in both fetal SC and granulosa cells [73] and shared type of somatic cells (KRT19 + cells) [82] reflected the homogeneity between genders, while the divergent somatic expression pattern of DMRT1 (high in SC but not granulosa cells) suggested the diversity [73].
Bioinformatic usages including algorithm developing and database/portal establishing Since developmental biology is one of the most important areas in which scRNA-seq analysis is applied [173], so far there has been a huge accumulation of related raw data deposited online. Therefore, the raw data could then be used to develop or validate scRNA data-related bioinformatic tools, such as bioinformatic algorithms, packages, softwares and databases. Pont et al. developed a method for single-cell based scoring system and visualization of gene set signature, and a scRNA-seq set of normal human spermatogenesis was employed to show the function of this method [174]. Recently, Stow et al. developed another scRNA-seq-based bioinformatic tool called SCIFER to analyze long interspersed element-1 (L1) mRNA level from individual L1 loci in single cells, and further employed human testicular scRNA-seq data to show its function [175]. Theoretically, the role of scRNA-seq data could either be a developer (to build up the tool) or a validator (to confirm the tool works well or to show how to use it), while the current studies seemed to prefer to use testicular scRNA-seq data as a validator. Moreover, for many life scientists (especially clinicians) who would like to use scRNA-seq analysis to validate their findings or interested genes in human testes, it's too hard for them to start from the raw data due to the high demand for coding skills [176]. Therefore, it's more convenient for researchers to use some interactfriendly or "zero-code" online websites/portals. So far there have been several popular interactive online portals containing human testicular scRNA-seq data (listed in Table 5), some of which were testis-specific [31,72,78,177], two were human gonad-based (with female's samples and in vitro differentiated gonadal cells) in terms of scRNA-seq data [79,[178][179][180], while the rest were based on various organs with testicular scRNA-seq data included. The function of these online portals was basically Uniform Manifold Approximation and Projection (UMAP)/t-distributed stochastic neighbor embedding (t-SNE) visualization of cells and illustration of the expression patterns of certain genes. Some advanced functions such as pseudotime trajectory of testicular cells or enrichment analysis are only supported by few portal [78]. As a result, bioinformatic engineers should pay more attention to the realization of integrated analysis and comprehensive analysis through interact-friendly (non-code) website tools. Table 5. Popular interaction-friendly online portals/websites containing scRNA-seq data of human testicular samples
Conclusions and future perspectives
The main purpose of the current review is to gather the updated evidence on human testicular samples revealed by scRNA-seq data, in order to gain a "single-cell" view of both normal and pathological development of the testis. So far, there have been a large number of novel findings in terms of normal spermatogenesis, age-related testicular microenvironment's development and changes, testis-related male infertility, as well as new bioinformatic methods suitable for testicular scRNA-seq data analysis. In terms of normal status, the whole development process of spermatogenic cells (from the most naïve primordial germ cells to testicular spermatozoa), together with their transcriptional patterns at each stage were largely revealed (including their similarity and differences among species). The researches on normal somatic cells were basically restricted to SCs, LCs and PTMs, while fewer scRNA-seq-based findings about human testicular macrophages, ECs or other rare cells (e.g., T cells) were reported. Additionally, numerous transcriptional changes and potential mechanisms have been revealed in disease data including spermatogenic dysfunction, OA, ejaculatory disorder and chromosomal abnormalities, giving us insight into the heterogenous pathogenesis of male infertility and providing us with innovative targets of treatments. Last but not least, more and more publicly available testicular scRNA-seq datasets give researchers more opportunities to do their individualized data analysis for different research needs. We believe this review will not only pave the way to the transcriptionally understanding of both normal and abnormal spermatogenesis, but also be a reference for the future usage of scRNA-seq analysis in the human testis.
One of the most prospective research highlights in this area is the combination of testicular scRNA-seq analysis with spatial transcriptomic analysis of the testis. Due to the dissociation of testicular cells in scRNA-seq, it could neither spatially illustrate spermatogenic cells in seminiferous tubules, nor analyze the spatial interaction between somatic and spermatogenic cells [134]. To solve this problem, Chen et al employed the novel technology "spatial transcriptomic analysis" to create spatial atlas that spatially illustrate testicular gene expression at near-single-cell resolution in the human testis [134]. More importantly, Garcia-Alonso et al did integrated analyses of the development of male gonads by combining scRNA-seq and spatial transcriptomics [74], which was more comprehensive and in-depth. We believe that the integration of these two approaches will improve future researches of the human testis and male infertility, allowing for a more complete and nuanced understanding of the functional/dysregulated spermatogenesis in the human testis. So far, there has been a limited number of such integrated studies, leading to a novel "Blue Ocean" of testicular researches.
Although scRNA-seq analysis has proved the existence of immune cells, such as macrophages (both M1 and M2), T cells, mast cells, or even B cells, in testicular microenvironment, the role of these minor cells and their transcriptional patterns in the development of male infertility remain unknown.
Meanwhile, more types of testicular diseases, including cryptorchidism, AZFc deletions, azoospermia after chemo/radiotherapy, as well as teratospermia, should be focused on and will benefit from the further usage of scRNA-seq technologies. Due to the high cost of scRNA-seq, few of the above-reviewed articles were directly out of clinical purpose. With the development and cost reduction of this technology, clinical usages such as diagnosis, classification, prediction of sperm retrieval or the effect of hormone treatment might come true. | 2023-05-06T05:05:11.597Z | 2023-04-09T00:00:00.000 | {
"year": 2023,
"sha1": "d5eaf3a4841eb6d47e977721ebcbbed6f5d0baf5",
"oa_license": "CCBY",
"oa_url": null,
"oa_status": null,
"pdf_src": "PubMedCentral",
"pdf_hash": "d5eaf3a4841eb6d47e977721ebcbbed6f5d0baf5",
"s2fieldsofstudy": [
"Medicine",
"Biology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
117916919 | pes2o/s2orc | v3-fos-license | Observable consequences of a hypothetical transient deviation from Quantum Mechanics
The conflict between Quantum Mechanics (QM) and the intuitive concepts of Locality and Realism (LR) is manifest in the correlation between measurements performed in remote regions of a spatially spread entangled state. In this paper, it is hypothesized that transient deviations (from the values predicted by QM) occur if the correlation is measured in a time shorter than L/c, where L is the spatial spread of the entangled state and c is the speed of light. In this way, the conflict is solved by changing QM minimally. Under general assumptions, it is obtained a mathematical model of the process that reproduces the QM value after a time longer than L/c has elapsed. One of the predictions of this model is that oscillations of the rate of coincidences should exist, with a main frequency lower than c/4L. An experiment able to reveal these oscillations is shown to be accessible, by placing stations at about 5 Km and reaching a coincidence rate about 3E5 1/s (a value already obtained at the laboratory scale). This means a test of QM vs LR of a completely new type, with several practical and theoretical advantages.
Introduction.
The Copenhagen interpretation of Quantum Mechanics (QM) has faced debate since its early years. A.Einstein objected the non-determinism intrinsic in the theory, and claimed that it was incomplete. One generation later, J.S.Bell showed that QM is in contradiction with intuitive notions about the locality of physical phenomena, or with the existence of a Universe whose properties are independent of the observer. These notions usually receive the shorthand of Local Realism (LR) [1,2]. J.S.Bell also showed that experiments with entangled states of two particles can decide whether QM or LR is valid in the Nature. The resolution of this controversy is essential to the foundations of Physics. In the proposed experiments, the statistical correlation between the results of measurements performed on the two entangled particles is obtained. LR imposes a limit on this correlation, while QM predicts a correlation higher than that limit, regardless of the distance between the particles. Many versions of these experiments have been performed, generally confirming the violation of the averaged correlation limit imposed by LR.
Nevertheless, giving up LR is so contrary to the common sense that the search for alternative explanations is, at some extent, defensible. Many alternative theories holding to LR, or LR theories (also named "hidden variable theories"), have been proposed to reproduce the observed results. These theories usually exploit the imperfections of the experiments, through one or more of the so-called logical loopholes. In general, the proposed LR theories are not claimed to describe a real physical situation. The effort is focused in showing that they are indistinguishable from QM in the practice, and hence, that LR has not been disproved yet. Despite the technical improvements achieved after many years in the measurement of the violation of the correlation limit, some loophole always survives, and the controversy remains undecided. It has even been speculated that, for deep fundamental reasons, the controversy is intrinsically not decidable [3].
It is pertinent to note that most of the proposed LR theories, if they were demonstrated true, would force a complete rewriting of the currently accepted description of the world at the microscopic scale, and to fully discard QM. The overwhelming success of QM in many different physical problems suggests that it is improbable that it ever fails in the prediction of statistically averaged values. Little attention has been devoted instead to the possibility of deviations in time (i.e., transient deviations) from the QM predictions, what is known as non-ergodic (LR) theories. This paper explores this possibility, and searches its observable consequences. A model is proposed (it is named "Transient Quantum Mechanics", TQM), by following a path somehow inverse to most LR theories. It does not assume a microscopic world completely different from QM. Instead, it modifies QM as little as possible. Besides, no effort is made to obtain results indistinguishable from the QM predictions. On the contrary, I look for the deviations from QM that arise naturally from the general properties of the model, and that can be effectively observed. An observable deviation is found indeed: quasi-periodical oscillations of the rate of coincidences after the analyzers (see Fig.1) with a main frequency < c/4L. The observation of these oscillations implies a test of a completely new type, independent of the violation of a statistical correlation, and with several practical and theoretical advantages. Besides, TQM is not a fully arbitrary manufacture, but it is inferred from a scenario or framework that is (in my opinion) physically plausible.
That framework is named here "Non-ergodic contextuality" (NEC). NEC is not indispensable to define and work with TQM, so that its details are left for the Appendix A. It is anyway convenient mentioning here that NEC assumes that the results of the observations at the quantum scale are not random but determined, in the statistical average at least, by the states of the atoms in the setup and its environment (i.e., contextuality). Besides, these atoms evolve in time (i.e., non-ergodicity) in a coupled and chaotic way. As a result of these features, NEC (and hence TQM) is not ruled out by the impossibility theorems or by the experiments performed until now (Appendix A).
Two hypotheses.
In this Section, I state and discuss the two hypotheses on which TQM is based.
The conflict QM vs LR arises mainly from the correlation, higher than allowed by LR, between the results of observations performed in remote regions of a spatially spread entangled state. In order to solve the conflict with a minimal modification of QM, it is proposed the following First hypothesis: H1: The correlation between the results of the observations performed in remote regions of a spatially extended entangled state of size L needs a time >L/c (after an "unpredictable change" in the setup, see the Second Hypothesis below) to reach the value predicted by QM. If the correlation is measured in a time <L/c, then the value obtained holds to LR. Figure 1: An EPRB setup. The source S emits pairs of photons towards stations placed at a distance L. The angle settings of the analyzers are changed in a way unpredictable by the system while the photons are in flight. In the TQM model, such unpredictable change in the value of α triggers forces acting only on the environment E (grey arrow). The reaction forces of E (black arrows) are described by Lindblad operators L (i) acting on the field. The spatial spread is taken into account by making the L (i) to act on the state the field had a time τ ≈L/c before the change. The field is always in a LR-limited state.
It is evident that H1 solves the paradoxes related with the apparent "spooky action at a distance" of entangled states. H1 means that entangled states of arbitrary large size L do not exist, although thinking in terms of such entangled states is a convenient approximation to calculate average rates. The approximation is valid if the time required by the measurement of the correlation is much longer than L/c. It is speculated that this is valid to all scales. For example: the correlation between particles on Earth and the Moon would not show entanglement unless the measuring process took longer than ≈1 sec, but entanglement at the atomic scale could be observed after ≈10 -19 sec. H1 is not refuted by any of the experiments performed to date (see Sections 4 and 5). H1 is the only change assumed from usual QM in this paper. All the concepts and the tools of QM remain valid, as far as they are not in conflict with H1. It is worth stating here that H1 has negligible impact on Quantum Computation, because quantum computers will be small and the algorithms will run during a time much longer than L/c. In the application known as Quantum Key Distribution, instead, H1 may open a window to an eavesdropper in the case the flux of pairs is larger than c/L (see Sections 4 and 5).
In what follows, because of its relevance and simplicity, I will focus on the case of two photons entangled in polarization: the Einstein-Podolsky-Rosen-Bohm setup (EPRB, see Fig.1). In this case, H1 can be read as a sort of semi-classical hypothesis: the matter is quantized, but the field is classical. "Classical" here does not mean that photons do not exist, but that the photon states hold to LR. In other words: the photon states are separable, non-entangled or, in general, "LR-limited". I will call "system" = setup + environment E. The "setup" = source + analyzers + detectors.
An important requirement of the experiments aimed to test QM vs LR is that the values of the angles {α,β} of the analyzers in Fig.1 must be unpredictable. This is to close the so-called locality or contextual loophole [4,5]. A key detail is that the values must be unpredictable by the system. This leads to several questions: even if {α,β} are changed during the flight of the photons from the source to the detectors, how to be sure that the system cannot predict, at least partially, that change? And why leaving {α,β} fixed should be less predictable by the system than changing them? Closing this loophole is difficult, for there can always be some hidden correlation between the source of the changes and the system [6]. This possibility cannot be simply excluded, even if only partial contextuality is suspected to be valid in the Nature, because the degree of statistical correlation sufficient for a contextual LR theory to reproduce the QM values is surprisingly low [6][7][8]. There is no reliable way to ensure that the settings are unpredictable enough. The importance of this problem has been pointed out by several authors [8][9][10]. It is an intricate issue that has even metaphysical consequences, leading to debates on the existence of free will [11] or a strictly deterministic Universe [12]. From the point of view of an experimentalist, the problem seems a dead end impossible to escape without making at least one unverifiable and controversial assumption. The NEC framework shows a way out that is, in my opinion, reasonable. The basic idea arises from the observation that, in some time scale, all physical systems become unpredictable.
In NEC, the coupled evolution of the different parts of the system is the cause of the high statistical correlation observed. But, the exponential increase of the distance between initially neighbor points in phase space, which is characteristic of chaotic dynamics, eventually destroys that coupled evolution. These spontaneous "losses of track" are equivalent, from the point of view of the system, to an unpredictable change of the settings. It is then natural, inside the NEC picture, to propose the following Second hypothesis: H2: Changes unpredictable by the system occur spontaneously (at an average rate µ≠0).
Between the two extremes (absolute free will or strict determinism) H2 means an intermediate position: any physical system evolves deterministically during periods, which are interrupted by unpredictable events. The origin of the unpredictability would be then physical, not metaphysical, and besides, it would be unavoidable. The precise value of µ would be an attribute of the system's dynamics, probably related with the value of its highest Lyapunov exponent [13]. An important practical consequence of H2 is that events equivalent to unpredictable changes should occur even if the analyzers' settings in the Fig.1 are kept fixed.
It is evident that the decisive way to test whether H1 is valid in the Nature is to measure the violation of the LR correlation limit in a time shorter than L/c after an unpredictable change. As it is shown later, this is an impossible task nowadays. In order to find an alternative, accessible test, the gap from the general idea to a mathematical model providing definite numerical predictions must be filled. In the next Section, some assumptions are made, in compliance with H1 but as unspecific and simple as possible, to get that mathematical model.
The TQM model is built in three steps:
First step: LR-limited state of the field.
According to H1, there are no spatially spread entangled states. Hence, the spatially spread state of the field in the Fig.1 must be LR-limited. This state must also be able to display the QM correlation, at least in some condition. In order to fulfill the two requirements, note that the probability values predicted by QM can be reproduced, regardless the value of β, by a mixture of photon pairs polarized parallel and orthogonal to the axis α, or: this state is classical, separable, or LR-limited. It is able to reproduce the QM predictions only because the value of α is known. When α=0 (π/4), ρ α=π/4 (ρ α=0 ) generates no correlations, i.e., P ij (α,β)= ¼ ∀i,j,α,β. Thus, it also reproduces the observable effect of the unpredictable change (or the spontaneous loss of track) discussed in the previous Section.
Second step: Lindblad approximation.
The unpredictable change in α assumed in H2 triggers forces of reaction inside the system. The evolution that follows is complex. In order to describe this evolution without making any specific assumption on the composition of the system and the involved interactions, only the observable effects can be taken into account. I assume then that the forces of reaction acting on the field come only from E, and describe their effect with the Lindblad operator: where the L µ describe the action of E on ρ α , as it is often done to describe decoherence. The Lindblad approximation means that the reaction forces inside E dissipate in a negligible short time. The eq.(2) is hence valid only in a certain "coarse" time scale and for a Markovian E [14]. Let take: where the parameter g measures the strength of the interaction. Its value is unknown. From the NEC picture, it is conceivably related with the distribution, density, composition and temperature of the atoms in E. The operator L (1) makes the LR-limited state ρ α=π/4 , which does not reproduce the QM predictions for α=0, to evolve into the LR-limited state ρ α=0 , which does (see Appendix B). The operator that produces the other required evolution, ρ α=0 → ρ α=π/4 is: The appropriate operator is L (1) or L (2) depending on the value of α. The interaction described by the eqs. (2)(3)(4) fulfills the condition of being unspecific, for it is defined by its observable effects only. No attempt is made to describe its details or mechanisms, to keep it as general as possible.
The probabilities of observing a coincidence are computed as Tr[ρ α .Q α (a) ⊗Q β (b) ], being the Q α (a) the operator of projection for the passage through an analyzer oriented at an angle α and acting on the subspace a. Thanks to the form of ρ α and the Q's [1] the following equalities hold ∀α,β: The probabilities can be then written in terms of a single parameter ρ d : where (see Appendix B): where ρ target = ¼ (or ½) if the setting is α=π/4 (or α=0).
As an illustration, let assume an initial condition corresponding to α=π/4 and that an unpredictable change to α=0 occurs at t=0. The probability of a double-passage coincidence evolves then as: that is, P ++ (α=0,β,t) goes from the uncorrelated value ¼ at t=0 to the QM prediction as t→∞. The operator L (2) produces the analog evolution when the unpredictable change is from α=0 to α=π/4. In this way, the QM predictions are reproduced by a LRlimited state for t→∞. This is, precisely, what is assumed by H1.
Note that the matrices obtained from tracing out one photon: Tr[ρ α ] partial = ½I ∀t, so that no time evolution is observed in the detection of single photons.
Third step: the state of the field is delayed.
The forces represented by the L (j) act over a spatially spread region. The information on the value of α (which defines which one of the L (j) is acting) needs a time >L/c to reach the other side of E. The consequence is that the reaction to an unpredictable change must be delayed a time >L/c. This is taken into account in the model by using the value of ρ α as it was a time τ ≈ L/c before the change. In other words: in the statistical average, the state of the field is delayed a time τ with respect to the instantaneous value of α. This delay is the formal way to introduce nonergodicity (i.e., dependence on the history of the system, see Appendix A). The eq.(8) then becomes: where the function ρ target (t) is externally defined (it represents the "unpredictable changes"), ρ target = ½ (¼) if α=0 (π/4). The eq.(10) completes the mathematical formulation of TQM. Note that the usual QM result is retrieved in the limit τ→0, g 2 →∞.
The eq.(10) is a delay differential equation. It evolves in a phase space of infinite dimensions. This is a convenient property, for it is intended to embed the high dimensional dynamics implicit in NEC. On the other hand, the features of its solutions are difficult to find in the general case. A glimpse, valid if µτ<<1 (i.e., low rate of unpredictable changes) is obtained by studying the case ρ target (t)= constant. In this case, eq.(10) can be written: where x(θ) ≡ ρ d (θ)-ρ target , θ ≡t/τ, and Γ = 4g 2 τ. In the fig.3, the solution of eq.(11) is plotted if x(t<0)=¼, Γ=1 and an unpredictable change occurs at t=0. The change is followed by a damped quasiperiodical oscillation towards the target state, with decay time ≈3.5τ and period ≈4.5τ. The physical cause of the oscillations is that the action of E into the field "overshoots" because of the delay. Enforcing locality (of the propagation of the information on the value of α by making τ≠0) is therefore sufficient to obtain the oscillations.
In the fig.4, the variation of the decay time and the (approximate) period of the solutions of eq.(11) are plotted as functions of Γ. The amplitude of the oscillations diverges if Γ > π/2. Note that for Γ >0.6 the period is nearly constant, and equal to ≈4.5τ. This is an important result regarding an experimental test.
Testing TQM.
A numerical simulation of an EPRB experiment with unpredictable changes in the analyzers' settings is performed. The evolution of ρ d (t) is calculated from eq.(10), where the value of ρ target (t) is changed in the following way: think a coin is tossed at random times, at an average rate µ. Depending if the result is head or tail, the setting is adjusted to α=0 (then ρ target = ½) or α=π/4 (then ρ target = ¼) with a negligible short time of transition. As a result, ρ target (t) jumps between the values ½ and ¼, and ρ d (t) follows it after some delay and damped oscillations (see Fig.5).
The observable, averaged value of S CHSH is obtained by numerical integration. Using the eqs.(5-7), it is: For fixed Γ, S CHSH fits the QM prediction if µτ<1, and (as it can be expected) it falls below 2 as µτ>>1. Yet, the result S CHSH = 2√2 can be obtained for any value of µτ, by tuning the (unknown) value of Γ. F.ex., for µτ=13 (the apparent value in the experiment in [5]), S CHSH =2.828 for Γ=1.549. The eq.(12) assumes that the rate of detected pairs is arbitrarily high, so that the details of the evolution of ρ d (t) are exactly followed. A more realistic simulation, where that rate is only 10 -2 or 10 -3 τ -1 , provides a poorer discrimination between TQM and QM. In summary: a procedure different from the measurement of S CHSH is necessary to test TQM. Note now that for the parameters' values in the fig.5 (µτ=0.2 and Γ=0.9), S CHSH = 2.79, too close to 2√2 to allow a discrimination between QM and TQM in the practice, but that the damped oscillations produce a broad, yet clearly visible, peak in the FFT power spectrum at the period ≈4.5τ ( fig.6). In the case that µτ>>1 (i.e., highly unpredictable settings) and Γ tuned so that S CHSH ≈ 2√2, the peak is sharper and higher than in the case µτ<1, but it remains in the same position. It only shifts, towards lower frequencies, if Γ<0.6. The accessible test that was looked for is evident now: looking for an oscillation of period >4L/c. As P ++ (α,β) is linearly dependent of ρ d (t) (eq.7), the oscillation can be detected in the number of coincidences after the analyzers. This is a test completely different from all the ones attempted until now. It is independent of the violation of a statistical correlation limit. The precise value of µ is irrelevant: it suffices that µ≠0 (H2). The (often cumbersome) evaluation of the number of accidental counts has negligible impact, and hence, the experiment is also more robust against noise than measuring S CHSH . The so-called coincidence-loophole, if it existed, would involve time shifts of the order of the time coincidence window, too short to appreciably distort the oscillations (see Appendix A). A definite result of the test is obtained even if the detectors are not perfect and eventready signals [1] are not available: if oscillations (with a period varying linearly with L) are observed, the usual form of QM is disproved. If the oscillations are not observed instead, TQM is disproved.
In the next Section, the conditions for an experimental test of TQM are discussed. But, before going on: the violation of the LR limit S CHSH ≤ 2 obtained even when µτ>>1 is possibly surprising, so that it deserves some comment. When Γ is tuned close to π/2 (i.e., the point where the amplitude of the oscillations of eq.11 diverges), the P ij (α,β) obtained from eq.(10) can take, transitorily, values outside the [0,1] interval. It is well known that extended probability values allow a LR theory to reproduce the QM predictions, although it is arguable that such situation actually holds to LR [15]. A possible interpretation for the appearance of probabilities outside [0,1] is, precisely, that non-stochastic processes (as the ones involved in any of the loopholes) are active. Anyway, the important result at this point is this: regardless of the particular mechanism, "conspiracy" or loophole the LR model exploits to fit the QM results in the average, a peak at frequency ≈c/4L (or lower, depending of the unknown value of Γ) is present.
Experimental requirements for the test.
In order to reveal the oscillations predicted by TQM, it is necessary to record the time value at which each photon is detected. This technique is named time stamping or time-tag. It implies some instrumental complication, but it is anyway unavoidable to close the coincidence-loophole [16]. Revealing a frequency peak by random sampling a noisy series is an involved issue. In order to get an idea of the situation, the usual criterion of recording at least two samples in the period of interest suffices. A sample of the value of P ++ (α,β,t) requires the detection of least 10 coincidences. Therefore, one needs 20 pairs after the analyzers in a period ≈4.5τ (the lowest value for the period), or ≈5 pairs in τ. In what follows, I consider only the most unfavorable situation, τ = L/c.
Let see now how far the performed experiments are from the goal of a rate of 5 detected pairs in τ. The highest reported rate of entangled pairs was obtained by matching the pumped volume in the non-linear crystal (used to generate parametric fluorescence) with the mode of collecting single-mode optical fibers [17], reaching ≈3×10 5 s -1 in a laboratory environment. This number cannot be improved, for the best currently available single photon detectors (avalanche photodiodes) cannot be used reliably with a rate detection above 10 6 s -1 . Therefore, in order to detect 5 pairs (with the rate reported in [17]) in a time L/c, L>5 Km. Detection of pairs at 13 Km [18] and even 144 Km [19] has been achieved, but the coincidence rate was much lower than needed, between 50 and 8 s -1 . Nevertheless, these experiments were performed under the unfavorable conditions of free air propagation. The numbers would improve if the propagation is through a controlled environment or optical fibers. Of course, translating the brightness and purity of a source tested in the lab to a field installation of a size of several Km is a difficult challenge, but it does not seem impossible. A practical advantage is that it is not necessary to achieve S CHSH ≈2√2 to detect the oscillations. Even a poor value of the average correlation would suffice. If a test were attempted by measuring S CHSH in t<τ instead, L scales to 1000 Km, and a good value of the state's purity should be achieved.
It must be mentioned that dynamics faster than the "coarse" time imposed by the Lindblad approximation may produce peaks at higher frequencies. They cannot be predicted by the TQM model, but of course they may exist in the experiment. More important from the point of view of the test, peaks at lower frequencies may also exist, caused by "revivals" of the coherence in a realistic, imperfectly Markovian E. Besides, the numbers discussed above are for the most unfavorable value of the unknown parameter Γ. As a consequence of all this, the criterion of 5 coincidences in L/c is the most stringent condition to observe the oscillations. A lower value may suffice. An early search performed on the time-stamped data of the experiment in [5] revealed no oscillations [20], but the rate in that experiment was only 2×10 -3 τ -1 [21]. The rate of detected pairs should be thus increased a factor about 10 3 with respect to the currently achieved values (in EPRB setups with remote stations) to enter the range where the oscillations can be expected to be detectable. As it was stated, this is difficult, but not unattainable.
Summary.
A hypothesis (H1) that reconciles QM with LR is proposed. It minimally changes QM. From H1 and the Lindblad approximation, a simple and unspecific model (TQM) is deduced, which has the mathematical form of a delay differential equation. If the Nature is not strictly deterministic (i.e., if H2 is valid) TQM predicts the existence of quasi-periodic fluctuations of the rate of coincidences detected after the analyzers in an EPRB setup.
That prediction leads to a test of QM vs LR of a new type, independent of the violation of a correlation limit. In the usual test, threshold values of the detectors' efficiencies and the unpredictability of the changes, as well as the knowledge of the moment of emission of the pairs, are necessary to close all the known loopholes [1,10,22,Appendix A]. This is difficult to achieve, and the controversy remains undecided. In the test proposed here, instead, a definite answer is obtained even if those threshold values are not reached and the moment of emission is unknown: if oscillations with a period varying linearly with L are observed, the usual form of QM is disproved. If the oscillations are not observed instead, TQM is disproved.
The test should be performed in a large-size (L≈5 Km) EPRB setup with a rate of pair detections after the analyzers ≈5 c/L. This requires an increase ≈10 3 of the currently available rates in large-size EPRB setups, up to the values already reached at the lab scale. The required time stamping resolution is ≈10 -5 s, which is trivially achievable. The analyzers can be kept fixed and the average correlation does not need to be high, these being significant practical simplifications.
I describe here the physical picture or conceptual framework supporting H1 and H2. Recall that TQM is deduced from H1, H2 and the Lindblad approximation only. I mean: what follows is not necessary to obtain the results presented before.
NEC arises from the following reasoning: one photon is an amount of energy so small, that it is natural that the result of any process involving it (say, the passage through an analyzer) is influenced by the surrounding atoms. The number of these atoms is enormous, but it is conceivable that the result of their influence converges to a value determined by the boundary conditions or symmetries of the whole. Note that the same happens in the process of frequency conversion in a nonlinear crystal.
NEC assumes that the observed correlations are caused by the influence of all the atoms in the system (i.e., the context) and besides, that this influence evolves in time depending of the system's history (i.e., non-ergodicity). Let review briefly the meaning of contextuality and non-ergodicity: Contextual theories were defined by J.S.Bell: "The result of an observation may reasonably depend not only on the state of the system (including hidden variables) but also on the complete description of the apparatus [i.e., the context]". Contextual theories are not ruled out by the impossibility theorems as Gleason's or Kocher and Specker's ones (see [1] p.1924-25). It is precisely this fact what led J.S.Bell to obtain his inequalities. These theories define the contextuality, predictability or locality logical loophole. In short, contextuality means that the usually reliable approximation of isolating a part of the Universe for its study (neglecting the rest) has found its limit in the EPRB experiment.
The non-ergodic theories [22] state that the ensemble averages assumed in QM are not equivalent to the time averages obtained in an actual experiment. It was imagined that "a field, medium or ether with a relatively stable states or memory exists" that influences the probabilities of detection from one particle to the next, and that acts so as to reproduce the QM predictions in the average over long times (in NEC, the role of the ether is played by the system itself). They are usually related with the coincidence, trapping or memory logical loophole [3,10,[23][24][25].
The coincidence-loophole allows a LR theory to fit the QM predictions by shifting the photons' detections in time, according to the analyzer's angle, in or out of the time window that defines a coincidence. This makes the number of coincidences to depend on the settings of both analyzers, although the entire process is local. The theories exploiting this loophole cannot be tested even in a setup with 100% efficiency of detection and 100% unpredictable settings. Not a single photon is lost: all what happens is that its detection is delayed or advanced. These theories can only be tested in a setup where the possible time of emission of each pair and the time of detection of each single photon are recorded. The simplest way of understanding these requirements is as follows: imagine the photon detection is delayed until the analyzer changes to the position convenient to fit the QM results. If the expected moment of arrival of the photon is unknown, the delay is unnoticed. There are mechanisms more complex than the one just described, but the delays involved in the coincidence-loophole always are, necessarily, a fraction of τ. Hence, they do not distort the oscillations predicted by TQM, which have a period > 4.5 L/c.
Because of the large number of atoms involved, the evolution assumed by NEC takes place in a phase space of high dimension. Because of the type of interactions involved, the dynamical equations have many crossed, nonlinear terms. The result is a high dimensional chaotic dynamics (sometimes called hyperchaos). As the long-range interactions are mostly electro-magnetic, if an unpredictable (or caused externally to the system) change in the distribution of the atoms (or a spontaneous loss of track as well) occurs in a spatially spread setup, the system requires a time >L/c to adjust its evolution to the new distribution of atoms. This is, precisely, H1.
It is reasonable to ask here how a LR model can reproduce the observed non-classical correlations after a time >L/c. There are several ways to do it. Possibly the first one (historically) is a mechanical model [24]: an array of classical oscillators, linked by nearest neighbor interactions, produced one photon detection when the value of a certain dynamical variable exceeded a threshold. Among the most recent proposals, the deterministic learning machines (DLM) [26] assume that the optical elements in the setup act as units following a simple and adaptive program to process information carried by the photons, to decide when, where, or if, the photons are detected. These approaches, as many others alike, are able to reproduce the QM correlations after a time >L/c. But, they make specific hypotheses involving the system and the interactions, and the values of several parameters must be adjusted. The generality of their predictions is uncertain. It is desirable, instead, to find the most general possible features that follow from H1. This is the aim of TQM, and the reason why the Lindblad approximation is used.
I have just mentioned examples of how a LR model can reproduce the QM correlation. Next, I present a symmetry argument (i.e., not based in any specific model) aimed to make plausible why it does so.
A.2 A symmetry argument.
The aim here is to understand why, although entangled states of the field are assumed not to exist, the system evolves to produce the same averaged statistical correlations as if an entangled state of the field were actually present.
In general terms, a correlation is the consequence of some symmetry invariance. Classical correlations are mostly the consequences of the symmetry action = reaction. It is therefore convenient to look for the symmetry invariance, additional to the classical ones, whose consequence is the extra correlation observed in the Fig.2. Let consider the general form of the density matrix of two photons. After taking into account general restrictions and assuming rotational symmetry, that general form is (ρ d , ρ a ∈ ℜ): in the basis {|x a ,x b 〉, |x a ,y b 〉, |y a ,x b 〉, |y a ,y b 〉} from up to down and from left to right. Be aware that ρ α in eq.(1) does not hold to eq.A1. The matrix ρ is positive iff: The probability of double passage is: (note the difference with eq.7). The state ρ does not necessarily display a non-classical correlation. F.ex., a semi-classical theory of radiation (SCRT), that assumes that the two photons of the pair are emitted with the same well-defined polarization, which randomly changes from one pair to the next, holds to eqs.A1-A2 with values ρ d =3/8, ρ a =1/8. Then P ++ (α,β)= ¼ [½ + cos 2 (α-β)] and S CHSH = √2 < 2.
Let take now a fresh view to the Fig.1 when α=β. The setup is left-right symmetrical (also E, at least in the average), as in a mirror. Therefore, if one photon is detected at P a (+) , it is natural to expect that the same happens at P b (+) , for it is simply the image in the mirror. If this does not happen, then there must be some cause, different in each side of the setup. Say, a different value of a "hidden" variable invisible in the mirror (f.ex., the well-defined polarization of the photons in the SCRT mentioned above). But if there is no such hidden difference (precisely as it is assumed in QM), the result of each measurement performed at a must be identical to the one performed at b, and then: P ++ (α=β)= ½. When this condition is imposed to eqs.A1-A3, ρ d = ½ = ρ a is the only solution, and the eq.A1 becomes the matrix of the state |ϕ + 〉 [27]. This result is not a mere coincidence. If the setup is not well assembled and the mirror symmetry is not perfect, so that P ++ (α=β)= ½-ε (ε>0), then the Concurrence of the resulting state is 1-8ε. In other words: the deviation from the mirror symmetry, and the deviation from "perfect entanglement", both are measured by the same number.
In summary: the symmetry of the system implies that the averaged statistical correlation must be the one obtained as if the state |ϕ + 〉 existed. But the actual state of the field is LR-limited. Hence, if the settings of the analyzers are changed in an unpredictable way, the field is not able to reproduce that correlation. Nevertheless, the system's symmetry will bound the following evolution in such a way (regardless the details) to reach a state of the field able to reproduce that correlation after a transient has elapsed.
A.3: Final comments.
The Bell states are pure states, i.e., entities with no internal parts. They are "atoms" in the original (ancient Greek) meaning of the word. That's why the correlations they produce are higher than classically allowed: what is being observed at the remote stations in the Fig.1 is not a pair of related things, but the same thing (to stress this idea, D.Klyshko coined the term biphoton). NEC states that "atoms" of arbitrary large size do not exist. NEC states that a time >L/c is needed until the system evolves into a state showing statistical correlations equivalent to the presence of a large "atom". The cause the system evolves in this way is its symmetry as a whole. If this symmetry changes, the (averaged) correlation will change in accordance. QM would be then an approximation, valid in the average over long periods of time, to a very complex process that involves a macroscopic number of true atoms and LR-limited fields. A detailed mathematical description of this process is practically impossible. One has to rely on statistics, as in QM. But, there is an important difference: in QM, it is assumed that each consecutive measurement is independently performed on identical copies of the (entangled) state, being the differences among the observed results the consequence of the intrinsic randomness of the microscopic world. In NEC, instead, the differences observed between consecutive measurements are the consequence of the chaotic dynamics underlying. In QM, the statistical correlation is caused by an entity named entangled state. In NEC, the correlation is caused by the average restrictions on that chaotic evolution imposed by the system's symmetry as a whole.
NEC completes QM, as envisioned by Einstein. It can be understood as a form of the old "statistical interpretation" of QM, updated or enlarged with chaotic dynamics and symmetry arguments. But, faced with attaining a mathematical description, NEC is much more difficult to deal with than QM. In the overwhelming number of cases, the QM approximation is preferable because of easier mathematics and accurate averaged results (although it fails to describe the transients). The success of QM would be the consequence of its capacity for taking into account, in a simple and compact way (through the algebra of operators), the symmetries of the system and the measuring process.
Appendix B: Evolution due to the Lindbladian.
The L (i) operators produce the required evolutions also if the initial condition is the SCRT state mentioned in the Appendix A and the Fig.2 To calculate rotational invariance (matrix R), make θ = ϕ. To calculate twist invariance (matrix T), make θ = -ϕ.
The projection on the "transmitted" port of an analyzer acting in the a-subspace and oriented at angle α is represented The matrix Qa(β)⊗Qb(α) (necessary to calculate the probability of coincidence) is the product of these last two matrices, one with angle α and the other one with angle β. | 2019-04-13T04:39:03.901Z | 2012-12-22T00:00:00.000 | {
"year": 2012,
"sha1": "9e02be1e37534033288ee89a2cf405abc1d007d5",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "7354cd13001f52ecb749266a3261d2a79ad13d7b",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Physics"
]
} |
122754507 | pes2o/s2orc | v3-fos-license | Heavy fermion superconductivity: 5f vs 4f
Heavy electron materials are the only class of materials in which we know where to look for superconductivity. We discuss the relation between single ion and dense Kondo heavy electron physics, what sets the energy scale in these materials, the ways in which their low temperature exotic superconductivity is a prototype for all highly correlated electron superconductivity and the similarities between the 4f and 5f heavy Fermion materials.
Much of the interest in heavy Fermion and other highly correlated electron materials centers around their superconductivity, in particular when and where it occurs. There exists of course the desire to place all highly correlated electron superconductivity within a single framework, and our intent here is to compare 5f and 4f heavy Fermions with this in mind while also keeping in mind the much higher T C 's found among transition metal materials.
What one might call the standard model of heavy Fermion superconductivity begins with the observation that almost all Ce-based heavy Fermion superconductors appear in the vicinity of an antiferromagnetic quantum critical point. A well studied example of this is shown in Fig. 1. While one cannot say with certainty that proximity to a quantum critical point is true for all heavy Fermion superconductors, nevertheless it is remarkable that the heavy Fermions are perhaps the only class of materials where we have some good idea where to look for superconductors.
A useful way to think about heavy Fermion materials is from the standpoint of the dense Kondo lattice. Chemically ordered Ce compounds in the heavy Fermion class can be regarded as a set of Kondo centers which condense at low temperature into a coherent heavy electron liquid. This condensation is characterized by a coherence temperature, quite generally visible as a peak in the temperature dependent electrical resistivity. This coherence temperature corresponds to an energy which sets the scale for the low temperature physics as out- * E-mail: zfisk@uci.edu lined in Fig. 2. In many cases of interest the 4f 1 crystal field ground state of Ce is a doublet, and the essential problem the Ce intermetallics are dealing with at low temperature is shedding the associated Rln2 entropy per mole Ce. One way to quantify the coherence temperature is as that temperature by which this Rln2 entropy has evolved in the electronic system.
In the vicinity of an antiferromagnetic quantum critical point the specific heat typically varies as T ln(T * /T ) with coefficient of order Rln2/T * , which gives an entropy of Rln2 at T * . We observe that T * appears to the scale for the superconducting transition temperature T C in that T C /T * is of order 1/20, similar to the ratio of T C to the Debye θ one sees in phonon coupled BCS superconductors.
A tenet of the standard model of heavy Fermion superconductivity is that the so-called glue of Cooper pairing comes from the magnetic fluctuations residing in the dense Kondo liquid. This means that both the magnetic order and superconducting order which are seen competing in Fig. 1 utilize the same fluctuation spectrum. The interesting feature we see in Fig. 1 is that when the antiferromagnetic transition temperature T N > T C , superconducting order and antiferromagnetism can coexist at some lower temperature. However, the reverse is not true: there is no co-existence when T C > T N . The simple explanation for this is that the superconductor gaps out the low energy part of the magnetic fluctuation spectrum which favors the antiferromagnetic order.
Data [2] on the 2.3 K ambient pressure superconductor CeCoIn 5 support this assertion. The entropy developed to T C is approximately 0.15Rln2 per Ce. At T C , the jump ∆C/γT C = 5, much larger than weak coupling BCS. However, if one integrates γdT out to T C and compares this to ∆C, one gets a value much closer to weak coupling, ∆C/ γdT = 2. We also have the data of Sparn et al. [3] shown in Fig. 3 for the dependence of ∆C/γT C on pressure. This drop in the specific heat jump with pressure corresponds to the pressure variation of an intrinsic residual resistivity which essentially vanishes at the same pressure where T C becomes a maximum seen in Fig. 4 [4]. A consistent way to think about this is that one has a residual resistivity arising from the remaining uncondensed Kondo centers and that pressure reduces their concentration to zero near 1.5GPa. These centers appear to be pair breaking, and in fact Howard et al. [5] have developed in some detail the idea of some intrinsic scattering being responsible for a considerable reduction in the T C of CeCoIn 5 , although not precisely along these lines.
A further experimental result consistent with this line of thinking is the observation by Stock et al. [6] of an inelastic neutron resonance that develops below T C in CeCoIn 5 . There is considerable weight in this resonance, and it can plausibly be thought that it comes from the gaping out of the residual Kondo centers in the superconductor present at T C . This explains as well the large specific heat jump at T C . The decrease in the the specific heat jump with pressure just corresponds to the decreasing concentration of these centers with pressure. In this simple picture then we have the antiferromagnetism arising in Cd-doped CeCoIn 5 , for example, from the RKKY coupling of the Kondo centers that have not yet condensed into the heavy electron liquid. It is worth noting that in CeCoIn 5 , the single ion Kondo temperature of Ce measured in LaCoIn 5 is 1.7K [7]. The proximity of CeCoIn [5] to an antiferromagnetic quantum critical point has been demonstrated via Cd-doping experiments 8 where small substitution of In by Cd generates a phase diagram versus Cd concentration which closely resembles that of CeRhIn 5 versus pressure.
This appears however not to be the whole story as far as heavy Fermion superconductivity in Ce-based materials is concerned. Experiments 9 on CeCu 2 Si 2 Gedoped CeCu 2 Si 2 find two apparently separated domes of superconductivity suggesting the phase diagram of Fig. 5. The idea is that there is not only magnetically mediated superconductivity but superconductivity arising from charge fluctuations in an intermediate valence regime. Such a possibility goes back to Varma et al. [10] and was later developed in more detail by Holme et al. [11]. This calls to mind the superconductivity seen at high pressure in α -Ce and the recent studies of its phonon spectrum via inelastic x-ray scattering [12]. In these it was seen that α -Ce is on the verge of a lattice instability and possibly a charge density wave instability, suggestive of its superconducting properties arising from this intermediate valence background.
So our question is then, do we see anything like this in 5f-based heavy Fermion superconductors? We first look at elemental U and Ube 13 . Elemental U is not classed as a heavy Fermion superconductor, but it has some remarkable similarities to α -Ce discussed above. First, the superconducting phase α-U has the same crystal structure, and it has as well a set of charge density wave instabilities that interfere with superconductivity. Full bulk superconductivity is only achieved in α-U at pressure sufficient to quench the charge density waves [13]. Apparently it is not known at present whether a CDW can co-exist in superconducting α-U at pressures where superconductivity occurs before any evidence of a CDW is seen. Nevertheless, α-U appears a good candidate for charge fluctuation mediated superconductivity.
Ube 13 , the first heavy Fermion superconductor identified after CeCu 2 Si 2 , bears close similarities to other heavy Fermion superconductors that would be classed as close to a quantum critical point and we note here some parallels with the properties of CeCoIn 5 . The specific heat jump at T C is considerably larger than weak coupling BCS (Fig. 6) [14], but not nearly as large as in CeCoIn 5 . There is also seen below T C the development of an inelastic neutron resonance, although with considerably less weight than that seen in CeCoIn 5 [15]. There is present a substantial negative magnetoresistance that Fig. 8. TC of Th-doped Ube13 as function of field and composition [17].
shows that the residual resistivity in the normal state depends strongly on magnetic field, a result that indicates a large intrinsic residual resistivity just as in CeCoIn 5 (Fig. 7) [16]. These features of UBe13 all show striking similarity to those of CeCoIn 5 .
But Ube 13 has another life. When Th-doped, a second superconducting region of different nature appears, as strongly evidenced by the presence within it of two superconducting transitions. Further, when magnetic field is applied, the second superconducting region completely separates from the first, appearing very much like the schematic phase diagram in Fig. 8 [17]. However, that the second dome is coming from valence fluctuations seems unlikely, given the small variation in entropy recovered by T C in the various alloys. More likely seems the possibility that a changed Fermi surface topology unlies this unusual behavior.
The intensely studied heavy Fermion superconductor UPt 3 provides a strong case for belonging to a different class than CeCoIn 5 [18]. This material shows Fermi liquid properties at low temperature, with no obvious indication of being close to any magnetic quantum critical point. The entropy developed out to T C = 0.56 K is small by heavy Fermion standards, approximately 0.04Rln2 and the specific heat jump at T C is slightly smaller than weak coupling BCS, when allowance is made for the sizeable residual normal electron fraction at T = 0 K required by entropy balance. There are three different superconducting phases existing, two of them seen in zero applied magnetic field. It is believed that all these states have triplet pairing [19]. There are reasons as well for supposing that the 1.7 K superconductor U 2 PtC 2 with somewhat smaller electronic specific heat γ than UPt 3 may also fall into the charge fluctuation characterization. 20 The hexagonal UPd 2 Al 3 [21] and UNi 2 Al 3 [22] have antiferromagnetic order coexisting with a lower temperature superconductivity. Inelastic neutron scattering experiments [23] on UPd 2 Al 3 show a strong renormaliza- tion of the inelastic response below the superconducting T C , reminiscent of what was seen by Stock et al. in CeCoIn 5 . The hidden order phase of URu 2 Si 2 which also supports superconductivity well below the hidden order transition appears to be a unique material with properties which do not fit the pattern of other heavy Fermion systems [23]. There is as well the unusual ferromagnet URhGe [24] where superconductivity and ferromagnetism coexist, as they do similarly in the related ferromagnets UCoGe [25] and UGe 2 [26]. In the case of UCoGe, it is argued that a ferromagnetic quantum critical point exists under the dome of superconductivity in the temperature-pressure plane, but the different physics of this situation contains components that call for separate consideration.
The T C = 5.0 K Np-based superconductor NpPd 5 Al 2 ( Fig. 9) [27] also appears in the class with CeCoIn 5 . Not only is the large entropy of 0.15Rln2 developed by T C , the specific heat jump at T C is also considerably larger than the weak coupling BCS limit. We see as well that the entropy balance requires approximately a doubling of the electronic specific heat γ at T = 0 K. In this case ∆C/ γdT is very close to weak coupling BCS. The extra entropy that develops above the value of γT at T C is approximately. 08Rln2, quite similar to that seen in CeCoIn 5 .
In the case of the Pu-based superconductors, it has been suggested that the large T C = 18 K of PuCoGa 5 may arise in a charge fluctuation dome, with the 2 K superconductor PuCoIn 5 being of the CeCoIn 5 antiferromagnetic quantum critical type. There is not enough specific heat data to really make a good comparison between the materials based on this, but the comparison of lattice volumes for the two in relation to the volumes of the corresponding rare earth materials is interesting and gives support to the idea that these two compounds may be in different regimes (Fig. 10) [28]. It is known that the actinide elements when their 5f's become fully localized have volumes which lie close to those of the cor- responding rare earth elements. One expects this then to be true as well for the corresponding intermetallic compounds and for Pu this means that if it has fully localized 5f5 configuration it will correspond with Sm. In Fig. 10 we see that PuCoIn 5 and SmCoIn 5 have identical volumes, while that of SmCoGa 5 is considerably larger than that of PuCoGa 5 . That this difference is significant is further seen by the fact that PuCoGa 5 has smaller volume than AmCoGa 5 , indicating that the actinide contraction which corresponds to the lanthanide contraction has not yet set in at Pu within the series. Assuming that AmCoGa 5 is localized 5f 6 , the two points of SmCoGa 5 and AmCoGa 5 are approximately parallel to the line for the lanthanide contraction seen in the rare earth Co-In 115 sequence. This argument is quite suggestive of intermediate valence behavior in PuCoGa 5 .
It appears from the considerations above that the 4f and 5f superconductors fit into similar classification. The surprise then is that we have not found, for example, the 20 K Ce-based superconductor. It seems that it should be there. | 2019-01-17T19:27:42.170Z | 2013-08-01T00:00:00.000 | {
"year": 2013,
"sha1": "2a0665515e8b0af5b130941e46ede53214b2e117",
"oa_license": "CCBY",
"oa_url": "https://escholarship.org/content/qt1fv5025q/qt1fv5025q.pdf?t=oam2wv",
"oa_status": "GREEN",
"pdf_src": "SpringerNature",
"pdf_hash": "30b5d9fed76431b7d864cf0c509d1c02d1f3004f",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Physics"
]
} |
80592149 | pes2o/s2orc | v3-fos-license | Improving Patient and Employee Safety through Implementation of an Infection Risk Screening Process for International Patients at Boston Children’s Hospital—The “AIRSHIP” Protocol
Abstract Background Vaccine-preventable diseases and multi-drug-resistant organisms (MDROs) are common outside of the US, and multiple infectious outbreaks have been linked to travelers. Boston Children’s Hospital cared for 2796 international patients in 2016 but lacked an infection risk screening process for these patients, placing patients and staff at risk. We developed the Assessing Infection Risks for Safe Healthcare of International Patients (AIRSHIP) protocol to identify risks to guide infection prevention and control (IPC) measures. Methods A multidisciplinary team of IPC, infectious diseases, and International Health Services (IHS) experts assessed current IHS intake procedures and stakeholder engagement. We then developed AIRSHIP, devising standardized processes and forms to (1) assess underimmunization, MDRO and tuberculosis history, recent exposures, and current symptoms and (2) triage cases for catch-up immunization, urgent healthcare evaluation, and/or IPC intervention (Figure 1). We piloted incorporation of AIRSHIP into existing intake procedures. We tracked process, outcome, and balancing measures to evaluate feasibility, effectiveness, and acceptability to families (Figure 2) and made iterative improvements through Plan-Do-Study-Act (PDSA) cycles. Results For our first 13 cases, we completed pre-arrival family and referring provider interviews in 5 cases and on-arrival family interviews in 8 cases (in no cases were both pre-arrival and on-arrival interviews feasible). We were able to assign a risk category in all cases, identifying 5 patients with infection risks (38%) and 4 who were undervaccinated (30%). In 7 of 8 cases (88%) in which on-arrival interviews were performed, the interview and referring provider records yielded complete and reliable data. The average duration of family interviews was 18 minutes. All 13 families reported being “very satisfied” with AIRSHIP. Conclusion International patients often present with active infections and are commonly undervaccinated. A feasible and effective strategy for infection risk screening of international patients is review of records pre-arrival, together with on-arrival family interview to gather additional data and identify acute symptoms and exposures. Disclosures All authors: No reported disclosures.
Background. The national opioid epidemic has been accompanied by precipitous increases in overdose deaths and hospitalizations for infectious complications of injection drug use (IDU). Despite this, there is scant literature addressing the topic. We aimed to describe demographic characteristics, type of infection, healthcare utilization, disposition and outcomes among patients hospitalized for IDU-related infection over a multi-year period at a large tertiary care referral center in Boston, MA.
Methods. We conducted a retrospective chart review of patients hospitalized for IDU-related infection from 1/1/2012-9/30/2015. 901 charts were initially identified using administrative codes; 234 met the following inclusion criteria: 1) hospitalization within the study period for treatment of ≥1 of 6 selected infections and 2) IDU within 6-months preceding qualifying hospitalization. During the study period, 234 patients had 488 cumulative admissions. Admissions for IDU-related infection and ≤30-day readmission, all-cause, underwent detailed abstraction (N = 338; 69%).
Conclusion. Our study describes the characteristics of patients hospitalized with IDU-related infection over a multi-year period in a region highly impacted by the opioid epidemic. High rates of hospital readmission, prolonged antibiotic therapy and out-of-hospital death were common in this young cohort.
Disclosures. All authors: No reported disclosures. Background. Real-time data collection of respiratory disease is important for understanding the spatiotemporal dynamics of disease transmission in the US. Healthcare professionals use tools such as FluView to help identify local pathogen circulation; however, these tools are limited to syndromic surveillance, and track a limited set of pathogens. Understanding respiratory disease dynamics requires 1) a large, pathogen rich data set 2) geographically dispersed data sources, and 3) fine temporal resolution. Here we utilize FilmArray® Trend, a research epidemiology system containing exported data from FilmArray® Respiratory Panel (RP) tests, to investigate geographic patterns of 20 common pathogens.
A Cloud Based Epidemiology
Methods. Over 6,000,000 individual pathogen assays from 19 clinical sites were exported to the Trend database from 2013 to present. Trend data were smoothed and normalized to produce the time series of pathogen incidence. A cross-correlation analysis was performed to compare sites to one another and determine offset of pathogen incidence. The results were plotted on a map of the US with visual indicators of correlation strength and directional movement as defined by cross-correlation lag values.
Results. The respiratory pathogens detected by the FilmArray RP test show a diverse set of spatial and temporal behaviors Most striking was the spread of the virus Coronavirus OC43, and Respiratory Syncytial Virus (RSV), with RSV traveling from east coast sites to west coast sites across the US over 20 days. In contrast Parainfluenza virus 3 (PIV3) shows a small cross-correlation lag across all of the Trend sites during the regular summer season, indicating near simultaneous onset of detection nationwide. A localized cluster of PIV3 in the winter of 2016 was observed in the midwest and west, identifying the significance of localized regional trends.
Conclusion. FilmArray Trend shows great promise in deciphering spatiotemporal dynamics of these common respiratory pathogens. This system can identify localized outbreaks and directional movement of pathogens over time. Future work with finer geographic distribution of contributing sites will aide in making conclusions regarding spatial dynamics of all 20 RP pathogens. Other pathogen transmission models may also be explored using this data set. | 2019-03-17T13:07:05.071Z | 2017-10-01T00:00:00.000 | {
"year": 2017,
"sha1": "57c39a179346902be856dee597e6b767a3fc7495",
"oa_license": "CCBYNCND",
"oa_url": "https://academic.oup.com/ofid/article-pdf/4/suppl_1/S237/20426229/ofx163.498.pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "eee18611f5fa2cc8cf5fe74092ba2408a6367e5f",
"s2fieldsofstudy": [
"Medicine",
"Environmental Science"
],
"extfieldsofstudy": [
"Medicine"
]
} |
24474538 | pes2o/s2orc | v3-fos-license | Comparison between Herbst appliances with or without miniscrew anchorage.
BACKGROUND
Herbst appliance is largely used in orthodontics for the correction of Class II. The aim of this paper was to analyze dental and skeletal effects of a splints Herbst-miniscrews combined device in comparison to a mandibular splints Herbst appliance.
MATERIALS AND METHODS
Fifty Class II division 1 patients (27 males and 23 females with a mean age of 11.8 ± 1.7 years) were included in the study. Lateral headfilms of 25 patients with a mandibular resin splint and a miniscrew anchorage (test group) and of 25 patients with mandibular acrylic resin splints (control group) were analyzed before (T0) and after (T1) the Herbst treatment. The mean and standard deviation (SD) of each variable were calculated; paired t-test was used to evaluate statistical changes before and after the treatment, in each group and Student t-test was used to compare the two groups.
RESULTS
Significant differences were observed for P < 0.05. At the end of the Herbst treatment, mandibular incisor proclination was significantly lower in the test group (2.8°) in comparison to the control group (7.4°).
CONCLUSIONS
The miniscrew-Herbst system, described in the present study, allows correction of Class II malocclusion, with a lower anchorage loss, in form of mandibular incisor proclination, during the treatment, in comparison to mandibular acrylic splints Herbst.
INTRODUCTION
Herbst appliance is largely used in orthodontics for the correction of Class II. The main points in favor are the short time required for the treatment and the fact that it does not need patient compliance. [1,2] Its effects are dental, including a posterior displacement of the upper dental arch and anterior displacement of the lower dental arch, and skeletal, such as a reduced sagittal growth of the maxilla and an enhanced sagittal growth of the mandible. It should be kept in mind that these skeletal effects vary among subjects, between sexes and with time of the therapy. [2] Several genetic studies have been done during the years. [3][4][5][6] It is well known that a point in disfavor of the Herbst treatment is a proclination of lower incisors due to the forces exerted on the lower teeth by the same telescope device. [7] Various modifications of the original Herbst such as the use of class III elastics, reduced and total cast splints, have been proposed, but none has been able to completely stop the proclination of mandibular incisors. [8] Weschler and Pancherz stated that the mandibular anchorage loss in Herbst treatment is a reality with which the orthodontist has to live and up until now there has been an agreement that flaring of the lower incisors cannot be prevented by any kind of anchorage system. [7] Not only did the introduction of the skeletal anchorage allow the simplification of many procedures conventionally employed for the control of anchorage, but also the reduction of the undesirable effects of many appliances too. [8] Moreover, miniscrews present many advantages, including low cost, low invasive insertion procedures and great versatility. Many authors have demonstrated that they can be used as successful sources of anchorage during orthodontic therapy. [9][10][11][12][13] To our knowledge, the possibility of combining Herbst appliance with skeletal anchorage has not been previously described in literature.
Therefore, the aim of this study was to analyze dental and skeletal effects of an acrylic Herbst-miniscrews combined device in comparison to a acrylic cast splints Herbst appliance, in the correction of Class II malocclusion.
MATERIALS AND METHODS
Patients who could benefit from Herbst treatment, who had a bilateral Angle Class II division 1 malocclusion, ≥1/2 cusp width, who were in the permanent or late mixed dentition, whose parents signed an informed consent form, were eligible for inclusion in the study. Patients were not included in the study if any of the following exclusion criteria were present: Poor oral hygiene and motivation, tooth agenesis or premature loss of permanent teeth, presence of second molars, transverse or vertical discrepancies, and incomplete available records.
All parents received thorough explanations and a written informed consent form prior to being enrolled in the study. Each case was accurately evaluated by a unique operator (AM) to assess the inter-arch relationships, panoramic radiographs, and lateral head films.
A total of 56 subjects were considered eligible for this study. Patients were allocated to either a test (combination of Herbst appliance with reduced mandibular acrylic splint, from first molar to first molar, and miniscrews) or control (Herbst with mandibular acrylic splint) group using a computerized random allocation process. A computer generated restricted randomization list was created. Only one of the investigators, not involved in selection and treatment of the patients, was aware of the randomization sequence and could have access to the randomization list. The randomized codes were enclosed in sequentially numbered, identical, opaque, and sealed envelopes.
Six patients refused to take part in this study and the final sample consisted of 50 patients, including 27 males and 23 females with a mean age of 11.8 ± 1.7 years.
In the test group, the miniscrews were applied in mandibular bone at the level of marginal or attached gingiva or mucogingival junction, between the lower first molar and second premolar. [14] The miniscrews employed (MAS, Micerium, Avegno, Italy) were titanium, 11 mm long, and shaped like a truncated cone with a diameter of 1.5 or 1.3 mm (according to the bone level) at the point and 2.2 mm at the neck. The shank of the screws was 1 mm in diameter, the threaded part had a length of 8 mm, and the heads featured a hexagonal slot to house the head of the screwdriver or contra-angle hand piece.
The mouth of each test patient was rinsed with 0.1% chlorhexidine gluconate solution and predrilling was carried out, and the miniscrews were inserted by means of a manual screwdriver.
According to the randomization sequence, a metallic or elastic ligature (100 g) linked the miniscrews to metallic buttons bonded to the lower canines of each side.
All patients included in the present study were treated by the same orthodontist (AM).
Lateral cephalograms were obtained for all patients before (T0) and at the end (T1) of the Herbst treatment to evaluate the outcome of the orthodontic therapy. No patients dropped out during the study.
The Sagittal Occlusion analysis of Pancherz (analysis of changes in sagittal occlusion) [15] was carried out manually for each patient by the same researcher blinded to the type of treatment received by the patient (MP), in order to analyze quantitatively the skeletal and dental structures.
This methodology was chosen in order that the results of this study would be comparable with the effects of various other Herbst devices described by other authors. [7] Occlusal line (OL) and Occlusal Line perpendicular (OLp) were transferred from the first lateral head film to the second by superimposition of the radiographs on stable bone structures of anterior cranial base. Furthermore, other parameters, including mandibular incisor proclination and cranial base-mandible angle were considered [ Figure 1].
All linear and angular measurements were taken to the nearest 0.5 mm and 0.5°, respectively. Moreover, all these measurements were performed twice, with a 7-day interval between the two recordings, in order to calculate Dahlberg's formula Method errors of the cephalometric variables were less than 1 mm, for linear measurements, and less than 1° for angular measurements.
The mean and standard deviation (SD) of each variable were calculated and paired t-test was used to evaluate statistical changes before and after the treatment, in each group. Student t-test was used to compare the two groups and significant differences we observed for P < 0.05.
RESULTS
All subjects of both groups had been successfully treated to an overcorrected bilateral Class I molar relationship.
Pretreatment and posttreatment records are shown in Table 1.
Total treatment duration (T0-T1) was comparable in the two groups, being 7.6 months for the test group and 7.5 months for the control group. No miniscrews were lost, or replaced, or became mobile during the treatment.
Maxillary and mandibular base
A slight maxillary base retrusion was achieved in both groups at the end of the Herbst treatment. At T0, the test group had an average A/OLp of 79.1 ± 4.3 mm, which had decreased by 0.4 to 78.7 ± 4.4 mm at T1 (P > 0.05) and the control subjects had an average A/OLp of 76.6 ± 3.4 mm at T0, which had decreased by 1 to 75.6 ± 3.7 mm at T1 (P > 0.05).
Comparing the groups with the t-test, no significant differences were observed (P > 0.05).
In the test group, an advanced position of Pg/OLp by 2.2 mm (from 81.1 ± 5.1 mm at T0 to 83.3 ± 5.8 mm at T1) was found at the end of the treatment (P < 0.05).
Comparing the groups with the t-test, no significant differences were recorded (P > 0.05).
No significant difference between the two groups was found (P > 0.05).
Maxillary and mandibular incisors
Maxillary incisors showed at the end of the treatment a slight incisal edge retrusion both in the test group (from Mandibular incisal edge advanced more in the test group (by 3 mm, from 80 ± 5.1 mm at T0 to 83±4.9 mm at T1; P < 0.05), than in the controls (by 2 mm, from 76.9 ± 4.2 mm at T0 to 78.9 ± 4.9 mm at T1; P < 0.05), although the difference between the groups was not statistically significant (P > 0.05).
The Herbst treatment improved similarly the overjet in both groups. At T0, the test group had an average overjet of 6.6 ± 2.4 mm, which at T1 had significantly decreased to 3.4 ± 1.4 mm (P < 0.05).
The control subjects had an average overjet of 6.6 ± 2.2 mm at T0, which had significantly decreased to 3.6 ± 1.5 mm at T1 (P < 0.05).
Comparing the groups with the t-test, no significant differences were recorded (P > 0.05).
Flaring of the lower incisors was noticed in all subjects. However, the mean mandibular incisor proclination in the test group, at the end of Herbst treatment, was lower (by 2.8°, from 100.5° ± 6° at T0 to 103.3 °± 5.7° at T1; P < 0.05) compared with the controls (by 7.4°, from 94.5° ± 4.7° at T0 to 101.9° ± 7.4° at T1; P < 0.05) and the difference between the two groups was statistically significant (P < 0.05).
Maxillary and mandibular molars
A similar maxillary molar distalization was achieved in both groups at end of the treatment. At T0, the test group had an average Ms/OLp of 55 ± 4.9 mm, which at T1 had decreased to 53.9 ± 4.4 mm (P < 0.05) and the control group had an average Ms/OLp of 51.7 ± 4.2 mm at T0, which had decreased to 49.6 ± 4.9 mm at T1 (P < 0.05). Comparing the groups with the t-test, no significant differences were recorded (P > 0.05).
A mesialization of lower molars was found both in the test group (by 3.7 mm, from 53.9 ± 5.7 mm at T0 to 57.6 ± 5.1 mm at T1; P < 0.05) and in the control subjects (by 2.7 mm, from 49.8 ± 4.7 mm at T0 to 52.5 ± 5.4 mm at T1; P < 0.05). The difference between the groups was not statistically significant (P > 0.05).
At T0, the test group had an average molar relationship of 1 ± 1.8 mm, which at T1 had significantly decreased to -3.8 ± 2.6 (P < 0.05). At T0, the control group had an average molar relationship of 1.8 ± 1.8 mm, which at T1 had significantly decreased to -2.6 ± 2.9 (P < 0.05).
Considering molar relationship, differences between the groups were not statistically significant (P > 0.05).
Cranial base-mandible angle
A slight anterior rotation of the mandible was found in the test group (SN/GoMe decreased from 33.5° ± 6.4° at T0 to 32.6° ± 6.3° at T1; P > 0.05), whereas in the control subjects a posterior rotation of the mandible was observed (SN/GoMe increased from 32.8° ± 5.6° at T0 to 33° ± 6.6° at T1; P > 0.05). However, the difference between the groups was not significant (P > 0.05).
DISCUSSION
Our results highlight that both types of Herbst treatments are efficient in the correction of Class II malocclusion. At the end of the therapy a bilateral first Class molar relationship was achieved in all patients, with a significant decrease of the overjet and the skeletal discrepancy.
Several factors contributed to these changes including: A slightly backward movement of the maxillary incisors, a forward movement of the mandibular incisors, a restraint of the forward movement of the maxilla and, in the test group, a forward movement of the mandible. Lucchese et al. and other authors conducted studies about the prediction of third molar eruption. [3][4][5][6]12,16,17] The treatment did not determine significant alterations of the cranial base-mandibular angle, and this is in agreement with that of other Herbst studies. [18,19] Results of the present study showed that in both groups of patients there was an increase of lower incisor proclination that is a general side effect of Herbst appliance treatment.
However, the combination of Herbst and miniscrews allowed a significantly better control of mandibular incisor proclination, in comparison with the control patients.
The combined miniscrew system has been shown to be able to consistently reduce mandibular incisor proclination, in comparison with other studies. [7] Hansen et al. [20] found a mandibular incisor proclination of 10.8° as a result of the total mandibular cast splint Herbst treatment.
Ruf et al. [21] observed a mean lower incisor proclination of 8.9° in 98 Class II total mandibular splint Herbst patients.
Von Bremen et al. in 2005 [22] observed the anchorage loss with reduced and total mandibular cast splints, during Herbst treatment, and found a mean proclination of mandibular incisors of 11.8° and 9.3°, respectively.
El-Fateh and Ruf [23] analyzed 100 Class II patients treated with reduced mandibular splint Herbst and recorded at the end of the treatment a mean lower incisors proclination of 12.9° ±4.6°. [17] Recently, a lingual appliance and Herbst combination with full control over mandibular incisor was introduced by Wiechmann et al. [24] However, these preliminary results need further investigations because of the small number of subjects included and the retrospective nature of the study.
In our research, the test group had lower proclination of incisors but a slightly more protruded mandibular incisor position (Is/OLp) than the control group.
These findings are contradictory as it is expected that a higher incisor proclination will result in increased incisal edge protrusion.
One possible explanation would be that the test group showed a slightly enhanced sagittal position of the mandible (Pg/OLp) in comparison to the control group. Furthermore, in the test group a slight anterior rotation of the mandible was observed.
The amount of forward movement of the mandible recorded in the test group was similar with that of other studies: Wigal et al. observed twenty-two patients with Class II division 1 malocclusion treated with an edgewise crowned Herbst appliance in the early mixed dentition and found a mean forward movement of the mandibular base of 2.0 mm. [25] According to our results, it might be speculated that a better mandibular incisor proclination control would allow a slightly mesial displacement of the mandible.
Taira et al. evaluated the effects of mandibular advancement plus prohibition of lower incisor movement on mandibular growth in rats and found that mandibular growth was accelerated before and during the pubertal period by mandibular advancement with a fixed functional appliance combined with prohibition of labial movement of the lower incisor. [26] We also have to consider that skeletal effects vary between sexes and with time of the treatment. A point in favour of this study is that the two groups of subjects were similar for age and sex.
On the other hand a limitation of the study was the relatively small sample size.
Another sign of anchorage loss due to the Herbst appliance is the advancement of the lower molars observed at the end of the treatment, despite the fact that in both groups splints reached the first mandibular molars. Thus, an active sagittal displacement of the mandibular molars was not avoided by the use of dental or skeletal anchorage systems. It might be assumed that the lower arch mesial displacement is partly due both to the mandible advancement and dental anchorage loss.
CONCLUSION
It can be concluded that the miniscrew-Herbst system, described in the present study, allows correction of Class II malocclusion, with a slight lower incisor proclination during treatment.
Further investigations should be carried out increasing the number of patients involved in the survey, in order to confirm the present findings. | 2018-04-03T00:48:17.934Z | 2012-12-01T00:00:00.000 | {
"year": 2012,
"sha1": "1db70d8fa5293295eb222eebf9d3f55381505e6c",
"oa_license": "CCBYNCSA",
"oa_url": null,
"oa_status": null,
"pdf_src": "Adhoc",
"pdf_hash": "9bf6335b45f1ff11dc05bc5ff2fb5aca0040fed9",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
67771144 | pes2o/s2orc | v3-fos-license | Fabrication of AlCoCrFeNi High-Entropy Alloy Coating on an AISI 304 Substrate via a CoFe2Ni Intermediate Layer
Through laser metal deposition, attempts were made to coat AlCoCrFeNi, a high-entropy alloy (HEA), on an AISI 304 stainless steel substrate to integrate their properties. However, the direct coating of the AlCoCrFeNi HEA on the AISI 304 substrate was found to be unviable due to cracks at the interface between these two materials. The difference in compositional change was suspected to be the source of the cracks. Therefore, a new transition route was performed by coating an intermediate layer of CoFe2Ni on the AISI 304 substrate. Investigations into the microstructure, phase composition, elemental composition and Vickers hardness were carried out in this study. Consistent metallurgical bonding was observed along both of the interfaces. It was found that the AlCoCrFeNi alloy solidified into a dendritic microstructure. The X-ray diffraction pattern revealed a transition of the crystal structure of the AISI 304 substrate to the AlCoCrFeNi HEA. An intermediate step in hardness was observed between the AISI 304 substrate and the AlCoCrFeNi HEA. The AlCoCrFeNi alloy fabricated was found to have an average hardness of 418 HV, while the CoFe2Ni intermediate layer had an average hardness of 275 HV.
Introduction
As a novel metallic alloy system, high-entropy alloys (HEAs) have received considerable attention in the past decade. The name HEA indicates that the mixing of the principal elements in the alloy leads to a substantial change in entropy. This change in entropy promotes the formation of a simple solid solution instead of complex compounds. One of the extensively studied HEAs is equiatomic AlCoCrFeNi, which shows high hardness, good wear behavior but low tensile ductility [1][2][3][4][5][6][7]. As-cast AlCoCrFeNi alloy showed a tensile elongation of 1.0%, while post-heat treatment, the elongation was increased to 11.7% [4]. Wang et al. studied the compressive properties of AlCrFeCoNi HEA prepared by vacuum arc melting. They found that this alloy showed large strain hardening and compressive strength up to 2004 MPa with a 32.7% compressive plasticity [6]. Munitz et al. reported the impact of heat treatment of AlCoCrFeNi HEA, in which the BCC (Body-centered cubic) matrix transformation occurred between 650 and 975 • C. This transformation led to a substantial increase in microhardness [5]. Further modification of this alloy system through the addition of titanium, leading to AlCoCrFeNiTi x (x = molar ratios), was found to be promising for wear protection [1]. Further, AlCoCrFeNi HEA solidified with dendritic and interdendritic microstructures due to elemental segregation. Dendritic segregation regions were found to be Al-and Ni-rich, while interdendritic areas were Fe-and Cr-rich, and the distribution of Co was uniform. Body-centered cubic (BCC) Fe and Cr precipitates, and B2 (ordered BCC) Al-and Ni-rich matrices were observed in previous studies [2,4,6,8,9]. Most of these studies are based on material fabricated through processes such as casting and arc melting. Unlike these early studies, laser metal deposition (LMD) was implemented in this study.
LMD is capable of fabricating freeform three-dimensional metallic components [10][11][12] and has been used to fabricate several HEAs [12][13][14][15][16]. Chen et al. fabricated Al x CoFeNiCu 1-x (x = 0.25, 0.5 and 0.75 atom %, respectively) HEAs using elemental powders on the AISI 304 substrate. They reported an increase in hardness with an increase in aluminum content [16]. He et al. used laser cladding to produce FeCoCrNiAlTi x (x = 0, 0.25, 0.5, 0.75 and 1 atom %, respectively) coating on Q253 steel through the use of elemental powders. Addition of titanium was observed to improve the hardness and wear resistance of the HEA [15]. Similarly, FeCoCrAlCu HEA coating by laser cladding demonstrated good wear resistance under a dry sliding condition [17].
In this paper, the feasibility of coating an AlCoCrFeNi HEA on an AISI 304 stainless steel substrate was investigated. Sole LMD fabrication of AlCoCrFeNi HEA components is very costly due to the need for high-purity (i.e., 99.9%) raw powders of elements such as Co, Cr and Ni. AISI 304 stainless steel, on the other hand, is a low-cost structural material. However, AISI 304 is a soft material with low wear resistance. It is widely used in industrial facilities, transportation equipment and architectural applications. Therefore, by coating AlCoCrFeNi HEA on AISI 304, it can enhance the hardness of AISI 304 structures. This combination of materials could facilitate fabrication of components for applications that require both hardness and wear resistance.
However, direct coating of AlCoCrFeNi HEA on AISI 304 is difficult due to the change in chemistry, thermal expansion and residual stress of the dissimilar materials. For example, the measured coefficient of thermal expansion (CTE, 10 −6 /K) for AlCoCrFeNi HEA was 9.03 (293-303 K), 12.47 (368-378 K) and 13.54 (423-773 K) [18]. However, the CTE values of AISI 304 were 14.7 (293 K), 16.3 (400 K), 19.5 (700 K) and 20.2 (800 K) [19]. Harihar et al. observed crack formation at the bottom of an AlCoCrFeNi deposit when deposited on an AISI 304 substrate. Due to the brittleness of the deposited material, the deposit broke off from the AISI 304 substrate easily [12]. An extensive network of cracks occurred when a TiVCrAlSi HEA was cladded on a Ti-6Al-4V substrate. This was attributed to the difference between the thermal expansion coefficients and residual stresses associated with the high cooling rate in laser cladding [20]. Therefore, to facilitate the dissimilar material bond, an intermediate layer was necessary and could accommodate the residual stresses and variation in chemistry change [10,21,22]. Intermediate layers of Fe/Cr/V were used between AISI 316 stainless steeland Ti-6Al-4V to facilitate a similar material bond [10]. Currently, there are few studies available identifying the viable intermediate layer between AlCoCrFeNi HEA and AISI 304. In this study, an attempt was made to coat the equiatomic AlCoCrFeNi HEA on the AISI 304 substrate using LMD. The objective was to obtain a strong bond between the two materials. We first demonstrated the issues with direct-coating the HEA onto the substrate. Then we proposed a candidate intermediate material and proved its viability.
Materials and Methods
Elemental powders of gas-atomized aluminum (Al), chromium (Cr), cobalt (Co), nickel (Ni) and iron (Fe) from Atlantic Equipment Engineers Inc. were used as precursor materials. These powders, weighed in required ratios, were mixed using a Turbula mixer (Glen Mills Inc., Clifton, NJ, USA) for 1 h to obtain homogeneous blends. Commercially procured AISI 304 bar stock (dimensions: 2.75 inch × 2 inch × 0.25 inch) was used as the substrate material for the deposition. The particle size distribution of the elemental powders stated by the producer is as tabulated in Table 1. Elemental analysis of the elemental powders is listed in Table 2. Elemental compositions (atom %) of the as-blended CoFe 2 Ni intermediate layer and AlCoCrFeNi alloy are given in Table 3. The laser deposition process was performed in an LMD system whose schematic representation is as seen in Figure 1a. The heat source was a 1 kW continuous-wave YAG fiber laser (IPG Photonics, Oxford, MA, USA) with a 2 mm beam diameter. The powders were fed using a vibration X2 powder feed system procured from Powder Motion Labs. The powder was introduced into the melt pool through an alumina tube. A computer numerical control (CNC) table was used to facilitate the movement during the deposition. Argon gas was used to ensure an inert atmosphere and act as a carrier gas to deliver the powder mixture to the melt pool.
In the current setup, the 2 mm spot size is insufficient to attain a large capture efficiency of the powder. This is due to the scatter of the powder flow out of the powder feed tube. This scatter was suspected to vary with individual precursor powder. Therefore, in order to obtain as-deposited compositions that are close to as-blended compositions, the capture efficiency during the deposition process needed to be increased. A trochoidal toolpath (shown in Figure 1b) was designed to create a large enough melt pool to improve capture efficiency during deposition. This toolpath was inspired by "weave"-style toolpaths that are commonly used in welding.
The AISI 304 substrates were cleaned with acetone to remove the impurities such as dirt and oil from the surface. A preheating scan was conducted by running the laser across the substrate surface. To ensure a successful start, the power of the initial five layers of the deposition was carried out at 750 W and 8.5% (3.36 g/min) powder feed rate. The remainder of the deposit was run at a power level of 550 W and 8.5% (3.36 g/min) powder feed rate. The thickness of each layer is 1 mm.
by "weave"-style toolpaths that are commonly used in welding.
The AISI 304 substrates were cleaned with acetone to remove the impurities such as dirt and oil from the surface. A preheating scan was conducted by running the laser across the substrate surface. To ensure a successful start, the power of the initial five layers of the deposition was carried out at 750 W and 8.5% (3.36 g/min) powder feed rate. The remainder of the deposit was run at a power level of 550 W and 8.5% (3.36 g/min) powder feed rate. The thickness of each layer is 1 mm. After laser deposition, vertical transverse sections of the specimens were cut using a wire electric discharge machine (Hansvedt Industries Inc., Rantoul, IL, USA) and mounted in Bakelite for polishing and etching. The metallographic specimens were first ground using 240, 400, 600 and 800 grit silicon carbide papers and then polished using 15 µm, 9 µm and 3 µm diamond suspensions. The final step of polishing involved 0.05 µm colloidal silica suspension. To reveal the microstructure, the electrolytic etching was carried out in the nitric acid solution (70 mL nitric acid, 30 mL distilled water) at 5 V for 5 seconds. Scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDS) and electron backscatter diffraction (EBSD) were performed on Helios Nanolab 600 SEM (Thermo Fisher Scientific, Waltham, MA, USA). The SEM image was acquired by an Everhart-Thornley detector. The EDS element was analyzed by the factory standardizations provided in the Aztec software. The EBSD step size was selected to be 2.5 µm. EBSD data acquisition and analysis were conducted using Aztec and Channel 5 software, respectively. Grain size was measured by the line intercept method, and the misorientation angle was 10 • . Optical microscopy images were collected using a Hirox optical microscope. X-ray diffraction patterns were collected using Philips X'pert MRD using Cu anode. The Vickers hardness was measured using a Struers Duramin hardness tester (Struers Inc., Cleveland, OH, USA) at a 9.8 N load and a 10 s load duration. The reported hardness results were the average of three indentations.
Direct Coating of AlCoCrFeNi HEA on AISI 304 Substrate
The direct LMD of the AlCoCrFeNi HEA on the AISI 304 substrate will be discussed first. Figure 2a shows a portion of the vertical transverse section of the HEA deposit near the AISI 304 substrate. An area close to the crack zone, as marked in the dashed-line box, is shown in Figure 2b with high magnification. A network of cracks, mostly transverse and horizontal in orientation, were found to be prevalent. Cracking occurred at the bottom of this HEA deposit. This could be attributed to the mismatch between the thermal expansion coefficients. The CTE of this HEA was reported to be 9.03 (10 −6 /K, 293-303 K) while the value of AISI 304 was 14.7 (10 −6 /K, 293 K) [18,19]. 2a shows a portion of the vertical transverse section of the HEA deposit near the AISI 304 substrate. An area close to the crack zone, as marked in the dashed-line box, is shown in Figure 2b with high magnification. A network of cracks, mostly transverse and horizontal in orientation, were found to be prevalent. Cracking occurred at the bottom of this HEA deposit. This could be attributed to the mismatch between the thermal expansion coefficients. The CTE of this HEA was reported to be 9.03 (10 −6 /K, 293-303 K) while the value of AISI 304 was 14.7 (10 −6 /K, 293 K) [18,19]. The elemental composition distribution along the interface between the HEA deposit and the AISI 304 substrate is shown in Figure 3. At the bottom of the melted metal, the composition mixing was significant during the laser deposition process (see Figure 3). The bottom of the deposit had high susceptibility of cracking in the transverse cross-section, as seen in Figure 2. The elemental composition distribution along the interface between the HEA deposit and the AISI 304 substrate is shown in Figure 3. At the bottom of the melted metal, the composition mixing was significant during the laser deposition process (see Figure 3). The bottom of the deposit had high susceptibility of cracking in the transverse cross-section, as seen in Figure 2.
The variation in Vickers hardness across the HEA-AISI 304 direct coating is presented in Figure 4. The average Vickers hardness of the HEA deposits was 412 HV, while that of the substrate was 161 HV. Since the coefficients of thermal expansion are mismatched between HEA and the substrate, residual stresses were developed during the laser deposition process. The AISI 304 substrate had a high elongation rate from 28% to 50% in the temperature range of 300-500 • C [23]. However, the tensile elongations of the AlCoCrFeNi HEA were 1% (as-cast condition) and 11.7% (after heat treatment) [4]. A difference in ductility exists between the substrate and the HEA. Having an intermediate material to bridge these differences was deemed necessary. The variation in Vickers hardness across the HEA-AISI 304 direct coating is presented in Figure 4. The average Vickers hardness of the HEA deposits was 412 HV, while that of the substrate was 161 HV. Since the coefficients of thermal expansion are mismatched between HEA and the substrate, residual stresses were developed during the laser deposition process. The AISI 304 substrate had a high elongation rate from 28% to 50% in the temperature range of 300-500 °C [23]. However, the tensile elongations of the AlCoCrFeNi HEA were 1% (as-cast condition) and 11.7% (after heat treatment) [4]. A difference in ductility exists between the substrate and the HEA. Having an intermediate material to bridge these differences was deemed necessary.
A New Transition Route
A blend of Fe, Co and Ni powders was selected as the candidate intermediate material. Since they are among the constituents of the AlCoCrFeNi HEA, no special procurement was needed. A Fe-Co-Ni ternary phase diagram at 1073 K compiled from experimental data is shown in Figure 5 [24]. Fe, Ni and Co have excellent mutual solubility, and no brittle intermetallic phases are expected. From the phase diagram, an atomic composition ratio of Fe, Ni and Co of 50%, 25% and 25%, respectively, was chosen. The selected ratio is expected to bridge the material composition gap between the AlCoCrFeNi HEA and AISI 304. This new transition route, AISI 304 substrate → CoFe2Ni intermediate layer → AlCoCrFeNi HEA, was then carried out and characterized.
A New Transition Route
A blend of Fe, Co and Ni powders was selected as the candidate intermediate material. Since they are among the constituents of the AlCoCrFeNi HEA, no special procurement was needed. A Fe-Co-Ni ternary phase diagram at 1073 K compiled from experimental data is shown in Figure 5 [24]. Fe, Ni and Co have excellent mutual solubility, and no brittle intermetallic phases are expected. From the phase diagram, an atomic composition ratio of Fe, Ni and Co of 50%, 25% and 25%, respectively, was chosen. The selected ratio is expected to bridge the material composition gap between the AlCoCrFeNi HEA and AISI 304. This new transition route, AISI 304 substrate → CoFe 2 Ni intermediate layer → AlCoCrFeNi HEA, was then carried out and characterized. layer composition was theorized to avoid the formation of intermetallic compounds and bridge the large gap in strength differences. Figures 6a and 6b show the optical images of etched surfaces of transverse sections of these deposits. Unlike the HEA-AISI 304 direct coating, no apparent cracks were observed, which indicated an improvement in bonding. However, issues of microporosity persisted. A dendrite microstructure was observed along the interface between the intermediate layer and the HEA. A high-magnification secondary electron image of the AlCoCrFeNi HEA deposit is shown in Figure 7, where a two-phase dendritic microstructure was observed. The area fraction of the dendritic microstructure was ~52%, while the interdendritic area fraction was ~48%. The interdendritic region is named A, and the dendritic region is named B. The mean elemental compositions of A and B (average from three arbitrary points) were analyzed by EDS, and the results are listed in Table 4. It is shown that the atomic percentages of Al and Ni were ~29% in A and ~41% in B. The percentages of Fe and Cr were ~54 atom % in A and 43 atom % in B. These results indicate that Fe and Cr were rich in A, while Al and Ni were rich in B. The composition of Co did not show evident differences between A high-magnification secondary electron image of the AlCoCrFeNi HEA deposit is shown in Figure 7, where a two-phase dendritic microstructure was observed. The area fraction of the dendritic microstructure was~52%, while the interdendritic area fraction was~48%. The interdendritic region is named A, and the dendritic region is named B. The mean elemental compositions of A and B (average from three arbitrary points) were analyzed by EDS, and the results are listed in Table 4. It is shown that the atomic percentages of Al and Ni were~29% in A and~41% in B. The percentages of Fe and Cr were~54 atom % in A and 43 atom % in B. These results indicate that Fe and Cr were rich in A, while Al and Ni were rich in B. The composition of Co did not show evident differences between A and B. The mixing enthalpies between Fe-Cr, Fe-Ni, Fe-Co, Fe-Al, Cr-Ni, Cr-Co, Cr-Al, Ni-Co, Ni-Al and Co-Al were −1, −2, −1, −11, −7, −4, −10, 0, −22 and −19 kJ/mol, respectively [6,25]. The mixing enthalpy of Al and Ni was higher than other pairs, which indicated that Al and Ni tended to form atomic pairs and segregate. Similar results have been reported for the AlCoCrFeNi HEA, with this microstructure being attributed to the spinodal decomposition [2,[4][5][6]9]. Figure 8. The present phases and the corresponding crystallographic information are summarized in Table 5. The peak patterns of FCC were observed in the CoFe2Ni intermediate layer, while BCC peak patterns were detected in the AlCoCrFeNi alloy. Löbel et al. found BCC and B2 (ordered BCC) phases in AlCoCrFeNiTix (x = 0) when fabricated via arc melting [1]. A similar result was reported by Shiratori et al., when casting was employed to produce an AlCoCrFeNi HEA [26]. Due to the same basic lattice structure and lattice parameters, the B2 ordered structure is very hard to detect from XRD, as the peak patterns of B2 and BCC are the same [2,9]. However, the Figure 8. The present phases and the corresponding crystallographic information are summarized in Table 5. The peak patterns of FCC were observed in the CoFe 2 Ni intermediate layer, while BCC peak patterns were detected in the AlCoCrFeNi alloy. Löbel et al. found BCC and B2 (ordered BCC) phases in AlCoCrFeNiTi x (x = 0) when fabricated via arc melting [1]. A similar result was reported by Shiratori et al., when casting was employed to produce an AlCoCrFeNi HEA [26]. Due to the same basic lattice structure and lattice parameters, the B2 ordered structure is very hard to detect from XRD, as the peak patterns of B2 and BCC are the same [2,9]. However, the evidence of the existence of the B2 phase was found from the EDS analysis above. Previously, an AlCoCrFeNi HEA was reported to also contain the FCC crystal structure with preheating or post-heat treatment [5,13,26]. The FCC structure was not found in this work, which could be because the high cooling rate during LMD inhibited the formation of the FCC crystal structure [5,13,26]. Figure 9b. The constituents of the AlCoCrFeNi HEA were detected by EDS (Al: ~16-17 atom %, Co: 19-20 atom %, Cr: ~17 atom %, Fe: ~25 atom %, Ni: ~20-21 atom %). The difference between the as-blended (20 atom %) and as-deposited aluminum (~16-17 atom %) percentages is suspected to be a consequence of inconsistency in capture efficiencies of the constituent powders, and evaporation due to differences in melting point. Al and Cr were present in the intermediate layer as seen in Figure 9b, and their total content was ~4-5 atom %. The evolution in chemistry from the intermediate layer to the substrate was characterized by an EDS line scan first. The quantitative results are shown in Figure 9a. The EDS measured results of the AISI 304 substrate (Cr:~18-19 atom %, Fe:~70-72 atom %, Ni:~9-10 atom % in Figure 9a) did not vary from the nominal AISI 304 elemental compositions. Mn (~1-2 atom %) was detected in the AISI 304 substrate by EDS but is not shown in Figure 9. 20-21 atom %). The difference between the as-blended (20 atom %) and as-deposited aluminum (~16-17 atom %) percentages is suspected to be a consequence of inconsistency in capture efficiencies of the constituent powders, and evaporation due to differences in melting point. Al and Cr were present in the intermediate layer as seen in Figure 9b, and their total content was~4-5 atom %.
AISI 304 substrate (Cr: ~18-19 atom %, Fe: ~70-72 atom %, Ni: ~9-10 atom % in Figure 9a) did not vary from the nominal AISI 304 elemental compositions. Mn (~1-2 atom %) was detected in the AISI 304 substrate by EDS but is not shown in Figure 9. The percentages of Co (~17-22 atom %) and Ni (~21-23 atom %) reduced, while the Fe (~54-56 atom %) content increased from the intermediate layer to the AISI 304 substrate. A small amount of Cr (~3-5 atom %) was present in the intermediate layer, because the substrate was mixed with the intermediate layer. The composition distribution from the HEA to the intermediate layer is shown in Figure 9b. The constituents of the AlCoCrFeNi HEA were detected by EDS (Al: ~16-17 atom %, Co: 19-20 atom %, Cr: ~17 atom %, Fe: ~25 atom %, Ni: ~20-21 atom %). The difference between the as-blended (20 atom %) and as-deposited aluminum (~16-17 atom %) percentages is suspected to be a consequence of inconsistency in capture efficiencies of the constituent powders, and evaporation due to differences in melting point. Al and Cr were present in the intermediate layer as seen in Figure 9b, and their total content was ~4-5 atom %. Figure 10a shows the inverse pole figure (IPF) map obtained from the bottom of the HEA section of the specimen. The measured area was approximately 3.4 mm × 1.2 mm of the cross-section parallel to the build direction (BD), which spanned from the left to the right of the specimen. The difference in color indicates the different crystallographic orientations. From Figure 10a, the overall constitution can be classified into two zones-the edge zone (1 and 3) and the middle zone (2). In areas 1 and 3, the grains were observed to be elongated along the build direction (see 1 and 3 in Figure 10a). The distributions of the intercept lengths (using 100 horizontal lines) in different areas are depicted Figure 10b. The median linear intercept for areas 1 and 3 was 72.5 µm, while it was 127.5 µm for area 2. From the linear intercept distribution of area 2, 25% of the intercept values were greater than 300 µm, whereas only 14% of the intercept values were above 300 µm for areas 1 and 3. This grain morphology is likely to be a consequence of deposition toolpath and variation in cooling rate at edges and in the middle [27,28]. Figures 10c and 10d show the {100}, {110} and {111} pole figures of different areas, which give the distribution of the pole density along the build direction. The pole figure of the areas 1 and 3 (Figure 10c) suggests that the orientations of the grains were close to the <100> direction. However, the grains were random in orientation and did not appear with obvious texture in area 2 (Figure 10d). Further study is necessary to investigate the impact of this toolpath on the grain morphology. Figure 10a, the overall constitution can be classified into two zones-the edge zone (1 and 3) and the middle zone (2). In areas 1 and 3, the grains were observed to be elongated along the build direction (see 1 and 3 in Figure 10a). The distributions of the intercept lengths (using 100 horizontal lines) in different areas are depicted Figure 10b. The median linear intercept for areas 1 and 3 was 72.5 µm, while it was 127.5 µm for area 2. From the linear intercept distribution of area 2, 25% of the intercept values were greater than 300 µm, whereas only 14% of the intercept values were above 300 µm for areas 1 and 3. This grain morphology is likely to be a consequence of deposition toolpath and variation in cooling rate at edges and in the middle [27,28]. Figure 10c Figure 10c) suggests that the orientations of the grains were close to the <100> direction. However, the grains were random in orientation and did not appear with obvious texture in area 2 (Figure 10d). Further study is necessary to investigate the impact of this toolpath on the grain morphology.
EBSD
distributions of the intercept lengths (using 100 horizontal lines) in different areas are depicted Figure 10b. The median linear intercept for areas 1 and 3 was 72.5 µm, while it was 127.5 µm for area 2. From the linear intercept distribution of area 2, 25% of the intercept values were greater than 300 µm, whereas only 14% of the intercept values were above 300 µm for areas 1 and 3. This grain morphology is likely to be a consequence of deposition toolpath and variation in cooling rate at edges and in the middle [27,28]. Figures 10c and 10d show the {100}, {110} and {111} pole figures of different areas, which give the distribution of the pole density along the build direction. The pole figure of the areas 1 and 3 (Figure 10c) suggests that the orientations of the grains were close to the <100> direction. However, the grains were random in orientation and did not appear with obvious texture in area 2 ( Figure 10d). Further study is necessary to investigate the impact of this toolpath on the grain morphology. Table 6 lists the Vickers hardnesses of the AlCoCrFeNi HEA, annealed AISI 304, aged Inconel 625, and annealed duplex steel SAF 2205 [29][30][31]. The average Vickers hardness of the HEA deposit was in the Figure 11 gives the Vickers hardness distribution of the AlCoCrFeNi HEA deposited on the AISI 304 substrate with the CoFe 2 Ni intermediate layer. The Vickers hardness of the CoFe 2 Ni intermediate layer was around the 275 HV, which could be attributed to the solid solution strengthening. Table 6 lists the Vickers hardnesses of the AlCoCrFeNi HEA, annealed AISI 304, aged Inconel 625, and annealed duplex steel SAF 2205 [29][30][31]. The average Vickers hardness of the HEA deposit was in the range of 418 HV, because of the second-phase strengthening [4].
Vickers Hardness Analysis
According to the XRD results, the AISI 304 substrate and the CoFe 2 Ni intermediate layer had an FCC structure, while the AlCoCrFeNi HEA had a BCC structure. The transition from FCC to BCC structure is also expected to enhance the hardness. The high hardness is expected to correlate with good performance in strength and wear resistance [1,16]. According to the XRD results, the AISI 304 substrate and the CoFe2Ni intermediate layer had an FCC structure, while the AlCoCrFeNi HEA had a BCC structure. The transition from FCC to BCC structure is also expected to enhance the hardness. The high hardness is expected to correlate with good performance in strength and wear resistance [1,16].
Conclusions
An AlCoCrFeNi HEA was coated on an AISI 304 substrate by laser metal deposition (LMD) technology. The coating on the substrate without and with the intermediate layer was characterized and discussed. The main conclusions are as follows: • Cracking was found to be prominent when the AlCoCrFeNi HEA was directly coated on the AISI 304 substrate due to the compositional change between HEA and the substrate.
Conclusions
An AlCoCrFeNi HEA was coated on an AISI 304 substrate by laser metal deposition (LMD) technology. The coating on the substrate without and with the intermediate layer was characterized and discussed. The main conclusions are as follows:
•
Cracking was found to be prominent when the AlCoCrFeNi HEA was directly coated on the AISI 304 substrate due to the compositional change between HEA and the substrate. | 2019-02-05T16:12:42.127Z | 2018-12-20T00:00:00.000 | {
"year": 2018,
"sha1": "3cd0fc86c06af0b25fec112879ea252146a59002",
"oa_license": "CCBY",
"oa_url": "https://doi.org/10.3390/e21010002",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "3cd0fc86c06af0b25fec112879ea252146a59002",
"s2fieldsofstudy": [
"Materials Science"
],
"extfieldsofstudy": [
"Medicine",
"Computer Science",
"Materials Science"
]
} |
119356731 | pes2o/s2orc | v3-fos-license | A simple gate for linear optics quantum computing
We describe a simple scheme for implementing the non-linear sign gate of Knill, Laflamme and Milburn (Nature, {\bf 409}, 46-52, Jan. 4 (2001)) which forms the basis of an experiment underway at the University of Vienna.
It was recently shown [1] that efficient quantum computing is possible using only linear optics, single photon sources and single photon detectors. One of the fundamental gates of the proposed scheme is known as the non-linear sign (NLS) gate; it is a non-deterministic gate which implements the transformation (on states of photon occupation number) given by with probability of success 1/4. The scheme suggested in [1] for implementing a NLS gate used a complicated interferometer requiring beamsplitters of variable reflectivity. We have found a scheme more amenable to an experimental demonstration of linear optics quantum computing, and it is depicted in Fig. 1. The input computational mode begins in a state of horizontal polarization, i.e. in general |ψ in = α|0 H +β|1 H +γ|2 H . This mode passes through a polarization rotator, of rotation angle σ. As with the proposal of [1], our scheme makes use of a single ancilla photon, in this case prepared with vertical polarization. The ancilla photon and computational mode are mixed at a polarizing beamsplitter (PBS). One output of the PBS goes to a detector D 1 , and the gate's success is conditioned on no photons being detected at D 1 . The other output mode of the PBS passes through a polarization rotator set to an angle θ. This mode is subsequently subjected to a measurement, and the gate operation is successful if a single vertical photon is detected. This is indicated in the figure by the addition of a second PBS and detector D 2 . In general the detector D 2 would need to be able to distinguish one from multiple photons, and such detectors are not readily available. However for the purposes of a 4 photon coincidence experiment such multiple photon events are excluded by the conditioning process, and so a simple demonstration of an NLS gate can be performed using commonly available single photon detectors.
If we take the transformation corresponding to a polarization rotator of angle x to be a † H → cos x a † H +sin x a † V , a † V → − sin x a † H + cos x a † V , then the (unnormalized) state of the output mode, given the correct conditioning at D 1 , D 2 , is [2] |ψ out = α cos θ|0 H + β cos σ cos 2θ|1 H +γ cos 2 σ cos θ(1 − 3 sin 2 θ)|2 H .
gives the desired transformation (1) with probability of success A 2 ≈ 0.227. This is slightly lower than the probability of success (0.25) for the scheme presented in [1], however the experimental simplification of our scheme is considerable. Furthermore the central result of [1], namely that this probability of success can be boosted arbitrarily close to 1 using only linear optics, single photon sources and single photon detectors, is unaffected by this small decrease in success probability of the NLS gate.
This work was supported by the Austrian Science Foundation FWF, and the TMR programs of the European Union Project No. ERBFMRXCT960087.
Note added. Recently a similar simplification of the NLS gate was presented [3] which shares the same probability of success as the one presented here. The scheme presented here enjoys the slight practical advantage of not requiring a beamsplitter with unequal reflectivity/transmittivity. | 2019-04-14T03:17:58.115Z | 2001-08-11T00:00:00.000 | {
"year": 2001,
"sha1": "499ca0bf0c6258ce9e59a01880fbecb268fb4932",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "2819fe8c45e7b09953963bace185c8339856e722",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Mathematics",
"Physics"
]
} |
118847253 | pes2o/s2orc | v3-fos-license | Ising Spin Network States for Loop Quantum Gravity: a Toy Model for Phase Transitions
Non-perturbative approaches to quantum gravity call for a deep understanding of the emergence of geometry and locality from the quantum state of the gravitational field. Without background geometry, the notion of distance should entirely emerge from the correlations between the gravity fluctuations. In the context of loop quantum gravity, quantum states of geometry are defined as spin networks. These are graphs decorated with spin and intertwiners, which represent quantized excitations of areas and volumes of the space geometry. Here, we develop the condensed matter point of view on extracting the physical and geometrical information out of spin network states: we introduce new Ising spin network states, both in 2d on a square lattice and in 3d on a hexagonal lattice, whose correlations map onto the usual Ising model in statistical physics. We construct these states from the basic holonomy operators of loop gravity and derive a set of local Hamiltonian constraints which entirely characterize our states. We discuss their phase diagram and show how the distance can be reconstructed from the correlations in the various phases. Finally, we propose generalizations of these Ising states, which open the perspective to study the coarse graining and dynamics of spin network states using well-known condensed matter techniques and results.
I. INTRODUCTION
The search for a quantum theory of gravity remains one of the major challenge of theoretical physics. Even if a complete theory of quantum gravity is still elusive, numerous approaches have shed lights on the problem itself and gave insights on the possible structure of the final theory. In general relativity, gravity does not evolve in a pre-assumed background geometry but moulds the geometry itself: the gravitational field is encoded directly as a curved and dynamical geometry. This background independence, conjugated to the major difficulty of constructing local diffeomorphism-invariant observables (see e.g. [1][2][3]), leads to the problem of reconstructing the geometry in quantum gravity from the information stored in the quantum state of the gravitational field.
The line of research we pursue here is at the interplay between condensed matter and quantum information on one side and quantum gravity on the other: the goal is to understand how the distance can be recovered from correlation and entanglement between sub-systems of the quantum gravity state. The issue of the emergence of distance is of course one aspect of a more general program to understand the emergence of geometry at a semi-classical level from a purely background independent quantum theory of gravity. To reconstruct a notion of locality and the geometry, the distance seems nevertheless to be the first natural quantity to define. Thus focusing on the distance, one could extract it from the 2-point correlation function assuming that the inverse square law holds. This perspective is quite similar to the point of view of spectral geometry and non-commutative geometry, in which distances, and more generally the ge- * alexandre.feller@ens-lyon.fr † etera.livine@ens-lyon.fr ometry, emerge from the spectral analysis of the 2-points function, given respectively by the Laplacian and Dirac operators [4,5]. Like in condensed matter physics, we expect the behavior of the correlations to depend on the considered phase. While the typical behavior of correlations is an exponential decay, the interesting regime for gravity, namely an algebraic decay, would be attained in critical regimes or algebraic phases of some specific model like the XY model. In this paper, we would like to propose a simple framework to test these ideas in the context of loop quantum gravity and we will introduce toy quantum states, which we dub Ising spin network states, whose correlations and phase diagram can be mapped onto the standard Ising model in statistical physics. This will allow us to discuss the relation between correlation and geometry in a well-under-control environment.
Loop quantum gravity is one of the most developed approach to quantum gravity and proposes a nonpertubative canonical quantization of general relativity (for textbooks, see [6][7][8]). The physical picture that emerges from this theory describes quantum space-time as a discrete geometry where areas and volumes operators have a discrete spectrum. The formalism is based on a 3 + 1 formulation of general relativity in terms of the Ashtekar-Barbero connection and a densitized triad, thus describing gravity as SU (2) gauge field theory, in fact a topological constrained field theory. In the quantum theory, spin network states form a basis of the wave functions that describes the quantum state of 3d space at the kinematical level. Physical spin network states will be the solution of the Hamiltonian constraints, which implement the Einstein equations and generate the dynamics.
The core of our investigation will be on correlations and entanglement entropy on spin network states. In the long run, the aim is to analyze the typical -possibly universalstructure of spin network states with well-behaved algebraic correlations and satisfying an area law entropy.
Correlations and especially entropy are of special importance for the understanding of black holes dynamics. Indeed, understanding the microscopic origin of black holes entropy is one the major test of any attempt to quantify gravity and entanglement between the horizon and its environment degrees of freedom appears crucial (for recent developments in loop quantum gravity, see [9,10]). The usual spin network basis states are eigenstates of the area and volume operators, but they carry trivial correlations because of their factorized structure with no entanglement at all between intertwiners degrees of freedom. We therefore look for more involved spin network states, correlating the spin and intertwiner degrees of freedom living on the whole graph. Having in mind statistical and condensed matter physics where studies of correlations are fairly advanced both numerically and analytically [11], the method we propose is to construct pure spin network states from well-known condensed matter models and possibly highlights some relevant lessons for loop quantum gravity. A similar endeavor is currently being investigated by Bianchi and collaborators, similarly using condensed matter methods, who introduce squeezed spin network states by a Bogoliubov transformation and focus on the evaluation of their entanglement entropy [12].
Ideally, testing the relation between correlations and distance requires physical states solutions of the Hamiltonian constraints of loop quantum gravity. However, these constraints are rather complicated and, despite several proposals, no consensus has been reached on the structure of physical states solving those constraints at the quantum level (see e.g. [13] for a review of the recent research efforts on the quantum dynamics of loop gravity). To circumvent this issue, we propose a different strategy: build test spin network states on regular lattices, with an a priori notion of distance, and compare that natural lattice distance with the reconstructed distance emerging from the correlation carried by the quantum states. Working in such a controlled setting should allow us to thoroughly illustrate the link between distance and correlation, test the viability of the proposal of reconstructing the distance entirely from correlations and understand the structures of well-behaved states supporting this proposal.
A spin network state is defined on a graph, dressed with spins on the edge and intertwiners at the vertices. A spin on an edge e is a half-integer j e ∈ N/2 giving an irreducible representation of SU(2) while an intertwiner at a vertex v is an invariant tensor, or singlet state, between the representations living on the edges attached to that vertex. Spins and intertwiners respectively carry the basic quanta of area and volume. We build our spin network states based on three clear simplifications: 1. We work on a fixed graph, discarding graph superposition and graph changing dynamics for now, and we will focus on working with a fixed regular lattice.
2. We freeze all the spins on all the graph edges. We fix them to their smallest possible value, 1 2 , which correspond to the most basic excitation of geometry in loop quantum gravity, thus representing a quantum geometry directly at the Planck scale.
3. We restrict ourselves to 4-valent vertices, which represent the basic quanta of volume in loop quantum gravity, dual to quantum tetrahedra.
Since the spins are frozen, the only degrees of freedom left are the intertwiners living at the 4-valent vertices. At each vertex, these recouple between 4 spins 1 2 , which corresponds to a two-dimensional Hilbert space or two-level quantum system with pairs of spins recoupling to a spin 0 or 1. This is completely equivalent to working with a qubit at each vertex. Qubits, as the elementary physical systems encoding information, are the basic ingredients of quantum information and quantum computing and are thus the basic building blocks of condensed matter models. Therefore these simplifications provide us with the perfect setting to map spin network states, describing the Planck scale quantum geometry, to qubit-based condensed matter models. Such models have been extensively studied in statistical physics and much is known on their phase diagrams and correlation functions, and we hope to be able to import these results to the context of loop quantum gravity. One of the most useful model is the Ising model whose relevance goes from modeling binary mixture to the magnetism of matter. We thus naturally propose to construct and investigate Ising spin network states.
In three space dimensions, the natural 4-valent regular lattice is the diamond lattice, which is a honeycomb lattice and can be seen as dual to a tetrahedral discretization of space. It is different from the usual cubic lattice used in lattice gauge theory, discretized gravity (à la Regge) or numerical general relativity, but we believe it is more natural and better suited to the loop quantum gravity context. This is nevertheless a brand new proposal to construct spin networks on the regular 3d diamond lattice. We will define 3d Ising spin network states on that lattice and analyze their properties. For mathematical simplicity and representation purposes, we will first focus on their two-dimensional counterpart and define 2d Ising spin network states on a square lattice. In fact, many obtained results are straightforwardly generalized to 3d. The 2d Ising model on a regular square lattice without magnetic field has been solved exactly [14] and exhibits a phase transition with algebraically decaying correlations at the critical coupling. The phase diagram can be studied using quantum field theory methods. Since away from the transition the correlations decay exponentially, the relevant scenario for gravity is indeed at the critical coupling which also has the major advantage to possess a continuum limit.
Thus this state appears to fall into the category of the test states we are looking for. Moreover, this method of investigation allows us to understand the typical structure of such states in terms of the holonomy operators of loop quantum gravity: non-trivial correlations seem to require holonomies around arbitrarily extended loops. We also show that this state is a unique solution of a set of local Hamiltonian constraints illustrating how local constraints can give rise to extended holonomies and to long range correlations, a common feature of condensed matter systems.
The paper is structured as follows. Section II reviews the definition of spin network and analyze the structure of 4-valent intertwiners between spins 1 2 leading to the effective two-state systems which we use to define the Ising spin network states. Different equivalent definitions are given in terms of the high and low temperature expansions of the Ising model. The loop representation of the spin network is then obtained and studied as well as the associated density which gives information about parallel transport in the classical limit. Section III introduces a set of local Hamiltonian constraints for which the Ising state is a unique solution and elaborates on their usefulness for understanding the coarse-graining of spin network [15] and the dynamic of loop quantum gravity. Section IV discusses the phase diagram of our Ising states and their continuum limit as well as the distance from correlation point of view. Section V end this paper with discussion and perspective.
A. Definition
In loop quantum gravity, kinematical states are cylindrical functionals of the Ashtekar-Barbero SU(2) connection that depend only on its holonomies along a set of paths which define an oriented graph Γ. We call E the number of edges of te graph Γ and V its number of vertices. These wave-functions of the geometry depend on E group elements in SU (2). We further require those states to be gauge invariant, which means an invariance under SU(2) transformations at each vertex. The Hilbert space of wave-functions 1 on the graph Γ is then L 2 (SU (2) E /SU (2) V , dµ) where the scalar product is defined by the SU(2) Haar measure dµ. A basis of this Hilbert space is given by spin networks states using the Peter-Weyl theorem stating that a basis of L 2 (SU(2), dµ) is given by the Wigner matrices of the SU(2) group element in all possible irreducible unitary representations. On each link e of the graph lives a spin j e associated to the area degrees of freedom. At 1 In fact, the true states of geometry in loop quantum gravity are defined as a sum over all possible graphs Γ thanks to a projective limit [16] of graphs H kin = lim Γ→+∞ H Γ . The measure on this space of generalized connections A is given by the Ashtekar-Lewandowski measure dµ AL . So H kin L 2 (A, dµ AL ). This projective limit allows to make sense rigorously of superposition of states living on different graphs. each vertex v lives an SU (2) invariant tensor called intertwiner i v ∈ Inv SU (2) ⊗ e v V je , with V je is the (2j e + 1)dimensional Hilbert space for the spin j e representation. These spin network basis states, as illustrated on fig.1, define the basic excitations of the quantum geometry and they are provided with a natural interpretation in terms of discrete geometry with the spins giving the quanta of area and the intertwiners giving the quanta of volume. This interpretation is most clear in the twisted geometry picture [17]. More explicitly, the spin network state |Γ, j e , i v , based on the graph Γ and dressed with spins j e and intertwiners i v , defines a wave function ψ(g e ) on the space of discrete connections of where the trace contracts the intertwiners with the Wigner matrix D je (g e ) representing the group element g e along the links between vertices. The scalar product between those states are obtained using the orthogonality relation of the Wigner matrices: Those spin network states, being tensor product of intertwiners, have trivial correlations. We would like to build quantum superposition of those basis spin network states that would cary non-trivial correlation. What we will develop below is a method to construct such states whose correlation structure maps onto the Ising model. We will then be able to translate the known quantities and results of statistical physics to the context of gravity.
As outlined in the introduction, we propose to define Ising spin network states by freezing the spins to their smallest value, j e = 1/2, that is the smallest quanta of area at the Planck scale, and to focus on 4-valent intertwiners. These 4-valent vertices will be organized along a regular lattice. We will consider the 3d diamond lattice and the 2d square lattice. We will focus on the square lattice as first step of analyzing the 3d case. Since most of the results obtained in 2d are straightforwardly generalized to 3d as we will see, we postpone the discussion of the 3d Ising spin network to the later section IV D. Those regular lattices carry a natural a priori notion of distance but nothing says it corresponds to a physical distance. We will strive to compare this lattice distance with the emergent distance reconstructed from the physical correlation carried by the quantum state. In this setting, the space of 4-valent interwiners between four spins 1/2 is two dimensional: it can be decomposed into spin 0 and spin 1 states by combining the spins by pairs, as on figure 2. Different such decompositions exist and figure 3 gives a graphical representation of them. There are three such decompositions, depending on which spins are paired together, which we dub s, t and u channels as in particle physics and quantum field theory. Writing V for the 2-dimensional Hilbert space of a spin 1/2, we denote its basis states (|↑ , |↓ ). We call |0 s and |1 s the spin 0 and 1 states in the s channel. They can be explicitly written in terms of the up and down states of the four spins: Those two states obviously form a basis of the intertwiner space. We give can the transformation matrices between this basis and the two other channels: The intertwiner basis that actually interests us is not attached to one of these channels but is defined in terms of the square volume operatorÛ v of loop quantum gravity. Since the spins, and thus area quanta, are fixed, the only freedom left in the spin network states are the volume quanta defined by the intertwiners. This will provide the geometrical interpretation of our spin network states as excitations of volumes located at each lattice node. For a 4-valent vertex, this operator is defined as: where J i are the spin operators acting on the i link. For clarity, we will for the remaining of this paper use the natural quantum gravity units for which G = c = = 1. This operator is Hermitian but not positive; it also registers the space orientation (or more precisely the orientation of the bivector space over R 3 , which is itself isomorphic to R 3 ). The volume itself can then obtained by taking the squareroot of the absolute value ofÛ . Geometrically, 4-valent intertwiners are interpreted as representing quantum tetrahedron, which become the building block of the quantum geometry in loop quantum gravity and spinfoam models [18,19]. Using the change of basis relations (4), U v takes the following form in the s channel basis: The eigenstates |u ↑,↓ can be obtained directly has 4 . This (square) volume is the smallest non trivial possible value of a chunk of space. The only freedom is the orientation of the chunk of volume corresponding to the intertwiner state.
We consider these two oriented volume states |u ↑,↓ as the two levels of an effective qubit. Let us point out that these two level states are a priori only defined up to phases, and we could actually choose arbitrary phases |u ↑,↓ → e iθ ↑,↓ |u ↑,↓ . Choosing √ 2|u ↑,↓ = |0 s ∓ i|1 s fixes those phases and actually selects the s-channel over the t or u channels.
We can now define a pure spin network state which maps its quantum fluctuations on the thermal fluctuations of a given classical statistical model -for instance the Ising model-by where we sum over all the possible spin σ v configurations modulated by an arbitrary phase Θ[σ v ] and an amplitude A[σ v ] which has been chosen to be an Ising nearestneighbors model with a coupling constant J and magnetic fields B v . We have mapped the Ising spin σ v = ± onto the space orientation of the square volume eigenvalues |u ↑,↓ . The state ψ σv1/2 represents a particular configuration of the spin network and the full state is a quantum superposition of them all. Defined as such, the state is unnormalized but its norm is easily computed to be the Ising partition function Z Ising : using the usual condensed matter notation v, v for nearest neighbor vertices or the usual loop gravity notation s(e), t(e) respectively for the source and target vertices of every (oriented) edge e of the graph.
In principle we could define such states with any amplitude involving Ising spins and generalize our Ising spin network states to any other condensed matter models built from 2-level systems.
The intertwiner states living at each vertex are now entangled and carry non-trivial correlations. More precisely this state exhibits Ising correlations between two vertex i, j: Those correlations are between two volume operators at different vertex which are in fact components of the 2point function of the gravitational field. So understanding how those correlations can behave in a non-trivial way is a first step toward understanding the behavior of the full 2-point gravity correlations and for instance recover the inverse square law of the propagator. We see that in order to emulate correlations of a classical statistical system, the amplitudes appearing in the pure state must be the square root of the classical Boltzmann factor. More generally, a spin dependent phase Θ[σ v ] could be introduced in the definition of the state itself as done in Eq.7. This phase could for example consist in complex valued local magnetic fields, which occur in the Lee-Yang zeroes theorem and are relevant to some decoherence models [20]. For instance, if we change the phases in the definition of the two states |u ↑,↓ , this would clearly change the Ising spin network state defined above, but the phases would entirely be re-absorbed into the Ising Hamiltonian as purely imaginary magnetic fields. Such a phase term doesn't actually change the spin correlations, · Θ = · Θ=0 . They will nevertheless affect the expectation values of dual operators that shift the spins and will affect the Hamiltonian constraints satisfied by the state as we will see below. For now, we consider this phase to be equal to unity for the sake of simplicity.
Similar states have been studied in a quantum information approach to condensed matter physics and appear to have nice properties with respect to entanglement entropy or even for quantum computation purposes [21,22]. The spirit of this approach consists in constructing quantum states with well controlled physical properties and then understand them as a ground state of a particular dynamic. This is the same perspective we are advocating for loop quantum gravity here.
B. Low and high temperature expansion definition
In order to better understand the structure of the Ising spin networks and how to build them from the basic loop quantum gravity operators, we look at them for the perspective of the low and high temperature loop expansions of the Ising model. This is also interesting from the point of view that the duality between the loop expansions at low and high temperatures allows to characterize exactly the critical point of the Ising model, especially for the 2d square lattice, which is self-dual (see e.g. [14]).
a. Low temperature expansion The partition function of the Ising model admits different representations which are more relevant in different temperature regimes. We restrict ourselves here to the case with no local magnetic field. In the low temperature regime, typically below the critical temperature, the ground state of the system is an ordered phase. It is then natural to extract the ground state contribution from the partition function and focus explicitly on the excitation contributions. Such a representation is called the low temperature expansion, or cluster expansion. For a planar graph, as the square lattice, it reads = 2e where z is the valence or order of the graph's nodes, fixed here at z = 4 for the square lattice. We are summing over all clusters C, that is all subsets of vertices of the graph Γ. The expansion is obtained by distinguishing the up spins from the down spins. A cluster C is equivalent to a even subgraph γ * ⊂ Γ * of the dual lattice, that is such as the valence of each node of the subgraph is even, i.e. 0 (the node does not belong to the subgraph), 2 (the subgraph goes through that node) or 4 (which is the maximal value). This even subgraph on the dual graph is simply the boundary of the cluster, as illustrated on figure 4. P C = P γ * is the number of edges of the dual subgraph or equivalently the total perimeter of the cluster. One can decompose an even subgraph in terms possibly intersecting loops, with loop intersection corresponding to the 4-valent nodes of the subgraph. This low temperature expansion leads to an alternative definition of the Ising spin network, which we note |ψ LT . We start from the totally ordered state |u ↑ u ↑ · · · and act with an operator that flip all spins inside a given cluster C. This reads: whereτ is the notation used for the Pauli matrices.τ x is the flip operator that sends the up spin |u ↑ on a down spin |u ↓ and vice versa. It can be constructed in terms of geometric operators, from area and scalar product operators acting on the intertwiner space, as shown in appendix A.
b. High temperature expansion In the high temperature regime, their exist various expansions and we focus on the loop expansion of the partition function. It is constructed using the following well-known linearization identity for Z 2 variables: Expanding the nearest neighbor exponentials leads to a loop expansion of the Ising partition function: We are summing over all even subgraphs γ of the initial lattice Γ. These subgraphs, as said earlier, can be seen as sets of possibly intersecting loops on the square lattice. Then P γ is the number of edges, or equivalently the perimeter, of γ. We proceed similarly from (7) and we define the high temperature form of the Ising spin network state |ψ HT as where we have switched the intertwiner basis to: We are now summing over all possible subgraphs γ (defines as arbitrary subsets of edges of the initial graph Γ) and not restricting ourselves to even subgraphs. The index γ v actually registered the parity of the valency of a vertex v with respect to the subgraph: γ v = +1 if v is attached to an even number of edges of the subgraph γ, while γ v = −1 if it is attached to an odd number of edges. It is a non-trivial consistency check to show directly that the norm ψ HT |ψ HT is actually simply the Ising partition function.
This high temperature state is constructed not form the full spin-up state but from the plus state |+ = 1 √ 2 (|u ↑ + |u ↓ ) which actually sums over all possible combinations of spins up and down. This seems natural since the high temperature regime of the Ising model is totally disordered.
From the perspective of loop quantum gravity, the |+ state is actually |0 s and theτ z operator acting at a vertex v is simply the normalized squared volume operator τ z = 4 √ 3Û v . This allows to define this high temperature expansion of the Ising spin network entirely in terms of geometric operators. The interested reader will find more detail in appendix A.
C. Loop representation
Loop quantum gravity is based on the reformulation of general relativity as a SU(2) gauge field theory. Its basic observables, and then operators at the quantum level, are the holonomies. From the viewpoint of the spin network wave-functions, the gauge invariance is ensured by the intertwiners. It is always enlightening to understand how to reconstruct some spin network states from the basic holonomy operators and we will investigate below how to derive such a loop decomposition for our Ising spin network states.
So the Ising spin network wave-functions for a vanishing phase Θ[σ v ] = 0 reads: where D 1/2 (g e ) are the spin-1/2 Wigner matrix representation of the SU (2) group element g e associated to every oriented edge e. Changing the orientation of an edge simply simply switches g e to its inverse g −1 e . Using the relation between the eigenstates of the squared volume operator and the spin zero intertwiner states in the three pairing channels, |u ↑,↓ = 2 3 |0 s + e ∓ iπ 3 |0 t e ∓ i2π 3 |0 u , we obtain a decomposition of the Ising states over tessellations T on the lattice: where the phase θ(T , [σ v ]) depend on both the tessellation and the Ising spins, and χ 1/2 is the character (i.e the trace) of the fundamental representation. We call here tessellation a set of loops covering every link of the lattice once and only once (or equivalently a partition of the edge set of the lattice in terms of closed loops). Defining an auxiliary variable t v = 0, 1, 2, as a value for the three channels s, t, u respectively, see figure 3, the phase takes a simple form, Here, we have omitted a factor (2/3) V in the normalization of the wave-function (17) coming from the decomposition of the up and down states |u ↑,↓ in terms of the spin-0 states in the s, t and u channels.
It is interesting to note that the amplitude associated to a given tessellation is a partition function of a 2D Ising model with local imaginary magnetic fields. For a given tessellation, or equivalently channel values t v for every vertex, the effective Hamiltonian is An analytical solution of the 2d Ising model with magnetic fields is actually still unknown. In one dimension, for B v = ih, there exist an infinite number of couples (J, h) for which the correlation length diverges [23]. In fact, studying models with imaginary magnetic fields gives information on the onset of phase transitions, relaxation or decoherence timescales [24][25][26]. Indeed, the Lee-Yang theorem explores the zeros of the partition function which in turn are related to the divergence of thermodynamical quantities like the free energy. It would be interesting in future investigation to look at that potential interplay between the Lee-Yang zeroes and the structure of the Ising spin network states in the thermodynamical limit. Figure 5 gives an example of a particular tessellation which as defined above is a set of loops passing once and only once through each link of the lattice. As a potential candidate for the generic solutions of the Hamiltonian constraint, this loop decomposition gives us information on the loop structure of physical states. Generically, the emergence of non trivial correlations is a consequence of the quantum superposition of all possible sets of extended loops covering the spin network. Having the loop representation, it is interesting to check the results mentioned in the previous section and see that, apart from a normalization factor, we have the same expressions of the norm and correlations. The method here is to write all the loop contributions as a product at each vertex v of a matrix element σ v |M |σ v . This matrix is quite reminiscent of the transfer matrix method used in statistical physics. In our case, it is proportional to the identity as is expected from the very construction of the state.
We give here some details on this calculation. Even if this specific evaluation can be done in a more straightforward way, the purpose here is to illustrate the method. We write the norm N N = We then basically define the matrix M as follows This matrix M is obtained by multiplying a phase matrix P and an intertwiner matrix I whose elements are simply the different overlaps t v |t v , M = t P IP , such that The same computation gives the expected correlation functions for the intertwiner spins.
D. Density
With this loop decomposition of the Ising spin networks at hand, we can look at the structure of its corresponding density ρ(g e ) = |ψ(g e )| 2 , which defines its probability profile.
Squaring the Ising wave-function ψ(g e ) dressed with spins 1/2 on all edges of the lattice, we get tensor products of two spins 1/2 on each edge, which recouple to a spin 0 or a spin 1. Therefore the probability density ρ(g e ) = |ψ(g e )| 2 will once again decompose as a gaugeinvariant function on SU(2) E as a superposition of spin networks dressed with spins 0 or 1 on the edges and admissible intertwiners at the vertices.
The spin configuration consisting in only spins 0 everywhere on all edges actually corresponds directly to the calculation of the norm N of the density computed above. Up to a normalization factor, we have shown that we recover the Ising partition function. We generalize this method to compute the probability of the spin configurations, which also involve some spins 1, and we will show how they are related to correlation functions of the Ising model.
Using the same notation as previously, namely |t which represented the tensor product of a particular configuration of intertwiner states depending on some given values t v , we can consider the probability density ρ(g e ) as a spin network state itself.
We would like to compute the overlap between the state |ρ(g e ) and the state |S associated to a chosen configuration S of spins and intertwiners: This defines a transfer matrix M S between the Ising spins σ v and σ v : We cut the spin configuration S into little pieces S v defining the spin configurations around each vertex v. The various cases are given on fig.6. Then we realize that the transfer matrix M S can be factorized vertex by vertex: Omitting the unambiguous vertex index v, we proceed as in the previous section and write the transfer matrix as M S = t P I S P where P is the phase matrix defined in the previous section in eqn. (21) and I S the 3 × 3 matrix interpolating between t v and t v and whose elements are the scalar products S v |t v , t v .
The task is then to evaluate those matrices. On the one hand, we have the states |t, t , which are simply tensor products of the states |0, 1 s and in the two other channels as defined in section II A, eqn.(A2). On the other hand, we have the spins 0 and 1 and we need the intertwiners between spins 1. There are three different cases, whether we recouple between 2,3 or 4 spin-1 at the considered vertex, as illustrated on figure 6. The bivalent and trivalent cases define a unique intertwiner. The four-valent is more involved since the intertwiner space is 3-dimensional. For the purpose of clarity, we give explicit details on a particular example. The other cases can be treated similarly. We focus on the case with spin 1 on the third and fourth edges, as the first case of figure 6. This intertwiner, denoted |I 0011 , is equal to |0 12 |1 34 , with The column notation mirrors the presence of two copies of spin 1/2, the top line representing the state one copy and the bottom one the other. The matrix I 0011 then consists in the scalar product 0 12 1 34 |0 s,t,u 0 s,t,u . Calculating those overlaps is now straightforward. The interested reader will find the explicit expressions of the states |0 s,t,u in appendix. We obtain the intertwiner matrix I 0011 : The last step is to perform the multiplication by the phase matrix P to finally have the specific 2 × 2 transfer matrix M 0011 , Let us remember that this matrix acts as an operator insertion between the Ising spins σ v and σ v . Looking at the probability amplitude S|ρ(g e ) defined in eqn.(24), we will have to insert all of the transfer matrices for each vertex. We give in appendix B the complete set of transfer matrices and operators associated to each type of configuration of spins and intertwiners. Having those, we can see that for a particular configuration S the amplitude is given by a correlation function of the Ising model: where each of the operator insertion σ v |M S v |σ v will consist in a Pauli matrix combination, switching or phasing the Ising spins at every vertex v.
This shows how to express the probability density defined by the Ising spin network in terms of the Ising correlation functions.
A. Definition and algebra
Up to now, we have focused on defining Ising spin network states such that their n-point functions map onto the correlations of the standard Ising model and on how to generate them using geometric and holonomy operators, but they remain kinematical states of loop quantum gravity. It would be more interesting and physically relevant if we could build physical states solving the Hamiltonian constraint operators or if we could interpret our Ising spin networks as at least approximate physical states in some regime. Indeed, the Hamiltonian constraints are crucial in loop quantum gravity, they generate the dynamics of the theory by implementing its invariance under space-time diffeomorphism and are the quantum Einstein equations for quantum gravity. However there isn't yet a systematic method to determine physical states, or at least a perturbative approximation scheme, in loop quantum gravity despite several recent lines of research (especially using spinfoam models [27]) since Thiemann's original proposal [28] (see e.g. [29] for a review of the various approaches through toy models).
Here we will build some local Hamiltonian constraint operators using the basic geometric operators of loop quantum gravity, such that the Ising spin network states are their only unique solutions. Of course, these do not have any a priori link with actual gravity or any proposal of dynamics for loop quantum gravity. Moreover, we are working on a fixed graph and one might expect the quantum gravity dynamics to act on the spin network graph. Nevertheless, despite all these shortcomings, there might be some lessons to draw for such a toy construction.
First it illustrates the type of Hamiltonian constraints that would lead to Ising-like spin network states and thus to admitting non-trivial long range correlations in a continuum limit. Second we will build our constraint operators from the volume operators acting on nearest neighbor vertices, or equivalently around loop on the dual lattice. This suggest a change of perspective from the usual methods to construct regularized Hamiltonian constraint operators in loop quantum gravity, which focus on the holonomy operators around loops on the actual lattice.
Let us thus construct some Hamiltonian constraint operators characterizing the Ising spin networks. From the condensed matter perspective, a first approach could be to look at the parent Hamiltonian technique and build a Hamiltonian operator whose ground state is our considered Ising state [21]. We will follow a simpler path, looking at the action of the basic Pauli matrix operators on the Ising spins and using them to define simple local Hermitian Hamiltonian constraint operators characterizing uniquely our Ising states.
Let us consider the Pauli operatorτ x,y,z v acting on the Ising spin at the vertex v and first focus on the switching operatorτ x |u ↑ = |u ↓ ,τ x |u ↓ = |u ↑ . Its action on the Ising state can be re-absorbed in the amplitude: Up to the magnetic field term, the change of sign of the Ising spin at the vertex v translates into the insertion of an additional factor involving its nearest neighbor spins, e −J v,w σwσv . We turn this factor in its linearized form, e −Jσvσw = cosh(J) − σ v σ w sinh(J), which allows us to define the following Hamiltonian constraint operator at the vertex v:Ĥ where we use for simplicity the auxiliary operators 2 : 2 These auxiliary operators are easily invertible: We have just showed that these constraints annihilate our Ising spin network state: Let us remember that the operatorτ z v is simply the normalized square volume operator acting at the vertex v.
Similarly, the action of the other Pauli matrix operatorτ y v leads us to define another set of Hamiltonian constraint operatorsĤ y v : Since the operatorsτ x,y,z v can all be constructed from the area and volume operators, those Hamiltonian constraints are entirely expressed in terms of geometric operators. This differs slightly from the usual construction of the Hamiltonian constraint operators in loop quantum gravity, based on Thiemann's trick [28] and which involves the holonomy operator (as regularization for the curvature) around loops of the spin network graph. Here, the product of operators v,w ĥ vw over the nearest neighbors of the vertex v can be also considered as living on a loop but on the dual lattice, as illustrated on fig.7. We will discuss more this shift of perspective below in section III C.ĥ We note here that the possible local spin dependent phases that could be added in the definition of the state (Eq.7) which do not change the correlation functions would change the Hamiltonian constraints previously defined.
By construction, the Ising spin network state is solution of all these constraints,Ĥ x,y v |ψ = 0. To check that there are not any other constraints, we check the algebra they generate. The constraints for two vertices which are not nearest neighbor vanish. The commutation relations on a single site v or for nearest neighbors v, w read: Every other commutation relations are zero. So the constraint algebra does not generate any further constraints satisfied by our Ising spin network states. Our Hamiltonian constraintsĤ v exist for all values of the Ising coupling J. They do not depend on the specific phase of the Ising model, it doesn't see a priori the structure of the correlations or the phase transition. We think that this setting would be the perfect testing ground for any coarse-graining scheme for spin network states in loop quantum gravity (e.g. [15,[30][31][32][33]). Indeed the phase diagram, coarse-graining and phase transition of the 2d Ising model is entirely under control and we know what to expect from the coarse-graining flow. The coarse-graning procedure for loop quantum gravity should reproduce the same flow on the correlations of the Ising spin network states. It will then be enlightening to understand what happens at the level of the Hamiltonian constraint operators, how the emergence of large scale correlations is taken into account and what triggers or signals the phase transition at the critical Ising coupling.
Another approach would be to have Hamiltonian constraints specifically tuned to the coarse-graining properties of the Ising partition function, which would not determine the Ising spin network states for arbitrary values of the coupling J but that would select specifically the critical point. Such Hamiltonian operator would reflect the exact coarse-graining flow of the Ising model. This would be much more complicated to realize than our present proposal. It would either involve a graph changing dynamics or implement a self-duality property of the Ising partition function at the critical temperature, probably through the square-star or triangle-star relations [14]. We believe that this is a very interesting line of research, but yet out of the scope of the present work and we postpone it to future investigation.
B. Unicity
So far, we have introduced a set of constraintŝ H x,y v |ψ = 0 of which the Ising spin network state is a solution. We now show that is it the unique solution of those constraints (up to a global phase factor).
The state of the intertwiners has the following general form where α [σv] are the coefficients of the state on the intertwiner basis. Imposing that the state solves the con-straintsĤ x,y v |ψ = 0 leads us to a detailed balance type condition between a configuration and another with only one spin flip difference at a vertex v, α σ1,··· ,−σv,··· ,σ N = e −J v,w σvσw α σ1,··· ,σv,··· ,σ N (39) The solution of this relation is found as follows. Beginning at a particular configuration, namely the one with all spins up u ↑ , we get the amplitude of an arbitrary configuration by flipping the relevant spins in that balance equation. The exponential factor that appears only depends on the number of pairs of anti-parallele spins. Using the low temperature expansion notation of section II B with down spin clusters C, we obtain that where the amplitude α associated to the totally ordered configuration with only u ↑ spins is fixed by the normalization condition . We recognize the low temperature expansion of the Ising partition function. So we conclude that the amplitude is then proportional to the Boltzmann factor of the nearest neighbors Ising model with coupling J/2. Thus the constraints have a unique solution up to a global phase given by the Ising spin network state. We note that because of the relation between the constraintŝ H x v andĤ y v , we only require either one to reach this conclusion. For more complex qubit models, this might not be the case.
This discussion also enlightens the action of the Hamiltonian constraints and the dynamics they could create. Their role is to impose a detailed balance condition between two different configurations of intertwiners. Such behavior is often encountered when studying stochastic systems relaxing toward equilibrium such as for instance Glauber dynamics for Ising models which consists exactly at looking at local spin flips.
C. Insight for Loop Quantum Gravity
The primary purpose of our Ising spin network states is to provide a toy model framework to study coarsegraining in loop quantum gravity. Working in the controlled environment of the Ising model with its explicitly known correlations seems to be a perfect testing ground to investigate coarse-graining procedures, continuum limit definitions and phases transitions. But beyond this aspect, it turns out that they could also bring some insight into the structure of the dynamics of loop quantum gravity.
Indeed, as we have pointed out earlier in section III A, the Hamiltonian constraint operators that we introduce for the Ising states are different from the typical construction in loop quantum gravity, based on holonomy operators around loops of the spin network graph or creating such loops [28,34] (also see [35] for a more recent reformulation of the holonomy and grasping operators in spinorial terms). These holonomies usually enter the constraints as a regularization of components of the curvature tensor. Here the natural structure of our Hamiltonian operators involve volume operators on dual loops (living on the dual graph) acting on all nearest neighbors of a given vertex. Although our present construction has clearly no a priori link with gravity, it seems closer to the intuition of gravitational waves deforming volumes and shapes from vertices to vertices. This suggests to look for a reformulation of the loop quantum gravity dynamics in such terms, or more generally to investigate the relation (e.g. under the form of a dual expansion) between the two types of Hamiltonian. This might help seeing gravitation waves (or at least geometry deformation waves) emerge in loop quantum gravity directly at the level of the Hamiltonian constraint algebra and not only in a large scale semi-classical limit.
From our viewpoint, this would require, first unfreezing the spins on the spin network edges to allow from holonomy operators to shift those spins, second to work out how our Hamiltonian constraints for the Ising spin network states can be re-written as some fixed-spin projection of more general operators built of holonomies. To this purpose, our decomposition of the Ising states in terms of basic holonomies given in section II C might be a good starting point. Then we hope to generalize this discussion to the full loop quantum gravity framework outside the very restricted toy model of the present Ising spin network states.
A. Orientation alignment
The effective spins we used were defined using the two eigenstates of the square volume operator which physically correspond to two different possible orientations of the fundamental volume. It is interesting to identify a regime in which all orientations would be aligned, for instance either all in the positive sector of theÛ v operators or in their negative sector. This is exactly what happens for the Ising model, in both 2d and 3d, at low temperature or equivalently at high enough coupling J in our setting. Indeed beyond the critical coupling, J c = ln(1+ for the regular square lattice [14], the Ising mode predicts an ordered state, with all the Ising spins align with each other. More precisely, a phase transition occurs for the Ising model from a disordered to an ordered state. Figure 8 shows the generic behavior of the magnetization σ v as a function of the coupling J (see eqn.(42) for exact formulas in the 2d case). Above the critical coupling, the system acquires an average orientation direction, randomly picked when passing the phase transition. Of course the system is perfectly ordered only in the infinite coupling limit (zero temperature in statistical physics) and gets quantum fluctuations away from it.
As for the Ising spin network states, in the infinite coupling limit, the state reduces to the sum over the two limit states, the one where the intertwiners are all in the positive volume orientation state plus the one where the intertwiners are all in the negative volume orientation state. As the coupling decreases, opposite spin clusters appear and their typical size increases with the corre- lation lengths. Towards the phase transition, we have clusters of all sizes with no apparent ordering or preferred orientation. Beyond the phase transition, at lower coupling, we are in the disordered phase, with no orientation alignment. An ordered universe, with aligned volume orientation, would therefore live in the ordered phase at high coupling J (i.e low temperature).
B. Distance from correlations
In a non-perturbative background-independent approach to quantum gravity, the very notions of distance, locality or metric need to be reconstructed from scratch in the absence of a background geometry. Understanding the emergent geometry from the quantum state is still an ongoing issue and we hope to be able to recover all this geometric information from the correlations and entanglement in the quantum state [30].
So the idea we pursue here is that a notion of distance emerges from the correlations in the spin network state. Using the known results on the Ising model, we get the Ising spin network correlations in terms of the natural lattice distance and then work backwards attempting to see to what extent the distance between two spins can get be extracted from the sole knowledge of the correlations between those spins. By construction, the Ising spin networks have exactly the same known correlation behavior than the classical Ising model, namely where ξ is the correlation length (in units of the lattice spacing) and C a positive constant. Here the distance between the vertices |i − j| is the graph distance as naturally defined on the 2d regular square lattice. The exact form of the magnetization and the behavior of the correlation length near the transition are known but won't be used in the following 3 . It is possible to invert the correlation formula so as to express the distance as a function of the correlation itself. The distance will be generically a monotonous decreasing function of the proper correlation between the two spins, σ i σ j p ≡ σ i σ j − σ i σ j , but the precise function will depend on the considered phase and regime. Away from the critical point, we have an exponential decay of the correlation and we can define the distance as d(i, j) ≡ −ξ log σ i σ j p up to an additive constant. The correlation scale ξ becomes the new length unit. At the critical point, things become especially interesting. The correlation length blows to infinity and we have an algebraic decay of correlation. We can now define the distance as d(i, j) ≡ 1/ σ i σ j 4 p . From the point of view of gravity, the most interesting case is clearly an algebraic decay of the correlations. Ideally, we'd like to derive a quadratic decay of correlation, just as in standard quantum field theory, in order to retrieve Newton's gravity law. In order to get this, we expect of course to have to change our Ising model state to another better suited statistical model, but also move to a 3d lattice or graph structures, unfreeze the spins of the spin network and so on. Nevertheless, the present Ising spin network state allows to illustrate a couple of important points: • Reconstructing the distance from the (2-point) correlation depends not only of the considered statistical model on the specific phase of that model.
• We naturally have a algebraic decay of the correlation in terms of the distance, and vice-versa, at the phase transition. At that point, the state also admits a non-trivial continuum limit.
Finally, in order to truly validate this notion of reconstructed distance, we need in general to check the triangle inequality satisfied by the distances between three points, which would involve the 3-point fluctuations. Here we know that the reconstructed distance is the initial graph distance, which satisfies this requirement. But if we would consider another spin network state, not necessarily related to a known local statistical model, this would have to be checked.
C. Coarse-graining
One of the main challenge of the loop quantum gravity program is to define and understand its renormalization flow, from the Planck scale to large scales and derive the low energy behavior as semi-classical general relativity with a systematic method to compute the perturbative quantum corrections. To this purpose, coarse graining procedures for the spin network states and the dynamics of the theory are essential tools [15,[30][31][32].
Understanding coarse-graining in loop quantum gravity is not an easy task because of the absence of a background geometry or structure and the complicated nature of the Hamiltonian constraints. In the present framework of the Ising spin network state, we have fixed a background graph, to a regular 2d square lattice up to now (we will deal with its three-dimensional generalization below in the next section IV D), and we can use this structure to define the coarse-graining flow as in standard statistical physics and condensed matter models. Thus this provides a neat toy model to test all coarse-graining procedures of loop quantum gravity.
In statistical and condensed matter physics, renormalization group and coarse graining methods [36] are widely used to understand critical phenomena. We wish to import these methods and results to quantum gravity to understand better the emergence of critical regimes. The usual formulation of loop quantum gravity focuses on the Hamiltonian constraints, their algebra and their flow generating the time evolution. Here, for our Ising spin network states, the Hamiltonian constraints that we introduced do not see a priori the phase transition: nothing obvious happens at the level of the algebra of the constraints when the coupling reaches its critical value. The information about the critical regime and phase transition is truly in the coarse-graining flow. We could, for example, apply the tensor network renormalization tools to our Hamiltonian constraints, such as it was done for scale (including the nearest and next nearest neighbor interactions) as a function of the smaller scale coupling J. We can follow the renormalization flow of the effective coupling by iterating this map and the critical point is identified as its non-trivial fixed point. In the critical regime, the state presents scale invariance which renders irrelevant the micro structure of the theory and allows for a well-defined continuum limit. However, this fixed point is repulsive for the Ising model: as we coarsegrain, the effective coupling runs aways from the critical regime towards either the low or high temperature fixed points. We hope to later identify a better statistical physics model such that the coarse-graining flow would run instead under the coarse-graining flow towards the critical regime with algebraic decaying correlation and continuum limit. The corresponding spin network state, built along the same lines as our Ising spin networks, would then be better suited to the definition of the continuum limit and semi-classical regime of loop quantum gravity.
D. Going tridimensional
So far, we restrain the discussion of the gravitational state to a two dimensional lattice for exposition simplicity but it happens that the generalization to 3d is straightforward. Keeping the requirement that the lattice be 4-valent (so as to keep a two-dimensional intertwiner space and the map to effective qubits), the natural regular lattice is the diamond lattice as illustrated on Fig.11. Under the usual geometrical interpretation of loop quantum gravity, this lattice can be seen as dual to a triangulation of the 3d space in terms of tetrahedra dual to each vertex. This can be seen as an extension of the more used cubic lattice better suited to loop quantum gravity. The Ising spin network state and the whole set of results which followed are then identical : the wave function eqn. (17), the Hamiltonian constraints and their algebra eqn.(37) are the same.
In 3d, the Ising model also exhibits a phase transition. Even if the exact solution is still unknown and is the subject of active research, much can be learn of its behavior around the critical coupling. Like its 2d counterpart, the Ising spin network state has an ordered phase which corresponds to an orientation alignment of the elementary chunk of space. Concerning the 2-points correlation functions, information about its long distance behavior at the phase transition or near it can be obtained using methods of quantum field theory. In d dimensions we have [14] σ where K(r) are modified Bessel functions and ξ is the correlation length. We mention that the correlation does not depend in the long distance regime on the "magnetic field" B (even local ones B i if we choose to define the state in this way). For the three dimensional case, we have the simple and exact expression Figure 12 represents the phase diagram of the 3d Ising spin network (in the presence of a "magnetic field"). At the critical coupling, the correlations have a one over the spacing power law decay at long distance, still not the inverse square law but less exotic then the 2d behavior. Understanding distance from correlations in 3d is much more vital than in the 2d setting since in 3d their exist loops that can wind around a vertex without visiting other distant vertex (in a relational sense). The expression for the correlation is obtained in a mean field setting where only a correlation length appears. However statistical systems possess another length scale, the lattice spacing, which leads to correction of this expression through the anomalous dimension. In quantum gravity the Planck length plays this role and it would be interesting to understand its influence on the behavior of the correlations.
The 3d phase diagram of the spin network state on the honeycomb lattice has a phase transition whose correlations behave with a one over the separation law.
V. CONCLUSION
In this paper, we have introduced a class of spin network states for loop quantum gravity on 4-valent graph. Such 4-valent graph allows for a natural geometrical interpretation in terms of quantum tetrahedra glued together into a 3d triangulation of space, but it also allows us to map the degrees of freedom of those states to effective qubits. Then we were able to define spin network states corresponding to known statistical spin models, such as the Ising model, so that the correlations living on the spin network are exactly the same as those models. For instance, we defined some Ising spin network states, both in 2d and 3d, on 4-valent regular lattice. In 2d, we use the usual square lattice, while the 3d case used the diamond lattice.
We want to propose those Ising spin network states and their generalizations (for instance to the XY model, Potts models and others) as toy models to test coarsegraining methods in loop quantum gravity and allow for explicit discussions on phase diagrams and transitions. As a start, we used them to illustrate the relation between correlations (2-point functions to be precise) and distances. Indeed, in a background independent framework with no a priori geometrical structure, all the geometry needs to be reconstructed from the quantum gravitational state and can only emerge from its correlation and entanglement structure. Here, using the exact phase diagram and correlation functions of the Ising model, we showed how the distance could be defined a posteriori as a monotonous decreasing function of the correlation, following the intuition that two systems are close if they interact much and consequently are strongly correlated. Now the explicit map between correlation and distance actually depend on the considered phase and regime: in the ordered or disordering phase, the distance would be minus the logarithm of the correlation, while we get an algebraic decay at the critical point. More generally, we hope that it will be possible to export the statisti-cal physics methods to study the detail of the coarsegraining and renormalization flow of those Ising spin network states.
We further described the Ising spin network states in details from the point of loop quantum gravity, defining them in terms of geometrical and holonomy operators, showing for instance that getting long-range correlations required holonomy operators around arbitrarily large loops on the graph (and not only local holonomies). We also introduced Hamiltonian constraints that characterize uniquely our Ising states. This opens the possibility of, on the one hand, investigate a potential reformulation of the actual loop gravity constraints in similar terms, and on the other hand, analyze the coarse-graining flow of such Hamiltonian constraints and algebra in the thoroughly-studied environment of the Ising model where everything is under control.
In fact, we view our construction as a first step towards studying the coarse-graining of spin network states and the loop quantum gravity dynamics from the condensed matter perspective. A similar angle of attack has been developed also in [12] towards the construction of spin network with controlled entanglement and in [31] with the application to quantum gravity toy models of statistical coarse-graining methods such as tensor network renormalization. We see a few lines of development of the present work. First, we should deepen the analysis of the proposal of the distance reconstruction from the correlations and go beyond the 2-point correlation to the 3-point and many-point correlation to check triangular inequalities and more generally to what extent we get a metric structure. Second, we should study the coarse-graining and renormalization flow of the Hamiltonian constraints and compare it to the standard flow of the Ising model. Third, we should generalize our construction to more generic states of loop quantum gravity, allowing for more freedom in the spins and intertwiners, and therefore extending our definition of spin network states to more complex statistical physics models with more interesting critical regimes. To conclude, we believe that our construction of spin network states as qubit statistical models offers a perfect arena as a toy model to investigate the structure of correlation and entanglement in spin networks and the emergence of phase transitions in loop quantum gravity, and we hope to use this tool to investigate further and test renormalization methods of the Hamiltonian constraint operators in that context. In the section we present a brief summary of the different states used in the core of the discussion and their relations. We consider the Hilbert space V of the spin 1/2 representation of SU (2). The natural basis of this two-dimensional space is denoted |↑, ↓ . We now look at the intertwiner space between four spins, i.e the SU(2)-invariant subspace in the tensor product space V ⊗4 . This is again a two-dimensional space, which we map onto the Hilbert space of an effective qubit: dim Inv SU(2) [V ⊗4 ] = 2 . (A1) We construct a natural basis of this intertwiner space by recoupling the spins by pair into an internal spin, which can take the value 0 or 1. They are three possible pairing between the four spins, which correspond to three different channels, which we called s, t, u in the main text and which are illustrated in fig.3. Recoupling the spins 1 and 2 together corresponds to the s channel and gives the following intertwiner basis states: We can similarly express the intertwiner states of spin 0 and 1 in the two other channels t and u in terms of the |↑, ↓ states. Then we compute the change of basis between the three channels: The Casimir operators, given by the scalar product between the SU(2) generators, are easily calculated in these various basis using the formula J. K = 1 2 ( J + K) 2 − J 2 − K 2 and the transfer matrices between basis. We compute in the (|0 12 , |1 12 ) basis: This allows to compute the action of the square volume operatorÛ as a commutator of those Casimir operators: We define the effective qubit states from the eigenvectors of this square volume operator: These will be the up and down spin states for the Ising model, now corresponding to the two geometrical orientations (of the 3d space (R 3 ) ∧2 ∼ R 3 ) leading to opposite signs of the squared volume.
In this qubit basis |u ↑,↓ , we act with the Pauli matrices τ x,y,z generating the su(2) algebra. The τ z operator is naturally defined as the normalized square volume operator: The τ x operator is defined from the scalar product operators: When discussing the low and high temperature definitions of the Ising spin network states, we needed the eigenstates of τ x , which we wrote |± = (|u ↑ ±|u ↓ )/ √ 2. In light of the definition above, these are simply the basis states in the s channel: | 2015-09-17T15:38:56.000Z | 2015-09-17T00:00:00.000 | {
"year": 2015,
"sha1": "0183ec8a41a3bbd44329b36e4cc1ae914c5cd548",
"oa_license": null,
"oa_url": "http://arxiv.org/pdf/1509.05297",
"oa_status": "GREEN",
"pdf_src": "Arxiv",
"pdf_hash": "0183ec8a41a3bbd44329b36e4cc1ae914c5cd548",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Physics"
]
} |
263151909 | pes2o/s2orc | v3-fos-license | Simple approximations to the positions of the Lagrangian points
The Roche potential is the sum of the gravitational and rotational potentials experienced by a massless body rotating alongside two massive bodies in a circular orbit. The Lagrangian points are five stationary points in the Roche potential. The positions of two of the Lagrangian points (L4 and L5) are fixed. The other three (L1, L2 and L3) are along the line joining the two masses: their positions depend on the mass ratio, $q$, and can be calculated numerically by finding the roots of a quintic polynomial. Analytical approximations to their positions are useful in several situations, but existing ones are designed for small mass ratios. We present new approximations valid for all mass ratios from zero to unity: \begin{eqnarray*} x_{\rm L1}&=&1 - \frac{q^{0.33071}}{0.51233\,q^{0.49128} + 1.487864} \\ x_{\rm L2}&=&1 + \frac{q^{0.8383} + 2.891\,q^{0.3358}}{1.525\,q^{0.848} + 4.046596} \\ x_{\rm L3}&=&-1 + \frac{q^{1.007}}{1.653\,q^{0.9375} + 1.66308} \end{eqnarray*} in a rotating frame of reference where the more massive body is at $x=0$ and the less massive body at $x=1$. The three approximations are precise to $6 \times 10^{-5}$ for all mass ratios.
The Roche potential is the sum of the gravitational and rotational potentials experienced by a massless body rotating alongside two massive bodies in a circular orbit.The Lagrangian points are five stationary points in the Roche potential.The positions of two of the Lagrangian points (L4 and L5) are fixed.The other three (L1, L2 and L3) are along the line joining the two masses: their positions depend on the mass ratio, q, and can be calculated numerically by finding the roots of a quintic polynomial.Analytical approximations to their positions are useful in several situations, but existing ones are designed for small mass ratios.We present new approximations valid for all mass ratios from zero to unity: x L1 = 1 − q 0.33071 0.51233 q 0.49128 + 1.487864 x L2 = 1 + q 0.8383 + 2.891 q 0.3358 1.525 q 0.848 + 4.046596 x L3 = −1 + q 1.007 1.653 q 0.9375 + 1.66308 in a rotating frame of reference where the more massive body is at x = 0 and the less massive body at x = 1.The three approximations are precise to 6 × 10 −5 for all mass ratios.
The Roche potential
The motion of two point masses in orbit around each other is a well-known and well-understood [1][2][3][4] solved problem in celestial mechanics.Its extension to three bodies was a major goal of 19th-century astrophysics, until Henri Poincaré demonstrated in 1887 that it was insoluble.
There is one situation where analytically tractable results can be obtained, known as the restricted three-body problem.In this case, termed the Roche model, two point masses are on a circular orbit and the third body is massless.The Roche potential is the sum of the gravitational and rotational potentials and can be used to describe the shapes of stars in close binary systems.The more massive star is defined to be at the origin of a Cartesian co-ordinate system (x = y = z = 0) and the less massive star is at (x, y, z) = (1, 0, 0).The coordinate system therefore rotates with the binary system.
The Roche potential experienced by a massless particle at point (x,y,z) is where G is the Newtonian gravitational constant, M 1 is the mass of the more massive star, M 2 is the mass of the less massive star, r 1 is the distance from the point to mass M 1 , and r 2 is the distance from the point to mass M 2 .Simple trigonometry gives and It is convenient to work with the normalised Roche potential which yields where q = M 2 /M 1 is the mass ratio.Thus the Roche potential has been simplified into a function which depends only on the position of the massless body, the mass ratio, and the scale of the system which is set by the semimajor axis of the relative orbit of the two massive bodies.
The Lagrangian points
There are five stationary points in the Roche potential where no net force is exerted on a particle, i.e. the gravitational and rotational forces balance.These are called the Lagrangian points and will be denoted L1, L2, L3, L4 and L5 below.The first three were discovered by Leohard Euler and the last two by Joseph-Louis Lagrange.All five Lagrangian points are in the x-y plane so have z = 0. Fig. 1 shows their positions in the case that q = 0.5.
The positions of the L4 and L5 points each form an equilateral triangle with the two masses at the other vertices.Their positions are thus fixed at and need not be discussed further.Orbits at the L4 and L5 points are stable for mass ratios q < 0.04004 (Prša 5 ).The L1 (inner Lagrangian), L2 (outer Lagrangian) and L3 points are all along the x-axis (so y = z = 0) but their positions depend on mass ratio and are less straightforward to determine.By setting the derivative of Eq. 5 to zero we find where r 1 → x and r 2 → 1 − x.Further algebra leads to the equation which is quintic so does not have a general solution.Eq. 9 can be solved using numerical methods, e.g.bisection, Newton-Raphson or grid search (see Leahy & Leahy 6 ).The positions of the L1, L2 and L3 points are shown as a function of mass ratio in Fig. 2.
FIG. 2: x-positions of the L1, L2 and L3 points as a function of mass ratio.
Existing approximations to the positions of the L1, L2 and L3 points
There are times when a precise solution of Eq. 9 is not necessary, and a quick and simple approximation is adequate.These cases include when making a plot, trying to put together the first version of a piece of code, or as a starting estimate for iterative refinement.Approximations are available * † , but in general are only valid for small mass ratios.
At small mass ratios the distances of the L1 and L2 points from the second mass become approximately equal to the size of its Hill sphere (e.g.Ref. 7 ): Another set of approximations was reported in the textbook by Prša 5 , and come from a perturbation analysis presented in the textbook by Battin 8 .In the geometry adopted in the current work these are: FIG. 3: Top: x-position of the L1 point (black filled circles) compared to the approximation in Eq. 14 (white line) on a logarithmic scale.Bottom: residuals of the approximation to the true values, with the values of the largest positive and negative residuals indicated with dotted lines.
where µ = q/(1+q) and ζ = (µ/3) 1/3 .These are designed for small mass ratios so are imprecise for larger values.The deviations for the positions of L1, L2 and L3 are less than 10 −5 for q < 0.01, 0.03 and 0.26, respectively.The largest deviations are −0.035 at q = 1, −0.0014 at q = 0.41, and 0.0012 at q = 1, respectively.The largest deviations can be decreased by optimising the numerical coefficients in eqs.11 to 13 at the expense of lowering the quality of the approximations for small mass ratio.Similar approximations to those in Eqs.11 to 13 were given in Murray & Dermott 9 (pp.78-80).They work well for small mass ratios, but deviate at higher mass ratios by much more than do Eqs.11 to 13, so will not be considered further.Other approximations suitable for small mass ratios are available and will not be summarised here.
New approximations to the positions of the L1, L2 and L3 points
We were motivated to obtain analytic approximations to the positions of the L1, L2 and L3 points with a higher precision that existing approximations.This required a grid of positions as a function of mass ratio and a fit to this grid of various analytical functions.
After some experimentation we set up a grid of positions as a function of FIG.4: As for Fig. 3 but for the L2 approximation in Eq. 15.
mass ratio containing two components.The first component consisted of 448 points which were spaced equally in log q in the interval q = [10 −5 , 0.295].The second component consisted of 141 points spaced equally in q in the interval q = [0.3,1.0].The final point at q = 1 was given ten times the weight of the other points, as this was found to decrease the amount of 'flailing' at this extremum.The positions of the Lagrangian points at each q were obtained to high precision using grid search methods.We then experimented with a range of possible analytical formulae including numerical coefficients which were optimised using the mpfit package from Craig Markwardt 10 .After some experimentation, we found that a good approximation for the position of L1 is where a i are the numerical coefficients to optimise.The values of the coefficients are a 1 = 0.33071, a 2 = 0.51233, a 3 = 0.49128 and a 4 = 1.487864, and the largest deviation from the true position is 5.9 × 10 −5 .For convenience, we have attempted to limit the number of significant figures required for these coefficients.Our choice of this function was inspired by that for the Roche lobe radius by Eggleton 11 .The fit and residuals are shown in Fig. 3.
Approximating the position of L2 required a more complex formula.We found FIG. 5: As for Fig. 3 but for the L3 approximation in Eq. 16.
We have again limited the number of decimal places required for all coefficients except b 6 ; a slightly more precise result (maximum deviation 5.5 × 10 −5 ) could be obtained without this restriction.For L3 we found the formula to be adequate, where c 1 = 1.007, c 2 = 1.653, c 3 = 0.9375 and c 4 = 1.66308.The largest deviation is 3.3 × 10 −5 (see Fig. 5) but this could be lowered to 3.2 × 10 −5 if the number of decimal places is not restricted for c 1 , c 2 and c 3 .
Final comments
The positions of the five Lagrangian points are of interest in many areas of celestial mechanics, in particular the positions of space missions such as Gaia and JWST.The L1, L2 and L3 points can be found by solving a fifth-order polynomial, which has no analytic solution so must be performed numerically.There are times when an approximate answer is adequate, but existing analytical approximations are designed for small mass ratios.
We have therefore derived new approximations for the positions of the L1, L2 and L3 points, which can be used for convenience when an exact answer is not needed.They are accurate to within 6 × 10 −5 (absolute deviations) for all mass ratios from 10 −5 to unity.Although we have taken care to limit the required precision of the numerical coefficients to four decimal places where possible, some must be specified to six decimal places to achieve the quoted precision.
Our new approximations for the L1 and L2 positions perform less well at the smallest mass ratios.For mass ratios less than 10 −5 we recommend that the Hill sphere approximations are used instead.
FIG. 1 :
FIG.1: Visualisation of the Roche potential for a system with q = 0.5.Lines of constant potential are shown with solid or dotted lines.The positions of the two masses are shown with filled circles.The position of the centre of mass is shown with an open diamond.The Lagrangian points are labelled. | 2023-09-28T06:42:35.476Z | 2023-09-27T00:00:00.000 | {
"year": 2023,
"sha1": "0c3f01ca06a6a8437088622fb653e0d6c336047e",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "0c3f01ca06a6a8437088622fb653e0d6c336047e",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Physics"
]
} |
202766152 | pes2o/s2orc | v3-fos-license | Low-cost Appliance Switching Circuit for Discarding Technical Issues of Microcontroller Controlled Smart Home
: As technological advancement in recent years is bringing ease to our everyday life, smart home systems based on wireless networks and sensors are rapidly gaining importance. Research on microcontroller controller smart home systems has become popular due to its low cost and reliability. However, appliance switching capacity of microcontroller is quite low which limits the widespread use of such sort of systems. This paper introduces an external appliance switching circuit for microcontroller controller automated home which has the ability to control 208 appliances. This large appliance switching capability has been obtained by demultiplexing digital signals obtained from microcontroller board using a complex array of 4 to 16 line decoders. The proposed appliance switching circuit can be implemented with 60-70% less wiring cost in real time compared to the typical wiring technique in microcontroller controlled smart homes. Also, hardware implementation expenditure in this research has been reduced significantly compared to the existing designs. Moreover, the issue associated with existing microcontroller appliance switching scheme in case of incorporating manual switch has been solved. The proposed circuit is incorporated with a smart phone application by which the user sends commands to the microcontroller board using Bluetooth as the medium of wireless transfer. The proposed microcontroller controlled smart home system with external appliance switching circuit is quite cheaper than the existing systems which allows consumers to afford smart living with limited budget. Due to reduction of implementation cost and solution of major technical design issues incorporated with the existing microcontroller-based system been solved in this research work, microcontroller based smart home systems is expected to become a potential product in the market in near future.
Introduction
Smart Home Systems or simply Smart Home means automation of our electrical and electronic devices of our house. Smart Home has achieved a great interest in the recent years and has become a popular topic of interest for researchers. By implementing Smart Home System in our home, numerous household activities of appliances can be controlled [1]. Depending on user's demand, various types of Smart Home Systems have been developed [2].
Early versions of Smart Home Systems used Dual Tone Multi Frequency (DTMF) technology [3]. However, DTMF technology-based appliance control system in home is not quite reliable since the system often result in false appliance switching. Recently, due to technological advancement in wireless sensors and smart devices, Smart Home System is heading towards a new dimension. Using smart phones to control home appliances is the most recent trend in Smart Home Systems [4]. Moreover, hand gesture controlled systems [5,6] and voice controlled smart home system [7] are getting popularity nowadays. However, gesture controlled and voice controlled smart hoe systems require huge installation cost which makes these systems less affordable for people having low budget. Moreover, accuracy is a major problem in such sort of systems. Global System for Mobile (GSM) and Internet have made operating range of home automation systems global [8][9][10]. Internet based systems nowadays offers a wide range of smart functions. But GSM and Internet based smart home systems are sometimes unreliable due to speed issues. Moreover, operating cost of GSM and Internet based smart home systems are high due to SMS and Internet bills. Bluetooth based smart homes offer users smart living within low budget but they have limited range of operation [11]. Energy management in smart home have also become popular to optimize energy consumption which reduces electricity bills significantly [12].
Microcontroller controlled smart home systems have recently gained much interest among researchers [13]. Various smart home methodologies such as Gesture Controlled, Bluetooth based, GSM technology based, voice recognitionbased use microcontrollers for appliance switching [5,11,14,15]. Microcontroller offers a low-cost implementation of smart home appliance control system. But the number of appliances that can be controlled by a microcontroller is very limited. Remote switching of appliances is the most basic and major function of a smart home. Due to having limited number of pins, microcontrollers don't have the ability of controlling a large number of appliances. Due to this reason, widespread use of microcontroller controlled cheap smart homes is halted. Hence, it is necessary to enhance the appliance switching capacity of microcontroller controlled smart homes. This paper introduces a microcontroller controlled smart home system with an external circuit which has the capacity of controlling 208 appliances. The external circuit only needs 4 pins of microcontroller to handle this large number of appliances. The system incorporates a smart app with user-friendly Graphical User Interface (GUI) by which user can send commands to the microcontroller board. Bluetooth was used as the wireless medium from smart phone to microcontroller board.
Literature Review of Existing Microcontroller Controlled Appliance Switching Circuits
Microcontroller controlled Smart Home Systems are gaining much importance nowadays due to its simplicity and low-cost implementation. Various smart features and complicated functions such as voice-controlled lighting, flame and gas monitoring, automatic ventilation etc. can be implemented by microcontroller controlled Smart Home Systems [16].
Typical microcontroller-controlled appliance switching systems consist of a smart phone app, a wireless device or system such as Bluetooth, Wi-fi, Internet etc. and relays to switch on/off appliances [11,17]. Block diagram of conventional microcontroller controlled smart home for appliance switching is presented in figure 1. Microcontroller based appliance switching other than using smart app such as Gesture Controlled, GSM controlled, voice controlled etc. utilizes the same technique for appliance switching by connecting microcontroller pins directly to relay [5,14,15]. The major problem for such sort of systems is appliance switching capacity since microcontroller has limited number of pins. For instance, ATmega328 microcontroller-controlled system presented [16] can control only 12 appliances. ATmega2560 microcontroller-based system presented in the study [18] can control 52 appliances which provides a better option.
Another major problem of microcontroller-based appliance control is inability of switching appliance manually. These systems cannot incorporate manual switches by which appliances can be switched on/off manually. Hence, it is necessary to overcome this problem.
Design Approach of Proposed Microcontroller Controlled Appliance Switching Circuit
The proposed circuit for switching on/off appliance consists of IC4067 (4-16 line decoder IC) and IC4013 (D Flip-flop IC). The proposed circuit requires only 5 back to back 4-bit digital input signals to control 208 appliances. Microcontroller board Arduino UNO R3 is used to send the 4-bit input signal to the circuit. Arduino UNO R3 uses ATmega328 as the microcontroller. Smart phone app has been developed using MIT AppInventor by which the user can send commands. MIT AppInventor is an online and open source software by which smart phone applications can be developed without any sort of programming language [19]. Bluetooth HC-05 module has been used to receive commands from the smart phone app wirelessly.
Block diagram of the proposed external circuit is represented by figure 3. The circuit divides the entire home into 13 different branches, locations or rooms. One decoder IC works as the central controller and the other decoder ICs work in the branches to control appliances. Each Branch Decoder IC has the capability of controlling 16 appliances. Hence, a total of 208 appliances can be controlled using the circuit.
Method of Operation of the Proposed Circuit
The To have a clear understanding of the operation of the proposed circuit, an example is provided using figure 4. If the user wants to switch on/off the appliance number 13 in branch 9, at first, he/she needs to give command from the android app by touching the particular button. The command is sent to the microcontroller board wirelessly by Bluetooth module. Since the Branch IC number is 9 for the command, the microcontroller sends the binary signal 1001 to the circuit.
Design Implementation
The proposed appliance switching circuit for microcontroller controlled smart home system has been practically implemented and tested. Picture of the implemented circuit is represented by figure 5. To show different modules clearly, the circuit contains control mechanism for only 9 appliances located in two different locations. Hence, two Branch Decoder ICs has been used. More appliances in different locations can be controlled by adding more Branch Decoder ICs and D Flip-Flop ICs. Outputs of D Flip-Flops for appliance are connected to LEDs which indicate outputs of our circuit. These output pins are required to be connected to relay for appliance switching. For controlling appliances manually, manual push switches can be connected to the clock of D Flip-Flops connected to relay.
Comparison Between Proposed and Existing Designs
The proposed appliance switching circuit can control large number of appliances compared to the conventional microcontroller systems. The circuit can incorporate manual switch which removes another major problem of existing systems. Detailed analysis between proposed and existing systems is discussed in the following sub-sections. Microcontroller Controlled Smart Home
Appliance Switching Capacity
The proposed ATmega328 microcontroller controlled have the capability of controlling 208 appliances whereas typical systems [11,16,19] can control only 12 appliances. ATmega2560 based system in [18] has the ability of controlling 52 appliances but still it would require 4 such kind of microcontroller to match the appliance switching capability of the proposed circuit.
Practical Implementation in Home
According to theoretical analysis, the microcontrollercontrolled systems [4,5,11,[13][14][15][16][17][18][19] might look very effective. However, these systems have a common major problem. These systems cannot incorporate manual switches with them. This problem is explained using figure 5. If we look fig. 6 carefully, we can observe that the microcontroller-controlled appliance switching technique won't work if the manual switch is turned off. On the contrary, if the relay switch is not energized, it is not possible to switch on the appliance using manual switch. Therefore, appliance switching has to be completely depended smart phone app, gesture, voice or sending SMS. But appliance switching cannot be completely depended on the system and adding manual switches with the system is compulsory.
The proposed circuit has the ability of adding manual switches with it. For manual control of appliances, manual push switches are needed to be connected to the clock of the D Flip-Flops for appliances. Hence, the proposed circuit overthrows a major limitation of existing microcontroller-controlled systems.
Comparative Cost Analysis
The proposed circuit requires only 5 wires to control 16 appliances in a location. But the existing systems need 1 wire for controlling an appliance. Hence, wiring cost is reduced by around 60-70%. Moreover, hardware cost of the proposed system is also quite less than the existing microcontroller-controlled systems. A detailed cost of hardware of the proposed system and existing systems is listed in Table 1.
System Size and Area
After observing Table 1, the thing that might come in one's mind that the proposed circuit contains too much D Flip-Flop ICs [20]. But with careful analysis, it can be figured out that the conventional systems [4,11,16,18] actually require more hardware since they will require additional hardware and circuit techniques to incorporate manual switches. However, the proposed circuit already has the hardware by which in can be incorporated with manual push switches. Moreover, the proposed circuit requires just 4 pins of a microcontroller for 208 appliances whereas the existing system [18] requires 4 microcontroller boards and existing systems in the studies [4,11,16,19] require 7 microcontroller boards. Hence, size of the proposed system is actually less than the conventional microcontroller-controlled systems.
Conclusion
Microcontroller based smart home systems offer consumers a low-cost implementation of automated home. However, due to the major problems such as low appliance switching capability, problems in adding manual switches, inefficient wiring technique etc. limited widespread use of such sort of systems. The proposed circuit for appliance switching has overthrown theses major problems and have made microcontroller controlled smart home systems more reliable, more practical and cheaper. Therefore, people having little budget would be able to afford smart home systems. Since the proposed circuit can handle large number of appliances, it can be also used in industries and offices for remote switching of appliances. We believe that with the proposed appliance control circuit, road to widespread use of low-cost microcontroller controlled smart home systems will be opened. | 2019-09-26T16:24:48.546Z | 2019-08-26T00:00:00.000 | {
"year": 2019,
"sha1": "0e33e46f8990afaa57fa730bb0dd785e42ef781e",
"oa_license": "CCBY",
"oa_url": "http://article.sciencepublishinggroup.com/pdf/10.11648.j.ijssn.20190702.11.pdf",
"oa_status": "GOLD",
"pdf_src": "MergedPDFExtraction",
"pdf_hash": "59e33b2a4b9ecdb8d57a88b1e0053f8a707ec788",
"s2fieldsofstudy": [
"Engineering",
"Computer Science"
],
"extfieldsofstudy": [
"Computer Science"
]
} |
33099855 | pes2o/s2orc | v3-fos-license | Role of activation-induced cell death in pathogenesis of patients with chronic hepatitis B.
AIM
To study and compare the difference of activation-induced cell death (AICD) in peripheral blood T-lymphocytes(PBL-Ts) from patients with chronic hepatitis B (CHB) and the normal people in vitro, and to explore the role of AICD in chronic hepatitis B virus (HBV) infection and the pathogenesis of CHB.
METHODS
Twenty-five patients and fourteen healthy people were selected for isolation of PBL-Ts. During cultivation, anti-CD3 mAb, PMA and ionomycin were used for AICD of PBL-Ts. AICD ratio of PBL-Ts was detected with TdT-mediated dUTP nick end labeling and assessed by flow cytometry.
RESULTS
When induced with anti-CD3, PMA and ionomycin in vitro, AICD ratio of PBL-Ts from CHB patients was significantly higher than that from healthy control (17.24+/-1.21 vs. 6.63+/-1.00, P<0.01) and that from CHB patients without induction (17.24+/-1.21 vs. 9.88+/-1.36, P<0.01). There was a similar AICD ratio of PBL-Ts between induction group and without induction group, but no difference was found before and after induction in healthy control. The density of INF-gamma in culture media of induction groups of CHB was lower than that of other groups (P<0.01). There was no difference between these groups in density of IL-10 (P>0.05).
CONCLUSION
When induced during cultivation in vitro, PBL-Ts from CHB have AICD very commonly. This phenomenon has a potentially important relation with pathogenesis of CHB and chronicity of HBV infection.
INTRODUCTION
Currently, the exact pathogenesis of chronic hepatitis B (CHB) and the reason of chronic hepatitis B virus (HBV) infection are still not completely understood. Activation-induced cell death (AICD) is related with lymphocytes decrease and functional defect. This phenomenon can cause decrease of immune clearance. Alloreactive T cells can effectively be depleted from allogeneic T cells by induction of AICD to prevent graft-versus-host disease [1] . AICD is essential for the function, growth and differentiation of T-lymphocytes [2] . This may be an important reason of persistent infection of HBV. AICD in peripheral blood T-lymphocytes (PBL-Ts) of CHB in vivo has been approved by some reports, but does AICD occur more easily in PBL-Ts of CHB than in those of healthy control? In order to explore the role of AICD in chronic HBV infection, we studied and contrasted the difference of AICD in PBL-Ts from patients with CHB and from normal people in vitro. 10 mL peripheral blood was taken and heparin was added for anticoagulation. After equivalent Ficoll-Paque (from Amersham-Pharmacia, USA) was gently added, peripheral blood monocytes (PBMCs) were isolated by density gradient centrifugation (600 g, 20 min). Then PBL-Ts were purified with negative selection technique using immune-magnetic beads as follow. After PBMC was washed, mouse-anti-human anti-CD 14 , anti-CD 16 , anti-CD 19 (2 µg·mL -1 , DAKO Company, Denmark) were added and incubated for 30 min at 0 , centrifuged for removing the uncombined antibody. Then goatanti-mouse CD3 mAb coating with magnetic beads (1 cell vs. 30 beads, Promega Company, USA) was added and incubated for 30 min at 0 . B cells, NK cells and monocytes were all linked with immune-magnetic beads and absorbed by magnetic stock (DAKO Company, Denmark). After the liquor was gently extracted by centrifugation, PBL-Ts were purified. The viability (95 %) of the cells was confirmed by trypan blue staining. When detected by flow cytometry, the purity of PBL-Ts was over 97 %.
Observation by fluorescence microscope
Some PBL-Ts were put on the carry sheet glass, dried naturally and fixed by 4 % formaldehydum polymerisatum. Then, all the cells were stained with TdT-mediated dUTP nick end labeling (TUNEL, procedure according to clarification of the kit) (Promega Company, USA). The positive cells of TUNEL staining were detected by fluorescence microscope (BG-12, Olympass, Japan). Cytokine detection 100 µl supernatant of medium was collected respectively from each group after cultured for 14 h and frozen in -20 refrigerator for detection. The contents of IFN-γ and IL-10 were detected by using an ELISA kit. The parallel sample was set up for each sample. The OD450 value of each sample was measured with an enzyme label meter (550 model, Bio-RAD Program, USA), and then the content of each sample was converted according to the standard curve.
Statistical analysis
The data were presented as x±s. ANOVA was used to compare the means.
Observation by fluorescence microscope
The apoptotic PBL-Ts presented DNA breakage. The breakage DNA could be linked by fluorescence labeling dUTP when TUNEL staining was adopted. The apoptotic cells took on kelly fluorescence under fluorescence microscope (Figure 1). This was named positive TUNEL staining. The plasm of PBL-Ts with AICD took on red fluorescence and the nuclei took on kelly fluorescence when TUNEL and PI double staining were adopted. But the cells without AICD only took on red fluorescence ( Figure 2). The positive cells of TUNEL staining in PBL-Ts of CHB (with and without anti-CD3mAb, PMA and ionomycin) were more excessive than that of healthy control.
Results of flow cytometry detection
After cultivated for 14 h with induction, the PBL-Ts of CHB patients displayed distinct apoptosis. Apoptosis was also found in groups without anti-CD3 and other inductions, but their apoptotic ratio was lower. There was a similar AICD ratio of PBL-Ts between induction group and healthy control without induction. AICD ratio of PBL-Ts from CHB patients (with or without induction) was significantly higher than that from healthy control ( Table 1).
Results of cytokine detection
Activated T lymphocytes may produce plentiful endogenous cytokine. Th 1 mainly produces IFN-γ, IL-2 and TNF-α. But Th 2 mainly produces IL-4, IL-5, IL-6, IL-10, etc. Cytokine IFN-γ, IL-10 were detected in this test. In all groups, the density of INF-γ in culture media of healthy control with induction group was the highest, and the patient in groups with induction was the lowest. But there was no difference among these groups in density of IL-10 ( Table 2).
DISCUSSION
Chronic HBV infection is mainly related to the immune function of patients. In a large degree, immune tolerance, especially neonatal immune tolerance, results in persistence of chronic HBV infection. Because naive T cells are sensitive to Fas-mediated AICD and more easily deleted by Ag restimulation than primed T cells [4] . AICD of PBL-Ts plays a key role in central and peripheral immune tolerance [5,6] . AICD is one kind of apoptosis of reactivated lymphocytes when these lymphocytes are induced by activation signals (especially by complex of TCR/CD 3 ). Ashwell and his colleagues first detected the AICD phenomenon in 1987 when they studied T lymphocyte hybrid tumors. AICD plays an important role in the negative selection of T lymphocytes in thymus, peripheral elimination and clearance of T lymphocytes that have already cleaned the foreign antigens. Therefore, AICD is an important mechanism in maintaining immunoregulation and achieving immune system homeostasis [6,7] . If one's AICD mechanism is disordered (up-regulation or down-regulation), immune tolerance or autoimmune disease would occur.
In this experiment, AICD of PBL-Ts was successfully induced using anti-CD3 mAb, PMA and ionomycin. The responses of PBL-Ts from CHB and healthy control were different. The results indicated that when induced with anti-CD3, PMA and ionomycin in vitro, AICD ratio of PBL-Ts from CHB patients was significantly higher than that from healthy control and that from CHB patients without induction. But there was a similar AICD ratio of PBL-Ts between induction group and healthy control without induction. The results imply that AICD exists in PBL-Ts of CHB and causes decrease of T lymphocytes especially Th 1 cells and functional defect. Specific immune response aiming directly at HBV should not occur. Finally, immunology tolerance to HBV would occur. Ji et al using staphylococcus aureus enterotoxin B and rHBcAg proved that AICD of PBMCs in patients would lead to persistent infection of HBV [8] .
Reduction of deferent cytokines in culture medium implies apoptosis of deferent subtype T lymphocytes, because the types of cytokine secreted by Th1 and Th2 are different. The detection results revealed that the density of INF-γ in culture media of induction groups from CHB was lower than that of other groups (P<0.01). There was no difference between these groups in density of IL-10 (P>0.05). These results imply AICD cells are mainly Th1 cells.
After infection of HBV, the virus elimination depends on specific cell immunity of the body. Mostly, specific cell immunity responses are induced by Th1 lymphocytes, but humoral immunity responses are induced by Th2 type lymphocytes. The sensitivity of the two types of T lymphocytes is not equal. The occurrence of AICD is easily induced by Th1 but not Th2 when induced by Anti-CD3 and corresponding antigen [9][10][11] . Fan et al have proved that enhanced Th2 responses are present in chronic HCV infection, and this should be responsible for the persistent HCV infection [12][13][14] . So, if specific PBL-Ts of CHB are reactivated by HBV antigens, AICD would occur mostly in Th1 type lymphocytes. Thus, specific cell immunity response aiming directly at HBV would be defective, and HBV permanent infection would occur. However, it would be a possible method to surmount immune tolerance and to clean HBV of CHB patients that we have managed to block the apoptosis of activated T lymphocytes [6,15] and raise the amount of specific T lymphocytes. | 2018-04-03T02:18:00.480Z | 2003-10-15T00:00:00.000 | {
"year": 2003,
"sha1": "290fe40de7c3d28da7fbb0939e4788d3b9e0f8f9",
"oa_license": "CCBYNC",
"oa_url": "https://doi.org/10.3748/wjg.v9.i10.2356",
"oa_status": "HYBRID",
"pdf_src": "Adhoc",
"pdf_hash": "f349d603806d5ecc3f538569186b3e828fd8e733",
"s2fieldsofstudy": [
"Medicine",
"Biology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
211259184 | pes2o/s2orc | v3-fos-license | Constraining Conformal Theories in Large Dimensions
In this paper, we analyze the constraints imposed by unitarity and crossing symmetry on conformal theories in large dimensions. In particular, we show that in a unitary conformal theory in large dimension $D$, the four-point function of identical scalar operators $\phi$ with scaling dimension $\Delta_\phi$ such that $\Delta_\phi/D<3/4$, is necessarily that of the generalized free field theory. This result follows only from crossing symmetry and unitarity. In particular, we do not impose the existence of a conserved spin two operator (stress tensor). We also present an argument to extend the applicability of this result to a larger range of conformal dimensions, namely to $\Delta_\phi/D<1$. This extension requires some reasonable assumptions about the spectrum of light operators. Together, these results suggest that if there is a non-trivial conformal theory in large dimensions, not necessarily having a stress tensor, then its relevant operators must be exponentially weakly coupled with the rest.
It is usually believed that there are no non-trivial conformal field theories (CFTs) in greater than six dimensions.This belief stems from thinking of CFTs as infrared fixed points of RG flows that initiate at the free theory.The RG flow is triggered by a relevant or marginally relevant operator about the free theory.The number of relevant and marginal operators decreases with the dimensionality of space-time.In dimensions greater than six free theories do not have any relevant or marginal operators.This leads to the commonly held belief 1 .
With the discovery of a plethora of the so-called non-Lagrangian superconformal field theories (SCFTs) in the supersymmetric context, one can debate whether a conformal field 1 The usual quartic coupling of scalars (φiφi) 2 is irrelevant in D > 4 dimensions, nevertheless for those dimensions, one can try to find the associated UV fixed point.In [1], authors identified this fixed point as a theory of φi and σ interacting with φiφiσ + σ 3 potential.Unfortunately, the dimension of operator φiφi goes below the unitary bound for D > 6.
theory can always be thought of as an endpoint of the RG flow emanating from a free theory.These non-Lagrangian SCFTs do not provide counterexamples in dimension greater than six because SCFTs do not exist in dimensions greater than six.However the reason for this is strictly a kinematical one: non-existence of superconformal algebra (such that supercharges transform in the spinor representation of the Lorentz group) [2,3].Hence the space of SCFTs is a poor diagnostic of the space of CFTs in large dimensions.The question of the existence of CFTs in large dimensions is only meaningful for non-supersymmetric CFTs.Nevertheless, SCFTs yield examples that are perhaps truly non-perturbative in nature.This means we need to analyze the space of CFTs in large dimensions using non-perturbative methods.In this paper, we do so using conformal bootstrap.This approach that started with the work of [4] has turned out to be very effective in putting numerical constraints on the CFT data, for example, on the critical exponents in 3d Ising model [5] as well as obtaining analytical results about the spectrum at large spin [6,7].See [8] for a review of the subject and a comprehensive list of references.
If CFTs do exist in large dimensions then using AdS/CFT correspondence they define non-perturbative quantum gravity in large dimensions.If all quantum theories of gravity come from string theory, then this suggests that string theory, albeit strongly coupled, admits large dimensional AdS vacua.Thus the question of the existence of CFTs in large dimensions is an important one also from the point of view of string theory/quantum gravity.
Let us now briefly discuss what we mean by a non-trivial CFT.First, note that theories of free massless scalars and free massless fermions are unitary conformal field theories and they exist in arbitrary integer dimension.The theory of massless (D − 2)/2 forms in D dimension (for even D) is also a unitary conformal field theory.There are no other free theories that are conformal.Free scalars, free fermions and free (D − 2)/2 forms all have stress tensor and hence admit a local coupling to gravity.If we relax the condition of existence of stress tensor then a family of "generalized free field theories" (GFFTs) can be easily constructed.In GFFT a higher point correlation function is defined as the sum of products of two-point function (which is fixed by conformal symmetry) i.e. as a Wick contraction.Although GFFT can be defined for fields transforming in any representation of the Lorentz group, only the case of scalars is relevant for this paper.For us, GFFTs (including the genuine free theories that have the stress tensor) are trivial conformal theories.
As we search for non-trivial theories we may either want to impose a) unitarity and b) existence of stress tensor or not.The physical expectation of having no non-trivial CFTs above a critical dimension is when one imposes both the conditions2 .If we relax either of the two then the existence of a non-trivial theory is plausible.Families of non-trivial conformal field theories have been constructed and studied in [10][11][12][13][14][15][16] that exist in high dimensions but are non-unitary.On the other hand examples of unitary theories but without stress tensor can potentially be constructed by coupling two GFFTs with a relevant operator and flowing down the renormalization group 34 .Hence, one would like to conjecture that there are no non-trivial unitary conformal field theories with a stress tensor in sufficiently large dimensions.However, we can constrain the space of conformal theories with certain properties even without requiring a stress tensor.In particular, we will show that the unitarity and crossing symmetry constrain the four-point function of scalar operators with scaling dimension ∆ φ /D < 1 to be that of the GFFT in a particular Lorentzian diamond.The use of space-time dimension D as an approximation parameter is not new.In the past, it has been used to study quantum gravity in large dimensions [19].In recent years, the large D limit has also been applied to general relativity obtaining a dramatic simplification in the black-hole dynamics e.g.see [20,21].
Structure of the argument
In this subsection we give a quick overview of the paper, highlighting the structure of our argument.In large D limit, unitarity forces ∆ also be O(D).We define ∆ = δD and also spin as = ωD.With this scaling, the correlator which is expressed as the sum over conformal blocks can be approximated by an integral over conformal blocks multiplied by OPE coefficient "density".We argue that this integral is of Laplace type and can be performed by saddle point approximation.The positivity of the OPE coefficient density can be exploited to argue that the real saddle points must lie in the unitary domain.Remarkably it turns out that this condition is incompatible with unitarity for δ φ < 3/4 (for δ φ < 1 if one assumes sparseness of the low lying spectrum) except for the saddle point that corresponds to the GFFT.
Outline
In section 2 we motivate the scalings and compute the conformal block in the large D scaling limit.Here the large D solution of [22] plays an important role.We independently check that our solution satisfies the conformal Casimir equation.In section 3 we approximate the conformal block expansion by an integral and find the region in the cross-ratio space where the conformal blocks are positive and both s-channel and t-channel OPE are convergent.The consequence of crossing symmetry and unitarity are then analyzed in section 4 leading to the main conclusion.In section 5, we summarize our conclusions with outlook.The paper is supplemented with three appendices.Appendix A gives explicit formulas for conformal blocks in our large D scaling limit.Appendix B illustrates some features of saddle point integrals that are relevant to the discussion in the paper.In appendix C we present a detailed analysis of the constraints of unitarity and crossing.
Conformal blocks at large D
In this paper we will be concerned with unitary CFTs.The conformal dimension of local operators in bounded from below by unitarity ∆ ≥ D 2 − 1, for = 0, ∆ ≥ + D − 2, for = 0. (2.1) In any large D limit, a unitary conformal field theory has local operators with dimensions that scale linearly with D. We take ∆ = δD with δ fixed.The eigenvalue of the conformal Casimir for the conformal multiplet with primary of dimension ∆ and spin is ∆(∆−D)+l(l+D−2)).
In order for the spin to contribute to conformal block we also take spin to scale linearly with D, = ωD with ω fixed.Conformal blocks with finite spin can be obtained by setting ω to be O(D −1 ).The conformal blocks satisfy the conformal Casimir equation.In the large D limit, this equation can be separated and solved.This was done in [22].We have reproduced their result below Here (u, v) are the standard conformal cross-ratios defined as u = x 2 12 x 2 34 /x 2 13 x 2 24 and v = x 2 14 x 2 23 /x 2 13 x 2 24 .Above approximation is valid for y + − y − 1/D or equivalently v 1/D 2 .To compute the blocks in the scaling limit ∆ = δD and = ωD we need to approximate the hypergeometric function at large values of parameters.This can be done by expressing the hypergeometric function in the Euler integral form and performing the integral using the saddle point.The result is of the form where f δ , f ω and g δ , g ω are complicated functions of the arguments and labels.We do not give their explicit form here as it does not offer much insight.It is given in appendix A. Although we have not checked explicitly, we believe the perturbative corrections in 1/D to N δ,ω form a convergent series and the non-perturbative corrections i.e. e −D corrections are absent.This means a given block contributes a specific cross-ratio dependent exponential piece e D (g δ (u,v)+gω(u,v)) to the correlator.If the two blocks have either δ or ω that is O(1) different then their exponential contributions are distinct.However, if one considers blocks for (δ , ω ) and (δ, ω) such that δ = δ + O( 1 D ) and ω = ω + O( 1 D ) then their exponential contribution can not be distinguished.This is because, e Dg δ ,ω (u,v) = c δ,ω (u, v)e Dg δ,ω (u,v) , for some c δ,ω (u, v) ∼ O(1). (2.4) These considerations are important as our arguments will essentially involve matching distinct exponential contributions of blocks.
The expression for the block simplifies if we look at the dependence in a small neighborhood of size 1/D around a certain point.We do this by substituting (u, v) → (a 2 e σ/D , b 2 e τ /D ) and focusing on the dependence on (σ, τ ).where k ± (a, b, This is the large D approximation to the conformal blocks that we will work with in the rest of the paper.When the block is thought of as a function of (σ, τ ), N δ,ω is interpreted as the "normalization".The (σ, τ ) dependence is the imprint of the function g δ,ω (u, v) in an O(1/D) neighborhood.Hence matching of exponential contributions is tantamount to matching the (σ, τ ) dependence in the exponent.This is what we will do in imposing crossing symmetry.
The (σ, τ ) dependence of the conformal block in the large D limit can be verified by checking that it satisfies the conformal Casimir equation in the scaling limit described above.Recall that the conformal Casimir equation is The eigenvalue C ∆, = ∆(∆ − D) + ( + D − 2).In terms of the conformal cross-ratios (u, v) this reduces to the following coupled second-order differential equation.
To see why the conformal block is dominated by a monomial of (e σ/D , e τ /D ), it is instructive to consider the series expansion of a scalar conformal block that is known in arbitrary dimension.
n ≡ Γ(x + n)/Γ(x).(2.9) In our large D scaling limit the sum over descendants becomes a saddle point integral.This is performed in appendix A. At a given value of (a, b), a single descendent with dimension k + (a, b, δ)(1 + 1 b )D dominates sum and hence the (σ, τ ) dependence of the conformal block takes the simple form (2.5).
Conformal block expansion at large D
In this paper we will be concerned with four point function of identical scalar operators of dimension ∆ φ = Dδ φ .The stripped correlation function i.e. φ(x 1 ) . . .φ(x 4 ) (x 2 12 x 2 34 ) ∆ φ is only a function of cross-ratios, G(u, v).It is expanded in terms of s-channel conformal blocks as follows The range of the sum is controlled by unitarity.We have divided the sum into two parts, the first part is supported on the range of ∆ for scalar operators (D − 2)/2 ≤ ∆ < D − 2 and the second part is the range (D − 2) ≤ ∆ − for all .This is the unitary domain.The first part of the sum is supported over what we call the D 1 domain and the second part, over the D 2 domain.The factor of 1 + (−1) picks only the contribution of the even spins as desired.
For now, let's replace the factor 1+(−1) by 1 in taking the large D limit.We will account for this error in section 3.3.In the large D limit, ∆ = Dδ and = Dω, it is convenient to replace the sums over ∆ and to integrals over δ and ω and replace the OPE coefficients by OPE coefficient density C δ,ω = CDδ,Dω .The OPE density consists of a collection of Dirac delta functions at (δ, ω) of all operators appearing in the OPE with then strength given by the OPE coefficient.At leading order, The explicit expression for B is given in equation (2.5).We have grouped the terms independent of (σ, τ ) in the brackets.In the large D limit, the domains of integration are In terms of (δ, ω) coordinates we have graphically presented the unitary domain in figure 1.
Consider the unlikely situation where the Dirac delta functions in the OPE density are O(1) spaced either in δ or in ω.Then as argued below equation (2.3), each operator in the block expansion yields a distinct exponential cross-ratio dependence to the correlator.Of course, more likely the OPE density consists of closely spaced i.e.O(1/D) spaced Dirac delta functions.This motivates the definition of a "smeared OPE density" C δ,ω .We define it to be the OPE density C δ,ω averaged over squares of size ∼ O(1/D) in (δ, ω) space.We will pick to be the smallest such that log(C δ,ω ) is smooth to leading order in D at the scale of 1/D 1 2 in (δ, ω) space.This means log(C δ,ω ) is piecewise smooth at scale of O(1) in (δ, ω) to leading order in D. Note that the unitarity condition that the OPE coefficient C∆, is positive implies that log(C δ,ω ) to leading order in D is real.The reason we demand the smoothness for logarithm and not the OPE density itself will become clear soon.Note that our condition is weaker than the smoothness for the function itself.Let us replace the very jagged function C δ,ω with the function with piecewise smooth logarithm C δ,ω .We will quantify the error associated with this replacement shortly.Now we are in the position to see why the analysis of (3.2) is viable.The key property is the exponential dependence in D in (2.3).This makes the integral in (3.2) in each piecewise smooth region of Laplace type.The feature of such integrals is that they are dominated by discrete points.These points could either be the saddle points of the integrand or be points on the boundary of the region.We call all such points "locally dominant points".We direct the interested reader to appendix B for more discussion of the generalities of such integrals.
In addition to the conformal block N δ,ω having e D... dependence, we will now argue that generically the smeared OPE coefficient density C δ,ω also has e D... dependence.If the OPE coefficients go as e D α ... where α > 1 then the saddle point will be determined only by the OPE coefficient and the large D correlator will simply be a single conformal block evaluated at that saddle point (δ * , ω * ) of C δ,ω .On the other hand, if α < 1 then the saddle point will be determined completely kinematically by the e D... factor in the conformal block N δ,ω .Most general saddle points are obtained if OPE coefficient density also goes as e D... .In the rest of the paper, we assume that to be the case 5 .All in all, the position of the saddle point depends on the factor inside the bracket.This is precisely the part that is independent of (σ, τ ).That is what makes the correlator simpler to compute in a small neighborhood of (a 2 , b 2 ) of size 1/D.
The position of the saddle point, more generally of the point that dominates the integral, is inside the integration range, in particular in the unitary domain D 1 ∪ D 2 , if the integrand does not have a rapidly oscillating phase i.e. of the type e iD... where . . .stand for a function of (δ, ω).The rapid oscillations are precisely what we dropped when we replaced C δ,ω → C δ,ω .The contribution from such rapid oscillations is a subleading saddle point that will lie in the complex domain of (δ, ω).Hence, the error associated with this replacement is an exponentially subdominant one.We ignore this error as we are interested in constraining only the exponentially dominant part of the correlator.
The conformal block N δ,ω (a 2 , b 2 ) could also be rapidly oscillating.This would give rise to locally dominant points that are outside the integration range.To avoid such possibility then we need to consider only that region of cross-ratio space where all conformal blocks are real and non-negative6 with no rapid oscillations 7 .In such a region, the integral (3.2) is dominated by discrete points which lie in the integration range i.e. the unitary domain D 1 ∪ D 2 .Moreover, each locally dominant point yields a distinct exponential cross-ratio dependence.As we move in the cross-ratio space, these points move in a continuous way.If a saddle point moves to the boundary of the smooth region, it dominates the integral as a boundary point.This seemingly unremarkable constraint on the location and movement of the dominant points will turn out to be extraordinarily powerful when paired with crossing symmetry.In the next two subsections, we will look for such region in the cross-ratio space where all s-channel, as well as all t-channel conformal blocks, are positive.Needless to say that we will also demand that the s-channel and t-channel expansions be convergent in this region.
Positivity of conformal blocks
It is known that all the unitary conformal blocks have positive coefficients when expanded in terms of variables (z, z), where u = z z and v = (1 − z)(1 − z).This was first shown in [7] and then in [23], in both by essentially using positivity of the norm of certain descendent states.Because, a m,n ≥ 0 where the conformal block is positive for real z, z ≥ 0. In our analysis of the crossing equation, we would also like to demand the t-channel conformal block to be positive.This forces 1 − z, 1 − z ≥ 0. To summarize, both the s-channel and the t-channel conformal blocks are positive in the diamond 0 ≤ z, z ≤ 1 with real (z, z).The conformal blocks are explicitly computed in appendix A and indeed they are positive and with no rapid oscillations.
Convergence of the OPE
In addition to having positive s-channel and t-channel conformal blocks, we are also interested in having a convergent OPE expansion both in s-channel and in t-channel.The positivity of the blocks discussed in the above subsection also helps determine the regions of convergence of the OPE.In fact, both in [7] and in [23], the positivity of the blocks was used to do precisely that.From state operator correspondence it follows that the s-channel OPE is convergent over the entire range of cross-ratios in the Euclidean regime except for the halfline z = z ≥ 1.Similarly, the t-channel OPE is convergent in the Euclidean regime except for the half-line z = z ≤ 0. In particular, the s-channel and the t-channel OPE are convergent for 0 < z = z < 1.From here, the positivity of the conformal block expansion coefficients in z, z helps us deduce that both the s-channel and t-channel OPEs are convergent in the diamond of interest 0 ≤ z, z ≤ 1.This is as follows, let z < z in the diamond, This is because of the coefficients a m,n appearing in the expansion (3.3) are positive.This means, Again, this is because the OPE coefficients C∆, are positive.The right-hand side is simply the s-channel expansion in the range 0 < z = z < 1 which makes the sum convergent, implying that the left-hand side is also convergent.The same argument can be repeated if z > z.Even the convergence of the OPE in the t-channel in the region 0 < z, z < 1 follows from this argument in the same way.Hence, from now on we will totally confine ourselves in the region 0 < z, z < 1, or the diamond for short.In terms of (a, b), the diamond is a, b > 0, a + b < 1.
u-symmetry
In this section we account for the error that we made near equation (3.2) of summing over all spins rather than summing over only even spins.
Let us call the first and second term on the right hand side t 1 and t 2 respectively.As the OPE coefficients are positive and the conformal blocks are positive in the diamond , it is clear that t 1 > t 2 in the diamond.Also note that the conformal blocks are u-symmetric in the following way, This makes t 2 the u-symmetric image of t 1 .Written as a sum of t 1 and t 2 , the correlator is manifestly u-invariant.When we take the large D limit, both t 1 and t 2 are saddle point integrals.The t 1 integral is the one in equation (3.2).The t 2 is the same integral except for an additional insertion of e iπωD .In the diamond, the saddle point of the t 1 integral lies in the unitarity domain D 1 ∪ D 2 but the saddle point for t 2 will generically be elsewhere in the complexified (δ, ω) plane.Due to rapid phase oscillations in ω, t 2 is exponentially smaller than t 1 .This is the same reason why rapid oscillations in δ give rise to exponentially smaller contributions.In the absence of u-symmetry, we will not be able to say anything about the location of this saddle point.However, we know that the t 2 saddle point is the u-symmetric image of the t 1 saddle point.
Solving Crossing symmetry
The crossing equation is, In the large D limit, the sum over conformal blocks is approximated as integrals 8 , 1 + We have naturally defined ∆ φ ≡ δ φ D. The explicit expression for B is given in equation (2.5).As discussed in section 3.3, the integrals on both sides are split into two terms t 1 and t 2 .The term t 1 comes from the 1 and the term t 2 comes from e iπωD in (1 + e iπωD ) respectively.Every correlator of identical operators has a universal contribution, namely the contribution from the identity operator.As the contribution of the universal saddle must be crossing dual to identity operator, it is perhaps not surprising that the universal saddle fixes the OPE density and its contribution to the correlator to be that of the GFFT.We show this explicitly in section 4.2.For now let us assume that there exists a pair of points (δ * s , ω * s ) and (δ * t , ω * t ) in the s-channel and t-channel expansion respectively that dominate over this universal part for some (a, b).If we find that the existence of such a saddle is inconsistent with unitarity and crossing symmetry then we must conclude that the correlator is given by 1 and its universal dual saddle i.e. by that of the GFFT.For δ φ < 1, we will show that it is indeed the case.
Let's proceed to a proof by contradiction.Let the points (δ * s , ω * s ) and (δ * t , ω * t ) in the schannel and t-channel expansion respectively be globally dominant points at (a, b).Both these contributions must necessarily come from t 1 part of the integral (because t 1 is exponentially dominant over t 2 ) and hence lie in the unitary domain D 1 ∪ D 2 .Focusing only on the (σ, τ ) dependent part in exponent on both sides, 8 We have suppressed the factors of D outside the integrals for compactness of the expression.
where # is some constant that is independent of (σ, τ ) (but does depend on (a, b)).Matching the coefficients of σ and τ in the exponent, we get where This equation expresses the dominant point in the t-channel (δ * t , ω * t ) in terms of the dominant point in the s-channel (δ * s , ω * s ).As we move around the cross-ratio space, then these points may cease to be globally dominant but continue to be locally dominant and will continue to exist in their respective unitary domain.We ask if this is possible i.e. we ask if a pair of the s-channel and t-channel points that satisfies equation (4.4) exists such that both of them lie in the unitary domain D 1 ∪ D 2 .In other words, we ask if the image of the unitary domain D t 1 ∪ D t 2 for the t-channel locally dominant point overlaps with the unitary domain D s 1 ∪ D s 2 for the s-channel locally dominant point under the crossing map (4.4).Cartoons of overlaps of the s-channel and t-channel unitary domains are given in figure 2. They serve to set terminology for possible types of overlaps.Understanding these types helps us understand the nature of the solution space.It is also important for the analysis in section 4.1.For actual overlap diagrams 9 , for various values of (a, b) and δ φ , see appendix C. Now that we have defined the types of overlaps that can occur, we are ready to summarize the result.In figure 3, we have colored the regions in the diamond with the kind of overlaps that are allowed by unitary and crossing as a function of δ φ .Remarkably we find that for δ φ < 3/4, there are regions of the diamond, however small, where the crossing symmetry constraint (4.4) does not have a solution in the unitary domain.Lack of unitary solution to the crossing equation (4.4) means that, to leading order in large D, the correlator of identical scalar fields with δ φ < 3/4 is identical to that of the GFFT!It turns out that this argument can be extended to make this result applicable to a wider range of external conformal dimension, δ φ < 1 with some reasonable assumptions about the OPE density.This is as follows.
Extension to δ φ < 1
In what follows we will assume that log(C δ,ω ) in non-differentiable at (δ, ω) = (1, 0) i.e. at the point where D 1 connects with D 2 .This includes the case where there are isolated operators i.e. operators spaced by O(1) from the point δ = 1.This assumption can be thought of as an assumption of sparseness for low lying operators.
For 3/4 ≤ δ φ < 1, there is no light-pink region in the diamond where any type of overlap doesn't exist.However, we note that the diamond does have regions where type IV overlap + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > (b) Type I: + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > (c) Type II: doesn't exist.These regions are light-green and light-blue regions in figure 3.In light-green region the only overlap is of type II i.e.D s 2 ∩ D t 1 while in the light-blue region the overlap is of type III i.e.D s 1 ∩ D t 2 .If a solution to crossing equation (4.4) were to exist for 3/4 ≤ δ φ < 1 then it must go from being of type II in the light-green region to being of type III in the light-blue region.
Let us first consider the light-green region.The locally dominant point in the t-channel (δ * t , ω * t ) must lie in D t 1 .Due to the discontinuities in log(C δ,ω ), as we move in the cross-ratio space, this point moves but at most to the boundary of the smooth region and as this smooth region lies inside D 1 , the locally dominant point (δ * t , ω * t ) continues to lie inside D t 1 .This is true even as we transit to the light-blue region.There the s-channel locally dominant point (δ * s , ω * s ) must necessarily be in the D s 1 .As (δ * t , ω * t ) lies inside D t 1 as well, the intersection that we are looking for is of type I.But the type I solution does not exist in the light-blue region.This means even for 3/4 < δ φ < 1, we do not find a pair of locally dominant points that is consistent with crossing and unitarity.
The universal saddle
Now that we have ruled out all possible pairs s-channel and t-channel locally dominant points that live in their respective unitary domains for δ φ < 1, let us turn our attention to the unique physical saddle, say in s-channel, that can exist.This is dual to the contribution of the identity operator in the t-channel.The dual of the t-channel identity contribution is (u/v) ∆ φ .This must come from the integration over OPE density in the s-channel.
.5)
Here we have focused only on the t 1 terms as that gives the leading contribution.As mentioned earlier, the t 2 term will give its u-symmetric image namely a 2δ φ D e σδ φ .Let us assume that the integral on the right hand side is dominated by the saddle (δ * s , ω * s ).Matching the coefficients of (σ, τ ) in the exponent yields, This could also be obtained directly by using the crossing equation (4.4) with (δ * t , ω * t ) = (0, 0) as expected.It is not difficult to see that this saddle belongs to the unitary region for δ φ > 1/2 when (a, b) are in the diamond i.e. a, b > 0, a + b < 1. Matching the two sides after doing the saddle point integral and after setting (σ, τ ) = (0, 0) gives, a b )s( This is an analytic function of (δ, ω) in the unitary domain.We have plotted it in figure 4. Thus the knowledge of the saddle point as a function of (a, b) has allowed us to fix the OPE density completely at large D. After determining gδ,ω , we can match the O(1) function that multiplies e D... terms on both sides.This allows us to compute fδ,ω .Simplifying this function to a closed-form turns out to be difficult.We have computed it numerically.
As the universal saddle point is dual to identity operator, we expect the OPE density f e Dg obtained above to be the large D limit of the OPE density of GFFT.The GFFT OPE coefficients are known in closed form in arbitrary dimensions [24].We reproduce them below for reference.
In the second line we have separated all the D dependence used to take the scaling limit.The large D limit is taken using Stirling's approximation for the Gamma function, Γ and it agrees with fδ,ω computed numerically using crossing symmetry.The u-symmetric contribution u ∆ φ is subleading in the diamond and comes from a saddle point that lies in the complex (δ, ω) space.One can indeed check by taking the OPE density (4.8), (4.11) that the t 2 saddle precisely gives this contribution.
Discussion and outlook
In this paper, we have argued that, at large D, the four-point function of identical scalar operators with δ φ < 1 is the same as that in the GFFT to leading order.First, we note that it is unreasonable to expect this result to extend beyond δ φ < 1.This is for the following reason.Consider a tensor product of GFFT of N fields χ i such that they have the same conformal dimension ∆ χ = ∆ φ /2.This theory has SO(N ) global symmetry.Importantly, as δ φ > 1, this GFFT is unitary.Consider the four-point function of flavor singlet operator χ i χ i .We define this to be the operator φ whose four-point function we consider.By construction, the conformal dimension of φ is ∆ φ .This four-point function of the composite operator is computed by all the Wick contractions.The stripped four-point function is (5.1) Here we have normalized the two-point function of φ to be 1.First three terms come from the disconnected diagrams.This is the same as what would appear in the GFFT of φ itself.The next three terms come from connected diagrams and can have a relative factor compared to the disconnected piece.At large D, the disconnected piece comes from the identity operator and the two saddle point that are discussed in section 4.2 while the connected terms come from three new saddle points.Among the three terms, (u/v) ∆ φ /2 is leading in the diamond.Interestingly, it is self-dual under crossing symmetry.Substituting (u, v) = (a 2 e σ/D , b 2 e τ /D ) and matching coefficients of σ and τ in the exponent with the form (2.5), we see that the saddle point is at (k s− , k s+ ) = (k t− , k t+ ) = (δ φ /4, δ φ /4).In terms of (δ, ω) this means, Interestingly this saddle point lies in the unitary domain for δ φ ≥ 1 and when (a, b) are in the diamond i.e. for a, b > 0, a + b < 1.This is consistent with our analysis because precisely for δ φ > 1, crossing and unitarity allow other solutions apart from the universal one and (5.2) is one of them.However, one may wonder whether this is the only other solutions to the crossing equation (4.4) for δ φ > 1 (and δ φ < 3/2) 10 .This would mean that for δ φ > 1, the GFFT of φ and GFFT of χ (where φ is the composite operator χ 2 ) are the only two solutions at large D. This is an interesting possibility but one about which we can't say anything currently.Of course, it is entirely possible that for δ φ > 1, things may not be as simple as that.Note that we are asking this question without imposing the existence of stress tensor.Having a stress tensor could provide additional constraints, more severely constraining the solutions at large D.
In this paper, we have constrained the unitary solutions to the crossing for δ φ < 1 only to leading order at large D. At sub-leading order, the solution may receive perturbative 1/D corrections in addition to the non-perturbative ones.The non-perturbative corrections correspond to subleading saddle points.As discussed below equation (3.2), we already know of a mechanism by which these could appear, namely, from the rapidly oscillating part of the OPE density (which we have chosen to smear over).We believe that these corrections would be very difficult to control.However, the perturbative corrections can be accounted for relatively straightforwardly.They come from 1/D corrections to the conformal blocks as well as 1/D corrections to the smeared OPE density.The corrections to the conformal blocks can be computed from the conformal Casimir equation.The question of computing 1/D corrections to the correlator then is actually the question of controlling the 1/D corrections to the smeared OPE density.As the contribution of the saddle point that is dual to the identity operator is completely fixed to all orders in 1/D, the 1/D corrections to the smeared OPE coefficient density must also be fixed.It would be interesting to compute these corrections explicitly and match them with the GFFT OPE coefficients in 1/D expansion.
Our arguments in the paper have uniquely fixed the leading order correlator only in the Lorentzian diamond.Elsewhere in the cross-ratio space, the conformal blocks could rapidly oscillate in phase and hence the dominating saddle could be anywhere in the complex (δ, ω) space.That is why it would seem difficult to extend the results outside the diamond.However, in addition to fixing the correlator in the diamond, we are also able to fix the leading order OPE density (see section 4.2).This strongly suggests that the theory of operator φ is perhaps GFFT itself which would make φ decoupled from the rest of the theory.Phrased another way, our results suggest that OPE coefficients of relevant operators in a non-trivial conformal theory 11 in large dimensions must be exponentially suppressed at large D.
Even though we are considering D a formal parameter in solving conformal bootstrap equations, there are indications that the unitary solution space to crossing is drastically different for non-integer D compared to integer D, see [25,26].In our analysis, we have not made any assumptions about integrality of D and the large D limit could as well be taken with D an integer.It would be useful to understand how the nature of the solution space changes as we make D fractional in our approach.
We would also like to point out the paper [27] where authors show that the number of subtractions to write a dispersion relation goes to infinity as D goes to infinity.This is perhaps an indication that such theories prefer to be free [28].
In this paper, the object that played an important role is the smeared OPE density.A more rigorous analysis of the constraints on the smeared OPE density may be possible using the so-called "Tauberian theorems".These techniques have recently been applied to conformal field theories to essentially estimate the errors associated with such a smearing [29][30][31][32].It would be nice to put our treatment on a more solid footing with the use of similar techniques.Finally, it would be interesting to approach the problem of CFTs in large dimensions in other ways, for example using the Lorentzian inversion formula [33] or using the method of extremal functionals which has provided optimal analytic bounds on OPE coefficients in the large ∆ limit [34].
with, A = 1 + 4(1 − ŷ+ )δ(δ − 1) and ŷ+ = a 2 (1+b) 2 .Similar steps can be done to find the spin dependent part of the block at y ± = a The integral of the first type is called the saddle point integral.Of course, the integral could have other local maxima and the integral receives similar but subdominant contribution from those.If the function g(x) is piecewise continuous, we divide the integration range in these pieces and the above discussion applies to the locally dominant point from each piece.We want to highlight the non-analytic behavior of the integral as the dominant point transits across the integration range.To that end, consider a Laplace type integral with a parameter λ.
It is clear that I(λ) is non-analytic at λ = 0 i.e. at the point where the saddle point enters the integration range.In fact the integral increases in magnitude as the maximum transitions inside the integration range.On the other hand, if g(x) has a varying imaginary part in the integration range then the integrand has a rapidly oscillating phase.Naturally such an integral is difficult to estimate by staying on the real line as the oscillating phase is expected to cancel huge numbers to give tiny remainders.The way out is to use the analyticity of the integrand to deform the contour into the complex plane such that the new contour of integration is on the path of stationary phase i.e. the path of constant Im[g(z)].This is also the path of steepest descent for Re[g(z)].
Here we want the reader to note that as a result of contour deformation, the saddle point could be outside the integration range, even in the complex plane.For some mixed types of overlaps see figure 9.Here also we have taken δ φ = 0.6.
1 <
F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e
2 <
4 8 5 b 1 4 7 9 7 H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > 1 l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b
1 <
F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e H N U c 6 L 8 + 5 8 L F p L T j F z D H / g f P 4 A k Y y P H A = = < / l a t e x i t > !< l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7 2 < l a t e x i t s h a 1 _
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > 1 b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b
1 < l a t e x i t s h a 1 _ b a s e 6 4 =
F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e F R + f i k r e N U M W y g b N S P 9 W Y U D a h I + x Z K m m E 2 s 8 W h 8 7 I h V W G J I y V L W n I Q v 0 9 k d F I 6 2 k U 2 M 6 I m r F e 9 e b i f 14 v N e G N n 3 G Z p A Y l W y 4 K U 0 F M T O Z f k y F X y I y Y W k K Z 4 v Z W w s Z U U W Z s N i U b g r f 6 8 j p p X 1 U 9 t + o 1 r y v 1 2 z y O I p z B O V y C B z W o w z 0 0 o A U M E J 7 h F d 6 c R + f F e X c+ l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e F R + f i k r e N U M W y g b N S P 9 W Y U D a h I + x Z K m m E 2 s 8 W h 8 7 I h V W G J I y V L W n I Q v 0 9 k d F I 6 2 k U 2 M 6 I m r F e 9 e b i f 14 v N e G N n 3 G Z p A Y l W y 4 K U 0 F M T O Z f k y F X y I y Y W k K Z 4 v Z W w s Z U U W Z s N i U b g r f 6 8 j p p X 1 U 9 t + o 1 r y v 1 2 z y O I p z B O V y C B z W o w z 0 0 o A U M E J 7 h F d 6 c R + f F e X c+ l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e F R + f i k r e N U M W y g b N S P 9 W Y U D a h I + x Z K m m E 2 s 8 W h 8 7 I h V W G J I y V L W n I Q v 0 9 k d F I 6 2 k U 2 M 6 I m r F e 9 e b i f 14 v N e G N n 3 G Z p A Y l W y 4 K U 0 F M T O Z f k y F X y I y Y W k K Z 4 v Z W w s Z U U W Z s N i U b g r f 6 8 j p p X 1 U 9 t + o 1 r y v 1 2 z y O I p z B O V y C B z W o w z 0 0 o A U M E J 7 h F d 6 c R + f F e X c+ l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e F R + f i k r e N U M W y
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7 2 <
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > 1 l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > 1 < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > 1 < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1v b O 8 X d 0 t 7 + w e F R + f i k r e N U M W y x W M S q G 1 C N g k t s G W 4 E d h O F N A o E d o L J 3 d z v P K H S P J Y P Z p q g H 9 G R 5 C F n 1 F i p 6 Q 3 K F b f q L k D W i Z e T C u R o D M p f / W H M 0 g i l Y Y J q 3 f P c x P g Z V Y Y z g b N S P 9 W Y U D a h I + x Z K m m E 2 s 8 W h 8 7 I h V W G J I y V L W n I Q v 0 9 k d F I 6 2 k U 2 M 6 I m r F e 9 e b i f 1 4 v N e G N n 3 G Z p A Y l W y 4 K U 0 F M T O Z f k y F X y I y Y W k K Z 4 v Z W w s Z U U W Z s N i U b g r f 6 8 j p p X 1 U 9 t + o 1 r y v 1 2 z y O I p z B O V y C B z W o w z 0 0 o A U M E J 7 h F d 6 c R + f F e X c+ l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e F R + f i k r e N U M W y x W M S q G 1 C N g k t s G W 4 E d h O F N A o E d o L J 3 d z v P K H S P J Y P Z p q g H 9 G R 5 C F n 1 F i p 6 Q 3 K F b f q L k D W i Z e T C u R o D M p f / W H M 0 g i l Y Y J q 3 f P c x P g Z V Y Y z g b N S P 9 W Y U D a h I + x Z K m m E 2 s 8 W h 8 7 I h V W G J I y V L W n I Q v 0 9 k d F I 6 2 k U 2 M 6 I m r F e 9 e b i f 1 4 v N e G N n 3 G Z p A Y l W y 4 K U 0 F M T O Z f k y F X y I y Y W k K Z 4 v Z W w s Z U U W Z s N i U b g r f 6 8 j p p X 1 U 9 t + o 1 r y v 1 2 z y O I p z B O V y C B z W o w z 0 0 o A U M E J 7 h F d 6 c R + f F e X c+ l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = "> A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e F R + f i k r e N U M W y x W M S q G 1 C N g k t s G W 4 E d h O F N A o E d o L J 3 d z v P K H S P J Y P Z p q g H 9 G R 5 C F n 1 F i p 6 Q 3 K F b f q L k D W i Z e T C u R o D M p f / W H M 0 g i l Y Y J q 3 f P c x P g Z V Y Y z g b N S P 9 W Y U D a h I + x Z K m m E 2 s 8 W h 8 7 I h V W G J I y V L W n I Q v 0 9 k d F I 6 2 k U 2 M 6 I m r F e 9 e b i f 1 4 v N e G N n 3 G Z p A Y l W y 4 K U 0 F M T O Z f k y F X y I y Y W k K Z 4 v Z W w s Z U U W Z s N i U b g r f 6 8 j p p X 1 U 9 t + o 1 r y v 1 2 z y O I p z B O V y C B z W o w z 0 0 o A U M E J 7 h F d 6 c R + f F e X c+ l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e F R + f i k r e N U M W y x W M S q G 1 C N g k t s G W 4 E d h O F N A o E d o L J 3 d z v P K H S P J Y P Z p q g H 9 G R 5 C F n 1 F i p 6 Q 3 K F b f q L k D W i Z e T C u R o D M p f / W H M 0 g i l Y Y J q 3 f P c x P g Z V Y Y z g b N S P 9 W Y U D a h I + x Z K m m E 2 s 8 W h 8 7 I h V W G J I y V L W n I Q v 0 9 k d F I 6 2 k U 2 M 6 I m r F e 9 e b i f 1 4 v N e G N n 3 G Z p A Y l W y 4 K U 0 F M T O Z f k y F X y I y Y W k K Z 4 v Z W w s Z U U W Z s N i U b g r f 6 8 j p p X 1 U 9 t + o 1 r y v 1 2 z y O I p z B O V y C B z W o w z 0 0 o A U M E J 7 h F d 6 c R + f F e X c+ l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > (e) Type IV:D s 2 ∩ D t 2 < l a t e x i t s h a 1 _ b a s e 6 4 = " A U / v q + D A g M m 1 x 9 R h U c 2 0 a A 7 B J w Y = " > A A A B 7 X i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E 0 G P R i 8 c K9 g P a U D a b T b t 2 k w 2 7 E 6 G E / g c v H h T x 6 v / x 5 r 9 x 2 + a g r Q 8 G H u / N M D M v S K U w 6 L r f T m l t f W N z q 7 x d 2 d n d 2 z + o H h 6 1 j c o 0 4 y 2 m p N L d g B o u R c J b K F D y b q o 5 j Q P J O 8 H 4 d u Z 3 n r g 2 Q i U P O E m 5 H 9 N h I i L B K F q p 3 Q + 5 R D q o 1 t y 6 O w d Z J V 5 B a l C g O a h + 9 U P F s p g n y C Q 1 p u e 5 K f o 5 1 S i Y 5 N N K P z M 8 p W x M h 7 x n a U J j b H N U c 6 L 8 + 5 8 L F p L T j F z D H / g f P 4 A k Y y P H A = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " A U/ v q + D A g M m 1 x 9 R h U c 2 0 a A 7 B J w Y = " > A A A B 7 X i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E 0 G P R i 8 c K 9 g P a U D a b T b t 2 k w 2 7 E 6 G E / g c v H h T x 6 v / x 5 r 9 x 2 + a g r Q 8 G H u / N M D M v S K U w 6 L r f T m l t f W N z q 7 x d 2 d n d 2 z + o H h 6 1 j c o 0 4 y 2 m p N L d g B o u R c J b K F D y b q o 5 j Q P J O 8 H 4 d u Z 3 n r g 2 Q i U P O E m 5 H 9 N h I i L B K F q p 3 Q + 5 R D q o 1 t y 6 O w d Z J V 5 Ba l C g O a h + 9 U P F s p g n y C Q 1 p u e 5 K f o 5 1 S i Y 5 N N K P z M 8 p W x M h 7 x n a U J j b H N U c 6 L 8 + 5 8 L F p L T j F z D H / g f P 4 A k Y y P H A = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " A U/ v q + D A g M m 1 x 9 R h U c 2 0 a A 7 B J w Y = " > A A A B 7 X i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E 0 G P R i 8 c K 9 g P a U D a b T b t 2 k w 2 7 E 6 G E / g c v H h T x 6 v / x 5 r 9 x 2 + a g r Q 8 G H u / N M D M v S K U w 6 L r f T m l t f W N z q 7 x d 2 d n d 2 z + o H h 6 1 j c o 0 4 y 2 m p N L d g B o u R c J b K F D y b q o 5 j Q P J O 8 H 4 d u Z 3 n r g 2 Q i U P O E m 5 H 9 N h I i L B K F q p 3 Q + 5 R D q o 1 t y 6 O w d Z J V 5 Ba l C g O a h + 9 U P F s p g n y C Q 1 p u e 5 K f o 5 1 S i Y 5 N N K P z M 8 p W x M h 7 x n a U J j b H N U c 6 L 8 + 5 8 L F p L T j F z D H / g f P 4 A k Y y P H A = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " A U/ v q + D A g M m 1 x 9 R h U c 2 0 a A 7 B J w Y = " > A A A B 7 X i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E 0 G P R i 8 c K 9 g P a U D a b T b t 2 k w 2 7 E 6 G E / g c v H h T x 6 v / x 5 r 9 x 2 + a g r Q 8 G H u / N M D M v S K U w 6 L r f T m l t f W N z q 7 x d 2 d n d 2 z + o H h 6 1 j c o 0 4 y 2 m p N L d g B o u R c J b K F D y b q o 5 j Q P J O 8 H 4 d u Z 3 n r g 2 Q i U P O E m 5 H 9 N h I i L B K F q p 3 Q + 5 R D q o 1 t y 6 O w d Z J V 5 Ba l C g O a h + 9 U P F s p g n y C Q 1 p u e 5 K f o 5 1 S i Y 5 N N K P z M 8 p W x M h 7 x n a U J j b H N U c 6 L 8 + 5 8 L F p L T j F z D H / g f P 4 A k Y y P H A = = < / l a t e x i t > !< l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " q U g 8 i X I 0 1 z U D f 6 d o e a e Q G g C g J 1 c = " > A A A B 7 X i c b V D L S g N B E O y N r x h f U Y 9 e F o P g K e y K o M e g F 4 8 R z A O S J c x O e p M x 8 1 h m Z o U Q 8 g 9 e P C j i 1 f / x 5 t 8 4 S f a
4 8 5 b 1 4 7 9 7 2 <
H o r X g 5 T P H 8 A f e 5 w + Q M Y 8 b < / l a t e x i t > 1 l a t e x i t s h a 1 _ b a s e 6 4 = " R l u O 4 v E g n 5 G D X l m i f g 6 c 9 8 H j b 0 4 = " > A A A B 7 n i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 m K o M e i F 4 8 V 7 A e 0 o W y 2 m 3 b p Z h N 2 J 0 I J / R F e P C j i 1 d / j z X / j N s 1 B W x 8 M P N 6 b Y W Z e k E h h 0 H W / n d L G 5 t b 2 T n m 3 s r d / c H h U P T 7 p m D j V j L d Z L G P d C 6 j h U i j e R o G S 9 x L N a R R I 3 g 2 m d w u / + 8 S 1 E b F b f X m d d B p 1 z 6 1 7 D 1 e 1 5 m 0 R R x n O 4 B w u w Y N r a M I 9 t K A N D K b w D K / w 5 i T O i / P u f C x b S 0 4 x c w p / 4 H z + A K f / j x s = < / l a t e x i t > 1 < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9 e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F 4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e + l q 0 F J 5 8 5 h T 9 w P n 8 A e p 2 M t Q = = < / l a t e x i t > < l a t e x i t s h a 1 _ b a s e 6 4 = " l 9e I m v Y c F O K p z E D j i / n 9 j P D e W b 8 = " > A A A B 6 H i c b V B N S 8 N A E J 3 U r 1 q / q h 6 9 L B b B U 0 l E q M e i F4 8 t 2 A 9 o Q 9 l s J + 3 a z S b s b o Q S + g u 8 e F D E q z / J m / / G b Z u D t j 4 Y e L w 3 w 8 y 8 I B F c G 9 f 9 d g o b m 1 v b O 8 X d 0 t 7 + w e
Figure 2 .
Figure 2. In these figures we have denoted the unitary domain in the s-channel the (δ * 1 , ω * 1 ) plane by blue color and have superimposed the image of the unitary domain in the t-channel under the map (4.4) in red.In (a) we do not have any overlap.In such a scenario there are no solution to the crossing equation compatible with unitarity.In subfigures (b), (c), (d), (e) we have shown possible elementary overlaps of D 1 and D 2 parts of the unitary domain in the s and t channel.Of course, combinations of such overlaps is also a possibility.For example in (f), overlaps types II, III and IV occur.
Figure 3 .
Figure 3.The plots show regions of (a, b) for different values of δ φ .The color coding is as follows.Light-pink: no solution, gray: type I, light-green: type II, light-blue: type III, green: type I + II, blue:type I + III, light-yellow: type II + III, yellow: type I + II + III, orange: type II + IV, brown: type III + IV, red: type II + III + IV.Note that the light-pink region is present for δ φ < 3/4.
. 7 )
Here we have made manifest the e D... dependence of the OPE density by taking C δ,ω ≡ fδ,ω e D gδ,ω .The notation det(g ) is a shorthand for the Hessian (the determinant of the matrix of second order partial derivatives) of (g δ + g ω + gδ,ω ) at (δ * s , ω * s ).The quantity multiplying D in the exponent in the conformal block is g δ + g ω .The functions g δ and g ω are given explicitly in appendix A. Inverting the relation (4.6) to express (a, b) in terms of (δ * s , ω * s ) and matching the function in the exponent multiplying D, we get the exponential dependence in the OPE density.gδ,ω = log 4 2δ φ −δ s(δ)s(ω + 1 2 )s( δ−ω+2δ φ −2 2
Figure 7 .
Figure 7. Plot (a) shows the overlap of type-II.Orange point is inside the two-dimensional blue region as required.Plot (b) shows the light-green region in (a, b) plane where type-II overlap can exist.The green and dashed green lines are a = 4δ φ +1 4δ φ −1 b − 1 and a = 2 2δ φ −1 4δ φ −1 (4δ φ − b − 1) respectively.The black point is (a = 0.15, b = 0.4) for which part (a) is plotted.
Figure 8 .
Figure 8. Plot (a) shows the overlap of type-I.Blue and red one dimensional regions intersect each at green point as required.Plot (b) shows the gray region in (a, b) plane where type-I overlap can exist.The blue, green, dashed orange and dashed brown lines are the same ones as shown in the previous figures.The black point is (a = 0.4, b = 0.4) for which part (a) is plotted. | 2020-02-25T02:01:23.049Z | 2020-02-24T00:00:00.000 | {
"year": 2020,
"sha1": "b4c8fa8b1631c4cbfa441b12761d3925802877e2",
"oa_license": "CCBY",
"oa_url": "https://link.springer.com/content/pdf/10.1007/JHEP02(2022)035.pdf",
"oa_status": "GOLD",
"pdf_src": "Arxiv",
"pdf_hash": "b4c8fa8b1631c4cbfa441b12761d3925802877e2",
"s2fieldsofstudy": [
"Mathematics"
],
"extfieldsofstudy": [
"Physics"
]
} |
235368950 | pes2o/s2orc | v3-fos-license | Optimising geographical accessibility to primary health care: a geospatial analysis of community health posts and community health workers in Niger
Background Little is known about the contribution of community health posts and community health workers (CHWs) to geographical accessibility of primary healthcare (PHC) services at community level and strategies for optimising geographical accessibility to these services. Methods Using a complete georeferenced census of community health posts and CHWs in Niger and other high-resolution spatial datasets, we modelled travel times to community health posts and CHWs between 2000 and 2013, accounting for training, commodities and maximum population capacity. We estimated additional CHWs needed to optimise geographical accessibility of the population beyond the reach of the existing community health post network. We assessed the efficiency of geographical targeting of the existing community health post network compared with networks designed to optimise geographical targeting of the estimated population, under-5 deaths and Plasmodium falciparum malaria cases. Results The per cent of the population within 60-minute walking to the nearest community health post with a CHW increased from 0.0% to 17.5% between 2000 and 2013. An estimated 10.4 million people (58.5%) remained beyond a 60-minute catchment of community health posts. Optimal deployment of 7741 additional CHWs could increase geographical coverage from 41.5% to 82.9%. Geographical targeting of the existing community health post network was inefficient but optimised networks could improve efficiency by 32.3%–47.1%, depending on targeting metric. Interpretations We provide the first estimates of geographical accessibility to community health posts and CHWs at national scale in Niger, highlighting improvements between 2000 and 2013, geographies where gaps remained and approaches for optimising geographical accessibility to PHC services at community level.
Table of Contents
Female ASC in 2013, n=353. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. ***Gender for 2 ASC was not recorded, and these ASC were excluded from the gender analysis. secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. ***Gender for 2 ASC was not recorded, and these ASC were excluded from the gender analysis. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. ASC in 2013 trained on promotion of childhood immunization, n=546. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands.
Supplementary Figure 2 (H). Geographic accessibility (travel time in minutes, walking in dry conditions) to the nearest ASC in 2013 trained on family planning at 100m x 100m resolution.
ASC in 2013 trained on family planning, n=183. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. ASC in 2013 trained on promotion of maternal health, n=300. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. ASC in 2013 trained on AMTL, n=15. ASC=Agent de santé communautaire. AMTL=Active management of the third stage of labor. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. ASC in 2013 trained on essential newborn care, n=108. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands.
Supplementary Figure 3 (A). Geographic accessibility (travel time in minutes, walking in dry conditions) to the nearest CS in 2013 with a nurse or ASC at 100m x 100m resolution.
CS in 2013 with a nurse or ASC, n=1739 (***does not include 13 CS that met this criteria but did not have geocoordinates). CS=Case de santé. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. CS with no stockout of any iCCM commodity lasting longer than seven days, n=640 (***does not include 8 CS that met this criteria but did not have geocoordinates). iCCM commodities = RDT and AL for malaria, low osmolarity ORS and zinc sulfate for diarrhea, cotrimoxazole (pill or syrup) for pneumonia. A stockout of any of these commodities lasting longer than 7 days resulted in the CS being considered as a CS with a severe stockout of any iCCM commodity. CS=Case de santé. ASC=Agent de santé communautaire. iCCM=integrated community case management. RDT=rapid diagnostic test for malaria. AL=artemether-lumefantrine. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. CS in 2013 with a nurse or ASC and no stockout of any iCCM commodity lasting longer than seven days, n=591 (does not include 7 CS that met this criteria but did not have geocoordinates). iCCM commodities = RDT and AL for malaria, low osmolarity ORS and zinc sulfate for diarrhea, cotrimoxazole (pill or syrup) for pneumonia. A stockout of any of these commodities lasting longer than 7 days resulted in the CS being considered as a CS with a severe stockout of any iCCM commodity. CS=Case de santé. ASC=Agent de santé communautaire. iCCM=integrated community case management. RDT=rapid diagnostic test for malaria. AL=artemetherlumefantrine. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. CS in 2013 with a nurse or ASC and no stockout of RDT or AL lasting longer than seven days, n=1038 (does not include 11 CS that met this criteria but did not have geocoordinates). A stockout of >= 7 days was considered a severe stockout. CS=Case de santé. ASC=Agent de santé communautaire. RDT=rapid diagnostic test for malaria. AL=artemether-lumefantrine. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. CS in 2013 with a nurse or ASC and no stockout of ORS or zinc lasting longer than seven days, n=1159 (does not include 9 CS that met this criteria but did not have geocoordinates). A stockout of >= 7 days was considered a severe stockout. CS=Case de santé. ASC=Agent de santé communautaire. ORS = low osmolarity oral rehydration solution. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. CS in 2013 with a nurse or ASC and no stockout of cotrimoxazole (pill or syrup) lasting longer than seven days, n=1172 (does not include 7 CS that met this criteria but did not have geocoordinates). A stockout of >= 7 days was considered a severe stockout. CS=Case de santé. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. CS in 2013 with a nurse or ASC and no stockout of RUTF lasting longer than seven days, n=1463 (does not include 9 CS that met this criteria but did not have geocoordinates). A stockout of >= 7 days was considered a severe stockout. CS=Case de santé. ASC=Agent de santé communautaire. RUTF=ready-to-eat therapeutic food. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. Female ASC in 2013, n=353. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. ***Gender for 2 ASC was not recorded, and these ASC were excluded from the gender analysis. road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. ***Gender for 2 ASC was not recorded, and these ASC were excluded from the gender analysis.
Supplementary Figure 5 (A). Contribution of ASC to additional geographic accessibility beyond the existing CSI and CS (without ASC) networks in 2013 (walking scenario) at 100m x 100m resolution.
ASC in 2013, n=1457, walking scenario. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. Figure 5
(B). Contribution of ASC to additional geographic accessibility beyond the existing CSI and CS (without ASC) networks in 2013 (walking + motorized transportation scenario) at 100m x 100m resolution.
ASC in 2013, n=1457, walking + motorized transportation scenario. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands.
Supplementary Figure 5 (C). Contribution of ASC trained on iCCM to additional geographic accessibility beyond the existing CSI and CS (without ASC) networks in 2013 (walking scenario) at 100m x 100m resolution.
ASC in 2013 trained on iCCM, n=1214, walking scenario. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands.
(D). Contribution of ASC trained on iCCM to additional geographic accessibility beyond the existing CSI and CS (without ASC) networks in 2013 (walking + motorized transportation scenario) at 100m x 100m resolution.
ASC in 2013 trained on iCCM, n=1214, walking + motorized transportation scenario. ASC=Agent de santé communautaire. *For visualization purposes road classes limited to motorway, trunk, primary, secondary, and tertiary. **Other water bodies from landcover layer included permanent water bodies, temporary water bodies and herbaceous wetlands. Merged land cover at 100m x 100m and 1km x 1km resolutions (latter not shown) derived using the " erge land over" tool in ess od v 10 . oad lasses " onstru tion" and "bridle ay" not sho n due to spa e limitations. We obtained vector shapefiles for administrative boundaries 0-3 developed by the Institut Géographique National Niger (IGNN) and OCHA in 2017 with updates from the REACH Initiative in 2018, accessed 14 February 2018, at https://data.humdata.org/dataset/niger-administrative-boundaries. 12 We reprojected the shapefiles for the BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) Health system pyramid and health service delivery networks The health system of Niger included a public and private sector organized in a decentralized, pyramidal structure with three administrative levels: a central level composed of the cabinet of the Minister of Public Health, the Secretary General and General/National Directorates, responsible for strategy and managing national hospitals, and national maternities and referral centers; a regional level composed of Regional Directorates, responsible for managing regional hospitals, and regional maternities and referral centers; and a district level, composed of District Health Teams, responsible for managing district hospitals, a network of first-level health facilities called centre de santé intégré (CSI), a network of community health posts called case de santé (CS)attached to the network of CSI as well as a small network of private clinics and practices. 14 Structures at the central and regional levels, as well as district hospitals provided referral, counter referral, specialist, and emergency services not available at the peripheral level through the centre de santé intégré (CSI) and Case de santé (CS) networks. 15 As of December 2012, there were 856 CSI, offering a minimum package of services, focused on primary health care, referral from and counter-referral to the CS, and supervision of the CS. CSI were typically staffed by nurses and in certain large communes by a generalist doctor and midwives. 14 According to national norms, CSI in rural areas (CSI Type I) serve a maximum population of 10000 and a maximum population of 5000 in rural areas with low population density, while CSI in urban areas or areas with high population density (CSI Type II) serve a maximum population of 15000. 15 As of December 2012, there were 2451 CS. 14 According to national norms, CS were attached to the CSI in the hierarchy of the health system, were intended to be situated beyond 5km from a CSI, and served a population of 2500 to 5000. 15 CS provided a minimum package of activities, focused on primary health care: case management for common infectious diseases, including acute respiratory illness, diarrhea, and malaria, referral services for severe or complicated cases, reproductive health services (family planning, antenatal care, assisted delivery and referral for pregnancies at elevated risk or with complications) and health promotion. 15 CS typically were staffed by a cadre of full-time agent de santé communautaire (ASC), community health workers, who were typically contracted, paid a monthly salary of roughly $100 USD, had completed at least secondary education, and received a six-month pre-deployment basic training on the minimum package and a sixday training on iCCM after deployment. 11,15-17 ASC typically provided services from the CS (i.e. fixed site service delivery) and did not typically provide mobile services or household visits. In 2013, there were 1535 ASC (1154 male and 381 female). 11 In addition 21.6% of CS had at least one nurse in 2013 (232 nurses were deployed at CS in 2013) and 42.0% of CS had at least one relais communautiare (RC)a network of volunteer community health workers attached to the CS and providing health promotion and prevention interventions in the communities within the catchment area of the CS (2672 RC were supporting CS in 2013). 11
Centre de santé intégré network
Through a data sharing agreement with UNICEF, we obtained a vector point shapefile dataset in the CRS EPSG:4326, WGS 84 with the global positioning system (GPS) coordinates and basic identification information for all CSI (n=849) in Niger collected through a national, georeferenced census of CSI, CS and ASC conducted in 2013 by the National Institute of Statistics of Niger (INS), Ministry of Public Health (MoPH) of Niger, and UNICEF. 11 We found that 10 records were misclassified as CSI and these were removed, leaving 839 CSI. We triangulated the CSI dataset with the CS dataset (below) to ensure no duplication or misclassification of CSI as CS and vice versa.
We reprojected the CSI shapefile to the CRS EPSG:32632 -G 84 / T one N, using the G " arp" tool in QGIS 3.12.0-Bu areşti. 13 For our analysis of geographic coverage, the maximum population capacity of a CSI was set at a population of 10000 for both CSI Type I CSI Type II to simplify the analysis and because we deemed a maximum population capacity of 15000 for CSI in urban areas (as noted above) unrealistic.
Case de santé network
We obtained, through the data sharing agreement with UNICEF noted above, a vector point shapefile dataset in the CRS EPSG:4326, WGS 84 with the GPS coordinates and basic identification information for 2409 CS in Niger BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) 11 Data was collected for 2432 structures (n=1703 structures with geocoordinates and complete interviews with the responsible health agent, n=294 structures with geocoordinates and partially complete interviews with the responsible health agent, n=3 structures with geocoordinates where the responsible agent declined to be interviewed, n=3 structures with geocoordinates but missing interviews with the responsible agent, and n=429 structures with geocoordinates that were closed at the time of the census or the responsible agent was absent. We excluded 23 CS records due to miscoding of CSI as CS, leaving 2409 CS records. In our analysis we included all CS records (n=2409) with geocoordinates, including those that were closed at the time of the census or the agent was absentwith the understanding that closures of CS and absences of responsible agents are typically temporary and vary from year to year 11,18 . We reprojected the CS shapefile to the CRS EPSG:32632 -WGS 84 / T one N, using the G " arp" tool in QG . . -Bu areşti. 13 For analyses at 1km resolution (geographic coverage, targeting and scale-up analysis) we adjusted GPS coordinates, where necessary, for barriers at 1km resolutionthese changes were maintained for analyses at 100m resolution (geographic accessibility). Detailed data on the availability of human resources for health and stockouts was available for a subset (n=1997) of the 2409 CS and our analysis of geographic accessibility to CS with available professional/trained human resources for health (e.g. had a nurseregistered nurse, certified nurse, state registered nurse, senior nursing technicianor ASC) and CS without severe stockouts of key commodities for case management of malaria, pneumonia and diarrhea was based on this subset of the data. "severe sto out" as defined as a sto out lasting seven days or longer. e considered key commodities for the case management of malaria (rapid diagnostic test, Artemether/lumefantrine 20/120 mg) pneumonia (cotrimoxazole in pill or syrup form), diarrhea (low osmolarity oral rehydration salt sachets and zinc sulfate 20 mg) and acute malnutrition (ready-to-use therapeutic food (RUTF)). For our analysis of geographic coverage, the maximum population capacity of a CS was set at a population of 2500.
Agent de santé communautaire network
In 2013, there were 1535 ASC (1154 male and 381 female). 11 We obtained, through the data sharing agreement with UNICEF noted above, a vector point shapefile dataset in the CRS EPSG:4326, WGS 84 with GPS coordinates of the work location of the ASC and detailed information for 1468 ASC (95.6% of the 1535 expected ASC) from 1421 CS, including socio-demographics, year of deployment, initial training and refresher training for specific interventions collected through a national, georeferenced census of CSI, CS and ASC conducted in 2013 by the INS of Niger, MoPH of Niger, and UNICEF. 11 We found 11 ASC without GPS coordinates and excluded them from analysis, leaving 1457 ASC (94.9% of the 1535 expected ASC). We reprojected the ASC shapefile to the CRS EPSG:32632 -G 84 / T one N, using the G " arp" tool in QG 3.12.0-Bu areşti. 13 For analyses at 1km resolution (geographic coverage, targeting and scale-up analysis) we adjusted GPS coordinates, where necessary, for barriers at 1km resolutionthese changes were maintained for analyses at 100m resolution (geographic accessibility).
We prepared separate vector point files using CRS EPSG:32632 -WGS 84 / UTM zone 32N for the ASC network, according to gender of the ASC, year of deployment and training on iCCM. We found that 1316 (90.4%) of ASC were located at the CS to which they were attached but 141 (9.6%) had unique GPS coordinates greater than 100m from the nearest CS. For our analysis of geographic coverage, the maximum population capacity of an ASC was set at a population of 2500. For ASC based at a CS, the maximum population capacity was maintained at 2500 (i.e. they were considered as contributors to the maximum population capacity of the CS).
CS-ASC network
We prepared a vector point shapefile with CRS EPSG:32632 -WGS 84 / UTM zone 32N that combined the CS (n=2409) and ASC with unique GPS coordinates (n=141) into a single CS-ASC network (n=2550). For our analysis of geographic coverage, the maximum population capacity of a CS-ASC was set at a population of 2500.
Optimized CS-ASC network
For our targeting analysis, we prepared three vector point shapefiles for hypothetical CS-ASC networks: 1) optimizing geographic coverage of the estimated residual population in 2013 beyond the geographic coverage of the CSI network (60-minutes walking considering maximum population capacity) 2) optimizing geographic coverage of the estimated residual under-five deaths in 2013 beyond the geographic coverage of the CSI network and 3) optimizing geographic coverage of the estimated residual Pf malaria cases among all ages (0-99) in 2013 beyond the geographic coverage of the CSI network to compare against the existing CS-ASC network, given the same number of CS-ASC as the existing CS-ASC network (n=2550), at 1km x 1km resolution. The optimized CS-ASC networks were prepared using the following steps: 1. Using the population beyond the geographic coverage of the CSI network --i.e. the population beyond the 60minute catchment of the CSI network, with maximum population capacity of 10000 population per CSI, we used the " aster al ulator" in QG . . -Bu areşti 13 to create a dummy raster containing cells at 1km x 1km resolution with greater than or equal to 500 people. The cut-off of greater than or equal to 500 people was chosen as it reflects 25% of the maximum population capacity (2500 people) of a CS-ASC 16 and we assumed deployment of CS-ASC to cells with less than 500 people would be contrary to country norms 16 .
2.
e ve tori ed the raster from step using the " olygoni e" tool in QG . . -Bu areşti, 13 resulting in a point vector shapefile of 5796 potential CS-ASC sites.
See the section below on the Targeting analysis for further details on preparation of these datasets.
Scaled-up relais communautaire network
The MoPH in Niger plans to scale-up the network of CHWs called relais communautaire (RC) for two contexts: 1) rural contexts at a ratio of 2 RC per 1000 population in communities beyond 5km of the CS-ASC or CSI networks to provide preventive, promotional and curative (e.g. iCCM) interventions and 2) in urban/peri-urban contexts at a ratio of 1 RC per 1000 population in communities within 5km of the CS-ASC and CSI networks to provide preventive and promotional interventions. For our scale-up analysis, we focus on the former. We prepared a hypothetical "optimi ed" net or n= 9 to over the population in ells ith at least people in beyond the geographic coverage of the existing CSI and CS-ASC networks at 1km x 1km resolution using the following steps: 1. Using the population beyond the geographic coverage of the existing CS-ASC networki.e. the population beyond the 60-minute catchment of the CS-ASC network, with maximum population capacity of 2500 population per CS-e used the " aster al ulator" in QG . . -Bu areşti 13 to create a dummy TIFF raster containing cells at 1km x 1km resolution with greater than or equal to 500 people.
2.
e ve tori ed the raster from step using the " olygoni e" tool in QG . . -Bu areşti, 13 resulting in a point vector shapefile of 3521 candidate RC sites, with 7042 RC at a ratio of 2 RC per site or 2 RC per 1000 people based on the national norm. 16 3. In our scale-up analysis (described below) we filtered out candidate sites with a realized capacity (i.e., the population covered within the catchment) of less than 500 population, leaving 3296 candidate sites for the scale-up analysis.
Roads
We obtained a vector line shapefile for the road network in Niger developed by the Humanitarian OpenStreetMap Team, accessed on 1 August 2018, at https://data.humdata.org/dataset/hotosm_niger_roads. 7 To prepare the final roads file, e hanged the olumn " igh ay" to "label"; re lassified the road types using the standard OpenStreetMap categories described at https://wiki.openstreetmap.org/wiki/Key:highway; simplified the road typology by excluding road types with very few segments or of little importance/relevance to the study; added a " lass" variable in order to enable lin ing ith the travel time s enarios; and reproje ted the from G:4 -WGS84 to EPSG:32632 -WGS 84 / UTM zone 32N in alignment with the final DEM (see files t_NER_reclass_roads_OSM.xls, v_NER_roads_100m_final.shp and v_NER_roads_1km_final.shp in Supplementary Appendix 1c at https://doi.org/10.6084/m9.figshare.13536779.v6). As described below in the section on the merged land cover raster, for our analysis at 100m x 100m resolution we uploaded the vector line shapefile for the road network into our Accessmod v5 project at 100m x 100m resolution and used the merge land cover tool in Accessmod v5 to rasterize the vector line shapefile for the road network as part of the merged land cover raster at 100m x 100m resolution. For our analysis at 1km x 1km resolution (geographic coverage, targeting and scale-up analysis) we repeated the above within our Accessmod v5 10 project at 1km x 1km resolution.
Rivers and Other Waterbodies
Rivers and other waterbodies were considered barriers to movement, where they were not crossed by a road. We obtained vector line shapefiles for rivers from HOT Open Street Map (HOTOSM), accessed on 15 January 2018, at https://data.humdata.org/dataset/hotosm_niger_waterways. 8 For our analysis at 100m x100m resolution (geographic accessibility), we reprojected the CRS from EPSG:4326 -WGS84 to CRS EPSG:32632 -WGS 84 / UTM zone 32N in alignment with the final DEM see file "v_N _rivers_final.shp" in Supplementary Appendix 1c at https://doi.org/10.6084/m9.figshare.13536779.v6). As described below in the section on the merged land cover raster, for our analysis at 100m x 100m resolution we uploaded the vector line file for rivers into our Accessmod v5 10 project at 100m x 100m resolution and used the merge land cover tool in Accessmod v5 10 to rasterize the vector line shapefile for rivers as part of the merged land cover raster at 100m x 100m resolution. For our analysis at 1km x 1km resolution (geographic coverage, targeting and scale-up analysis) we repeated the above within our Accessmod v5 10 project at 1km x 1km resolution. Data on other water bodies (permanent and temporary) were already included as part of the land cover raster described above.
Merged land cover
For our geographic accessibility analysis, we prepared a merged land cover raster at 100m x 100m resolution using the " erge land over" tool in ess od v 10 see file "r_NER_land_merged_100m_final.tif" in Supplementary Appendix 1c at https://doi.org/10.6084/m9.figshare.13536779.v6). The process is described in detail in Ray et al, 2008. 10 n brief, the " erge land over" tool sta s, orders, and merges the road net or , barriers rivers and other waterbodies, the later from the land cover), and land cover files into a single raster dataset. For our analysis at 1km x 1km resolution (geographic coverage, targeting and scale-up analysis) we prepared a merged land cover raster at 1km x 1km resolution using the process described above within our Accessmod v5 10 project at 1km x 1km resolution see the file "r_NER_land_merged_100m_final.tif" .
Travel scenario tables
We developed travel scenario tables for the following scenarios walking in dry conditions and walking to the nearest road and then using motorized transportation in dry conditions see files "t_NER_walk_dry. ls" and "t_NER_walk_veh_dry. ls" in Supplementary Appendix 1c at https://doi.org/10.6084/m9.figshare.13536779.v6). We set traveling speeds by mode of transportation (walking or walking + motorized transportation) for each land cover class and road class. Travel speeds were adapted from previous studies. 18,19 Population
Data preparation of population raster layers for the year 2013
We obtained a GeoTiff raster for the estimated population count for Niger in 2015 adjusted to UN population estimates at roughly 30m x 30m resolution, the High Resolution Settlement Layer (HRSL) from https://data.humdata.org/dataset/highresolutionpopulationdensitymaps-ner, courtesy of Facebook Connectivity Lab and Center for International Earth Science Information Network (CEISIN) at Columbia University, accessed 6 August 2020. 2 The 2015 HRSL was developed with computer vision techniques and supervised machine learning applied to high resolution commercial satellite imagery from the DigitalGlobe, courtesy of Maxar 20 to identify and classify human-built structures, combined with population estimates from the Gridded Population of the World v4 21 . Further details are provided elsewhere 22 . We also obtained a GeoTiff raster for the estimated population count for Niger in 2013, adjusted to UN population estimates, at roughly 100m x 100m resolution in Geographic Coordinate system WGS84 from Worldpop, accessed 3 March 2020. 3 A random forestbased dasymetric redistribution approach was used to develop the Worldpop dataset and is described in detail elsewhere. 23 We prepared a GeoTiff raster file for the estimated population count in 2013 at 100m x 100m resolution that adjusted the HRSL GeoTiff of the estimated population count in 2015 to the GeoTiff from Worldpop for the estimated population count in 2013 [ner_ppp_2013] at the lowest administrative level (adm3) but maintained the population settlement footprint of the 2015 HRSL. We kept the footprint of the 2015 HRSL because we deemed it more appropriate for our purposes (analysis of geographic accessibility to health services) than the footprint of the Worldpop raster based on visual inspection against satellite imagery for Niger and recent assessments of its accuracy 23 . The population footprint of the orldpop raster is "un onstrained", that is, smoothed a ross spa e, 24 including cells where there are no settlements, 25 whereas the HRSL is confined to cells with settlements. 2 We note that since the time of our analysis, an additional datasetthe World Settlement Footprint 2015 26has been made publicly available and Worldpop has developed population count datasets constrained to population settlement footprints. 27 Recent analyses suggest that modelling of population counts constrained to settlement footprints is improved through the use of multiple settlement footprints. 28,29 We adjusted the counts of the 2015 HRSL dataset to the Worlpop counts for 2013 to align with our analysis for the year 2013. The raster layer for the estimated population count in the year 2013 at 100m x 100m resolution was used in our analysis of geographic accessibility, geographic coverage and scale-up. We used the following steps to prepare the raster layer for the population count in 2013 at 100m x 100m resolution:
Data preparation of population raster layers for the years 2000-2012
We obtained GeoTiff rasters for the estimated population count for the years 2000-2012 in Niger, adjusted to UN population estimates, at roughly 100m x 100m resolution in Geographic Coordinate system WGS84 from Worldpop, accessed 3 March 2020. 3 e rasteri ed this ratio at m resolution using the " asteri e" tool in QGIS 3.12.0-Bu areşti 13 [r_NER_ratWP00OtFB13F_100m_unadj_barriers] with the ratio from step 3 as the burn and the extent of the DEM at 100m x 100m resolution as the extent. Using raster calculator in QGIS 3.12.0-Bu areşti, 13
Data preparation of live birth raster layer for the year 2013
We obtained a GeoTiff raster [NER_births_pp_v2_2015] for the estimated live birth count in 2015 for Niger, adjusted to UN population estimates, at roughly 1km x 1km resolution in Geographic Coordinate system WGS84 from Worldpop, accessed on February 18, 2021. 3 We prepared a GeoTiff raster layer for the estimated count of live births in 2013 at 1km x 1km resolution to be used in our targeting analysis for under-five deaths. We used the following steps:
Estimated under-five deaths
We used the following steps to prepare the raster layer for the estimated count of under-five (0-5 years old) deaths in Niger in 2013 at 1km x 1km resolution to be used in our targeting analysis:
Estimated Plasmodium falciparum malaria cases
We used the following steps to prepare a GeoTiff raster layer for the estimated count of Plasmodium falciparum malaria cases among all ages (0-99 years) in Niger in 2013 at 1km x 1km resolution to be used in our targeting analysis: 1. We obtained a GeoTiff raster file for modelled pixel-level estimates of the annual mean incidence of Note that we did not need to adjust for the estimated Pf malaria cases on barriers because this step was conducted when preparing the raster for the estimated population in 2013.
We repeated the steps above using GeoTiff raster files for the 95% lower bound estimate for mean incidence of a. What percentage of the population was within 30 min and 60 min of a CS-ASC in 2013, assuming a walking scenario in dry conditions? How did this vary across geographies? How did this vary by availability of trained human resources (nurse, ASC) and essential commodities? b. What percentage of the population was within 30 min and 60 min of a CS-ASC in 2013, assuming a scenario of walking to the nearest road and then using motorized transportation in dry conditions? How did this vary across geographies? How did this vary by availability of trained human resources (nurse, ASC) and essential commodities?
Methods for Geographic Accessibility question 1
We define accessibility coverage as the estimated percentage of people within a given travel time to the nearest health service delivery location of a given health service delivery network, accounting for travel speeds of different modes of transportation over different land cover classes and slope, with the direction of travel toward the health service delivery location. 10 We estimated accessibility coverage at 100m x 100m resolution for the CSI, CS and ASC networks in 2013and for the ASC network by gender, year of deployment (2000-2013), training, and availability of essential commoditiesusing 30-minute and 60-minute cutoffs for administrative levels 0-3 and the two travel scenarios. We used 30-minute and 60-minute cutoffs as previous analyses have shown careseeking decays as a function of travel time after these cutoffs 30 and they are clinically relevant (e.g. for prompt treatment of severe illness). 31 The analysis was constrained to national borders but allowed for travel across subnational administrative boundaries. e used the "geographi a essibility" module ithin ess od v . .48 10 to calculate travel time layers and the " onal statisti s" module to al ulate the onal statisti s for ea h travel time layer by administrative level. Analysis 1. We conducted a geographic accessibility analysis of the existing CSI network in 2013 based on a travel scenario of walking in dry conditions scenario at 100m x 100m resolution using Accessmod v5. a. We used the following data inputs: i. Population: raster_population_r_NER_FB13_100m_final ii. Land cover merged: raster_land_cover_merged_r_NER_land_merged iii. Scenario Table 1 for zonal statistics from these travel time rasters and detailed results for administrative layers 0-3 in Supplementary Appendix 2.
We repeated steps 1-4 above for the travel scenario walking to the nearest road and then taking motorized transportation, resulting in the a travel time raster for the areas with additional geographic accessibility beyond 60 minutes walking + motorized transportation in dry conditions of a CSI or CS (without an ASC) due to the contribution of ASC in 2013 at 100m resolution in CRS EPSG:32632 -WGS 84 / UTM zone 32N with the final DEM at 100m as the extent [r_NER_ga_additional_contribition_ASC_le2013_walkvehd_100m] (see Supplementary Appendix 1b at https://doi.org/10.5281/zenodo.4482969). See Table 1 for zonal statistics from these travel time rasters and detailed results for administrative layers 0-3 in Supplementary Appendix 2.
We repeated steps 1-4 above for ASC trained on iCCM, resulting in the a travel time raster for the areas with additional geographic accessibility to iCCM services beyond 60 minutes walking in dry conditions of a CSI or CS Table 1 for zonal statistics from these travel time rasters and detailed results for administrative layers 0-3 in Supplementary Appendix 2.
Finally we repeated steps 1-4 above for ASC trained on iCCM, using the walking + motorized transportation travel scenario, resulting in the a travel time raster for the areas with additional geographic accessibility to iCCM services beyond 60 minutes walking + motorized transportation in dry conditions of a CSI or CS (without an ASC) due to the contribution of ASC trained on iCCM in 2013 at 100m resolution in CRS EPSG:32632 -WGS 84 / UTM zone 32N with the final DEM at 100m as the extent [r_NER_ga_additional_contribition_ASC_le2013_iCCM_walkvehd_100m] (see Supplementary Appendix 1b at https://doi.org/10.5281/zenodo.4482969). See Table 1 for zonal statistics from these travel time rasters and detailed results for administrative layers 0-3 in Supplementary Appendix 2.
e used the "Zonal statisti s" tool in Accessmod v5 to calculate the percent of the population beyond 60 min of a CSI and CS (without an ASC) that were within 30 minutes and 60 minutes of an ASC in 2013, using walking and walking + motorized transportation travel scenarios. See Table 1 for zonal statistics from these travel time rasters and detailed results for administrative layers 0-3 in Supplementary Appendix 2.
Methods for Geographic Accessibility question 4
We repeated steps 1-3 from research question 1, using the following facility inputs: 1. CS-ASC network 2. CS-ASC network without a severe stockout of any iCCM commodity (severe stockout=stockout of any iCCM commodity lasting longer than seven days; iCCM commodities = RDT and AL for malaria, low osmolarity BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance ORS and zinc sulfate for diarrhea, cotrimoxazole (pill or syrup) for pneumonia. A stockout of any of these commodities lasting longer than 7 days resulted in the CS being considered as a CS with a severe stockout of any iCCM commodity.) 3. CS-ASC network with trained human resources (nurse and/or ASC) 4. CS-ASC network with trained human resources (nurse and/or ASC) and no stockout of any iCCM commodity (severe stockout=stockout of any iCCM commodity lasting longer than seven days; iCCM commodities = RDT and AL for malaria, low osmolarity ORS and zinc sulfate for diarrhea, cotrimoxazole (pill or syrup) for pneumonia. A stockout of any of these commodities lasting longer than 7 days resulted in the CS being considered as a CS with a severe stockout of any iCCM commodity.) 5. CS-ASC network with trained human resources (nurse and/or ASC) and no severe stockout of RDT or AL (severe stockout=stockout of any iCCM commodity lasting longer than seven days; iCCM commodities = RDT and AL for malaria) 6. CS-ASC network with trained human resources (nurse and/or ASC) and no severe stockout of ORS or zinc (severe stockout=stockout of any iCCM commodity lasting longer than seven days; ORS = low osmolarity oral rehydration solution) 7. CS-ASC network with trained human resources (nurse and/or ASC) and no severe stockout of cotrimoxazole (pill or syrup) (severe stockout=stockout of any iCCM commodity lasting longer than seven days; cotrimoxazole was the first-line antibiotic for pneumonia) 8. CS-ASC network with trained human resources (nurse and/or ASC) and no severe stockout of RUTF (severe stockout=stockout of any iCCM commodity lasting longer than seven days; RUTF=ready-to-eat therapeutic food) See Supplementary Appendix 1b at https://doi.org/10.5281/zenodo.4482969). See Table 1 for zonal statistics from these travel time rasters and detailed results for administrative layers 0-3 in Supplementary Appendix 2.
Geographic coverage
We defined geographic coverage as the theoretical catchment area of a health service delivery location, within a maximum travel time, accounting for the mode of transportation and the maximum population capacity of the type of health service delivery location. 10 We used the "geographic coverage" module of AccessMod 5 (v5.6.48) 10 to estimate geographic coverage for the CSI and CS-ASC networks in 2013 at 1km x 1km resolution for the two travel scenarios. The maximum travel time was set at 60 minutes. The maximum population capacity was set at 10000 for CSI and 2500 for CS-ASC based on the norms of the MOPH of Niger. 15 The maximum extent of a catchment was therefore delimited by the maximum travel time of 60 minutes except in cases where the estimated population in the catchment exceeded the maximum population capacity of the health service delivery locationin which case the extent of the catchment was smaller than the maximum travel time and was defined by the area containing the estimated population, up to the maximum population capacity.
Research questions
1. What percentage of the estimated population was covered by the CSI network in 2013? 2. What percentage of the estimated residual population beyond the geographic coverage of the existing CSI network was covered by the CS-ASC network in 2013? 3. What percentage of the estimated population was covered by the combination of the CSI and CS-ASC networks in 2013?
Methods for Geographic Coverage research question 1
We conducted a geographic coverage analysis of the estimated population covered by the existing CSI network in 2013, with each CSI catchment defined by a maximum travel time of 60 min (walking or walking + motorized vehicle) and maximum population capacity of 10000.
Analysis BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance i. Type of analysis: anisotropic ii.
Direction of travel: towards facilities iii.
Facilities processing order according to: The population living within a given travel time from the facilities 1. Travel time (minutes) for the processing order: 60 iv.
Processing order: Descending v.
Methods for Geographic Coverage research question 2
We conducted a geographic coverage analysis of the estimated residual population beyond the geographic coverage of the existing CSI network in 2013 that were covered by the existing CS-ASC network in 2013, with each CS-ASC catchment defined by a maximum travel time of 60 min (walking or walking + motorized vehicle) and maximum population capacity of 2500. Data analysis 1. We conducted a geographic coverage analysis for the existing CS-ASC network in 2013 considering maximum population apa ity using the variable " apa ity" set at population per O norms and pro essing order based the estimated residual population within a 60-minute catchment (walking) of each CS-ASC. This provided the final outputs for the analysis of geographic coverage for the existing CS-ASC network.
a. We used the following data inputs: Variable "amPopCoveredPercent_ResidPopBeyondCSI" in the tab " op_ -ASC" of upplementary ppendi 3 provides the cumulative geographic coverage of the estimated residual population covered by the CS-ASC network.
Methods for Geographic Coverage research question 3
The zonal statistics from Geographic Coverage research question 2 defacto provide the geographic coverage of the combined CSI + CS-ASC network.
Research questions
1. How many community health workers are needed (and where) to optimally cover the population beyond the 1hour catchment of the existing network of CS + ASC and CSI?
The MoPH in Niger has planned to scale-up RC in communities beyond 5km of CS or CSI to provide a standard package of preventive, promotive and curative services, including iCCM. We conducted a geographic coverage analysis to determine how many RC would be needed (and where) to optimally cover the estimated residual population beyond the geographic coverage of the existing CSI and CS-ASC networks in 2013, within a maximum travel time of 60 min walking from/to the RC and maximum population capacity of 1000 for each RC. This analysis BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance aimed to provide information (or at least a methodology) that could be used to inform a rational scale-up of the RC that would maximize geographic coverage of the residual population beyond the geographic coverage of the CS-ASC and CSI networks in 2013.
Methods for Scaleup research question 1
Data preparation 1. Identification of potential RC sites for scaleup: a. Given the norm for the RC-to-population ratio is 1 per 1000, we used a 500 people as a minimum cutoff to identify cells for potential RC sites because it would be inefficient and impractical to place RC in all communities beyond the geographic coverage of the existing CS-ASC and CSI networks, regardless of population si e. e used the " aster al ulator" tool in QG . . to prepare a GeoTiff raster that identified cells from the residual population raster of the geographic coverage analysis of the existing CS-ASC network in 2013 [raster_population_residual_r_NER_gcCS_ASC_60min_1km_prioritize_rFB13TT] with greater than or equal to 500 people. Note that the cells identified here were also beyond the geographic coverage of the CSI network, since the geographic coverage analysis for the existing CS-ASC network used the residual population from the geographic coverage analysis of the existing CSI network as the input population dataset. This resulted in 3521 cells identified as potential RC sites for scaleup. b.
e used the " olygoni e" tool in QG . . Because we did not know the actual order or scale-up of the existing CS-ASC network and because we wanted to ensure a conservative estimate of the efficiency of geographical targeting, for the comparison of geographic coverage of the population (comparison a above) we assumed the prioritization order for the existing CS-ASC network based on the estimated residual population within a 60-minute catchment (walking) of an existing CS-ASC (as with the hypothetical network in a above). For comparison of geographic coverage of the estimated residual U5 deaths (comparison b above) we assumed the prioritization order for the existing CS-ASC network based on the estimated residual U5 deaths within a 60-minute catchment (walking) of an existing CS-ASC (as with the hypothetical network in b above). For comparison of geographic coverage of the estimated residual Pf malaria cases (comparison c above) we assumed the prioritization order for the existing CS-ASC network based on the estimated residual Pf malaria cases within a 60-minute catchment (walking) of an existing CS-ASC (as with the hypothetical network in c above). This is likely to overestimate the slope (efficiency) for the existing network and result in a conservative (underestimated) estimate of the efficiency gains of the hypothetical network over the existing network. This conservative approach to estimating the efficiency gains of geographical targeting of the hypothetical network over the existing network is justified given the absence of knowledge of the true criteria and/or factors that determined the scale-up order the existing network. indicating the modelled catchment area of each health service delivery point.
Research questions
3. Comparison of geographic coverage of the existing CS-ASC network and the hypothetical CS-ASC network: ee tabs "rPop13_ isting", "rPop13_ ypotheti al" and " omparison_rPop13" in Appendix 6. We compared the percentage of the estimated residual population beyond the geographic coverage of the existing CSI network in 2013 that was covered by the existing network of CS-ASC (from Geographic Coverage research Question 2) with the percentage of the estimated residual population beyond the geographic coverage of the existing CSI network in 2013 that was covered by the hypothetical network of CS-ASC that prioritized the estimated residual population in the processing order (from Targeting research question 1, Data Analysis step 1 above) given the same number of potential CS-ASC sites as in the existing network of CS-ASC (i.e. 2550) as well as for the total number of potential CS-ASC sites (i.e. 5796).
a. n tab "rPop13_ isting" of Supplementary Appendix 5, we sorted the results using variable "amPopCatchmentTotal" from highest to lowest, assuming maximal efficiency of the existing network. aintaining this order, the last value for the variable "amPopCoveredPercent_ResidPopBeyondCSI" provided the percentage of the estimated population beyond the 1hr catchment of the existing CSI network in 2013 that was covered by the existing network of CS-ASC. b. n tab "rPop13_ ypotheti al" of Supplementary Appendix 5, we sorted the results using variable "amPopCatchmentTotal" from highest to lowest, assuming maximal efficiency of the hypothetical network (as was done above for the existing network). Maintaining this order, the value for the variable "amPopCoveredPercent_ResidPopBeyondCSI" for the 2550 th potential CS-ASC (distributed in 1523 1km locations) provided the percentage of the estimated population beyond the 1hr catchment of the existing CSI network in 2013 that was covered by the hypothetical network of CS-ASC that prioritized the residual population in the processing order, using the same number of sites as the existing CS-ASC network. The value for the variable "amPopCoveredPercent_ResidPopBeyondCSI" for the 5796 th potential CS-ASC provided the percentage of the estimated population beyond the 1hr catchment of the existing CSI network in 2013 that was covered by the full hypothetical network of the 5796 CS-ASC, prioritizing the residual population in the processing order. c. n tab " omparison_Population" of Supplementary Appendix 5, we compared the results from 3a for the network of existing CS-ASC to the results from 3b for the first 2550 hypothetical CS-ASC in terms of the absolute and relative difference in geographic coverage of the estimated residual population beyond the 60-minute catchment of the CSI network and estimated residual population covered beyond the 60-minute catchment of the CSI network.
Methods for Targeting research question 2
Data preparation ee tabs " d_ isting", " d_ ypotheti al" and " omparison_ deaths" in Supplementary Appendix 5. We compared the percentage of the estimated residual U5 deaths beyond the geographic coverage of the existing CSI network in 2013 that was covered by the existing network of CS-ASC, prioritizing the estimated residual population in the processing order, with the percentage of the estimated residual U5 deaths beyond the geographic coverage of the existing CSI network in 2013 that was covered by the hypothetical network of CS-ASC, prioritizing the estimated residual U5 deaths in the processing order given the same number of potential CS-ASC sites as in the existing network of CS-ASC (i.e. 2550). There is no MOPH norm for the ratio of ASC per U5 deaths and thereby no maximum capacity limit of the ASC for U5 deaths. Rather than make the unrealistic assumption that one CS-ASC could cover all U5 deaths within their catchment regardless of population size, we calculated the number of CS-ASC required in both the existing CS-ASC network and hypothetical CS-ASC network to completely cover (saturate) the estimated residual population in each catchment based on the MOPH ratio of one CS-ASC per 2500 population. For the existing CS-ASC network, this resulted in 2550 CS-ASC in 1924 CS-ASC catchments see variable "rPop13Cum_CS-ASC_saturate" in the tab "r d _ isting" in upplementary ppendi . For the hypothetical CS-ASC network, this resulted in 2550 CS-ASC in 2044 CS-ASC catchments see variable "rPop13Cum_CS-ASC_saturate" in the tab "r d _ isting" in upplementary ppendi . a. n tab "r d _ isting" of upplementary ppendi , e sorted the results using variable "am op at hmentTotal" from highest to lo est, assuming ma imal effi ien y of the e isting net ork. aintaining this order, the value for the variable "am op overed er ent_ esid opBeyond " for the 2550 th CS-ASC see ell see ell V 9 4 in tab "r d _ isting distributed in 1924 catchments provided the percentage of estimated U5 deaths beyond the 1hr catchment of the existing CSI network in 2013 that was covered by the existing network of 2550 CS-ASC. b. n tab "r op _ ypotheti al" of upplementary ppendi , we sorted the results using variable "amPopCatchmentTotal" from highest to lowest, assuming maximal efficiency of the hypothetical network (as was done above for the existing network). Maintaining this order, the value for the variable "amPopCoveredPercent_ResidPopBeyondCSI" for the 2550 th potential CS-ASC (see cell V 44 in tab "r d _ ypotheti al" distributed in 2044 catchments provided the percentage of the estimated U5 deaths beyond the 1hr catchment of the existing CSI network in 2013 that was covered by the hypothetical network of CS-ASC that prioritized the residual U5 deaths in the processing order, using the same number of CS-ASC as the existing CS-ASC network. c. n tab " omparison_U5deaths" of Supplementary Appendix 5, we compared the results from 3a for the network of existing CS-ASC to the results from 3b for the first 2550 hypothetical CS-ASC in terms of the absolute and relative difference in geographic coverage of the estimated residual U5 deaths beyond the 60-minute catchment of the CSI network and estimated number of U5 deaths covered beyond the 60-minute catchment of the CSI network.
Uncertainty analysis
We assessed the potential effect of uncertainty of the estimates for under-five deaths on targeting as follows. We used the "Zonal statisti s" tool in QGIS 3.12.0-Bu areşti 13 to extract the estimated mean and 95% confidence intervals for the number of under-five deaths for each catchment area defined by the geographic coverage analysis for the hypothetical network from step 2 of targeting research question 2. We sorted the catchments by the estimated mean number of under-five deaths from largest to smallest, as this reflected the prioritization order of the geographic coverage analysis used for the targeting analysis (step 2 of targeting research question 2). Because policy makers and planners typically support scale-up of facilities and CHWs in groups we identified five potential groups of CS-ASC for consideration. Group 1 included the 500 CS-ASC with the highest estimated mean number of under-five deaths, (median of means across catchments = 128, median of lower 95% confidence interval = 108, and median of upper 95% confidence interval = 149). Group 2 included 500 CS-ASC with the next highest estimated mean number of under-five deaths (median of means across catchments = 43, median of lower 95% confidence interval = 36, and upper 95% confidence interval = 51). Group 3 included 500 CS-ASC with next highest estimated mean number of under-five deaths (median of means across catchments = 24, median of lower 95% confidence interval = 20, and median of upper 95% confidence interval = 28). Group 4 included 500 catchments with the next highest mean number of under-five deaths (median of means across catchments = 16, median of lower 95% confidence interval minimum = 13, and median of upper 95% confidence interval = 19). Group 5 included 550 catchment CS-ASC with the next highest estimated mean number of under-five deaths (median of means across catchments = 11, median of lower 95% confidence interval = 9, median of upper 95% confidence interval = 13.0). Based on the medians of the 95% confidence intervals, decision makers could confidently prioritize Group 1 over Groups 2-5; Group 2 over Groups 3-5; Group 3 over Groups 4-5; and Group 4 over Group 5 (see Supplementary Appendix 5 -Targeting un ertainty, tabs " ummary_un ertainty_r d " and "Groups_un ertainty_r d ").
Methods for Targeting research question 3
Data preparation 1. Geographic coverage analysis of the estimated residual Pf malaria cases among all ages (0-99 years) by the existing network of CS-ASC: We conducted a geographic coverage analysis for the estimated residual Pf malaria cases (all ages) beyond the geographic coverage of the existing CSI network, using the existing network of CS-ASC sites in 2013, with the processing order based on the capacity of the CS-ASC sites, and setting the maximum population capacity variable "Capa ityN" at 100000 to effectively not consider maximum population capacity as a constraint to the CS-ASC catchment areas. The analysis removed the estimated Pf malaria cases (all ages) within each catchment area at each iteration (calculation of each catchment area) to avoid double counting estimated Pf malaria cases (all ages) where the 60 min catchment areas overlap. This provided the final outputs for the analysis of geographic coverage of the estimated residual under-five deaths by the existing CS-ASC network. a. We used the following data inputs: i. Population: r_NER_rCases13_final_1km ii.
Select existing health facilities layer (vector): v_NER_cells_at_100m_with_CS_or_ASC_adj_barriers_1km v.
ID field: id vi.
Facility name field: cat vii.
Select zones layer (vector): adm3 1. Select zones unique ID (integer): objectid 2. Select zone name (text): nom_com b. We used the following analysis settings: i. Type of analysis: anisotropic ii.
Direction of travel: towards facilities iii.
Fa ilities pro essing order a ording to: field in the fa ility layer "capacity" iv.
Processing order: Descending v.
Maximum travel time (minutes): 60 vi. Add short tag: r_NER_gc_Existing_CS_ASC_rCases13_60min_1km_prioritize_rCases13Aire 2. Geographic coverage analysis of the estimated residual Pf malaria cases among all ages (0-99 years) by the hypothetical network of CS-ASC, prioritizing estimated residual Pf cases in the processing order: We conducted a geographic coverage analysis for the estimated residual Pf malaria cases among all ages (0-99 years) beyond the geographic coverage of the existing CSI network, using the existing network of CS-ASC sties in 2013, with the processing order based on the estimated residual Pf malaria cases in 2013 within 60-minute of each CS-ASC site and setting the maximum population capacity at 100000 to effectively not consider maximum population capacity as a constraint to the CS-ASC catchment areas. The analysis removed the estimated Pf malaria cases within each catchment area at each iteration (calculation of each catchment area) to avoid double counting estimated Pf malaria cases where the 60 min catchment areas overlap. This provided the final outputs for the analysis of geographic coverage of the estimated residual under-five deaths by the existing CS-ASC network. a. We used the following data inputs: i. Population: r_NER_rCases13_final_1km ii.
Scenario provided the percentage of estimated Pf malaria cases (all ages) beyond the 1hr catchment of the existing CSI network in 2013 that was covered by the existing network of 2550 CS-ASC. b. n tab "r ases _ ypotheti al" of upplementary ppendi , e sorted the results using variable "am op at hmentTotal" from highest to lo est, assuming ma imal effi ien y of the hypotheti al network (as was done above for the existing network). Maintaining this order, the value for the variable "am op overed er ent_ esid opBeyond " for the th potential CS-ASC (see cell V in tab "r ases _ ypotheti al" distributed in at hments provided the per entage of the estimated Pf malaria cases beyond the 1hr catchment of the existing CSI network in 2013 that was covered by the hypothetical network of CS-ASC that prioritized the residual Pf malaria cases in the processing order, using the same number of CS-ASC as the existing CS-ASC network. c. n tab " omparison_ alaria" of upplementary ppendi , e ompared the results from a for the network of existing CS-ASC to the results from 3b for the first 2550 hypothetical CS-ASC in terms of the absolute and relative difference in geographic coverage of the estimated residual Pf malaria cases (all ages) beyond the 60-minute catchment of the CSI network and estimated number of Pf malaria cases (all ages) covered beyond the 60-minute catchment of the CSI network.
Uncertainty analysis
We assessed the potential effect of uncertainty of the estimates for under-five deaths on targeting as follows. We used the "Zonal statisti s" tool in QGIS 3.12.0-Bu areşti 13 to extract the estimated mean and 95% confidence intervals for the number of Pf malaria cases for all ages (0-99 years) for each catchment area defined by the geographic coverage analysis for the hypothetical network from step 2 of targeting research question 3. We sorted the catchments by the estimated mean number of Pf malaria cases for all ages (0-99 years) from largest to smallest, as this reflected the prioritization order of the geographic coverage analysis used for the targeting analysis (step 2 of targeting research question 3). Because policy makers and planners typically support scale-up of facilities and CHWs in groups we identified five potential groups of CS-ASC for consideration. Group 1 included 500 catchments with the highest estimated mean number of Pf malaria cases, (median of means across catchments = 12865, median of lower 95% confidence interval = 4568, and median of upper 95% confidence interval = 21303). Group 2 included 500 catchments with the next highest estimated mean number of Pf malaria cases (median of means across catchments = 3668, median lower 95% confidence interval = 1637, and median upper 95% confidence interval = 5414). Group 3 included 500 catchments with next highest estimated mean number of Pf malaria cases (median of means across catchments = 2417, median of lower 95% confidence interval = 1027, and median of upper 95% confidence interval = 3647). Group 4 included 500 catchments with the next highest mean number of Pf malaria cases (median of means across catchments = 1842, median of lower 95% confidence interval minimum = 824, and median of upper 95% confidence interval = 2738). Group 5 included 500 catchments with the next highest estimated mean number of Pf malaria cases (median of means across catchments = 1462, median of lower 95% confidence interval = 619, and median of upper 95% confidence interval = 2240). Based on the medians of the 95% confidence intervals, decision makers could confidently prioritize Group 1 over Groups 2-5; Group 2 over Groups 3-5; Group 3 over Groups 4-5; and Group 4 over Group 5 (see Supplementary Appendix 6 -Targeting uncertainty, tabs " ummary_un ertainty_r ases " and "Blo s_un ertainty_r ases "). | 2021-06-09T06:18:29.602Z | 2021-01-08T00:00:00.000 | {
"year": 2021,
"sha1": "2dc30423160b06db22fb8693a5e4b1b56937b448",
"oa_license": "CCBYNC",
"oa_url": "https://gh.bmj.com/content/bmjgh/6/6/e005238.full.pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "9baa9ebc386fe73e923e379a310de5fa689b9fc2",
"s2fieldsofstudy": [
"Geography",
"Medicine",
"Environmental Science"
],
"extfieldsofstudy": [
"Geography",
"Medicine"
]
} |
239081126 | pes2o/s2orc | v3-fos-license | Associations of COVID-19 Risk Perception, eHealth Literacy, and Protective Behaviors Among Chinese College Students Following Vaccination: A Cross-Sectional Study
Background In spite of strict regulation of coronavirus disease 2019 (COVID-19) preventive measures and containment in China, there are still confirmed cases sporadically occurring in many cities. College students live in groups and have active social activities so that it will trigger a serious public health event once an infection event occurs. Thus, identifying the status and related factors of protective behaviors among them after receiving vaccination will be crucial for epidemic control. This study aimed to gather information on the protective behaviors and to identify the associations of COVID-19 risk perception, eHealth literacy, and protective behaviors for Chinese college students following vaccination. Methods A cross-sectional survey of college students engaged in protective behaviors post vaccination was conducted using the COVID-19 risk perception scale, eHealth literacy scale, and protective behaviors following vaccination questionnaire in one of the groups. Multiple linear regression analysis was used to confirm the correlation among the COVID-19 risk perception, eHealth literacy, and protective behaviors for Chinese college students. Results A total of 5,641 Chinese college students were included. Male students comprised 59.01% with an average age of (21.39 ± 2.75) years and most students rating their health as very good (44.85%) or pretty good (46.98%). A smaller percentage (13.76%) believed that they would likely or most likely be infected with COVID-19 after getting vaccinated. In addition, more than 1 in 10 (10.35%) college students had ever suspected to suffer from post-vaccination reactions following the COVID-19 vaccination. The mean score of protective behaviors was 26.06 ± 3.97. Approximately one-third (30.42%) of the students always or often did not wear a mask when going out. Some college students (29.25%) did not maintain distance of at least 1 m from others in social situations. Older female college students who were in good health and perceived as being at a low risk of getting infected with COVID-19, and those never suspected to suffer from post-vaccination reactions expected to engage in post-vaccination protective measures. Those with a higher level of perceived risk, severe risk perception and eHealth literacy, and a lower level of unknown risk perception were more likely to engage in further protective behaviors after getting vaccinated. Conclusions Overall, the level of protective behaviors among the Chinese college students following vaccination could be improved, especially for male, younger college students in poor health. This study revealed the predictive effects of risk perception and eHealth literacy on protective behaviors, recommending that the negative and positive effects of risk perception should be balanced in epidemic risk management, and eHealth literacy promotion should also be emphasized for public health and social measures.
Background: In spite of strict regulation of coronavirus disease 2019 (COVID-19) preventive measures and containment in China, there are still confirmed cases sporadically occurring in many cities. College students live in groups and have active social activities so that it will trigger a serious public health event once an infection event occurs. Thus, identifying the status and related factors of protective behaviors among them after receiving vaccination will be crucial for epidemic control. This study aimed to gather information on the protective behaviors and to identify the associations of COVID-19 risk perception, eHealth literacy, and protective behaviors for Chinese college students following vaccination.
Methods: A cross-sectional survey of college students engaged in protective behaviors post vaccination was conducted using the COVID-19 risk perception scale, eHealth literacy scale, and protective behaviors following vaccination questionnaire in one of the groups. Multiple linear regression analysis was used to confirm the correlation among the COVID-19 risk perception, eHealth literacy, and protective behaviors for Chinese college students.
Results: A total of 5,641 Chinese college students were included. Male students comprised 59.01% with an average age of (21.39 ± 2.75) years and most students rating their health as very good (44.85%) or pretty good (46.98%). A smaller percentage (13.76%) believed that they would likely or most likely be infected with COVID-19 after getting vaccinated. In addition, more than 1 in 10 (10.35%) college students had ever suspected to suffer from post-vaccination reactions following the COVID-19 vaccination. The mean score of protective behaviors was 26.06 ± 3.97. Approximately one-third (30.42%) of the students always or often did not wear a mask when going out. Some college students (29.25%) did not maintain distance of at least 1 m from others in social situations. Older female college students who were in good health and perceived as being at a low risk of getting infected with COVID-19, and those never suspected to suffer from post-vaccination reactions expected to engage in post-vaccination protective measures. Those with a higher level of perceived risk, severe risk perception and eHealth literacy, and a lower level of unknown risk perception were more likely to engage in further protective behaviors after getting vaccinated.
Conclusions:
Overall, the level of protective behaviors among the Chinese college students following vaccination could be improved, especially for male, younger college students in poor health. This study revealed the predictive effects of risk perception and eHealth literacy on protective behaviors, recommending that the negative and positive effects of risk perception should be balanced in epidemic risk management, and eHealth literacy promotion should also be emphasized for public health and social measures.
INTRODUCTION
Coronavirus disease 2019 (COVID-19) is renowned for causing an infectious pneumonia which broke out at the end of year 2019 (1). Subsequently, the COVID-19 epidemic upgraded into a pandemic in January, 2020, and it was declared a public health emergency of international concern by the WHO. In March, 2020, the WHO had declared COVID-19 to be a pandemic. COVID-19 is characterized as a highly infectious and strongly pathogenic condition which posed serious threats to global health. As of June 25, 2021, there have been 179.69 million cases of COVID-19 diagnosed globally, of which about 3.90 million people have died (2). College students, a group of welleducated young people with a high Internet penetration rate, are characterized by active social contacts and intense crossregional mobility. Additionally, college students live mainly in school groups, and have frequent contact with each other, which can easily lead to public health emergencies once a COVID-19 case is found. Thus, college students are the key group we should pay more attention for COVID-19 prevention and control and are among those who should be vaccinated. During the outbreak of COVID-19, college students showed good protective behaviors. Over two-thirds of American college students reported washing their hands at least six times a day (3) and most Sherubtse college students (93.5%) had good practice toward COVID-19 (4). Almost all Chinese college students were highly in favor of epidemic containment strategies and showed high adherence to them. Specifically, well over three quarters of college students performed well in frequent hand washing or hand hygiene (86.9%), in wearing face masks (92.8%), and in avoiding going out in public and hosting gatherings (91.2%) (5). The female students and those enrolled in post-graduate studies were inclined to take preventive measures (5); however, it is uncertain that how they will behave and respond to regular COVID-19 prevention and control at this phase after receiving COVID-19 vaccination, and the Abbreviations: WHO, World Health Organization; CSM-SR, the Common-Sense Model of Self-Regulation; IPC, infection prevention and control; PHERPS, the public health emergency risk perception scale; eHEALS, the eHealth Literacy Scale; VIF, Variance inflation factor. differences among individuals with varying sociodemographic characteristics require further analysis.
Risk perception is defined as the individual feelings and recognition to exterior objective risk which drive vital decisions on the behaviors (6). During the period of COVID-19, the risk perception for citizens was highly associated with their individual preventive measures against COVID-19. Savadori et al. found that the risk perception exerted positive effects on the preventive measures upon infection prevention for citizens (7). Specifically, those who felt anxiety toward COVID-19 were more inclined to engage in washing hands, wearing face masks, and maintaining social distance from others. Additionally, those who perceived themselves to be highly susceptible were inclined to reduce social contact (7). Alegria et al. identified positive correlation between frequency of washing hands with self-perceived effectiveness of washing hands and epidemic risk perception based on the Common-Sense Model of Self-Regulation (CSM-SR) (8). Chinese college students were associated with a relatively high risk perception toward COVID-19 in the study of Ding et al., especially in female students and non-medical students as 92.5% of them believed that those in good health condition may be infected with COVID-19, and 85.1% of the college students were concerned about members of their family becoming infected (9). Those with a higher level of risk perception were better informed with regards to COVID-19, as well as carrying out more preventive actions (9). The pandemic has been contained effectively as the research work continues on COVID-19, disease diagnosis and treatment go into standardization along with the regular prevention and control measures being placed into effect (10). At present, China has entered the phase of regular prevention and control, with the government offering free vaccines for COVID-19 for all of its citizens (11). The national vaccination plan has observed great progress due to the diligence of the government and its personnel combined with the active cooperation of Chinese citizens; however, sporadic cases have emerged in several regions in recent days (12). Under these circumstances, we sought to identify the risk perception of COVID-19 among the Chinese college students after receiving vaccination. In addition, we sought to confirm the association between the risk perception and preventive actions on infection prevention post vaccination.
During the time of COVID-19, home quarantine and limiting in person social contact were recommended and enforced globally. Public outings and social contact were confined on a large scale; thus, the internet and social media have become the main channels for citizens to grasp dynamic information and maintain social connections with one another. However, misinformation about COVID-19 shows up frequently on the internet. eHealth literacy plays a critical role in rapidly targeting high-quality information from the confusing online environment that is flooded with mixed messages, as well as in making correct decisions for practices. eHealth literacy is a fundamental skill from which individuals can benefit from eHealth services.
Norman et al. defined eHealth literacy as the ability to search, find, understand, and produce critical analyses of the targeted health information from online health resources, followed by making the correct decision to address health problems (13); it makes up an integral driving factor behind individual health behaviors (14,15). Do et al. discovered that during the time of COVID-19, higher eHealth literacy was closely associated with the increased adherence to the actions on infection prevention and control (IPC) and maintenance of a better lifestyle (16). Nevertheless, frequent exposure to social media has its downside, which may lead to the reluctance of an individual to obtain or even refusing to obtain the COVID-19 vaccine. This may be the result of online spreading of false information about the COVID-19 vaccines (17). Still, as individuals with higher eHealth literacy perform better in critically evaluating online information, their decision-making is less likely to be influenced by misinformation. Li et al. have confirmed a positive moderating effect of eHealth literacy in social media use and preventative actions. Therefore, higher eHealth literacy is strongly correlated to an increased adherence to infection preventive actions (18). Li et al. found that Chinese college students showed a relatively high eHealth literacy during the time of COVID-19, and the higher eHealth literacy predicted better preventive behavior at a significant level (19). However, the eHealth literacy for Chinese college students under the phase of regular prevention and control remains to be seen. In addition, how eHealth literacy is associated with the preventive actions of individuals after receiving COVID-19 vaccines requires further investigation and analysis.
Based on the research questions, the present study sought to: gain knowledge of the preventive actions on infection defense, risk perception of COVID-19, and the eHealth literacy among Chinese college students, as well as discussing links between preventive actions with risk perception and eHealth literacy. This study presented not only as a basis for implementing measures of COVID-19 prevention and control among the Chinese college students, but also, as a reference for public health management on the COVID-19 prevention and control among Chinese college students from the perspective of risk perception and eHealth literacy.
Research Design and Recruitment of Participants
This cross-sectional study with onsite survey methods was conducted at a comprehensive university in Changsha city in the province of Hunan in China. The university, as one of key universities and colleges in China, offers 106-degree programs at the bachelor's level and enrolls more than 50,000 full-time students. The questionnaire QR code, downloaded from the online platform (Questionnaire Star, URL: https:// www.wjx.cn/), was distributed as a paper questionnaire. College students were invited to scan the QR code on the campus from June 10 to 15, 2021. An informed consent form was initially signed online by all participants. Before participants filled in the questionnaire, the introduction, i.e., research background, purpose, rules of anonymity and confidentiality, and precautions would be explained to them. The inclusion criteria for participants consisted of: (a) college students over 18 years of age; (b) Chinese students from the target university; (c) completion of the COVID-19 vaccination; and (d) consent to participate in the survey. A total of 6,312 college students responded with 6,282 responses having met the inclusion criteria (1 person was under 18 years and 29 people who did not consent to this survey). We declared that the minimum time for completing the survey was 90 s, and 641 invalid responses were excluded for the time limit. In total, 5,641 valid responses were included in the data analysis (effective response rate was 89.37%).
Demographic Information
Sociodemographic information included gender, age, major, education level, health condition, self-perception of susceptibility, and suspicion of suffering from post-vaccination reactions. Health condition status was based on self-report with response alternatives "Very good, " "Pretty good, " "in General level, " "Pretty poor, " and "Very poor." Suspicion to suffer from post-vaccination reactions was measured using the question: "Have you ever suspected to suffer from post-vaccination reactions following the COVID-19 vaccination?" and responses were dichotomized as follows: "yes" and "no."
COVID-19 Risk Perception
The COVID-19 risk perception was evaluated by the public health emergency risk perception scale (the PHERPS) compiled by Shen et al. in 2020 (20). The public health emergency in this survey was defined as COVID-19. The PHERPS included 9 items within 3 domains of dread risk perception (3 items), severe risk perception (3 items), and unknown risk perception (3 items). Each item was scored on a 5-point Likert scale from strongly disagree to strongly agree. The total score was the sum of 9 items with a range of 9-45. The higher the score, the higher the level of risk perception (Cronbach's α = 0.84).
eHealth Literacy
The eHealth Literacy scale (eHEALS) developed by Norman et al. in 2006 (21) was adopted by eHealth literacy. The Chinese version of eHEALS was developed from the English version and translated into Chinese by two graduate students who held an International English Language Testing System (IELTS) certificate. Afterward, the back-translation was performed by a professor who had previously studied abroad to ensure the accuracy of the translation. The eHEALS included 8 items with one domain. Each item was scored on a 5-point Likert scale from strongly disagree to strongly agree with a total score ranged from 8 to 40. The higher the score, the higher the level of eHealth literacy is (Cronbach's α = 0.96).
Protective Behaviors Following Vaccination
The questionnaire of protective behaviors was developed by the research team based on the COVID-19 advice for the public after getting vaccinated (22). It included 7 items within seven aspects, i.e., social distancing, mask-wearing, handwashing, sneeze protection, going-out limit, ventilating, and traveling limit. The score adopted a Likert 5 rating from never (1 point) to always (5 points). The item 2 was scored in reverse and the others were positive scores. The total score was the sum of the 7 items with a range of 7-35 (Cronbach's α = 0.73). The above instruments appeared in Multimedia Appendix 1.
Data Analysis
For data analysis, IBM SPSS 25.0 (IBM Corporation, NY, USA) was used. For descriptive statistics, categorical variables were presented as N (%), such as gender, age groups, major, education level, health condition, self-perception of susceptibility, and suspicion of suffering from post-vaccination reactions. Continuous variables were presented as mean ± SD, such as age, the level of COVID-19 risk perception, e-health literacy, and protective behaviors after COVID-19 vaccination. A univariate analysis was performed using Student's t-test or one-way ANOVA. A multiple linear regression was employed to test the determinant factors affecting the protective behaviors. The multiple linear regression analysis with stepwise method (α in = 0.05, α out = 0.10) was conducted with the score of protective behaviors after vaccination as a dependent variable and the variables with statistical significance in univariate analysis, three domains of risk perception, and eHealth literacy as independent variables. Variable assignments were as follows. They were gender (1 = male; 2 = female), age (1 = 18 ∼ 20 years; 2 = 21 ∼ 23 years; 3 = ≥24 years), education (1 = undergraduate; 2 = post-graduate; 3 = PHD), health condition (1 = very good; 2 = pretty good; 3 = in General level, pretty poor, very poor), self-perception of susceptibility (1 = Impossible; 2 = Not likely; 3 = Likely, most likely), and suspicion to suffer from post-vaccination reactions (1 = yes; 2 = no). VIF ranged from 1.02 to 1.80, indicating that there was no multi-collinearity among selected independent variables. Statistical testing was bilateral with the statistical significance at p < 0.05.
Socio-Demographic Characteristics
The results showed that among 5,641 investigated college students, male students accounted for 59.01% with the average age being (21.39 ± 2.75) years. The majority of this population were non-medical students (95.87%) and undergraduate students (73.11%), with 91.83% of them in very good or pretty good health condition. A small proportion (13.76%) thought they would likely or most likely be infected with COVID-19 after getting vaccinated. Besides, more than 1 in 10 (10.35%) college students had ever suspected to suffer from post-vaccination reactions after the COVID-19 vaccination ( Table 1).
Characteristics of COVID-19 Risk Perception, eHealth Literacy, and Protective Behaviors
The mean score for COVID- 19 Additionally, the mean score of protective behaviors for Chinese college students was 26.06 ± 3.97. The majority (60.96%) of college students always or often avoided going out into a crowd. Over two-thirds (70.74%) always or often canceled non-work or work-related trips. Some college students performed insufficient protective behaviors. Nearly one-third (30.42%) of the students always or often failed to wear a mask while going out. A fair portion of college students (29.25%) failed to maintain at least 1 m of distance from others in social settings ( Table 2).
The Factors Which Influence COVID-19 Protective Behaviors
As shown in Table 1, the univariate analysis indicated that the female college students, those above 24 years of age, those with a higher education level, those in superior health, those with a lower self-perception of susceptibility, and those never suspected to suffer from post-vaccination reactions were more likely to perform better in the personal protection (p < 0.05). Results ( Table 3) showed that nine determinant factors were reserved, i.e., gender, age, health condition, self-perception of susceptibility, suspicion of suffering from post-vaccination reactions, dread risk perception, severe risk perception, unknown risk perception, and eHealth Literacy, accounting for 14.2% of variation in predicting the level of protective behaviors.
DISCUSSION
The results showed that the protective behaviors after COVID-19 vaccination for Chinese college students were positive in general and the majority of college students maintained good protective habits. Dread risk perception, severe risk perception, and eHealth literacy would positively predict the protective behaviors, but unknown risk perception had a negative predictive effect. Female college students, those who were older, in good health condition, with a lower self-perception of susceptibility, and those never suspected to suffer from post-vaccination reactions were inclined to have a higher level of protective behaviors. Under the circumstance of sporadic diagnoses which occurred in many cities, the results indicated a sufficient level of protection awareness among the Chinese students post vaccination. The COVID-19 risk perception was strong, which was closely correlated with the enforcement of strict measures to control the epidemic. Overall, our findings emphasized the importance of risk perception and eHealth literacy, and prepare policymakers and health managers to develop the necessary prevention policies and target education measures.
Demographics and Protective Behaviors Following Vaccination
The results of this study suggested that female college students, those who were older, in good health condition, with less selfperception of susceptibility, those never suspected to suffer from post-vaccination reactions were more likely to perform better in the COVID-19 protection. This was confirmed via several former studies. Ferdous et al. (23) and Li et al. (24) found that the female, elder residents in good health condition will take precautions more frequently, which may be attributed to a higher level of cognition, a more precautious attitude toward COVID-19 as well as better compliance of the IPC guidance for those residents (25,26). A previous study (23) in Bangladesh showed that those residents with a high level of education would perform better in the protective behaviors, and education showed a statistically significant correlation with the level of protective behaviors in this study during the univariate analysis. This was similar to the results of Olaimat et al., i.e., undergraduates performed worse in protective behaviors compared with graduate students in Jordan (27). However, it was insignificant in the multivariate linear regression. This may be related to the characteristics of education among college students, i.e., with more advanced age, higher levels of education were more prevalent and the education factor was adjusted in the regression analysis. A small proportion of those who were likely or most likely to infect with COVID-19 after being vaccinated or suspected to suffer from post-vaccination reactions performed significantly worse in protective behaviors, which may be due to their negative attitudes toward COVID-19 as well as lack of trust in the efficacy of vaccine for these college students (28). Wang et al. pointed out that lower self-perception of susceptibility and good protective behaviors were two important protective factors for good mental health during the COVID-19 outbreak (29). Mo et al. found that much greater concerns and self-perceived susceptibility toward COVID-19 were the risk factors for anxiety and depression (30). This suggested that more attention should be given in assessing the psychological state to prevent the rising of psychological health issues.
COVID-19 Risk Perception and Protective Behaviors Following Vaccination
The results of this study showed relatively higher levels of dread risk perception and severe risk perception than that of unknown risk perception. Unknown risk perception was primarily associated with the level of cognition with regards to COVID-19 and the accuracy of detection and diagnosis (20), for which the results were determined by the development of COVID-19. During the pandemic, as a result of the low level of cognition and the lack of effective detection and diagnosis methods, the level of unknown risk perception for the college students was quite high. With the establishment of the global COVID-19 IPC guidance, the epidemic was gradually brought under control, the vaccination strategy was carried on methodically, and the level of unknown risk perception for the college students was improved side-by-side. In addition, the results suggested that different domains of risk perception (16). The results of this study showed that eHealth literacy positively predicted the protective behaviors of college students post vaccination. This corresponded to previous findings (39) and emphasized the importance of promoting eHealth literacy and the necessity of the related educational programs in epidemic prevention and control. However, the positive effects seemed to be weaker in this study (β = 0.125) than those in the COVID-19 global pandemic (16,40). There may be two reasons for this. The first reason may be that relatively stable protective habits were developed after the epidemic and the other was likely the more significant effect of protective attitude and self-efficacy on health behaviors when the epidemic was under control in China. Previous studies indicated that multiple factors affected the eHealth literacy. Shi et al. divided the influencing factors into individual-level factors (age, gender, education, economy, frequency of Internet use, and trust in online health resources), interpersonal-level factors (marital status, family caregivers, and the experience of studying looking for health information), and social/community level factors (language and cultural barriers) based on socioecological model with the systematic review methods (41). Levin-Zamir et al. indicated that eHealth literacy was affected by the complexity of the network system (42). When the accessibility and usability of electronic resources were good, that is, the complexity of the network system decreased, the public eHealth literacy would be greatly positively affected. Thus, the comprehensive factors should be considered to improve the level of eHealth literacy from multiple perspectives. Besides, although the eHealth literacy reported by college students was positive in this study, the ability of actual information application may not be optimistic. Kim et al. found that people with a greater ability to seek out information and make judgments had a lower accuracy rate when answering actual questions in Korea (43). This indicated a gap between the subjective self-reporting and objective application results. Neter et al. confirmed a weak correlation (r = 0.34) between subjective and objective eHealth literacy among Israeli adults, and proposed that different evaluation tools should be used for evaluation independently (44).
Study Limitations
This study may have the following limitations: first, data were drawn from one comprehensive university of China and thus the generalizability of this study was limited. Therefore, further multi-center or nationwide investigations were recommended to generalize the findings. Second, the data quality with the online questionnaire may have declined, as the insufficient investigators limited possible oversight throughout when the participants completed the questionnaires. The completion time was used to control the data quality. In addition, this study failed to include college students with lower than a bachelor's degree and broader education levels should have been included. Finally, self-reported eHEALS was used to assess the level of eHealth literacy in this study and it might differ from the objective levels to some extent. It is recommended to further explore the associations between subjective and objective eHealth literacy.
CONCLUSIONS
The results observed that the older female college students, those in good health condition, perceived to have little chance of being infected with COVID-19, and never suspected to suffer from the post-vaccination reaction were more likely to perform better in protective behaviors. This study confirmed the associations of COVID-19 risk perception, eHealth literacy, and protective behaviors. College students with higher eHealth literacy were more likely to engage in positive protective behaviors. Different levels of risk perception significantly predicted the protective behaviors. The dread risk perception and severe risk perception had positive effects in contrast with unknown risk perception which played a negative role. The results suggested that further attention should be given to male, young, and college students in poor health to conduct targeted educational measures following vaccination. Additionally, recommendations were provided by this study for COVID-19 risk management to minimize the negative effects and placing importance to eHealth literacy to support the effective infection control work post vaccination.
DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.
ETHICS STATEMENT
This study was approved by the Ethics Committee of the Third Xiangya Hospital (ID: I 21071). Online informed consent was obtained from the individuals for the publication of any potentially identifiable images or data included in this article.
AUTHOR CONTRIBUTIONS
NQ conceptualized the study, performed the statistical analysis, and drafted the original manuscript. SS, GM, XL, YD, and ZS designed the instrument, collected the data, and participated in revision of the paper. AL and ZZ participated in the design of the study, supervised all the process, and controlled the quality of this study. All authors read and approved the final manuscript. | 2021-10-20T16:40:16.743Z | 2021-09-13T00:00:00.000 | {
"year": 2021,
"sha1": "560376c3b0926985149270e546adc7b330f886e1",
"oa_license": "CCBY",
"oa_url": "https://www.frontiersin.org/articles/10.3389/fpubh.2021.776829/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "fea5c4389064afafb643a5f65ef393c46c9af865",
"s2fieldsofstudy": [
"Education"
],
"extfieldsofstudy": [
"Medicine"
]
} |
84186296 | pes2o/s2orc | v3-fos-license | Mask-guided Style Transfer Network for Purifying Real Images
Recently, the progress of learning-by-synthesis has proposed a training model for synthetic images, which can effectively reduce the cost of human and material resources. However, due to the different distribution of synthetic images compared with real images, the desired performance cannot be achieved. To solve this problem, the previous method learned a model to improve the realism of the synthetic images. Different from the previous methods, this paper try to purify real image by extracting discriminative and robust features to convert outdoor real images to indoor synthetic images. In this paper, we first introduce the segmentation masks to construct RGB-mask pairs as inputs, then we design a mask-guided style transfer network to learn style features separately from the attention and bkgd(background) regions and learn content features from full and attention region. Moreover, we propose a novel region-level task-guided loss to restrain the features learnt from style and content. Experiments were performed using mixed studies (qualitative and quantitative) methods to demonstrate the possibility of purifying real images in complex directions. We evaluate the proposed method on various public datasets, including LPW, COCO and MPIIGaze. Experimental results show that the proposed method is effective and achieves the state-of-the-art results.
INTRODUCTION
Recent appearance-based gaze estimation is performed under outdoor conditions by using an annotated large-scale real image training dataset. However, annotating training data sets requires a lot of manual labor. To solve this problem, a training model on a synthetic image is preferred because the annotations are automatically available. But this solution has a drawback, the distribution between the real image and the synthetic image is quite different. Traditionally, the solution is to use unmarked actual data to improve the authenticity of the synthetic image from the simulator, such as SimGANs [1], these methods only learn the global features without considering local features. In the gaze estimation task, after realization with simGANs, the shape of the pupil or the edge of the pupil might by changed , the gaze estimation error will be increased due to the wrong pupil center location. Thus, these methods cannot be applied to outdoor (field) scenes due to its weak training time and adaptability to different situations in the field. In a different manner, we try to purify real image by extracting discriminative and robust features to convert outdoor real images to indoor synthetic images. Synthetic images is more regular and easy to learn, meanwhile, the annotations are automatically available.
To avoid changing the shape of the pupil or the edge of the pupil, we propose an mask-guided style transfer network to learn both local and global features. The way to handle local features is to obtain the attention region (pupil or iris) by segmentation. Fortunately, with the rapid development of deep learning [2,3] based image segmentation methods including FCN [4],SegNet [5],U-net [6],Mask R-CNN [7], we can obtain much better mask. To learn the style and content information from synthetic images, we first introduce the segmentation masks to construct RGB-mask pairs as inputs, then we design a mask-guided style transfer network to learn style features separately from the attention and bgkd(background) region , learn content features from full and attention region. For feature extraction, our work is most directly related to the work initiated by Gates et al. [8]. The feature map of the deep convolutional neural network with differentiated training is used to achieve the breakthrough performance of the transfer of painting style. We train a feed-forward feature extraction network for image transformation tasks. Our network aims to learn as much as possible on the premise of synthetic distribution, to minimize the loss of content transmission, and to solve the problem of insufficient spatial alignment information caused by the gram matrix. To achieve this goal, we propose a loss network with a novel task-guided loss, the attention region ,background region and full image region will be calculated in different task.
Our contributions are presented in three folds: 1. We took the first step to consider the attention region in style transfer task and propose an mask-guided style transfer network to purify the real image, making it similar to indoor conditions while retaining annotation information. Different with previous work in refining the synthetic images with global features, we purified the real images with local and global features.
2.Our network not only considers the RGB color channel, but uses the segmentation masks to construct RGB-mask pairs as inputs. We learned style features separately from the attention and bkgd(background) region and learned content features from full and attention region.
3. We proposed a hybrid research method (qualitative and quantitative) for experiments on two tasks. The results show that the proposed architecture significantly purified the real image compared with the baseline methods. Meanwhile, We achieve the state-of-the-art results on gaze estimation task.
PROPOSED METHOD
As shown in Figure 1, there are three multi-scale stages and a loss net to learn final features. It contains three multi-scale stages and a loss net to learn final features. There are three main streams which extracted from different regions of image , i.e. , the full-stream f f ull , the attention stream f attetion , the background stream f bkgd . The full-stream f f ull learns features from the raw images. Meanwhile, the attention stream f attetion and the background stream f bkgd are learned attention features and background features with attention maps. The attention maps are generated by the attention subnet, the streams are designed to retain the content of input image I RGB and transfer the style from style image S RGB with the input image mask I mask and style image mask S mask .
Attention Subnet
Given the input(style) image pair (RGB-Mask) as inputs, the attention subnet then produces attention maps which can be denoted as where σ is the sigmoid function, weight and b are the convolutional filter weights and bias. In the contrary, the background maps denoted as att − , att + and att − constitute a contrastive attention pair, for each location (i,j) which in the pair of attention maps and backgrounds maps should meet the constraint: Thus, the stream of attention and background can be denoted as : where ⊗ means the spatial weighting operation.
Loss network with region-level task-guided loss
With the attention maps described in last subsection, we further introduce the region-level triplet loss to enhance contrastive feature learning. After the attention operation, features from three main streams can be denoted as f f ull ,f attetion and f bkgd , f f ull ,f attetion and f bkgd are used to calculate region-level task-guided loss for two tasks: keep the content and style transfer. Our loss network can be divided into two parts: Feature reconstruction loss (a) and Style reconstruction loss(b), feature reconstruction loss is denoted as content which is the summary of gc and lc , meanwhile, style reconstruction loss is denoted as style which is the summary of gs and ls .
Feature reconstruction loss
Traditional feature reconstruction loss which known as content loss only takes the input image I RGB as input and try to minimize the loss between the content of input image I RGB and output image O RGB without considering encoding content reconstructions. We address this problem with the image segmentation masks I mask for the input images, the local feature of pupil region can be addressed when calculating the loss of f f ull and f attetion . To visualise the image information that is encoded at different layers of the input image with masks, we perform gradient descent on a white noise image to find another image that matches the feature responses of the original image with mask. We then define the squared-error loss between the two feature representations where C is the number of channels in the semantic segmentation mask and l indicates the l-th convolutional layer of the deep convolutional neural network, F f ull [·] is the f f ull in each layer l with the channel c, F attention [·] is the f attention in each layer l with the channel c, λ g is the weight to configure layer preferences of global losses ,λ l is the weight to configure layer preferences of local losses .
Each layer with N l distinct filters has N l feature maps each of size M l , where M l is the height times the width of the feature map. So the responses in each layer l can be stored in ij is the activation of the i th filter at position j in each layer l. As minimizing content , the image content and overall spatial structure are preserved but color, texture, and exact shape are not.
Style reconstruction loss
Feature Gram matrices are effective at representing texture, because they capture global statistics across the image due to spatial averaging. Since textures are static, averaging over positions is required and makes Gram matrices fully blind to the global arrangement of objects inside the reference image. So if we want to keep the global arrangement of objects, make the gram matrices more controllable to compute over the exact region of entire image, we need to add some texture information to the image.
Instead of taking input image I RGB and style image S RGB as inputs, we take the input image I RGB and style image S RGB with their mask I mask and S mask as pair inputs. To learn the skin style and pupil style respectively, we denote the pupil region as attention region and extract attention maps f attention from both style image and input image, meanwhile, the skin region denoted as background region and product background maps f bkgd from style image and input image. We then define the squared-error loss between the two region feature representations where F f ull [·] is the f f ull in each layer l with the channel c, F attention [·] is the f attention in each layer l with the channel c, F bkgd [·] is the f bkgd in each layer l with the channel c, λ g is the weight to configure layer preferences of global losses ,λ l is the weight to configure layer preferences of local losses. We formulate the style transfer objective by combining both two components together: (14) where L is the total number of convolutional layers and l indicates the l-th convolutional layer of the deep convolutional neural network. α l and β l are the weights to configure layer preferences. content is the content loss (Eq.(10)) and style is the style loss(Eq.(13)). α l ,β l are scalars, α l = 10 2 ,β l = 10 4 , in all cases the hyperparameters α l ,β l are exactly the same. We find that unconstrained optimization of Equation 20 typically results in images whose pixels fall outside the range [0,255]. For a more fair comparison with our method whose output is constrained to this range, for the baseline we minimize Equation 20 using projected L-BFGS. Image O is generated by solving the problem where I is initialized with white noise. The advantage of this solution is that the requirement for mask is not too precise. It does not only retain the desired structural features, but also enhance the estimation of the pupil and iris information during the reconstruction of the style.
Style Transfer
The purpose of the style transfer is to generate an image that combines the content of the target content image as the real image content with the style of the target style image as the style of the synthetic image. We train an image transformation network for each of the several hand selection style goals and compare our results with the baseline methods of Gatys et al. [8] and Feifei Li et al. [9]. As a baseline, we reimplemented the method of Gatys et al. [8] and Feifei Li et al. [9]. In order to make a fairer comparison with our method whose output is constrained to [0, 255], for the baseline, we minimize the equation 1 and equation 6 by using the projected L-BFGS by cropping the image to the range [0, 255] at each iteration. In most cases, the optimization converges to satisfactory results in 500 iterations. Qualitative Results: We compare the proposed style transfer method with methods proposed by Gatys et al. [8] and Feifei Li et al. [9]. Figure 4 shows the qualitative results in the indoor and outdoor scenes for the UnityEyes [10] and LPW [11] datasets respectively. From the LPW dataset, we choose six different real images to represent six different conditions of outdoor scenes, which are images (a),(b),(c),(d),(e), and (f). From the UnityEyes dataset, we selected two synthetic images that have different distributions as the style images, style A, B, none of which has no same gaze angle as the six real images. As can be seen from (a),(b), and (c), the proposed method is less affected by external factors, such as illumination, and is similar to the results obtained by Gatys et al. [8] and Feifei Li et al. [9]. However, if look closely, we can find that the proposed method is more capable of preserving the color information of the style image. From (d),(e), and (f), it is obvious that Gatys et al. [8] and Feifei Li et al. [9] are so greatly influenced by light and other factors that the pupil and the iris regions cannot be completely separated and even in (e),there is a loss of pupil area. What's more, the distribution of pupil and iris regions is dramatically different from style image. In comparison, our proposed method can not only separate the pupil and the iris regions more easily but also the distribution of pupil and iris regions is more similar to style image. Speed: Table 2 compares the runtime of our method with Gatys et al. [8], Feifei Li et al. [9] for several image sizes. Across all image sizes, compared to 400 iterations of the baseline methods, our method is three orders of magnitude faster than Gatys et al. [8] and we achieve better qualitative results (Fig.6) compared with Feifei Li et al. [9] in tolerate speed. Our method processes images of size 512×512 at 20 FPS, making it feasible to run in real-time or on video.
Quantitative Results: There are five gaze estimation methods used as base-line estimation methods. Three of them are common methods, Support Vector Regression(SVR), Adaptive Linear Regression(ALR) and Random Forest(RF). The other two methods are reproduced for fairly comparison with state-of-the-art. The first method is a simple cascaded method [24][25] [26] which uses multiple k-NN(k-Nearest Neighbor) classifier to select neighbors in feature space joint head pose, pupil center and eye appearance. The second one aims to train a simple convolutional neural network (CNN) [27][28] [29] to predict the eye gaze direction by using l 2 loss. As is shown in table 2, we compare the performance of these two gaze estimation methods on different datasets, where the "Method" represents gaze estimation methods with different training sets. It can be observed that the accuracy of gaze estimation of each dataset has improved at least three degrees, which means that our proposed method has greatly improved the performance when testing the output. This improvement will have some practical value in human-computer interaction.
Furthermore, in order to prove that purifying real dataset can achieve better performance than refining synthetic dataset on gaze estimation task, we compare proposed method with state-of-the-art methods of refining synthetic dataset for gaze estimation,SimGANs [1]. In order to make a fairer comparison with our method, we reproduce SimGANs [1], different from traditional methods which training on synthetic UnityEyes dataset and testing on real MPIIGaze dataset, SimGANs [1] refining UnityEyes with its method, try to realistic synthetic image and test on real MPIIGaze dataset. In the contrary, proposed method purify the testing real MPIIGaze dataset without modifying training synthetic datset. From Table 3, we can see that purifying testing real dataset outper- Preserving Ground Truth: We manually quantify the gaze center of the 200 real and purified images by fitting the ellipse to the pupil, which can be seen as an approximation of the direction of gaze and is difficult for humans to label accurately, to quantify that there is no significant change in the ground truth gaze direction. The absolute difference of the estimated pupil center between the real and corresponding purified images is very small: 0.8 ± 1.1 (eye width=55px)
CONCLUSION
This paper purify the real image by weakening its distribution, which is a better choice than improving the realism of synthetic image. We have applied this method to style transfer and gaze estimation tasks where we achieved comparable performance and drastically improved speed compared with existing methods. Performance evaluation indicates that purified MPIIGaze dataset (purified by our proposed method) records smaller error angle when used for gaze estimation task as compared with the raw MPIIGaze dataset.
In the future, we intend to explore modeling the real-time gaze estimation system based on the proposed method and improve the speed of purifying videos. | 2019-03-19T03:54:55.000Z | 2019-03-19T00:00:00.000 | {
"year": 2019,
"sha1": "736714381cf6e4a177e7adc93f4ac614c6ba2905",
"oa_license": null,
"oa_url": "http://arxiv.org/pdf/1903.08152",
"oa_status": "GREEN",
"pdf_src": "Arxiv",
"pdf_hash": "01803639eff60b28db5ea79f4cd8351da3e5f474",
"s2fieldsofstudy": [
"Computer Science"
],
"extfieldsofstudy": [
"Computer Science"
]
} |
230667786 | pes2o/s2orc | v3-fos-license | Social Distancing Associations with COVID-19 Infection and Mortality Are Modified by Crowding and Socioeconomic Status
The SARS-CoV-2 virus is a public health emergency. Social distancing is a key approach to slowing disease transmission. However, more evidence is needed on its efficacy, and little is known on the types of areas where it is more or less effective. We obtained county-level data on COVID-19 incidence and mortality during the first wave, smartphone-based average social distancing (0–5, where higher numbers indicate more social distancing), and census data on demographics and socioeconomic status. Using generalized linear mixed models with a Poisson distribution, we modeled associations between social distancing and COVID-19 incidence and mortality, and multiplicative interaction terms to assess effect modification. In multivariable models, each unit increase in social distancing was associated with a 26% decrease (p < 0.0001) in COVID-19 incidence and a 31% decrease (p < 0.0001) in COVID-19 mortality. Percent crowding, minority population, and median household income were all statistically significant effect modifiers. County-level increases in social distancing led to reductions in COVID-19 incidence and mortality but were most effective in counties with lower percentages of black residents, higher median household incomes, and with lower levels of household crowding.
Introduction
The Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) virus has emerged as a worldwide public health emergency. By the beginning of February 2021, more than 105 million cases of COVID-19 had been documented worldwide, and over 2.3 million deaths had been recorded [1]. Although vaccinations have begun, social distancing remains one of the primary public health approaches to slowing disease transmission in the United States and other parts of the world [1]. Theoretically, increases in social distancingthrough prohibiting social gatherings, closing non-essential business, and ordering people to stay at home-may reduce disease incidence [2,3], but it cannot be implemented without personal and economic consequences. In a study of 134 countries, lockdowns devised to increase social distancing were shown to reduce COVID-19 transmission. The same study observed proportional decreases in Google mobility metrics and COVID-19 transmission rates [4]. A study modeling the impacts of social distancing in the 25 counties in the United States with the highest numbers of confirmed cases as of 16 April 2020 demonstrated that increases in cellphone-measured social distancing, even before stay-at-home orders, were correlated with a decreased growth in positive COVID-19 test rates [5]. Compelling 2 of 8 evidence suggests that in the United States there are clear disparities in disease burden, with higher rates of death from COVID-19 among racial and ethnic minorities [6][7][8]. For example, housing attributes are widely known to be associated with infectious disease through various processes, such as through the transmission of disease between occupants in crowded conditions [9]. The pandemic has already amplified and underscored various gaps in systemic preparedness, as well as a wide range of societal disparities related to race, ethnicity, and socioeconomic status. These disparities are likely driven by numerous risk factors that are independently and jointly concentrated among low-income individuals and families and within communities of color. Health information, beliefs, and behaviors regarding COVID-19 vary by race, ethnicity, and age [10]. However, some factors are removed from individual agency. For example, the ability to comply with social distancing is directly related to numerous correlates of socioeconomic status including occupation, neighborhood attributes, mode of transportation, housing, and household conditions. It is not only the condition of these spaces but the flexibility to re-craft and adapt daily life to a new situation without disruption that is a hallmark of privilege. In numerous reports, COVID-19 mortality rates have been elevated among United States. residents living in the most disadvantaged counties [4][5][6][7].
The effectiveness of social distancing interventions on morbidity and mortality in the United States has not been fully described [5], and their efficacy in different communities is unknown [11]. This evidence is critical to inform public policy decisions such as plans for reopening non-essential businesses, and for future pandemic management [12]. We sought to use data from the first wave of COVID-19 to examine the associations between objective social distancing and COVID-19 incidence and mortality, and to determine if these associations differed in counties with different sociodemographic characteristics and levels of household crowding.
Materials and Methods
Incidence and mortality counts per county through 29 April 2020 (to represent the time period before social distancing policies throughout the United States began to be removed) were collected from the COVID Tracking Project [13]. To determine objective social distancing for each county in the United States, we used nationwide, de-identified smartphone GPS data provided by Unacast ( Figure 1) [14]. For this analysis, we utilized data from 24 February 2020 to represent pre-COVID-19 levels of distancing. Objective social distancing, scored 0-5 (5 indicates increased distancing), was calculated based on (1) change in average distance traveled, (2) change in non-essential venue visitation, and (3) the probability that two users were in close proximity [14][15][16][17]. Unacast measures were shown to be correlated with Census data according to geography, income, age, and sex. Covariate data at the county or state level were obtained from the 2018 American Community Survey 5-year estimates, 2019 Robert Wood Johnson Foundation and University of Wisconsin Population Health Institute County Health Rankings, and The Atlantic COVID Tracking Project.
Generalized linear mixed models with a Poisson distribution accounting for counties nested within states were used to calculate incidence rate ratios (IRRs) and 95% confidence intervals (CIs) for a 1-unit increase in objective social distancing. Restricted cubic regression splines were used to test for deviations from linearity. Multivariable models were a priori adjusted for variables associated with incidence rates [18], case ascertainment, or physical features likely to impact distancing, including: county-level Hispanic ethnicity, minority race (includes non-Hispanic black, non-Hispanic Asian, non-Hispanic American Indian or Alaska Native, non-Hispanic Native Hawaiian or Other Pacific Islander, non-Hispanic other or two or more races), percent aged 50 years and older, percent males, median household income, population density, obesity prevalence, percent household crowding (>1 individuals/room) or percent extreme crowding (>1.5 individuals/room), and state-level cumulative COVID-19 testing rate. To determine if the associations between objective social distancing and COVID-19 incidence or mortality were modified, we used multiplicative interaction terms to determine statistical significance and present results stratified by each potential effect modifier. We examined modification by tertiles of county racial and ethnic composition, percent of the county over age 50 years, median household income, percent crowding, and percent extreme crowding. In a sensitivity analyses, we examined the alternate time windows of objective social distancing based on the published incubation periods and symptom onset windows (e.g., 5-day lag, 14-day lag [19]) to determine if the results were robust to our choice to use baseline exposures and scaled Poisson models based on the Pearson and deviance methods accounting for overdispersion. All tests were two-sided and p < 0.05 was considered statistically significant. Our study did not constitute human subjects research and was considered exempt from Institutional Review Board review. and state-level cumulative COVID-19 testing rate. To determine if the associations between objective social distancing and COVID-19 incidence or mortality were modified, we used multiplicative interaction terms to determine statistical significance and present results stratified by each potential effect modifier. We examined modification by tertiles of county racial and ethnic composition, percent of the county over age 50 years, median household income, percent crowding, and percent extreme crowding. In a sensitivity analyses, we examined the alternate time windows of objective social distancing based on the published incubation periods and symptom onset windows (e.g., 5-day lag, 14-day lag [19]) to determine if the results were robust to our choice to use baseline exposures and scaled Poisson models based on the Pearson and deviance methods accounting for overdispersion. All tests were two-sided and p < 0.05 was considered statistically significant.
Our study did not constitute human subjects research and was considered exempt from Institutional Review Board review.
Results
Objective social distancing data were available for 3054 counties (94%) in all 50 states and Washington, D.C. In multivariable-adjusted models, each unit increase in objective social distancing was associated with a 26% decrease in COVID-19 incidence and a 31% decrease in COVID-19 mortality (Table 1). Models were similar regardless of adjustment for percent crowding or percent extreme crowding. Percent crowding, minority population, percent of individuals over 50 years of age, and median household income were all statistically significant modifiers of the associations between objective social distancing and COVID-19 incidence and mortality (Table 2). In counties in the highest tertile of percent minority population (27.9-99.3%), increases in objective social distancing were not associated with COVID-19 incidence (IRR = 0.89; 95%CI: 0.77-1.04) or mortality (IRR = 0.98; 95%CI: 0.76-1.27). In contrast, higher levels of objective distancing were protective for incidence (IRR = 0.67; 95%CI: 0.54-0.85) and mortality (IRR = 0.51; 95%CI: 0.36-0.71) in counties with the lowest percent of minority residents (0.0-9.8%). Similar patterns were observed with county median income, where objective social distancing was more protective in the wealthiest counties. Objective social distancing was protective in the highest and lowest tertiles of percent of residents aged 50 and older, but there was no association among the middle tertile. As expected, in counties with the highest levels of crowded households, objective social distancing measures were associated with smaller decreases in incidence or mortality, although this pattern was less clear when looking at the percentage of households with extreme crowding. Results were similar in models using objective social distancing measures for periods other than baseline and in Poisson models accounting for overdispersion (data not shown). Table 1. Association of each one-unit increase in county-level objective social distancing with the risk of county-level COVID-19 incidence or mortality in the United States through 29 April 2020. Table 2. Effect modification of the association of each one-unit increase in county-level objective social distancing on the risk of county-level COVID-19 incidence or mortality in the United States through 29 April 2020.
Discussion
In this analysis of the impacts of county-level objective social distancing and countylevel COVID-19 incidence and mortality during the first wave of the pandemic in the United States, we observed that increases in objective social distancing were associated with decreased incidence and decreased mortality, even after adjusting for county-level sociodemographics, crowding, and obesity levels. Objective social distancing, however, was not equally protective in all counties. We observed that objective social distancing was less protective in counties with a larger minority population, lower median incomes, and more crowding. Social distancing was most beneficial in counties with the highest levels of median income.
Our results provide insights into the factors that may increase or decrease the efficacy of social distancing. The overall effectiveness of social distancing, however, is already clear. Our results for the entire United States are well in line with a recent modeling study of the correlations between daily cellphone-based mobility ratios and the rates of newly confirmed cases of COVID-19 in 25 United States counties with the highest case rates as of April 2020. Among these 25 counties, correlations between an 11-day lagged mobility ratio and the case-growth ratio ranged from 0.53 to 0.90, with an overall correlation of 0.71, indicating that increased distancing had a significant impact on case growth rates [5]. One analysis estimated, using national county-level data, that COVID-19 case counts would have been 35 times higher without any of the social distancing restrictions that were put in place in March and April of 2020 [20]. Social distancing, measured using Google's community mobility reports, was estimated to have led to approximately 10,000 fewer deaths from COVID-19 in the Sao Paulo area of Brazil [21]. Social distancing policies were also shown to lead to 65% reductions in transmission in a study of 134 nations [4]. Restrictions on social distancing focus on public behavior in public spaces, but some of the most important microenvironments and activities that influence transmission likely occur outside of the public domain. We know that crowded homes, like crowded workplaces, provide multiple mechanisms for transmission through airborne and surface transfer. In Italy, one factor thought to contribute to COVID-19 cases was the spread driven by the high proportion of multigenerational households comprised of family structures such as a relatively higher mean age of leaving the parental household (30.1 years) compared to the European Union (26 years), frequent physical contact with extended families, and a relatively large proportion of the population (approximately 25%) being aged 65 years or older [22][23][24][25]. There is also a substantial body of literature demonstrating crowding and housing conditions as a key risk factor for the transmission of various infectious agents [26][27][28][29][30][31][32][33][34].
Social distancing is an effective tool because it addresses a mechanistic and modifiable pathway for viral transmission. However, the persistence and scale of health disparities are largely shaped by factors that are not easily modified for individuals, especially on a timescale that can contain a pandemic. Health disparities are often driven by mechanisms that are hidden from view. Housing, transportation modes, and occupational settings are critically important settings that are also known to have strong socioeconomic determinants. Our results illustrating the differences in social distancing efficacy across different types of counties are in line with a growing body of literature demonstrating vast disparities in COVID-19 infection and mortality rates. A recent analysis of death records during the first four months of 2020 in Massachusetts showed that, compared to previous years, mortality rates were significantly higher in communities with higher poverty, higher household crowding, higher percentage of populations of color, and higher racialized economic segregation [7]. These results have been confirmed in analyses including all U.S. counties [35].
Our study has several key limitations. First, our measures of outcome, exposures, and effect modifiers are all aggregated to the county level. This lack of granularity restricts our ability to observe associations and likely incorporates a substantial measurement error into our analyses. However, this would likely decrease our ability to detect patterns. Second, to draw straightforward comparisons, we chose not to incorporate data on social distancing after states began to relax social distancing rules. This limits the generalizability of our findings but allowed us to clearly examine the impacts of social distancing measures without accounting for time lags after reopening. It also allowed us to study the effects of social distancing before the wide onset of "pandemic fatigue." Additional analyses using data spanning 2020-2021 and accounting for differences in the timing of and changes in stay-at-home orders and social distancing are warranted. Finally, we are unable to rule out the possibility that some other factor that varies at the county level may be driving the associations that we have observed. Other measures apart from social distancing have been implemented to mitigate the spread of infection including wearing masks [36], which also impacts the generalizability of our findings as different countries have utilized different public health control measures [37]. In addition, other variables that should be considered include testing coverage, contact tracing, quarantine of contacts, isolation of non-severe cases, social support, and environmental factors (e.g., humidity, temperature) that have been shown to be associated with the severity of illness and mortality due to COVID-19 [38]. Our models were generally robust to adjustment for many potential confounders, but given our ecological study design, this cannot be ruled out.
However, our study does have many strengths. We used multiple sources of publicly available data to assess the differential impact of social distancing on COVID-19 incidence and mortality for the entire U.S. Our results provide a framework for public health professionals to optimize social distancing guidance, and a way to identify communities that may require additional resources/strategies for transmission reduction. These may be more important as vaccination uptake varies across communities, to identify areas where social distancing is unlikely to be as efficient and vaccination uptake is low. Importantly, these results also demonstrate that known drivers of health disparities in the U.S. also lead to important gaps in prevention.
Conclusions
In conclusion, although we have demonstrated that in the first wave of the pandemic social distancing is an effective way to reduce COVID-19 infection and mortality, we have also shown that this strategy does not work evenly across populations. As local, state, and federal governments consider containment approaches and changes in approaches, it is important to determine if there are factors that may impact their efficacy across populations. A focus on developing effective strategies for populations experiencing crowding or with historical patterns of health disparities will be a key piece to reducing overall transmission and disease containment. Data Availability Statement: All data are publicly available from the referenced sources (the COVID Tracking Project at: https://covidtracking.com/data (accessed on 25 April 2021). and the Unacast Social Distancing Scoreboard at: https://www.unacast.com/covid19/social-distancing-scoreboard (accessed on 25 April 2021). Statistical code is available upon request from the authors. | 2020-10-28T18:34:40.028Z | 2020-09-16T00:00:00.000 | {
"year": 2021,
"sha1": "c963ae3b8691b90a4b2dc12d99cb9feccb6fa330",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/1660-4601/18/9/4680/pdf?version=1619591667",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "0cea79f3b6c7e5f44e5967df35742556f100a1b7",
"s2fieldsofstudy": [
"Economics"
],
"extfieldsofstudy": [
"Psychology",
"Medicine"
]
} |
225419341 | pes2o/s2orc | v3-fos-license | Augmenting the Sustainability of Vegetable Cropping Systems by Configuring Rootstock-Dependent Rhizomicrobiomes that Support Plant Protection
: Herbaceous grafting is a propagation method largely used in solanaceous and cucurbit crops for enhancing their agronomic performances especially under (a)biotic stress conditions. Besides these grafting-mediated benefits, recent advances about microbial networking in the soil / root interface, indicated further grafting potentialities to act as soil environment conditioner by modulating microbial communities in the rhizosphere. By selecting a suitable rootstock, grafting can modify the way of interacting root system with the soil environment regulating the plant ecological functions able to moderate soilborne pathogen populations and to decrease the risk of diseases. Genetic resistance(s) to soilborne pathogen(s), root-mediate recruiting of microbial antagonists and exudation of antifungal molecules in the rhizosphere are some defense mechanisms that grafted plants may upgrade, making the cultivation less prone to the use of synthetic fungicides and therefore more sustainable. In the current review, new perspectives o ff ered by the available literature concerning the potential benefits of grafting, in enhancing soilborne disease resistance through modulation of indigenous suppressive microbial communities are presented and discussed.
Introduction
Modern agriculture needs innovative strategies inspired by the principles of agroecology, aimed to guarantee soil conservation and fertility, to face the adversities that affect crop productivity and, generally, to increase the sustainability of intensive systems. One of the most important challenges for sustainability concerns the control of soil-borne diseases being considered a major limitation to crop production. The massive use of chemicals against plant pathogens is no longer a viable practice for environmental risks and human health care. The most commonly applied eco-friendly approaches include solarisation, biofumigation, crop rotation, tillage management practices, residue management and organic amendments [1]; additional strategies such as applications of plant growth promoting rhizobacteria (PGPRs), endo-and ectomycorrhizal fungi, cyanobacteria and other organisms can also improve plant resistance to soilborne pathogens. Several studies reported that plants can recruit a specific beneficial rhizosphere microflora which can contribute to reduce the activity of plant pathogens and to make plants more resistant to environmental stressors [2,3]. Investigations on rhizosphere microbiome carried out by next-generation sequencing technologies allowed to identify and quantify microorganisms associated with the root apparatus of vegetable crops highlighting an evolutionistic mechanism of microbial recruiting adopted by plants.
Vegetable grafting is a propagation method largely used in solanaceous and cucurbit crops to increase plant resistance to soilborne pathogens as well as other environmental stresses, and to enhance crop productivity and fruit quality [4]. In soils infected by highly destructive plant pathogens, the use of resistant grafted plants represents the main biological-based method that allows cultivation of the highly susceptible cultivars [5]. In the latest years, increasing studies on the ecological role of the grafting revealed new interesting opportunities of this technology to contrast and limit the activity of soilborne pathogens and their damages on vegetable crops, by modifying the presence of beneficial root-associated microbes. For instance, Poudel et al. [6] used a grafted tomato system to study the effect of rootstock genotypes and grafting on soil bacteria communities and their results highlighted an effect of rootstock genotype on bacterial diversity and composition in the rhizosphere. Moreover, Duan et al. [7] reported that some pepper grafting combinations significantly increased the populations of fungi and actinomycetes in the rhizosphere enhancing the activities of peroxidase, catalase, phosphatase, invertase, urease, and nitrate reductase in root rhizosphere soil. Similar results were also reported in grafted eggplants [8]. Besides plant species, additional environmental factors such as soil type can also affect the microbial communities in the rhizosphere indicating the need to evaluate the rootstock-mediated effects on rhizosphere microbiome under different environmental conditions [9,10]. The above findings indicate the potential of using plant rootstocks as a mean to recruit specific soil beneficial microorganisms for biotic stress management.
This review offers a novel perspective on the potentialities of vegetable grafting as sustainable practices to enhance crop resistance to environmental stresses through the modulation of microbial community structure in the rhizosphere.
Soil Microorganisms for Biological Control of Plant Diseases
Rhizosphere and bulk soil around growing plants constitute the habitat for a large number of microbial species with different lifestyles, interacting with one another, with the soil and plants. The culturable bacteria and fungi associated with rhizosphere can contain up to 10 11 microbial cells per gram of roots [11]. Some microorganims have a neutral effect on the plant, but many known microorganisms are beneficial to the plants for nutrient and carbon cycling, soil organic matter formation and stabilization, so influencing agricultural productivity. They are symbiotic nitrogen-fixing bacteria (Rhizobium leguminosarum), endo-and ectomycorrhizal fungi, plant growth-promoting rhizobacteria (PGPR) and other fungi [12]. Many of these microorganisms are even investigated for their ability to prevent or limit soil-borne plant pathogens (fungi and oomycetes are the most important) through mechanisms as hyperparasitism, antibiosis and competition for ecological niches and nutrients [13].
Microbial hyperparasites have a predatory behavior: they enter host cells of fungi and oomycetes helped by secreting cell-wall lytic enzymes (as chitinases, cellulases and proteases) and feed on the pathogen as long as it dies. Parasitic activity is quite common, and well documented for Trichoderma and Gliocladium against fungal pathogens as Rhizoctonia, Sclerotinia, Verticillium and Gaeumannomyces [14][15][16]. Even Coniothyrium minitans [17] and Sporidesmium sclerotivorum [18] are effective in controlling diseases caused by sclerotia-forming fungi. Chitinolytic activity of Pseudomonas spp. is responsible of antagonistic activity towards R. solani [12].
With the competition for space, pathogens are prevented from accessing root surface and plant tissue whereas the competition for nutrients, especially for carbon released as root exudates, affect the saprophytic phase of pathogens, and may cause lacking spore germination and microbiostasis [11]. Siderophore-producing Pseudomonas spp. are involved in the competition for iron, a micronutrient essential for growth and activity of the pathogens and are able to reduce disease incidence or severity of pathogenic fungi [23].
Literature survey reveals many pivotal examples of rhizosphere-competent microbial biocontrol agents of plant pathogens. Fungi belonging to Trichoderma genus, for example, have antagonistic properties towards a plethora of plant pathogens relying on all of the antagonist modes [24]. T. harzianum and T. asperellum showed antagonistic effects against Fusarium oxysporum of tomato [25][26][27] and melon [28] under field conditions. Gava et al. [28] tested different Trichoderma species (T. harzianum, T. viride, T. koningii, and T. polysporum) in a naturally infested soil and obtained the highest control of melon wilt using T. polysporum.
Malolepsza [29] observed a significant stimulation of systemic defenses (activation of antioxidant enzymes and enhancement of phenols) by T. virens inoculation of tomato plants, leading to lowered Rhizoctonia solani infection. Some authors verified the involvement of the proteins Sm1 and Epl1 in the systemic protection of tomato plants mediated by Trichoderma spp. [30]: in presence of these proteins it was observed an increase in disease resistance against Alternaria solani, Botrytis cinerea, and Pseudomonas syringae due to the increased expression of peroxidase and α-dioxygenase encoding genes [30], the elicitation of the salicylic acid pathway [31,32] and the ethylene pathway [31]. Even the accumulation of phenolic acids, flavonoids and de novo synthesis of catechins, enhanced by T. atroviride inoculation, is supposed to contribute to cucumber protection against R. solani [33]. In cucumber seedlings inoculated with T. harzianum inoculations, Chen [34] found alterations in nuclear DNA content and cell cycle-related genes expression that might maintain a lower ROS accumulation and higher root cell viability counteracting Fusarium disease in open field. Quantitative proteomics studies on black pepper plants primed with T. harzianum confirmed the plant defense response against the pathogen Phytophthora capsicum through an increase of ethylene synthesis and activating both the isoflavanoid pathway and lignin synthesis [35].
Coating of tomato seeds with T. asperellum and Bacillus subtilis decreased the susceptibility of plants to Pythium aphanidermatum [36]. Frequently the combination of different antagonists in microbial consortia improves their effectiveness for the involvement of diverse biocontrol mechanisms [37]. In watermelon, a systemic acquired resistance against F. oxysporum f.sp. niveum was related to the inoculation with a consortium of T. harzianum, Paenibacillus polymyxa and other antagonistic microorganisms that activated defense-related enzymes [38]. T. harzianum and Bacillus amyloliquefaciens together inhibited the growth and production of mycelia and sclerotia protecting over 80% of tomato, squash and eggplant seedlings against Sclerotinia sclerotiorum [39]. Similarly, to T. asperellum, even the arbuscular mycorrhizal fungus Rhizophagus irregularis (previously known as Glomus intraradices) lowered disease incidence of Fusarium wilt in tomato plants [40]. Both these agents increased plant height, chlorophyll content and Ca, Mg, S, Mn, B and Si uptake. T. harzianum combined with Glomus intraradices induced a plant basal resistance attenuating the hormone (ethylene and abscisic acid) disruption induced by F. oxysporum in melon [41] whereas associated with Glomus mossae is useful in cucumber against Phytophthora melonis increasing the transcription level of defensive genes as phenylalanine ammonialyase, cucumber pathogen-induced 4, lipoxygenase and galactinol synthase [42]. In field experiment the consortium mycorrhizae, Trichoderma and plant growth-promoting bacteria enhanced the pepper yield and modulated the activities of defense enzymes as polyphenol oxidase, peroxidase, superoxide dismutase, and catalase [43]. Even arbuscular mycorrhizal fungi alone can control plant soil-borne diseases [44,45] activating a systemic plant immune response. Mycorrhizal fungi reduced the disease severity index of eggplant Verticillium wilt [46]; the disease tolerance was related to a lower proline content and relative electrical conductivity in leaves, and to higher activity of browning related enzymes (phenylalanine ammonia-lyase, polyphenol oxidase and peroxidise). Panda et al. [47] found that pre-colonization of tomato roots with mycorrhizal fungus Piriformospora indica systemically induced resistance against early blight by Alternaria solani; after pathogen attack P. indica induced a rapid activation of jasmonic acid/ethylene-mediated basal defenses against pathogen infection by altering the expression of JA/ET related genes. Glomus mossae suppresses F. oxysporum development in the roots and rhizosphere of watermelon and modulates the composition of root exudates. Bacterial species of the Bacillus and Pseudomonas genus control plant disease by producing antibiotics or stimulating the host resistance [48]. Tomato plants inoculated with Bacillus amyloliquefaciens exhibited significantly low F. oxysporum f. sp. lycopersici infection compared with the control [49]. Clonostachys rosea counteracted gray mold disease not only by suppressing development and sporulation of Botrytis cinerea but also by inducing the resistance of tomato plants against B. cinerea [50].
How Plants Recruit Antagonist Microbes to Prevent the Infection by Pathogens
The rhizosphere microbiome has been recognized as the second genome of the plants [11], and it may reveal the greater ability of the plant in adapting to the external environment, including the protection degree against pathogenic agents. Plants through their roots are in dynamic communication with the surrounding bulking soil microbial communities [51]. Roots exudates are a main food source for microorganisms and a driving force for their assembling and activities. Carbon and nitrogen are exudates as simple molecules (sugars, organic acids, secondary metabolites) or complex polymers (mucilage) but their composition varies with the plant genotype, developmental stage, and the presence of (a)biotic stresses [52]. It has been seen that microbial communities in the rhizosphere of different plant species growing on the same soil are often different and vice versa [53,54] thus demonstrating that plants modulate exudates profile (alteration of biosynthesis and transport of molecules) and immune system activities to recruit specific beneficial microbes. Moreover, plants can detect communication signals between bacteria in the rhizosphere (quorum sensing signals) and produce molecules that stimulate or deactivate these signals so influencing the outcome of microbe-microbe and/or plant-microbe interactions [55]. L-malic acid is the small signalling molecule exudated from tomato plant roots that is responsible for biofilm formation and root colonization of the antagonistic Bacillus subtilis [56]. Similarly, Tan et al. [57] confirmed the important role of malic and citric acids in tomato root surface colonization of Bacillus amyloliquefaciens. Moreover, different organic acids in watermelon and cucumber roots exudates stimulate the beneficial Paenibacillus polymyxa [58] and Trichoderma harzianum [59], respectively. Glucose, succinic acid, p-hydroxybenzoic acid, p-coumaric acid and glutamic acid in cucumber root exudates recruit Trichoderma with the most relevant effect from glucose [60].
Recent advances reveal that multiple signals operate in the establishment and the maintenance of arbuscular mycorrhizal symbiosis including calcium spiking, reactive oxygen species and phytohormones [61,62]. The plant hormones strigolactones are actively exuded into rhizosphere as ex-planta signalling molecules that attract arbuscular mycorrhizal fungi affecting both pre-symbiotic and symbiotic phases [63]. Other important hormones involved in the onset of the AM symbiosis are jasmonates [64], gibberellins [65], ethylene [66], and auxins [67]. A possible crosstalk between auxin and strigolactones is also postulated in tomato plants [68]. Martínez-Medina et al. [69] studied the interaction between tomato roots and the AM fungus Rhizophagus irregularis showing a regulatory role of nitric oxide mediated by a specific phytoglobin during the symbiosis.
Defense Mechanisms in Grafted Plants against Soilborne Pathogens
Disease control-related mechanisms of grafting may rely on the genetic traits of the rootstock-type. Interspecific grafting, for example, is often secured with the non-host resistance of the rootstock against species-specific soil-borne pathogens, such as the wilting causal agents belonging to F. oxysporum group that harbour differentially host-compatible formae specialis [70,71]. Limitation in spread of tomato wilting-associated bacteria in resistant cultivars could avoid the generalization of infection to the entire vascular tissues of susceptible scion [72]. Graft-transmissible resistance to airborne disease has been observed for sweet pepper cultivar grafted on a resistant cherry pepper rootstock [73]. In this situation, Agronomy 2020, 10, 1185 5 of 11 the effects of rootstock on the scion is the major determinant of resistance. Disease resistance of grafted plants may also result from the enhanced vigour for which grafted plants grow more fortified and more efficient in resource utilization, as well as developing vigorous root apparatus that allow them to better withstand pathogenic attacks caused by the parenchymatous parasites [74]. For example, reduction of Verticillium wilting by effective eggplant grafting on S. lycopersicum × S. habrochaites rootstock is associated to promotion of plant vigour [75]. Another mechanisms of disease resistance promoted by grafting is the release in the rhizosphere of exudates having antifungal activity on pathogen propagules. Liu et al. [76], for example, found root exudates from grafted eggplants, suppressive against the Verticillium dahliae mycelial development, contrarily to those released by the non-grafted ones. Grafting can affect root ability to harbour rhizosphere-competent microbes and to regulate beneficial antagonistic functions by modifying the root-architecture and the exudate profiling [77,78]. Rootstock/scion interaction can modulate regulations of transcripts and metabolites affecting disease susceptibility to fungal pathogens [79]. For instance, rootstock can influence disease resistance of scion by modifying secondary metabolites in the sap flow moving through the vascular system [80]. Shibuya et al. [81] attributed the transient reduction of disease symptoms on cucumber scion to changes in morphology or physiology in response to the modified water relations immediately after grafting onto squash. Defense mechanisms in grafted plants against soil-borne pathogens are summarized in Figure 1.
Agronomy 2020, 10, x FOR PEER REVIEW 7 of 12 infection to the entire vascular tissues of susceptible scion [72]. Graft-transmissible resistance to airborne disease has been observed for sweet pepper cultivar grafted on a resistant cherry pepper rootstock [73]. In this situation, the effects of rootstock on the scion is the major determinant of resistance. Disease resistance of grafted plants may also result from the enhanced vigour for which grafted plants grow more fortified and more efficient in resource utilization, as well as developing vigorous root apparatus that allow them to better withstand pathogenic attacks caused by the parenchymatous parasites [74]. For example, reduction of Verticillium wilting by effective eggplant grafting on S. lycopersicum × S. habrochaites rootstock is associated to promotion of plant vigour [75]. Another mechanisms of disease resistance promoted by grafting is the release in the rhizosphere of exudates having antifungal activity on pathogen propagules. Liu et al. [76], for example, found root exudates from grafted eggplants, suppressive against the Verticillium dahliae mycelial development, contrarily to those released by the non-grafted ones. Grafting can affect root ability to harbour rhizosphere-competent microbes and to regulate beneficial antagonistic functions by modifying the root-architecture and the exudate profiling [77,78]. Rootstock/scion interaction can modulate regulations of transcripts and metabolites affecting disease susceptibility to fungal pathogens [79]. For instance, rootstock can influence disease resistance of scion by modifying secondary metabolites in the sap flow moving through the vascular system [80]. Shibuya et al. [81] attributed the transient reduction of disease symptoms on cucumber scion to changes in morphology or physiology in response to the modified water relations immediately after grafting onto squash. Defense mechanisms in grafted plants against soil-borne pathogens are summarized in Figure 1. Experimental evidences suggested that the above listed mechanisms act synergistically in the regulation of the disease resistance of grafted roots. Greenhouse trials carried out by Song et al. [77] allowed to define as changes in root exudates from watermelon grafted on Lagenaria siceraria and Cucurbita pepo rootstocks, have promoted the non-host holobiont resistance to F. oxysporum f. sp. niveum infections through additional pathogen exclusion effects. In particular, niche chemical shifts around watermelon grafted roots affected the sheltered microbial diversity, increasing, for example, biocontrol-associated bacteria populations, such as Sphingobacteriia and Bacillus [77]. Biochemical investigations on watermelon/bottle gourd rhizodeposits have allowed to detect the ex-novo formation of bioactive proteins associated with plant resistance to biotic and abiotic stresses [82] and the release of chlorogenic and caffeic acids involved in the microbiostasis against propagules of the wilting causal agent [83]. Findings suggesting that the ungrafted watermelon delivery of molecular signals that are normally used as stimuli by pathogens harbouring the rhizosphere [84], with the grafting, likely, it is interrupted. In this light, combining grafting with environmentally friendly soil treatments (e.g., soil solarization) could slow down the recovery of pathogen recrudescence, often observed in conductive soils with the continuous monoculture cropping [84]. Structural variations of Experimental evidences suggested that the above listed mechanisms act synergistically in the regulation of the disease resistance of grafted roots. Greenhouse trials carried out by Song et al. [77] allowed to define as changes in root exudates from watermelon grafted on Lagenaria siceraria and Cucurbita pepo rootstocks, have promoted the non-host holobiont resistance to F. oxysporum f. sp. niveum infections through additional pathogen exclusion effects. In particular, niche chemical shifts around watermelon grafted roots affected the sheltered microbial diversity, increasing, for example, biocontrol-associated bacteria populations, such as Sphingobacteriia and Bacillus [77]. Biochemical investigations on watermelon/bottle gourd rhizodeposits have allowed to detect the ex-novo formation of bioactive proteins associated with plant resistance to biotic and abiotic stresses [82] and the release of chlorogenic and caffeic acids involved in the microbiostasis against propagules of the wilting causal agent [83]. Findings suggesting that the ungrafted watermelon delivery of molecular signals that are normally used as stimuli by pathogens harbouring the rhizosphere [84], with the grafting, likely, it is interrupted. In this light, combining grafting with environmentally friendly soil treatments (e.g., soil solarization) could slow down the recovery of pathogen recrudescence, often observed in conductive soils with the continuous monoculture cropping [84]. Structural variations of a root-associated microbiome after grafting have been also reported in open field tomato production, where Poudel et al. [6] showed rootstock-specific filtering effects driven by vigorous rooting, on endosphere and rhizosphere microbiomes enriched by representatives of Firmicutes, Verrucomicrobia, Planctomycetes, Proteobacteria and Acidobacteria. Under greenhouse, the use of Solanum torvum as rootstook for eggplant, induced an incremental shift in Bacteria and Actinomycetes on the root system, and in the responsive rhizosphere soil enzymes were found both associated to the pathogen population lowering and Verticillium wilting decrease [85].
Plant domestication privileged the selection of agronomic traits strictly related to the edibility, quantity and quality of the yields, leaving on the way all those alleles that gave to wild relatives the ability to profitably deal with soil microorganisms, reducing the potential of modern cultivars to lead rhizosphere microbiome assembly and functions [86]. In order, to bridge this gap, only recently some breeding programs are including, among the genetic improvement objectives, the enhanced ability of crops in recruiting beneficial host-specific root microbiota too [87]. In the meantime, horticultural experiences indicated that, at the moment, grafting better than others available technologies allows to reach this phenotype more quickly depending on the specific influence of rootstock genotype [84,88]. The metabolic structure of rootstock-recruited microorganisms, for example, has been implemented for managing tree resistance to the apple replant disease complex in replanted orchards [89]. Thereby, also vegetable productions may alleviate detrimental risks hidden into the intensive cropping systems following a holistic approach, in which grafting improves telluric environment adaptation of cultivar through the reinforcement of partnerships with beneficial and biodiverse microbiota and the enhancement of the efficient resource utilization. A summary of the possible directional interactions occurring in the soil environments shared by grafted roots and microbiome aiming at reducing pathogens pressure is reported in Table 1. Table 1. Possible directional interactions occurring in the soil between grafted roots and microbiome aiming at reducing pathogens pressure.
To Grafted Roots Microbiota
Grafted roots Allelopathy Competition for space, water and nutrients Selection-filtering effects C-food providing Activation of the endophytism Microbiota Nutrients availability Growth promotion Induction of resistance to (a)biotic stresses Antibiosis Hyperparassitism Niche occupying Essential food depletion Relationships-mediated by rootstock between the vegetable scion and the infected soil open new opportunities for disease risks management in a more environmentally friendly way (Table 2). Moreover, the integration of grafting technology with other agronomic, chemical and/or biological control means can allow to further reduce the risks of soilborne diseases especially in a long term. However, the use of chemical inputs must be carefully evaluated to avoid depression of beneficial indigenous soil microflora in the rootstock rizhosphere. Under these boosted agroecological conditions monitoring of plant and soil health status over time is crucial.
Conclusion and Future Prospects
Grafting represents a relatively recent innovation in the vegetable production systems of Western Countries; this advanced technology allowed to reduce the negative impact of soilborne pathogens in solanaceous and cucurbit crops depending on the disease resistance level of rootstock. Grafting can enhance plant disease resistance through several multiple defense mechanisms in plant. Recent studies demonstrated that rootstock-mediated effects on rhizosphere microbiome can contribute to reduce the soilborne diseases by stimulating indigenous microflora able to compete for food and space with plant pathogens and to reduce their activity through antibiosis and hyperparasitism. Moreover, plant-beneficial microbes recruited by rootstock can enhance soilborne disease resistance indirectly through the increase of plant nutrient availability, the stimulation of plant growth and the induction of resistance to biotic stresses. Despite experimental evidences demonstrated a significant role of rhizosphere microbiome in enhancing soilborne disease resistance in grafted plants, more studies are necessary to better understand the scion-rootstock-rhizosphere microbiome interaction under different environmental conditions. Advanced technologies like metagenomics can help to identify and characterize the microbial strains in the rhizosphere of grafting combinations in order to link the changes in rhizosphere microbial community to enhanced plant resistance to specific soilborne pathogens. This knowledge could allow to develop new disease control strategies based on the combined application of selected microbial inoculants and specific grafting combinations.
Author Contributions: Conceptualization, M.C. and C.P.; writing-original draft preparation, M.C. and C.P.; writing-review and editing, M.C., Y.R., M.C.K., G.C. and C.P. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding. | 2020-08-20T10:09:30.905Z | 2020-08-13T00:00:00.000 | {
"year": 2020,
"sha1": "48fabb5dd43c409777b6ed3ec0bbfbf2abd6ba73",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/2073-4395/10/8/1185/pdf?version=1597303423",
"oa_status": "GOLD",
"pdf_src": "ScienceParsePlus",
"pdf_hash": "e52c95b0b7e3df0404ae07fb11b181ad67872927",
"s2fieldsofstudy": [
"Agricultural And Food Sciences"
],
"extfieldsofstudy": [
"Biology"
]
} |
18529423 | pes2o/s2orc | v3-fos-license | Effect of hydroxyethyl starch on blood glucose levels
Background Hydroxyethyl starch (HES), a commonly used resuscitation fluid, has the property to induce hyperglycemia as it contains large ethyl starch, which can be metabolized to produce glucose. We evaluated the effect of 6% HES-130 on the blood glucose levels in non-diabetic patients undergoing surgery under spinal anesthesia. Methods Patients scheduled to undergo elective lower limb surgery were enrolled. Fifty-eight patients were divided into two groups according to the type of the main intravascular fluid used before spinal anesthesia (Group LR: lactated Ringer's solution, n = 30 vs. Group HES: 6% hydroxyethyl starch 130/0.4, n = 28). Blood glucose levels were measured at the following time points: 0 (baseline), 20 min (T1), 1 h (T2), 2 h (T3), 4 h (T4), and 6 h (T6). Results Mean blood glucose levels at T5 in the LR group and T4, T5 in the HES group, increased significantly compared to baseline. There were no significant changes in the serial differences of mean blood glucose levels from baseline between the two groups. Conclusions Administration of 6% HES-130 increased blood glucose levels within the physiologic limits, but the degree of glucose increase was not greater than that caused by administration of lactated Ringer's solution. In conclusion, we did not find evidence that 6% HES-130 induces hyperglycemia in non-diabetic patients.
Jung et al.
and serious adverse clinical outcomes [10]. There are few publications describing the possible effects of HES on blood glucose level and these include studies suggesting that 6% HES-450 was able to trigger changes in glycemia [8,11].
We hypothesized that the administration of 6% HES-130 is also able to increase blood glucose level. For this reason, we examined whether 6% HES-130, which has a lower molecular weight and lower degree of substitution than the 6% HES-450, increase blood glucose level in non-diabetic patients undergoing lower limb surgery under spinal anesthesia. In addition, we compared the serial differences of mean blood glucose levels from baseline between 6% HES-130 and lactated Ringer's solution.
Materials and Methods
After approval by the Institutional Review Board, written informed consents were obtained from all patients after the study was carefully explained. A total of 60 patients who were scheduled to undergo elective lower limb surgery with spinal anesthesia were enrolled. Only patients who belonged to the American Society of Anesthesiologists (ASA) class I or II and were aged between 30 and 80 years, with a body weight between 40 and 75 kg, were included in the study [8]. Patients with diabetes mellitus, taking hyperglycemic drugs (acetaminophen, ascorbic acid, steroids, etc.), with possible allergies to experimental drugs (especially with an allergy to corn), coagulation disorders, renal or cardiac dysfunctions, and suspected hypervolemia, including pulmonary edema, were excluded from the study.
All patients were premedicated with 0.05 mg/kg midazolam, intramuscularly, 30 min prior to the induction of anesthesia. An intravenous access was achieved with an 18 gauge intravenous cannula and the patients were transported to the operating room. Basal monitoring devices such as electrocardiogram, pulse oximetry, and non-invasive arterial pressure were attached to the patients in the operating room, and blood was sampled from the fingertip for testing blood glucose level using glucometer (Accu-Check Inform II METER with RF card, Roche Diagnostics GmbH, Germany). The reading obtained at this point was considered as the baseline (T0). After recording of basal glucose level, calculated doses of intravenous fluid were administered according to the allocated group for 20 min before anesthesia. Then, the fluids were changed with physiologic saline as subsequent intravenous fluid until the end of the study. Together with blood glucose level, the heart rate (HR) and the mean blood pressure (MBP) were also measured. A total of six measurements were obtained: T0, baseline before administration of fluid; T1, 20 minutes after the termination of the designated fluid administration; T2, T3, T4, and T5 representing 1, 2, 4, and 6 hours after the termination of the designated fluid administration. After the blood glucose values were obtained, the mean blood glucose levels and serial differences between mean blood glucose level and baseline were calculated.
Patients' characteristics, fasting time, types of operation, operation time, the dosage of fluids administered during surgery, estimated blood loss, as well as the use of phenylephrine and total dosages, were recorded.
Statistical analysis
Sample size was calculated using the G*Power3 free software (available at http://www.gpower.hhu.de). The effect size was assumed according to the Cohen's guidelines for the social sciences and we selected a medium effect size of 0.3 [12]. Using α = 0.05 with a power of 80%, the total sample size was calculated as 54 for the statistical analysis of six consecutive time points measurement of blood glucose level with repeated measures ANOVA between different factors with a correlation of 0.5. After assuming a 10% drop out rate, 30 patients were allocated to each group.
Data were analyzed using SPSS (Windows ver. 12.0, SPSS Inc., Chicago, IL, USA) and all the results are presented as mean ± standard deviation (SD). Statistical analysis was performed after the normality assessment with Kolmogorov-Smirnov test. A Student's t-test was performed to evaluate statistical significance between the two groups for age, height, body mass index (BMI), the dosage of administered fluids during surgery, and fasting time. The non-parametrically distributed weight, operation time, estimated blood loss, and total dosage of phenylephrine used were analyzed by the Mann-Whitney U test. Fisher's exact test was performed for the analyses of gender, ASA score, type of operation, and the use of inotropics, which were non-parametrically distributed.
Two-way ANOVA with repeated measures for one factor was performed to compare the differences of MBP, HR, and blood glucose level between the groups. Values with P < 0.05 were considered statistically significant. Post hoc analysis was done with a t-test. Bonferroni's correction was applied and values with P < 0.0083 were considered statistically significant between the groups.
Results
A total of 67 patients were assessed for eligibility, and 60 patients decided to participate. A total of 58 patients were enrolled VOL. 69, NO. 4, AUGUST 2016 HES and blood glucose level and examined. Two patients were excluded as general anesthesia replaced the spinal anesthesia due to an unexpected prolongation of surgery (Fig. 1).
There were no significant differences in the patients' characteristics between the two groups (Table 1).
Mean blood glucose levels at T5 in the LR group and T4, T5 in the HES group increased significantly compared to baseline (Fig. 2).
Because the glucose levels of T0 were different between the two groups, and the mean blood glucose levels in the LR group were higher than those of the HES group at all 6 time-points, we performed a multilevel mixed model analysis and the results showed a significance value of 0.879. Thus, we assumed the observed changes of glucose levels over time were not significantly different between the groups. Since the significance value of Mauchly's test was 0.000 and Greenhouse-Geisser estimate of sphericity (ε) was 0.617, we used Greenhouse-Geisser test for sphericity correction. There were no significant differences in the serial differences between mean blood glucose levels and baseline after fluid preloading between the two groups (P = 0.908, Table 2).
Discussion
In this study, we assessed changes in blood glucose levels after the administration of 6% HES-130 and lactated Ringer's solution during lower limb surgery in non-diabetic patients. We found that there are no significant differences in the serial differences of mean blood glucose levels from baseline between the administration of 6% HES-130 and lactated Ringer's solution.
As mentioned earlier, the structure of HES is similar to that of glycogen as it contains conformationally different polysaccharides; therefore, it is metabolized by serum amylases [7]. The original form of starch is not used as volume expander because it is hydrolyzed fast by serum amylases [13]. HES is made by substituting hydroxyl residues of starch with hydroxyethyl residues in order to delay the hydrolysis by serum amylases [14].
Especially, the inhibition of plasma amylase and the consequent increase in the intravascular half-life of the HES solution is more prominent when the hydroxyethyl residues are bound to the C 2 position of glucose [14]. Thus, HES has the potential to increase blood glucose level after administration.
HES metabolism has been studied on animals and it was found that HES infusions did not change blood glucose levels [15,16]. Because HES metabolism results in the production of increasingly smaller molecules (down to about 40,000 to 50,000 D), which are excreted in urine, degraded molecules are rapidly excreted without further metabolism and glucose formation [17]. It is known that the initial rapid amylase-dependent break-down occurs immediately after HES infusion and within the first 24 h, 50% of the administered dose is excreted by the kidneys [14]. These might be the reasons why HES does not alter blood glucose levels significantly.
However, other studies reported that increased blood glucose levels were observed after the administration of HES in noninsulin-dependent diabetics; indeed, Murty et al. [8] reported that HES increased blood glucose levels significantly compared to lactated Ringer's solution. They also demonstrated that the rise in blood glucose levels started within fifteen minutes of HES administration, and the peak was reached after two hours of administration. Patki and Shelgaonkar [11] reported a sustained and statistically significant rise in blood glucose levels after HES administration. Our study demonstrated a statistically significant rise in blood sugar levels (still within the physiological HES and blood glucose level limits) from the baseline to 4 h after terminating HES administration. In addition, serial differences of mean blood glucose levels from baseline showed no differences between the HES and the LR groups. This discordance might be explained by the use of different HES solutions. In previous studies, 6% HES-450 was used as oppose to 6% HES 130/0.4, used in our study [8,11]. Compared to 6% HES-450, 6% HES 130/0.4 has a lower molecular weight and lower degree of substitution, and thus provides less plasma accumulation because of the shorter half-life and quick hydrolysis by serum α-amylase [17,18]. Less plasma accumulation and rapid elimination due to the lower molecular weight might play a role in diminishing the effect of HES on the changes of glucose levels. Moreover, Volulyte 6% TM contains acetate, which is quickly converted to acetyl-CoA by the enzyme acetyl-CoA synthetase [19]. Previous reports showed that intra-venous administration of sodium acetate had a large suppressive effect on fat and carbohydrate utilization and decreased plasma glucose concentration [19,20]. Hence, it is possible that the acetate contained in the HES solution used in the study suppressed the rise of glucose levels.
There are limitations to this study. Firstly, there were significant differences in the basal glucose levels between the two groups. Thus, it is possible that the patients' characteristics might be different between two groups, hence affecting the results of our study. However, there were no statistical differences in age, gender, BMI, fasting time, operation time, input, estimated blood loss, etc. The mean age was higher in the LR group than in the HES group (54.5 vs. 49.6). The prevalence of impaired fasting glucose (IFG) shows a tendency to increase with age after 40 years. Overall, 28.4% of patients > 50 years of age, had IFG whereas only 20.0% of subjects < 50 years of age had an abnormal fasting blood glucose level [21]. Based on this study, we would expect basal glucose levels to be higher in the LR group compared to the HES group. Therefore, we applied a multilevel mixed model and Greenhouse-Geisser test for the sphericity correction. According to the results with applied multilevel mixed model and sphericity correction, we calculated that there were no significant differences in the changes of glucose levels according to time, between the two groups. Secondly, the stress of surgery itself can produce hyperglycemia via catecholamine release [9]. However, local anesthesia can abolish adrenocortical and glycemic responses to surgery by blocking both the afferent input from the operative site to the central nervous system, and the efferent autonomic neuronal pathways to the liver and adrenal medulla [9]. In this study, all patients had spinal anesthesia and the surgery duration was relatively short with minimal blood loss. Therefore, most of the confounding factors | 2018-04-03T05:09:42.003Z | 2016-06-22T00:00:00.000 | {
"year": 2016,
"sha1": "22a1cc9d9f528a8f9bda1c20fd971d30ccb86db2",
"oa_license": "CCBYNC",
"oa_url": "http://ekja.org/upload/pdf/kjae-69-350.pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "22a1cc9d9f528a8f9bda1c20fd971d30ccb86db2",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
25458936 | pes2o/s2orc | v3-fos-license | Energetics of Transport through the Nuclear Pore Complex
Molecular transport across the nuclear envelope in eukaryotic cells is solely controlled by the nuclear pore complex (NPC). The NPC provides two types of nucleocytoplasmic transport: passive diffusion of small molecules and active chaperon-mediated translocation of large molecules. It has been shown that the interaction between intrinsically disordered proteins that line the central channel of the NPC and the transporting cargoes is the determining factor, but the exact mechanism of transport is yet unknown. Here, we use coarse-grained molecular dynamics simulations to quantify the energy barrier that has to be overcome for molecules to pass through the NPC. We focus on two aspects of transport. First, the passive transport of model cargo molecules with different sizes is studied and the size selectivity feature of the NPC is investigated. Our results show that the transport probability of cargoes is significantly reduced when they are larger than ∼5 nm in diameter. Secondly, we show that incorporating hydrophobic binding spots on the surface of the cargo effectively decreases the energy barrier of the pore. Finally, a simple transport model is proposed which characterizes the energy barrier of the NPC as a function of diameter and hydrophobicity of the transporting particles.
Introduction
Molecular transport between the cytoplasm and the nucleoplasm is gated by highly selective nuclear pore complexes (NPCs). The NPC is embodied in the nuclear envelope membranes and provides bidirectional pathways for passive transport of small molecules and active (facilitated) transport of larger molecules [1][2][3][4]. Small molecules and ions are passively transported through the pore by free diffusion, but large molecules are barred to enter the pore. The transport of large macromolecules requires binding to soluble transport factors, generally known as karyopherins or Kaps. During active import or export the Kap binds to cargo with a nuclear import or export signal, upon which the Kap-cargo complex is translocated through the NPC.
The NPC of yeast, as a model system, is composed of approximately 30 different proteins called nucleoporins (Nups) which are arranged with an eight fold-symmetrical pattern inside the pore [5,6]. The Nups that line the central channel of the pore have been found to be intrinsically disordered and contain many phenylalanine-glycine (FG) repeats [7]. These FG-Nups are essential for the viability of yeast and presumably all eukaryotes but their function in regulating the active and passive transport is not yet understood [8]. Several models have been proposed to explain the role of FG-Nups and Kaps during transport but no consensus has been reached so far on a prevailing model [3,[9][10][11][12][13][14].
The permeability barrier of the NPC has been characterized by studying the passive transport of a wide range of inert molecules of different size. Single molecule studies have revealed that cargoes up to 29 kDa can smoothly diffuse through the pore, while transport of cargoes larger than 61 kDa is prohibited [15]. Early experiments have estimated a diameter of 4. 5-5.4 nm for the diffusion channel of the NPC [16,17], which was refined to a diameter of % 5.3 nm in more recent experiments [18]. In addition to the size, the shape of the transporting species has been shown to be an important factor in passive transport. It has been observed that elongated non-spherical cargoes diffuse faster than spherical ones with the same mass [18]. Passive and active transport pathways have been widely discussed in the literature [12,15,18,19]. Several studies have suggested that passive and active transport take place through different spatially-separated pathways across the pore [15,19,20]. Recently, it has been demonstrated that passive transport mostly occurs through the central region of the pore [15,21].
Mutational analyses of Kaps suggest that the interaction between Kaps and FG-repeats is necessary for active transport [22,23]. It has been shown that Kaps have a greater surface hydrophobicity compared to other cytoplasmic proteins [24]. In addition, structural analyses have demonstrated that, during interaction, hydrophobic side chains of the FG-Nups closely interact with hydrophobic pockets on the surface of the Kap [22,25]. Naim and co-workers [26] have shown that a cargo that is normally blocked, can be triggered to enter by modification of its surface with hydrophobic amino acids. This suggests that a certain amount of surface hydrophobicity is necessary for cargo to actively translocate through the pore. Molecular dynamics simulations and experiments have revealed several binding spots on the surface of Kaps [27][28][29][30]. Recently, the effect of electrostatic interactions on active and passive transport has been studied through a high resolution fluorescence microscopy technique [31]. The results of this experiment suggest that electrostatic interactions are less important than hydrophobic interactions in nuclear transport through the NPC.
In addition to experimental studies, several theoretical and computational studies have been conducted to elucidate the mechanism of nuclear transport. These studies include investigations of single FG-Nups [12,32], their collective behavior in brush-like structures [33,34] and in the transport channel of the NPC [35][36][37] as well as modeling transport through the nuclear pore [38][39][40][41]. Regarding nuclear transport, Mincer et al. [38] have used a super-resolution approach which treats FG-Nups as flexible filaments with various binding spots to serve as FGrepeats. Even though many simplifying assumptions are made, the model can qualitatively predict several aspects of transport. Coarse-grained Brownian dynamics simulations have shown that the Kap-cargo complex interacts with a layer of FG-Nups formed close to the channel wall [39]. However, the specific amino acid sequence of the FG-Nups was not accounted for in these simulations. The density distribution of FG-Nups and colloidal particles inside a cylindrical axi-symmetric structure, mimicking the NPC, have been studied through a classical density functional theory approach [41]. The model provides valuable insight into several aspects of nuclear transport including the crowding effect of the cargoes inside the pore. The energetics of translocation of a model cargo through the pore has been investigated by Tagliazucchi and coworkers [40] using a theoretical model that accounts for different FG-Nups by distinguishing six families of amino acids. Their results suggest that the transporting cargo experiences an energy barrier at the center of the pore which is lowered through hydrophobic and electrostatic interactions. However, the size selectivity of the NPC during passive transport and the effect of binding spot distribution on active transport have not been investigated.
We have previously investigated the distribution of disordered FG-Nups in the transfer conduit of the NPC by means of coarse-grained implicit solvent molecular dynamics sumulations [37]. The developed one-bead-per-amino-acid model distinguishes between all 20 amino acids of the FG-Nups and takes into account hydrophobic and electrostatic interactions between the amino acids, the backbone stiffness of the Nups as well as the screening effect of free ions and polarity of the solvent through a modified Coulomb equation. The model has been calibrated against experimental Stokes radii of a wide range of FG-Nup segments [12]. Since the model is parametrized against the equilibrium properties of FG-Nups, it cannot be directly used to study kinetic aspects of nuclear transport. The goal of the present work is to use this model to study the energetics of passive and active transport through the disordered domain of the nuclear pore complex. We examine the size selectivity of the NPC by calculating the energy barrier for passively transporting cargoes. In addition, we elucidate the active transport mechanism by studying how the number of binding spots and their spacing on the surface of model Kaps affect the energy barrier.
Methods
Molecular dynamics simulations are performed using a one-bead-per-amino acid coarsegrained (CG) model [37]. The distance between neighboring beads is fixed at 0.38 nm using a harmonic potential and an average mass of 120 Da is assigned to each CG bead. The backbone stiffness of the FG-Nups is controlled through bending and torsion potentials extracted from the Ramachandran data of the coiled regions of protein structures [42]. The Gromacs molecular dynamics software [43] is used to perform Langevin dynamics simulations. The temperature is set to 300 K and the cut-off distance for Van der Waals and Coulombic interactions are set to 2.5 nm and 5.0 nm, respectively. A time-step of 0.02 ps is chosen and the Langevin friction coefficient is set to 50 ps −1 which is similar to the collision frequency of water molecules [44]. For details of the model the reader is referred to Ghavami et al. [37].
A simplified geometrical model of the NPC is built based on the geometry of the core scaffold of the yeast NPC and the FG-Nups are anchored at the predicted positions inside the pore [45,46]. The scaffold is modeled using hard-sphere beads with a diameter of 5.0 nm which are assumed to have no interaction with the FG-nups (see Fig 1A).
Passively transporting cargoes are modeled as rigid spheres with diameter D. Inert cargoes smaller than 5.0 nm in diameter are modeled as a single neutral bead with a purely repulsive potential at distances smaller than σ c = D/2+r b , where r b = 0.3 nm is the radius of the CG-beads for the FG-Nups (see Fig 1B). Cargoes larger than 5.0 nm in diameter are constructed with a number of neutral beads each having a repulsion distance of σ c = 2.5 nm as shown in Fig 1C. These overlaping beads are centered at a distance of r c = D/2 − (σ c − r b ) from the center of the composite cargo. A schematic representation of the inert cargoes used in this work is presented in Fig 1E. Kaps are elongated boat-like proteins with FG-Nup binding spots on their convex surface [3,13], while their concave surface is used to attach to the cargo. In view of this, the Kap-cargo complex is modeled as a rigid sphere of D = 10 nm with several binding spots on its surface. Since binding spots and FG-repeats are reported to have similar affinities with FG-Nups [47], the binding spots are represented by hydrophobic beads that are similar to Phenylalanin (F) amino acids (with a maximum interaction energy of −5.2 kJ.mol −1 ) and are distributed on the surface of the sphere along a stripe at a spacing d (see Fig 1D and 1E) [27].
We use umbrella sampling to calculate the potential of mean force (PMF) of the transporting cargoes [48]. The PMF is the effective potential that a particle experiences at a certain position due to the presence of all other particles, averaged over all conformations of the system. In the umbrella sampling method, the reaction coordinate is subdivided into several overlapping windows. The system is then simulated in the presence of a bias potential to enhance the sampling in each window. Ultimately, the information from the separate simulations are unbiased and recombined to obtain the PMF along the reaction coordinate (see the S1 Text).
Results and Discussions
Size selectivity of the NPC First, the PMF curve along the central axis of the NPC is calculated for inert cargoes of different size as shown in Fig 2. In all cases the PMF curve increases as the cargo relocates from its starting position inside the cytoplasm (z = 27 nm) to the center of the NPC and a peak value is observed near the central plane of the NPC. This continuous increase in PMF already starts inside the cytoplasm (i.e., z > 15.4 nm, see Fig 1), which indicates that some of the FG-Nups extend to the cytoplasmic environment and repel large cargo molecules by their entropic motion. This is consistent with our previous study showing that the FG-Nups located at the peripheries of the NPC have more conformational freedom and are spread over a larger volume compared to the FG-Nups located near the symmetry plane of the NPC [37]. In order to study the size selectivity of the NPC, the free energy barrier corresponding to different cargo diameters is calculated and plotted in Fig 3. The energy barrier is defined as the difference between the mean PMF at -5.0 nm < z <5.0 nm and 20 nm < z <27 nm and is an indication of the work required to translocate the cargo from the cytoplasm to the core of the NPC. The results show that the energy barrier of the NPC for passively transporting cargoes decreases as the diameter of the cargo decreases. The size selectivity threshold of the pore is calculated by comparing the energy barrier of the pore with the thermal energy k B T. If the energy barrier experienced by the cargo is larger than k B T, the probability that a cargo passes through the pore decreases. Thus, by defining k B T as a soft limit for transport, we find that cargoes larger than D = 5.0 nm have a small probability to pass through. This is within the experimental estimates in the range 4.5-5.4 nm for the size selectivity of the NPC [16][17][18]. A comparison between the energy barrier of the wildtype and a minimal viable NPC for a cargo with D = 10 nm is made in Fig 3. The minimal viable NPC corresponds to an NPC that has the least amount of FG-nups according to Strawn et al. [8], yet is viable. In the minimal viable NPC Nup42, Nup159, Nup1, Nup60, Nup100, Nsp1 and Nup145 are removed (see [37]). The results indicate that the minimal viable NPC is also able to screen large non-specific cargoes from entering the pore. However, the energy barrier has decreased by 22% in the minimal NPC compared to the wildtype NPC, which is probably due to the removal of half of the mass of the FG-Nups in the minimal viable NPC.
In order to gain more understanding on the energy barrier of the NPC, the density distribution of the FG-Nups is studied in the presence of a cargo with diameter of D = 10 nm at different vertical distances from the central plane of the NPC (see Fig 4). The two-dimensional density plots are obtained by averaging the density distribution of the FG-Nups in the circumferential direction for the umbrella simulations [37]. The results show that once the cargo approaches the center of the NPC, it has to push the high density FG-Nup region aside in order to pass through the pore. This will result in a high energetic penalty for the cargo to pass through the central region of pore, resulting in a total energy barrier of 10.5 kJ/mole (Fig 2).
Possible pathways
The calculated PMF curves in Fig 2 are obtained along the central axis of the pore (r = 0). In order to check whether the lowest energy route for passive transport passes through the center, radial PMF curves have been calculated. This is done by comparing radial free energy profiles for a cargo with D = 10 nm at the maximum point of the axial PMF curve (i.e., z = −2.5 nm, see Fig 2). The radial energy profiles are obtained in four different directions, θ = 0°, 90°, 180°and 270°(see Figs 1 and 5).
The difference between the PMF curves can be rationalized by the non-uniform distribution of the FG-Nups in the pore (see [37]). For all orientations the lowest energy pathway for an inert cargo with D = 10 nm passes through the center of the pore. This is consistent with single molecule tracking experiments which suggest that passively transporting particles pass through the center [15].
Lowering the barrier
The energetics of active transport is investigated by calculating PMF curves for Kap-cargo complexes through the central axis of the NPC (r = 0 nm). The Kap-cargo complex is modeled as a sphere of diameter D = 10 nm with 7 binding spots on its surface, but at different spacings d (see Figs 1 and 6). When the binding spots are spaced at d = 4.94 nm, Fig 6 shows that a reduction is observed in the energy barrier of the NPC from 10.5 kJ/mol in the absence of any binding spots (cf. Fig 2) to 7.3 kJ/mol. However, this reduction is not enough for transport of the model Kapcargo complex. When the spacing between the binding spots is further decreased to 2.5 nm, the energy barrier does not show a considerable difference from d = 4.9 nm. However, when the spacing is decreased to d = 1.3 nm, the obtained PMF curve shows a large reduction of the energy barrier to the order of k B T (see Fig 6). These trends are consistent with the experimental findings of Naim and co-workers [26], who showed that large inert cargoes were able to transport through the pore when hydrophobic amino acid side chains were attached to the surface of the cargo.
In addition, the critical spacing of 1.3 nm is in close agreement to the reported values of d = 1.1 ± 0.3 nm and d = 1.4 ± 0.3 nm for the distance between binding spots on the surface of Importin-β and NTF2, respectively [27]. For n = 7, the energy barrier is on the order of k B T. Interestingly, by increasing the number of binding spots to n = 11, the barrier completely disappears and the NPC forms a potential well for the Kap-cargo complex. In this case the complex is attracted towards the pore and tends to stay in the central region of the NPC. Clearly, the probability for transport would be strongly reduced compared to a complex with 7 binding spots. The energy barrier versus the number of binding spots n is plotted in the inset of Fig 7. The results suggest that there is an optimum number of binding spots for efficient transport of a Kap-cargo complex through the pore. It must be noted, however, that our simulations are for a cargo with D = 10 nm and the optimum number of binding spots might be different for a cargo with different size.
Transport model
Earlier theoretical and experimental studies propose that the free energy of insertion of particles into grafted polymer brush scales with either the volume (in high grafting densities or bad solvent conditions) or the surface area of the particle (in low grafting density or good solvent conditions) [49][50][51][52]. Since the density of the FG-nups is low in the central channel of the NPC, the free energy penalty for passage of cargoes scales with their surface area G * D 2 (see Fig 3). Upon addition of binding spots on the surface of the cargo, the free energy barrier will be reduced by an attractive energy gain [53] proportional to the contact area of the cargo and the FG-nups. Based on these considerations, a simple transport model is proposed, which describes the energy barrier of the NPC, G, as a function of the diameter and hydrophobicity of the Kapcargo complex, Using the proposed model, the energy barrier of the NPC is characterized for Kap-cargo complexes with different sizes and number of hydrophobic binding spots in Fig 8. The calculated energy map suggests that efficient transport occurs in a strip confined between two isolines of +k B T and −k B T which is shown as a gray area in Fig 8. The region below the +k B T line corresponds to the situation in which the number of hydrophobic bindings spots is too small to reduce the free energy barrier enough for transport to be possible. The area above the −k B T line represents a state in which the NPC turns into an energy well due to the presence of a large number of hydrophobic binding spots on the cargo. These spots have a high affinity to the FGnups which results into entrapment of the cargo inside the pore.
Conclusion
We have characterized the selective permeability barrier of the NPC by studying the energetics of transport through the pore. All 128 FG-Nups of the yeast NPC are accounted for in a onebead-per-amino acid coarse-grained model in which the FG-Nup amino acid sequence is explicitly represented [37]. The energy barrier for transport of model cargoes has been studied by calculating PMF curves through umbrella sampling. Our results indicate that the disordered FG-Nups inside the NPC form an energy barrier along the central axis of the NPC that does not allow the passage of inert cargo molecules larger than D * 5 nm in diameter. However, the PMF curves of Kap-cargo complexes show that the attachment of several hydrophobic binding spots to the surface of the cargo complex lowers the energy barrier below k B T, facilitating the transport of large cargo molecules. In addition, the effect of surface hydrophobicity and spacing between binding spots on active transport has been analyzed. Our results show that in addition to the number of hydrophobic spots the spacing between binding spots is a key feature in facilitated transport through the NPC. Additionally, we show that there is an optimal number of hydrophobic binding spots for efficient transport of Kap-cargo complexes of a certain size. Depending on the number and spacing of binding spots, a cargo can be expelled from, transported through or trapped inside the pore.
Supporting Information S1 Text. The umbrella sampling procedure. | 2018-04-03T00:58:31.527Z | 2016-02-19T00:00:00.000 | {
"year": 2016,
"sha1": "479daee7eb3395c564e340e12710dd086b2b6bf4",
"oa_license": "CCBY",
"oa_url": "https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0148876&type=printable",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "479daee7eb3395c564e340e12710dd086b2b6bf4",
"s2fieldsofstudy": [
"Biology",
"Physics"
],
"extfieldsofstudy": [
"Chemistry",
"Medicine"
]
} |
245250833 | pes2o/s2orc | v3-fos-license | Variability of Sagittal Plane Proximal Tibial Morphology and Its Effect on Stem Placement in Total Knee Arthroplasty
The tibial joint line is offset posteriorly relative to the tibial sagittal anatomic axis. This can have consequences when using stemmed implants during total knee arthroplasty. We retrospectively analyzed native knee lateral radiographs in 100 patients. The distance between the sagittal anatomic axis and the center of a simulated tibial resection was calculated as a percentage of overall tibial width. Analysis of 5 manufacturers’ baseplates showed that the tibial stem attached on average 10% anterior to the midline. We measured the impingement point of a 12-mm-diameter stem starting from this position. The tibial joint surface was offset posteriorly from the anatomic axis in all patients by an average of 23.5% of the tibial width (range: 13.1%-33.2%). A 12-mm tibial stem would impinge within 40 mm in 2% (2/100) of patients and within 60 mm in 19% (19/100). There was a weak but statistically significant correlation between proximal tibial offset and distance to impingement. During total knee arthroplasty, the center of the cut tibia is offset posteriorly from the sagittal anatomic axis. In patients with high offset, tibial stem extensions can impinge against the posterior tibia, causing baseplate malpositioning, diminished cement mantle, or fracture.
Introduction
Tibial stems are an important tool for complex total knee arthroplasty (TKA). Stem extensions are routinely utilized in revision TKA when proximal tibial bone is compromised as they significantly reduce compressive and shear forces on proximal tibial cancellous bone [1]. They are increasingly utilized in obese patients [2] to minimize the risk of catastrophic varus collapse associated with unstemmed implants [3]. Stemmed implants can also improve stability in patients with osteopenia or severe coronal plane deformities [4]. In revision surgery, outcomes are similar when either cementless or cemented stems are used in aseptic revision TKA [5]. However, because it is important for cementless stems to engage the diaphysis for success, variation in tibial bone morphology must be recognized in preoperative planning.
It is well recognized by cementless stem users that valgus bowing of the tibia in the coronal plane requires recognition and technical judgment to avoid an insertional fracture or malpositioning at the joint line [1]. In such situations, offset stems can be used to avoid medial overhang if a canal-filling, diaphysealengaging cementless stem is used. Alternatively, a shorter cemented construct could be utilized to allow a nonecanal-filling technique and avoid medial overhang. In either construct, recognition of tibial bone abnormalities is essential for avoiding insertional complications. While this valgus bowing of the tibia is commonly recognized by surgeons treating complex problems requiring stem extensions, the sagittal plane morphology of the proximal tibia is also an important consideration during the placement of a stemmed TKA implant. The impact of this variability in sagittal plane morphology has not been previously described.
The tibial diaphysis is triangularly shaped and relatively short, generally extending for approximately 6 cm [2]. While the anatomic axis is consistently centered over the ankle in the sagittal One or more of the authors of this paper have disclosed potential or pertinent conflicts of interest, which may include receipt of payment, either direct or indirect, institutional support, or association with an entity in the biomedical field whichmay be perceived to have potential conflict of interest with this work. For full disclosure statements refer to https://doi.org/10.1016/j.artd.2021. 10.018. plane [3], this is not the case at the knee where the sagittal joint center is posterior to the anatomic axis. This mismatch is one reason why Herzog curves are incorporated into the design of intramedullary nails utilized for fixation of tibia fractures [4]. This posterior offset of the proximal tibia relative to the diaphysis can influence the trajectory of a tibial stem in a similar fashion to an intramedullary nail. In patients with significant posterior offset at the joint line, this can cause the stem to impinge on the posterior tibial cortex, potentially leading to an insertional fracture. In cases where a cemented stem is used, it can lead to a diminished posterior cement mantle ( Figs. 1 and 2). When an uncemented diaphyseal-engaging stem is used, it can lead to significant anterior overhang ( Fig. 2a and b). Adding posterior slope to the tibial cut can mitigate some of these problems, although such an adjustment can effect flexion stability. Another strategy to avoid anterior overhang with diaphyseal-engaging press-fit stems is the use of an offset stem. Recognition of this problematic morphology preoperatively would allow surgeons to plan appropriately to address it.
In our practice we have noticed several cases where a long tibial stem contacted the posterior cortex during a complex primary or a revision TKA. In this study, we sought to define the normal amount of posterior offset of the tibial center at the cut surface for a TKA relative to the diaphyseal anatomic axis. We also sought to define the amount of posterior proximal tibial offset which would cause issues with impingement on the posterior cortex. Our hypothesis was that the distance between the anatomic axis of the tibia and the joint center would be clinically insignificant in most patients but that in a certain subset of patients, posterior offset in the sagittal plane would result in posterior cortical impingement with potential negative consequences when using a stemmed tibial implant.
Material and methods
We reviewed a series of consecutive patients presenting to a single arthroplasty surgeon's clinic (TKF) with a primary complaint of knee pain during 2019-2020. Patients were included if they had an appropriate lateral knee radiograph available for review. Radiographs had to be performed before TKA, include a marker ball, and have appropriate rotation. Patients with abnormal fibular overlap with the proximal tibia or malrotation of the femoral condyles on the lateral radiograph were excluded. Patients were also excluded if they had any history of tibial fracture or tibial osteotomy or any known neuromuscular or metabolic bone disorders.
Posterior proximal tibial offset was calculated for each patient as the difference between the diaphyseal sagittal plane anatomic axis of the tibia and the center of the tibia at the cut surface for a standard TKA ( Fig. 3 and b). All measurements were obtained using TraumaCad® software. The sagittal diaphyseal anatomic axis was estimated by a line bisecting the center of the tibia at the most distal point visible on the lateral knee radiograph and the distal extent of the tibial tubercle (line CD). A line was then drawn to simulate a tibial resection which was perpendicular to the anatomic axis and intersected the posterior-superior apex of the tibia (line AB). We simulated a resection which would be perpendicular to the mechanical axis of the tibia because revision TKA incorporating long tibial stems is most often used with posterior-stabilized components designed to be implanted with neutral tibial slope [5]. This was intended to simulate the minimum resection necessary to achieve a flat surface on which to seat a tibial baseplate on the proximal tibia. The center point of this cut surface was defined (line EF), and the percentage of posterior tibial offset was calculated as the difference between the anatomic axis and the center of the cut tibial surface as a percentage of overall tibial width (CE/AB).
Tibial baseplates for TKA systems which incorporate long stems were obtained from 5 companies. These stems were analyzed photographically to determine the position of the center of the stem in the sagittal plane. This was defined as the distance from the most anterior point of the baseplate to the center of the stem housing as a percentage of the total width of the baseplate in the Table 1 depicts these photographic measurements as well as the stem options for the respective companies.
The average position of the center of the tibial baseplate in the sagittal plane for those 5 companies was 10% anterior to the midline, or at the 40% mark in the anterior-posterior plane ( Table 1). Based on this, we measured the point at which a 12-mm tibial stem would impinge against the posterior endosteal surface of the tibia if stems of varying lengths were used. We did this by extending a line parallel to the sagittal anatomic axis from a point 6 mm posterior to the 40% position at the resected tibial surface and measuring the distance at which it intersected the posterior endosteal cortex. This 6-mm measurement was chosen to simulate the posterior half of a 12-mm-diameter stem extension. This diameter was chosen to simulate an average-sized stem based on our review of stem size options from 5 companies which make revision tibial baseplate sets.
Demographic data were obtained on all patients included in the study from a registry. Institutional review board approval was obtained before the initiation of the study. All data were input using RedCap [6], and analysis was performed using Microsoft Excel and SAS v. 9.4 (SAS Institute, Cary, North Carolina, USA). Normality testing was performed on all numeric data. No power analysis was performed as we did not perform any comparative analyses in this study.
Results
One hundred thirty-one patients were reviewed for this study. There were 31 patients excluded. One patient had undergone a prior tibial tubercle osteotomy, one had had a prior tibial fracture, and the other 29 lacked appropriate preoperative radiographs for review, because no preoperative radiographs were available, no marker ball was utilized, or the lateral radiograph was malrotated. Demographic data can be seen in Table 2.
The average posterior offset distance between the anatomic axis of the tibia in the sagittal plane and the center of the cut tibial surface was 23.5% (standard deviation: 3.8%, range: 13.1%-33.2%) of the overall width of the tibia at that point. The distribution of percent offset between the anatomic axis and the center of a resected joint line can be seen in Figure 5.
If a 12 mm stem was inserted at the average sagittal plane position of the tibial baseplates we analyzed, the majority of patients would impinge against the posterior endosteal surface between 60 and 80 mm (59%). However, 19% of patients would impinge with a 60-mm stem and 2% would impinge with a 40-mm stem (Fig. 4).
The shortest distance to impingement was 33.7 mm. and the longest distance to impingement was 128 mm (Figs. 5a, b and 6).
There was a weak but statistically significant correlation (Adj. R 2 ¼ 0.322, P < .001) between the percentage of posterior offset between the center of the resected joint surface and the estimated impingement point for a 12-mm tibial stem inserted 10% anterior to the center of the resected joint line (ie, at the 40% mark in the anterior-posterior plane) (Fig. 6).
Discussion
We have demonstrated the variability of sagittal plane morphology of the proximal tibia as well as its potential impact on stem placement during TKA. Tibial stems are an important tool for revision TKA [7] and complex primary TKA in obese patients [8] or patients with severe preoperative deformities [9]. In this study, a 60-mm-long stem with a diameter of 12 mm would impinge against the posterior endosteal surface in 19% of patients. In 2% of patients, a stem only 40-mm long would have impinged against the endosteal surface. This impingement could create a stress riser leading to either insertional or postoperative fracture. It could also tilt the tibial baseplate into inappropriate tibial slope, which could interfere with flexion gap balancing. Finally, intersection with the endosteal surface could result in a diminished posterior cement mantle if a cemented stem is used or increased anterior overhang of the baseplate relative to the anterior cortex of the proximal tibia if a canal-filling, diaphyseal-engaging stem is used.
This study has limitations. It lacks clinical correlates, as it was based on preoperative radiographs of the native knee in patients who did not eventually undergo TKA which incorporated a long stem. We therefore cannot address the clinical prevalence of issues arising from the mismatch of tibial stems and the joint center during placement of a stemmed TKA in this purely radiographic analysis. It was based on plain film radiographs, and although we did control for quality of the radiographs and eliminated 22% of the patients based on poor radiographic quality, a CT analysis would have provided more reliable results. Our study population selected for patients with painful knees presenting to an arthroplasty surgeon and may not necessarily represent population-wide norms.
Our methodology for measurement of the distance between the anatomic axis of the tibia and the center of a tibial resection was a novel method which has not been utilized in previous studies. Our study population was not racially diverse, which could be relevant as 3 previous studies have demonstrated increased variability in tibial morphology in patients of Asian descent [10][11][12] and there was only one patient who self-reported as Asian in this study.
Replication of this study methodology in a more diverse ethnic population could yield significantly different results than what we have reported here. We chose to analyze the depth at which a 12mm stem would impinge; however, the diameter of the stem would effect this distance, as thinner stems would impinge lower and thicker stems would impinge higher within the tibia. Additionally, all of our resections were based off of native knee radiographs, while in the revision setting, preexisting resections result in increased bone loss and could result in earlier impingement. Finally, this study only analyzed the sagittal plane morphology on lateral radiographs, meaning we cannot comment on the incidence of coronal plane malalignment. The most relevant literature on the relationship of the tibial diaphysis to the proximal tibia comes from an autopsy study published in 1995 which involved performing axial CT scans of the proximal tibia after placing a straight rod in 10 cadaver tibias. This allowed the examiners to calculate the relative position of the tibial diaphysis compared with the joint line. That study found that the tibial anatomic axis was anterior to the plateau center in 9 of 10 specimens, ranging from 15 mm anterior to 1.5 mm posterior to the center of the joint line [2]. These results are similar to those reported here, as the anatomic axis was anterior to the plateau center in all 100 patients. This is the first study to correlate the sagittal plane morphology of the proximal tibia with potential issues during insertion of stemmed TKA implants. This is only a concern among patients with above-average posterior offset of the proximal tibia. There are several strategies that could be utilized to avoid the issues caused by this impingement. Cemented nonecanal-filling stems can be utilized to allow slightly asymmetric positioning within the canal. Posterior slope can be increased with the downside of effecting flexion/extension gap balancing (Fig. 7). Offset stems allow baseplates to be positioned posteriorly relative to the center of the diaphyseal anatomic axis; however, if an offset stem is utilized with cement proximal to the offset junction, extraction can be extremely challenging. Finally, a shorter stem can be used (Fig. 8).
Overall, we have demonstrated the potential for issues related to stem impingement in patients with a proximal tibial joint line which is posteriorly offset relative to the diaphyseal anatomic axis.
In this study, the center of the diaphyseal axis and the center of the resected tibia were not aligned in any of the patients in the sagittal plane. Awareness of this fact could encourage surgeons to incorporate analysis of lateral radiographs into preoperative planning and thereby mitigate the risk of complications related to the variability described here.
Conclusions
The sagittal plane morphology of the proximal tibia can influence stem placement during TKA. Here we have demonstrated that long tibial stems will impinge earlier in patients in whom the center of the proximal tibial joint line is offset posteriorly relative to the sagittal plane anatomic axis. If long tibial stems are placed in these patients, impingement against the posterior tibia can result in baseplate malpositioning, diminished cement mantle, or fracture. Recognition of this morphology preoperatively would allow surgeons to implement strategies to avoid these consequences.
Conflicts of interest
Thomas Fehring receives royalties from, is a paid consultant for, and receives research support as a principal investigator from DePuy and A Johnson & Johnson Company; the other authors declare no potential conflicts of interest. | 2021-12-17T16:44:32.513Z | 2021-12-15T00:00:00.000 | {
"year": 2021,
"sha1": "112807e6331350c624a40c6f9a5254e8694d9d98",
"oa_license": "CCBYNCND",
"oa_url": "http://www.arthroplastytoday.org/article/S2352344121002107/pdf",
"oa_status": "HYBRID",
"pdf_src": "PubMedCentral",
"pdf_hash": "1df21a89dbcb53edd7bf6b506b2d691694dc927c",
"s2fieldsofstudy": [
"Medicine",
"Engineering"
],
"extfieldsofstudy": [
"Medicine"
]
} |
254766976 | pes2o/s2orc | v3-fos-license | A Metabolomic Profile of Seminal Fluid in Extremely Severe Oligozoopermia Suggesting an Epididymal Involvement
Male infertility has increased in the last decade. Pathophysiologic mechanisms behind extreme oligospermia (EO) are not yet fully understood. In new “omics” approaches, metabolomic can offer new information and help elucidate these mechanisms. We performed a metabolomics study of the seminal fluid (SF) in order to understand the mechanisms implicated in EO. We realized a targeted quantitative analysis using high performance liquid chromatography and mass spectrometry to compare the SF metabolomic profile of 19 men with EO with that of 22 men with a history of vasectomy (V) and 20 men with normal semen parameters (C). A total of 114 metabolites were identified. We obtained a multivariate OPLS-DA model discriminating the three groups. Signatures show significantly higher levels of amino acids and polyamines in C group. The sum of polyunsaturated fatty acids and free carnitine progressively decrease between the three groups (C > EO > V) and sphingomyelins are significantly lower in V group. Our signature characterizing EO includes metabolites already linked to infertility in previous studies. The similarities between the signatures of the EO and V groups are clear evidence of epididymal dysfunction in the case of testicular damage. This study shows the complexity of the metabolomic dysfunction occurring in the SF of EO men and underlines the importance of metabolomics in understanding male infertility.
Introduction
Recent years have seen an increase in the incidence of male factor infertility in France [1] and all over the world, with a notable decline in the sperm concentration of 1.4 to 1.6% per year, according to the geographical regions [2].
The diagnosis of male factor infertility is based on abnormalities in one or several of the objective sperm parameters. Oligospermia, which is characterized by a decrease in the sperm concentration, is among the most frequently encountered. It is considered mild when the sperm concentration is between 10 and 15 million/mL (M/mL), moderate if the concentration is between 5 and 10 M/mL, and severe when the concentration is lower than 5 M/mL [3]. Extreme oligospermia (EO) are defined by a sperm concentration in the ejaculate lower than 1 M/mL [4,5]. Among these, the most severe is non-obstructive azoospermia (NOA), which is characterized by the complete absence of sperm cells in the ejaculate, confirmed on two separate sperm samples. On the other hand, cryptozoospermia is defined by the absence of sperm cells on the direct examination of fresh preparation, but the presence of a very low number in the pellet after centrifugation [6]. Despite these two pathologies being separate entities, the distinction is not always straightforward [7] and there seems to be a continuum between the different disorders among EO. Indeed, Metabolites 2022, 12, 1266 2 of 11 studies have shown that around 20% of men with severe oligoasthenospermia will become crypto/azoospermic in a mean time of 3 years [3,8].
EO is frequently diagnosed during an infertility workup and requires specific management. Indeed, given the high intra-individual variability in sperm parameters, and the risk of progression towards azoospermia, these cases require the cryopreservation of several samples of sperm, and/or testicular biopsies, as part of the preparation for assisted reproductive technologies (ART). On the other hand, most cases of EO are idiopathic, even though genome studies increasingly show underlying genetic causes [9].
Recently, omics technologies, and most notably metabolomics, are being applied in the field of male infertility in an effort to understand its pathophysiology, with the seminal fluid (SF) being at the center of analyses. Several studies have shown that, in patients with oligoasthenospermia (OA), there is an alteration of the energetic metabolism-more specifically the beta-oxidation [10] and glycolysis [11][12][13] pathways-as well as a distortion of the oxidative balance due to a decrease in certain amino acids or biogenic amines [10,14,15], and an alteration of the seminal fluid composition [10,15]. Studies have also confirmed a correlation between the concentrations of certain metabolites and sperm parameters: carnitine and carnosine were linked to concentration [16,17] and fructose linked to mobility [13].
Despite showing interesting results, most of these studies were limited by small samples sizes, the use of multiple techniques targeting different metabolites, and the inclusion of heterogenous populations with varying alterations of sperm parameters. To the best of our knowledge, no study in the literature reported on the use of metabolomics in the seminal fluid to try and find a specific signature that would help understand the pathophysiologic mechanisms of EO. Based on that, we decided to analyze the metabolomic profile of the SF of patients with EO, and compare it with two control groups, one with normal semen parameters according to the WHO criteria, and one with a history of vasectomy. The latter group would allow to differentiate and separately analyze the metabolites in epididymal secretions.
Study Population
We performed a prospective observational study at the Reproductive Medicine Department of the Angers University between October 2018 and February 2021.
We included three groups of patients. The EO group included patients with extreme oligospermia, defined as < 1 M/mL, who presented to our department for an IVF/ICSI treatment, for the cryopreservation of sperm, or for semen analysis. The vasectomy (V) group included patients who presented to our center for a semen analysis ordered by their physician to confirm azoospermia following vasectomy surgery. The control (C) group included patients with normal semen parameters, according to the WHO criteria, whose partners were undergoing intrauterine inseminations (IUI) or IVF/ICSI for female factor infertility.
All patients in the EO group underwent a full urologic workup (physical examination, testicular ultrasound, and genetic testing) in order to rule out obstructive causes. Excluded from the group were all patients with a positive history suggestive of secondary causes of EO, such as orchiepididymitis, hypogonadotropic hypogonadism, or gonadotoxic treatment.
All patients who were followed at our department during the study period and who signed the consent form were eligible for inclusion. We excluded from the study all patients who refused to sign the consent form, patients who did not fit in any of the three study groups, and patients with associated medical conditions or under chronic medication. None of the included patients had any metabolic condition or had lifestyle habits associated with altered sperm parameters, and none were taking any medications or supplements to improve sperm quality.
Ethical Approval
All participants signed the consent form prior to inclusion, and the study was approved by the Ethics Committee of the University Hospital of Angers, France (Number DC-2014-2224 and AC-2017-2993).
Sample Preparation
All semen samples were collected by masturbation in the andrology laboratory, following an abstinence period of 3 to 5 days. A fraction of the sample was taken for the assessment of the sperm parameters and the planned ART, and the remainder taken for our study. The samples were centrifuged on a density gradient to isolate the seminal plasma, which was later centrifuged for 5 min at 10,000× g before being frozen at −80 • C for the metabolomics analysis.
After thawing, all seminal fluid samples were vortexed and centrifuged at 4 • C for 5 min at 5000× g. As per the recommendations of the kit, 10 microliters of each sample were then added to the filter on the upper wells of the 96-well plate for the FIA analysis, and 50 microliters for the LC analysis. Three quality controls (QCs) composed of human plasma samples at three concentration levels (QC1, QC2, and QC3) were added to validate our analysis.
Statistical Analysis
After validation of the kit quality controls, the raw metabolomic data were normalized with UV scaling and mean centering. For each metabolite, the mean concentration was substracted and the difference was divided by the standard error.
We performed an unsupervised multivariate principal component analysis (PCA), using Simca −P+ v 16.0.1 (Umetrics, Umea, Sweden), in order to visualize the distribution of the metabolomics data and detect the grouping of samples. Hotelling's T2 plot was used to detect aberrant samples.
We then performed a supervised orthogonal partial least-squares discriminant analysis (OPLS-DA) of the three groups followed by a comparison of each two groups separately. The quality and performance of the models were assessed using different variables: the Q 2 Y(cum) (goodness of the prediction), the R 2 Y(cum) (goodness of the fit), the crossvalidation analysis of variance (CV-ANOVA), and the permutation test (evaluation of the overfitting risk).
The significance of the metabolites was assessed using the variable importance in projection (VIP) plot. Only metabolites with a VIP value of at least 1 (VIP ≥ 1) were retained.
The metabolites found during the different comparisons were all presented as volcano plots using the Simca software −P+ v 16.0.1 (Umetrics).
We depicted the differences between the three groups on a Venn diagram using the software Venny 2.1.0.
The univariate analyses comparing certain ratios and the sum of metabolites between the groups were performed using the non-parametric Mann-Whitney test on GraphPad Prism v 8.0 (GraphPad Software, San Diego, CA, USA). All tests were considered significant at p-value < 0.05.
Population Characteristics
We included a total of 61 men: 19 in the EO group, out of which 4 had azoospermia and 15 cryptozoospermia, 22 men who had a vasectomy at least 3 months before inclusion in the V group, and 20 men with normal semen parameters according to the WHO classification in the C group (Table S1).
Metabolomics Signature
We were able to accurately measure 114 metabolites out of the 188 analyzed by the kit. The raw data (concentration of each metabolite in µM in each patient) are presented in the Supplementary Table S2. In total, the following metabolites were correctly measured in the SF of patients, each according to its measuring range: 46 glycerophospholipids out of the 90 analyzed by the kit (24/38 PCaa, 16/38 PCaa and 6/14 LysoPC), all the sphingomyelins (15/15), 20 out of the 21 amino acids, 11 out of the 21 biogenic amines, 21 out of the 39 acylcarnitines, and all the hexoses.
According to the unsupervised PCA approach, which allows us to visualize the dataset, we did not find any aberrant value or outliers according to Hotelling's T2 range ( Figure 1A). We obtained a discriminating OPLS-DA model that allowed to differentiate the three groups of patients and to confirm that the patients included in the EO group (azoospermia, cryptozoospermia, and severe oligospermia) were homogeneous ( Figure 1B). The OPLS-DA model had a good prediction capacity (Q2cum = 70%) and good performances on the permutation test (Q2cy = −0.45) and the CV-ANOVA (p-value = 1.33468 × 10 −19 ).
The significance of the metabolites was assessed using the variable importance in projection (VIP) plot. Only metabolites with a VIP value of at least 1 (VIP ≥ 1) were retained.
The metabolites found during the different comparisons were all presented as volcano plots using the Simca software −P+ v 16.0.1 (Umetrics).
We depicted the differences between the three groups on a Venn diagram using the software Venny 2.1.0.
The univariate analyses comparing certain ratios and the sum of metabolites between the groups were performed using the non-parametric Mann-Whitney test on GraphPad Prism v 8.0 (GraphPad Software, San Diego, CA, USA). All tests were considered significant at p-value < 0.05.
Population Characteristics
We included a total of 61 men: 19 in the EO group, out of which 4 had azoospermia and 15 cryptozoospermia, 22 men who had a vasectomy at least 3 months before inclusion in the V group, and 20 men with normal semen parameters according to the WHO classification in the C group (Table S1).
Metabolomics Signature
We were able to accurately measure 114 metabolites out of the 188 analyzed by the kit. The raw data (concentration of each metabolite in μM in each patient) are presented in the supplementary Table S2. In total, the following metabolites were correctly measured in the SF of patients, each according to its measuring range: 46 According to the unsupervised PCA approach, which allows us to visualize the dataset, we did not find any aberrant value or outliers according to Hotelling's T2 range ( Figure 1A). We obtained a discriminating OPLS-DA model that allowed to differentiate the three groups of patients and to confirm that the patients included in the EO group (azoospermia, cryptozoospermia, and severe oligospermia) were homogeneous ( Figure 1B). The OPLS-DA model had a good prediction capacity (Q2cum = 70%) and good performances on the permutation test (Q2cy = −0.45) and the CV-ANOVA (p-value = 1.33468 × 10 −19 ). We also obtained discriminating OPLS-DA models for the comparisons between each two groups separately (EO vs. C; EO vs. V; V vs. C) ( Table 1). These models allowed us to determine the metabolomic signature specific to each group. The metabolites allowing the distinction between the groups are represented as volcano plots (Figure 2), and the metabolites implicated in these models, two at a time, are the ones with VIP > 1 (list presented in Supplementary Table S3). The metabolites with a negative p(corr) are decreased, and those with a positive p(corr) are increased. The metabolites allowing the distinction between the groups are represented as volcano plots (Figure 2), and the metabolites implicated in these models, two at a time, are the ones with VIP > 1 (list presented in supplementary Table S3). The metabolites with a negative p(corr) are decreased, and those with a positive p(corr) are increased. All the results are globally represented in the Venn diagram (Figure 3), which shows the 23 metabolites distinguishing the C group (mainly 13 amino acids), the 15 metabolites distinguishing the V group (mainly sphingomyelins (n = 7) and free carnitine (C0)), and the 9 metabolites distinguishing the EO group. The complete list of metabolites is presented as supplementary data (Table S4). All the results are globally represented in the Venn diagram (Figure 3), which shows the 23 metabolites distinguishing the C group (mainly 13 amino acids), the 15 metabolites distinguishing the V group (mainly sphingomyelins (n = 7) and free carnitine (C0)), and the 9 metabolites distinguishing the EO group. The complete list of metabolites is presented as supplementary data (Table S4). Figure 4 represents all the metabolites and the groups of metabolites allowing to distinguish the three groups.
The amino acid level and the polyamines (spermine and spermidine) were signifi- Figure 4 represents all the metabolites and the groups of metabolites allowing to distinguish the three groups.
The amino acid level and the polyamines (spermine and spermidine) were significantly higher in the C group compared with the other two groups (Figure 4). The sum of polyunsaturated fatty acids (PUFA) and free carnitine progressively decreased between the three groups: they were significantly highest in the C group and lowest in the V group (Figure 4). The sphingomyelins were significantly lower in the V group compared with the C and EO groups (Figure 4). p-values were obtained using the non-parametric Mann-Whitney test. *: significant p-value < 0.05; ** significant p-value < 0.01; *** significant p-value < 0.001and **** significant p-value < 0.0001.
The amino acid level and the polyamines (spermine and spermidine) were significantly higher in the C group compared with the other two groups (Figure 4). The sum of polyunsaturated fatty acids (PUFA) and free carnitine progressively decreased between the three groups: they were significantly highest in the C group and lowest in the V group ( Figure 4). The sphingomyelins were significantly lower in the V group compared with the C and EO groups (Figure 4).
Discussion
The seminal fluid is made up of a mixture of secretions from the epididymis, prostate, and seminal vesicles, and reflects the exchanges between the genital tract and the sperm cells. It is essential for the survival of the sperm cells, since it contributes to their nutrition, maturation, and protection from their fertilizing capacity (decapacitation). The exchanges between the sperm cells and the SF occur via exosomes, which represent approximately 3% of its proteins, making it among the richest biologic fluids in exosomes. The exosomes are mainly proteasomes and epididymosomes [18], the latter being secretion vesicles of the epididymal epithelium that are essential for the acquisition of the fertilizing capacity of the sperm cells. They are rich in sphingomyelins and lipids, but contain also RNAs, proteins, and several metabolites such as amino acids [19,20].
In order to analyze the SF, we used a metabolomics approach targeting 188 metabolites with the objective of finding a specific signature that would differentiate men with EO from men with normal semen parameters. We also included a group of vasectomized men in order to have a more detailed understanding of the structures and mechanisms implicated. The signature we found is comprised mainly of amino acids (AA), polyunsaturated fatty acids (PUFA), free carnitine (C0), spermine, and spermidine.
We found a decrease in AA in the EO and V groups when compared with controls. Besides being essential components of proteins, AA play several important roles. It is well known that, in humans, they are abundant in the SF and that they play an essential part in reproduction [21]. On the other hand, in animals, supplementation with certain AA has been shown to improve sperm quality, the fertilizing capacity of sperm cells, and their resistance to cryopreservation [22][23][24]. In men, the administration of AA with antioxidant properties helps to maintain sperm DNA integrity, and the functional parameters of the sperm cells during the cryopreservation process [25].
In infertile men, the levels and composition of AA in the SF seem to be altered. Indeed, studies have shown a global decrease in AA in the semen of azoospermic men [26]. More recently, mass spectrometry studies analyzing the AA content of the SF have revealed an important imbalance in men with asthenospermia [21,27] and oligospermia [21], with a significant decrease in their levels compared with controls. Likewise, metabolomics studies have shown several modifications in the expression profile of AA in the SF according to the sperm quality or the patients' fertility (metabolism of branched AA) [14], asthenospermia [28], oligoasthenoteratospermia (OATS) [10,12,15], and unexplained infertility [11,[29][30][31]. In general, the majority of studies make the case for a decrease in the concentration of AA in the SF in cases of infertility, but there is no consensus on the expression profiles of AA that could be specific to certain sperm alterations, mainly because of the large heterogeneity in the profiles assessed. Our study has focused on EO, a very well-defined and severe phenotype of sperm alteration, and we found a significant decrease in the concentration of the majority of the AA compared with men with normal semen parameters. The presence of sperm cells in the ejaculate does not seem to modify the content in AA of the SF, and a glandular origin of these AAs has been proposed [26]. The fact that the concentration of AA was also lower in vasectomized men (V group) than in controls makes the case for an epididymal origin of the AA. Furthermore, the fact that they are likewise decreased in men with EO suggests an alteration in the "epididymal production" in men with secretory problems.
Our study has also found a significant and progressive decrease in the concentration of PUFA between the C, EO, and V groups, respectively. Lipids are the major components of the sperm membranes, which are mainly comprised of cholesterol and phospholipids that carry saturated and polyunsaturated fatty acids, distributed in an asymmetrical manner. The lipid composition of the spermatic membrane varies across the different developmental stages of sperm cells. During spermatogenesis, there is an integration of PUFA in the membrane via the transformation of essential nutritional fatty acids by elongation and desaturation steps in the germinal cells [32]. During epididymal maturation, there is an increase in the unsaturation level of the fatty acids [33], especially with the incorporation of Docosahexaenoic acid (DHA), which is the main PUFA of the spermatic membrane [34]. The addition of cholesterol during maturation and the upkeep of an asymmetrical distribution of phospholipids allow to ensure a certain membrane stability during periods of stasis and ejaculation [35]. In the female genital tract, and after the cholesterol efflux, they allow increases in the membrane fluidity, which is crucial for the fusion of the sperm and oocyte membranes [35,36].
The link between lipids and male fertility has been proven by several studies analyzing the plasmatic membrane and/or the seminal fluid. The DHA concentration in sperm has been correlated to morphology, motility, and sperm concentration [34,[37][38][39]. Many studies have also shown a decrease in the PUFA concentration in the sperm cell membranes in men with asthenospermia [38,[40][41][42] and oligospermia [39].
On the other hand, there is a correlation between the lipid composition of the sperm membrane and that of the seminal plasma [43]. Several metabolomics studies have shown an increase in the level of saturated fatty acids [28,44], as well as a decrease in the PUFA levels [34,38] in the seminal plasma of men with asthenospermia, a finding also noted in men with oligoasthenospermia [10]. Finally, it has been proposed that the sperm and the seminal fluid composition in fatty acids could be predictive markers of the success of the sperm freezing procedure [45]. The exact mechanisms of the epididymal rearrangement of the lipid composition of the sperm membranes are not yet fully understood, but they seem to involve the epididymosomes [46]. Indeed, studies have described a variability in the lipid composition of the epididymosomes in the epididymis itself, with the presence at the level of the head of vesicles rich in PUFA and at the level of the tail of vesicles rich in cholesterol and phospholipids [46]. In our study, the decrease in PUFA in the V group is an argument in favor of their epididymosomal origin. The concomitant decrease in sphingomyelins, which are important components of epididymosomes, further supports this hypothesis. Likewise, the decrease in PUFA in the EO group suggests an impairment of the epididymosomal function.
We have also found a progressive decrease in free carnitine (C0) between the control, EO, and V groups. C0 is well known to be primarily of epididymal secretion [47] and has been used as a seminal biochemistry marker. Studies have reported a decrease in free carnitine in infertile men, more specifically men with severe OA [10], and found a correlation between the concentration of C0 and several sperm parameters [48,49].
Finally, we have observed a decrease in several polyamines, such as spermine and spermidine, in the EO and V groups, compared with the controls. These polyamines, which are derived from arginine, play several roles in spermatogenesis and act on cellular proliferation [50]. Spermine also has a role in sperm cell decapacitation [51] and has antioxidant activity that neutralizes free radicals [52]. Polyamines are known to be of prostatic origin [53] but studies have also reported a testicular secretion by Sertoli and Leydig cells [54]. Other studies have indirectly confirmed that finding by showing an alteration of the polyamines concentration in men following vasectomies [55]. Metabolomics studies of the seminal fluid have also shown decreased concentrations of spermine in men with OATS [10,15] and asthenospermia [14].
Surprisingly, we noticed a great similarity between the signature of EO and vasectomized men. That overlapping raises the question whether there is also an epididymal damage associated to the testicular dysfunction (anomalies of spermatogenesis). This double dysfunction has two major consequences. The first is that it highlights the difficulty of finding seminal fluid biomarkers that could help distinguish excretory (non-testicular) from secretory (testicular and non-testicular) azoospermia. The second is that it could constitute an argument in favor of the hypothesis of the testicular dysgenesis syndrome, which states that, in many cases, exposure to environmental toxins, such as endocrine disruptors, and genetic factors can cause an alteration of the spermatogenesis but also malformations of the genital tracts, thus leading to the development of azoospermia and oligospermia [56,57].
Conclusions
In the current study, we have found a specific signature characterizing extreme oligospermia that includes certain metabolites already linked to infertility in previous studies. Globally, all these metabolites are decreased in men with EO when compared with controls, which rules out a signature linked to the consumption of metabolites by sperm cells. Indeed, the absence or the very low numbers of sperm cells could have led to a non-consumption of metabolites, which would have had higher concentrations than in men with normal semen parameters. This is why this signature most likely reflects a production problem, linked to a dysregulation of the functions and secretions of the genital tract. | 2022-12-17T16:04:27.625Z | 2022-12-01T00:00:00.000 | {
"year": 2022,
"sha1": "c5e4c5272ec6b88f70e84d0aa683a2145e42e960",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/2218-1989/12/12/1266/pdf?version=1671096044",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "738d536fc150c183dd5b215228dc7751a00e7fcd",
"s2fieldsofstudy": [
"Medicine",
"Biology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
210473760 | pes2o/s2orc | v3-fos-license | The Impact of Equilibria on the Shape of Hysteresis Loops
Hysteresis is a common phenomenon found in many dynamical systems. It is typically described as a looping behaviour in the systems input-output graph. For a dynamical system to exhibit hysteresis, it must have multiple stable equilibria. This work examines the impact that different types of equilibria can have on the shape of hysteresis loops exhibited in input-output graphs of ordinary differential equations.
The study of hysteresis is important for understanding the behaviour of a system given changes in the input. The appearance of hysteresis in a dynamical system can make modelling the system more difficult. Consequently, the more hysteresis is understood, the easier it can be to control the hysteresis arising in a given dynamical system. While there is much literature that mentions hysteresis to describe a particular system, the following discuss hysteresis in greater depth by examining its appearance in certain fields and attempts to provide a rigorous definition for it: Bernstein and Oh (2005); Brokate and Sprekels (1996); Ikhouane (2013); Ikhouane and Rodellar (2007); Morris (2011);Noori (2014); Visintin (2005).
Operators are commonly used to describe hysteresis, as well as using notions such as rate-independence, memory and path-independence (Brokate and Sprekels (1996); Cross et al. (2009);Visintin (2005)). However, these definitions are not always tractable, and may exclude systems that exhibit hysteresis. Rather consider the following definitions, which will be used in this paper.
Definition 1. (Morris, 2011, Definition 3) A hysteretic system is one which has (i) multiple stable equilibrium points and (ii) dynamics that are considerably faster than the time scale at which inputs are varied.
Definition 2. Bernstein and Oh (2005) Hysteresis is a nontrivial closed curve with periodic input in the input-output graph of a system that persists as the frequency of the input approaches 0. This nontrivial closed curve is called a hysteresis loop. Nickmand et al. (2013) examine how different porous materials affect the shape of hysteresis loops. Wang and Hui (2017) discusses hysteresis loops that are not simple; that is, loop crosses itself. But in general, there are not many studies in regards to what affects the shape of hysteresis loops. Since multiple stable equilibria are essential to a hysteretic system as noted in definition 1, there is likely a correlation between the equilibria of a system and the shape of the hysteresis loop. This relationship is explored in greater depth in this paper. Additional references that consider the relationship between equilibria and hysteresis are Afshar et al. (2016); Angeli et al. (2004); Chow and Morris (2014); Ikhouane (2013); Morris (2011).
In this paper, hysteresis in dynamical systems of ordinary differential equations (ODEs) are considered. These ODEs have been chosen because they exhibit different types of equilibria including: a continuum of stable equilibria, a finite number of discrete stable equilibria, an infinite number of discrete stable equilibria and the presence or absence of unstable equilibria. These variations affect the shape, such as symmetry, of the corresponding hysteresis loop.
To establish hysteresis in the examples presented in this paper, the following procedure is applied. The equilibria of each system is first determined and then their stability is established. This is a necessary condition of hysteresis as indicated in definition 1. Constructing the input-output graph of each system is also necessary in order to show definition 2 is satisfied. The output is the solution to the ODE. Let u(t) be the input, which is introduced into the system to construct the input-output graphs. These are generated using MATLAB. Figure 1 demonstrates what is meant by the presence versus the absence of looping in the input-output graph of dynamical systems. In this paper, the periodic input u(t) = sin(ωt) is used so that the frequency of the input is ω. It is important to note the presence of looping is not sufficient to conclude hysteresis but rather, as definition 2 indicates, the loops must persist as ω goes to zero. When the shape of the hysteresis loop is unchanged as the frequency of the periodic input changes, the arising hysteresis is said to be rate-independent; otherwise, the hysteresis is said to be rate-dependent (Bernstein and Oh (2005)). Rate independent versus rate dependent hysteresis loops are also discussed in this paper. Furthermore, some papers conclude hysteresis can only appear in nonlinear dynamical systems (Ikhouane (2013); Bernstein and Oh (2005)
LINEAR EXAMPLES
Consider the linear exampleẋ(t) = u(t). The dot notation represents the derivative with respect to time t. When there is no input; that is, u(t) = 0, and the time derivative is zero, x(t) is any constant for all t. Trivially this indicates a continuum of stable equilibria. To test for hysteresis as per definition 2, construct the input-output ofẋ(t) = u(t) by letting u(t) = sin(ωt) for various values of ω.
The results are depicted in Figure 2 and show persistent looping in the input-output graph. That means definition 2 is satisfied and suggestsẋ(t) = u(t) exhibits hysteresis.
Consider another linear examplë
(1b) This example can be found in Chow and Morris (2014) except the first derivative term has been rescaled by a factor of 1 3 . In order to verify that (1) has multiple stable equilibria, let x 1 (t) =ẏ(t) and x 2 (t) =ÿ(t), so that equation (1) can be rewritten equivalently as a system of coupled linear ODEs: If there is no input; that is, u(t) = 0, and setting the timederivative to be zero, leads to a continuum of equilibria of the form (x 1 , x 2 ) = (a, 0) where a ∈ R is a constant. The stability of these equilibria are determined by their corresponding eigenvalues, denoted λ. In this case, they are λ 1 = 0 and λ 2 = −5, which do not depend on a.
If the real part of all the eigenvalues of an associated equilibria is less than or equal to zero, the equilibria is said to be stable. Otherwise, it is unstable. This is well known and can be referenced from most undergraduate introductory level textbooks about ODEs. See for instance (Khalil, 2002, Chapter 2) or (Perko, 2002, Chapter 1). Therefore, the continuum of equilibria of (1) are all stable. To test whether (1) exhibits hysteresis, recall definition 2. Construct the input-output of (1) by letting u(t) = sin(ωt). The results are depicted in Figure 3 for various values of ω and shows persistent looping. This indicates the system in (1) exhibits hysteresis.
Both linear examples have a continuum of stable equilibria and the shape of their hysteresis loops are similar.
NONLINEAR EXAMPLES
Consider the following first order nonlinear ODĖ (3b) This example can be found in (Perko, 2002, Page 336). For u(t) = 0, the time derivative is zero when x(t) equals 1, 0 and −1. Therefore, these are the equilibria of (3). The phase portrait of (3) depicted in Figure 4 indicates 1 and −1 are stable, while 0 is unstable. Notice the symmetry of the equilibria; that is, the two stable equilibria are equidistant from the unstable equilibrium point. This indicates that 1 and -1 are stable equilibria while 0 is an unstable equilibrium point of (3). Figure 5 shows the input-output graph of (3) for different frequencies of the input u(t) = sin(ωt) with loops persisting as ω goes to zero. Notice the symmetry about the origin of the hysteresis loops for smaller values of ω. This correlates to the symmetry of the equilibria.
Consider another first order nonlinear ODĖ x(0) = 10. (4b) This example can be found in (Murray, 2002, Page 7-8), Ludwig et al. (1978) and represents the classic Spruce Budworm Model. In this model, r(t) depends on the birth rate of the population and the constant q depends on the carrying capacity. When r(t) is set to be the constant input 0.52 and q = 25, equation (4) exhibits multiple stable equilibria (see Figure 6). In particular, 22.90502477 and 0.9715485792 are stable equilibria while 0 and 1.123426650 are unstable equilibria. The non-zero equilibria are determined by solving for x(t) in x(t) 1 + (x(t)) 2 = 0 with r(t) = 0.52 and q = 25. for various values of ω. Equation (3) has two stable equilibria and one unstable equilibrium point. Fig. 6. The phase portrait of (4) with r(t) = 0.52 and q = 25, generated with Maple. This indicates that 22.90502477 and 0.9715485792 are stable equilibria while 0 and 1.123426650 are unstable equilibria of (4). When the birth rate is treated as the input of the system, the model exhibits hysteresis. That is, r(t) = u(t). Physically, the birth rate is always positive, therefore the input, sin(ωt) is placed in absolute value. The corresponding hysteresis loops of (4) are shown in Figure 7 and indicates persistent looping as ω goes to zero; however, there is no evidence of symmetry in the loop shape. Note: there is no symmetry betweeen the equilibria of (4).
(0, 0) is an unstable equilibria. Notice the symmetry of the equilibria; that is, the two stable equilibria are equidistant from the unstable equilibria. Figure 8 shows the input-output graph of (5) for different frequencies of the input u(t) = sin(ωt). Notice the symmetry about the origin of the hysteresis loops for smaller values of ω. This correlates to the symmetry of the equilibria.
(8b) has infinitely many discrete equilibria of the form nπ 2 which are stable when n is even and unstable when n is odd. This was determined by rewriting (8) into a system of first order ODEs as shown in detail for equation (1). For the stable equilibria, the eigenvalues are −1±i √ 3 2 and for the unstable equilibria, the eigenvalues are −1± √ 5 2 . When u(t) = sin(ωt), the hysteresis loops of (8) are illustrated in Figure 11 and are similar in shape and behaviour to the examples with a continuum of stable equilibria (see Figures 2 and 3). Fig. 11. The input-output graph of (8) with input u(t) = sin(ωt) for various values of ω. Equation (8) has infinitely many stable and unstable discrete equilibria.
The following examples demonstrate rate-independent hysteresis loops and are already well known rate-independent examples. They are included here for a more comprehensive discussion of loop shapes.
From Bernstein and Oh (2005), (9b) where B(t) represents magnetic flux, the input is the applied magnetic field, denoted H(t), and a, b and c are parameters of this magnetic model. This means the input is u(t) = H(t), where H(t) = sin(ωt). When there is no input or a constant input,Ḃ(t) = 0, which implies B(t) is a constant for all t, and hence (9) has a continuum of stable equilibria. Figure 12 depicts the input-output graph of (9) with a = 0.02125, b = 0.1 and c = 0.04361. The Duhem model (10d) is well known to exhibit hysteresis (Feng et al. (2009);Bernstein and Oh (2005)). Similar to (9), equation (10) has a continuum of stable equilibria. Figure 13 depicts the input-output graphs of (10) with u(t) = sin(ωt).
In both Figures 12 and 13, the shape of the hysteresis loop is unchanged regardless of the value of ω. This supports the fact that (9) and (10) exhibit rate-independent hysteresis as noted in Bernstein and Oh (2005) (1) Aside from Figures 12 and 13, all other figures depicted rate-dependent hysteresis loops; that is, the shape of the loops changed as the frequency of the input varied. "Fully" rate-dependent systems are shown in Figures 2, 3 and 11. By fully rate-dependent, this means the shape of the loop changes every time the frequency of the input changes. These fully ratedependent systems correspond to dynamical systems with either infinitely many discrete stable equilibria or a continuum of stable equilibria. For Figures 5, 7, 8, 9 and 10, rate-independence of the hysteresis loops emerged for small values of ω.
(2) Figures 5, 7, 8, 9 and 10 correspond to systems with two stable equilibria (with different types of unstable Fig. 13. Input-output graphs of (10) with input u(t) = sin(ωt) for various values of ω, which demonstrate rate-independent hysteresis loops. Equation (10) has a continuum of stable equilibria. equilibria). Their hysteresis loops are distinguished by two "horizontal portions" of stable areas and the vertical "jumps" between them as ω approaches zero. That is, the overall shape of the hysteresis loops are the same as the frequency of the input goes to zero. This may be due to the dynamical systems having exactly two stable equilibria. (3) The linear ODEs exhibit a continuum of stable equilibria, and the corresponding hysteresis loops shown in Figures 2 and 3 are smooth in that there are no vertical "jumps". This may be because the system is always transitioning between stable equilibria. (4) The hysteresis loops in Figures 5 and 8 are symmetric about the origin as ω goes to zero. These correspond to dynamical systems with two stable equilibria that are equidistant from the one unstable equilibrium point at the origin. On the other hand, Figures 9 and 10 do not have symmetry about the origin and the associated dynamical systems do not have equidistant stable equilibria.
These observations suggest shapes of hysteresis loops can be predicted if properties of the dynamical system are known; however, there is still much that needs to be established. For example, perhaps the symmetry of loops can be determined solely by measuring the magnitude of equilibria from one another as well as using the magnitude of the corresponding eigenvalues. The observations made about hysteresis loops here and the definitions available in the reference literature may in the future provide a way to predict or even standardize hysteresis loop shapes. This can also lead to future work on controller design to modify the shape of hysteresis loops by adjusting the types of equilibria that arise in dynamical systems. | 2020-01-15T02:01:00.117Z | 2020-01-13T00:00:00.000 | {
"year": 2020,
"sha1": "40a854947e289dfcebf7ed91afe5cf1384745adc",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "40a854947e289dfcebf7ed91afe5cf1384745adc",
"s2fieldsofstudy": [
"Economics"
],
"extfieldsofstudy": [
"Mathematics",
"Physics"
]
} |
236971549 | pes2o/s2orc | v3-fos-license | The Fibrosis-Targeted Collagen/Integrins Gene Profile Predicts Risk of Metastasis in Pulmonary Neuroendocrine Neoplasms
Recently, collagen/integrin genes have shown promise as predictors of metastasis mainly in non-small cell lung cancer and breast cancer. However, it is unknown if these gene expression profiling differ in metastatic potential of pulmonary neuroendocrine neoplasms (PNENs). In this study, we sought to identify differentially expressed collagen/integrin genes in PNENs in order to understand the molecular mechanisms underlying the development of stroma-associated fibrosis for invasion and metastasis. We compared collagen/integrin gene expression profiling between PNE tumors (PNETs) and PNE carcinomas (PNECs) using a two-stage design. First, we used PCR Array System for 84 ECM-related genes, and among them, we found COL1A2, COL3A1, COL5A2, ITGA5, ITGAV, and ITGB1 functionally involved in the formation of the stroma-associated fibrosis among PNENs histological subtypes. Second, we examined the clinical association between the six collagen/integrin genes in tumor tissues from 24 patients with surgically excised PNENs. However, the pathological exam of their resected tissues demonstrated that 10 developed lymph node metastasis and 7 distant metastasis. We demonstrated and validated up regulation of the six fibrogenic genes in PNECs and down regulation in PNETs that were significantly associated with metastasis-free and overall survival (P<0.05). Our study implicates up regulation of fibrogenic genes as a critical molecular event leading to lymph node and distant metastasis in PNENs.
INTRODUCTION
Neuroendocrine neoplasms (NENs) are classified into differentiated neuroendocrine tumors (NETs), also known as carcinoid tumors (typical carcinoid and atypical carcinoid), and poorly differentiated neuroendocrine carcinomas (NECs), including large cell neuroendocrine carcinoma (LCNEC) and small cell carcinoma (SCLC) (1). Patients with pulmonary neuroendocrine tumors (PNETs) have tumors sufficiently localized to be considered treatable by surgical resection, and among those whose tumors are successfully resected, approximately 90-98% of patients with typical carcinoid, and 50-60% of atypical carcinoid, survive 5 years (2,3). In contrast, only 20-30% of the patients with large cell neuroendocrine carcinoma survive 5 years after surgical resection and adjuvant chemotherapy (4), and only 10% of the patients with small cell lung carcinoma survive 5 years after Cisplatin + Carboplatin + Etoposide (5). Clearly, some PNETs and PNECs have developed occult dissemination beyond the lung even when they appear to have been completely removed or responsive to adjuvant chemotherapy. Since different PNENs are composed of different mutated neuroectodermic cells, their malignant potential and prognosis may vary greatly. Although some of these differences are known to physicians, not only is it still often difficult to predict which tumors will invade, metastasize, and shorten the patient's life, but effective adjuvant treatments still depend on identifying these tumors shortly after biopsy or surgery as well.
Due to tissue accessibility, genome-wide examination of biomarkers associated with metastatic progression and cancer specific death has primarily been based on observations made in the primary tumor behavior (6)(7)(8) and not their effects on the stroma-associated fibrosis more lethal, and more therapeutically relevant for metastatic lesion. In addition, genome-wide studies that have preliminarily explored in metastatic tumors have done so using small sample sizes (9)(10)(11)(12)(13). Thus, the molecular mechanisms that lead to PNENs metastasis remain largely unknown and require further study. The identification of fibrogenic genes in primary tumors and their effects on tumor microenvironment (TME) as new biomarkers and therapeutic targets for PNENs is promising.
To address these gaps in the knowledge, we identified fibrotic genes that support PNENs metastasis in localized surgically resected primary tumor. Overall, we performed an analysis of gene expression data generated using mRNA in two approaches where we first utilized gene expression microarray technology to identify candidate genes that are associated with PNENs metastasis, and subsequently, validated the candidate genes in a similar cohort of patients with PNENs tumors using in silico analysis.
Discovery Cohort
We identified 24 patients at A. C. Camargo Cancer Center, in São Paulo, Brazil and Hospital do Amor, in Barretos, Brazil, who were surgically resected with PNENs [10 carcinoid tumors (5 TC and 5 AC)], 4 LCNEC, 10 SCLC, and had fresh-frozen tissue available from their primary tumor.
Two pathologists (T.G.P. and V.L.C) carried out a blinded comprehensive review of all tumors to confirm histological subtype (14), the mitotic count, the presence of an organoid pattern (rosettes, pseudo rosettes, palisading, spindle cells) and necrosis. The neoplastic area was delimited during the frozen section procedure to ensure the exclusion of non-neoplastic tissue. Patient's demographics and clinicopathological characteristics were obtained from medical records and included age, sex, smoking history, tumor size, tumor stage (according to the International Association for the Study of Lung Cancer classification system, 8th edition), and follow-up information (14).
The internal ethics committees of all the participating institutions approved this study's protocol (process number 1.077.100) with a waiver for informed consent by their review boards.
Validation Cohort
To validate our data we performed in silico analysis using public database. Data included 19 normal lung tissues, 8 LCNEC tissues, 15 SCLC tissues, and 12 primary typical carcinoids. The GSE1037 gene expression profile was obtained from the National Center for Biotechnology Information Gene Expression Omnibus 1 (GEO) (15,16) based on the GPL962 platform. The mRNA expression raw data were analyzed by GEO2R online tools. In order to compare gene expression, we created two heatmaps using the Heatmapper platform 2 to investigate the gene expression of 5 different genes, namely COL1A2 (a2 chain COL I), COL3A1 (a1 chain COL III), COL5A2 (a2 chain COL V), ITGA5 (alpha 5-integrin) and ITGB1 (beta 1-integrin), in each PNEN histological type, comparing the profile seen in GSE1037 to that of our cohort. Then, we used the average distance and the Euclidean distance between elements to perform an unsupervised hierarchical grouping.
Gene Expression Profile Data
The neoplastic area was micro dissected during the frozen section procedure to ensure the inclusion of neoplastic tissue and distant non-neoplastic tissue as control. Total mRNA was extracted from fresh-frozen tumor and normal tissues using the QIAsymphony miRNA CT 400 kit (Qiagen, CA, USA) according to the manufacturer's instructions. RNA integrity and quality were determined using the Bioanalyzer 2100 (Agilent Technologies). Complementary DNA was synthesized using the c-DNA -RT² First Strand Kit (Qiagen Sample & Assay Technologies) according to the manufacturer's protocol. The difference of expression in EMT genes was evaluated by the real-time PCR method. Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was performed using the RT² Profiler PCR Array System (PAHS-090Z; Qiagen, Dusseldorf, Germany) kit for the human epithelial-to-mesenchymal transition (EMT) pathway with 84 target genes. The array includes a total of 84 EMT genes, 5 housekeeping genes (ACTB, B2M, GAPDH, HPRT1, RPLP0), 1 genomic DNA control (GDC) to assess contamination, 3 reverse transcriptase controls (RTC) that certify the efficiency of the reverse transcription step, and 3 positive PCR controls (PPC) consisting of an artificial DNA sequence certifying the test accuracy. Each 96-well plate includes SYBR ® Green-optimized primer assays for a thoroughly researched panel of 84 EMT genes, that also are included the collagen and integrin genes. Furthermore, the high-quality primer design and RT 2 SYBR ® Green qPCR Mastermix formulation enable the PCR array to amplify 96 gene-specific products simultaneously under uniform cycling conditions. The samples were amplified using Applied Biosystems Step One Plus (Applied Biosystems, California, USA). The cycling conditions were as follows: 95°C for 10 minutes, 40 cycles at 95°C for 15 seconds, 60°C for 1 minute, followed by the dissociation period. The data were then analyzed in the StepOne software (v. 2.0, Applied Biosystems) using the D threshold cycle (Ct) method (2 −DDCt ) (17). All data were normalized by the housekeeping genes, and normal lung tissue specimens were used as case control. Using the EMT expression analysis, we created a heatmap of EMT gene expression across PNENs histological subtypes, which showed different levels of expression between 6 EMT genes (FC≥ 2.0). Among them, we found COL1A2, COL3A1, COL5A2, ITGA5, ITGAV and ITGB1 that were differentially expressed among PNENs histological subtypes.
Functional Enrichment Analysis of Collagen and Integrin Genes
To further elucidate the function and signaling pathways involved in the enrichment of the collagen and integrin genes, we inputted the selected genes plus ITGAV, totaling 6 genes, into Metascape (18) to perform Gene Ontology (GO) function, KEGG, and REACTOME pathway analyses. The GO analysis was composed of 3 categories, namely, biological processes (BPs), cellular components (CCs), and molecular functions (MFs). Results that met the threshold value with P<0.05 were regarded as significant.
PPI Network Construction and Module Analysis
To reveal the functional interactions among the proteins encoded by these genes, the selected genes were uploaded into STRING tools to map their PPI network (18). Results that presented a combined interaction score of P>0.9 were considered to be significant.
Data Management and Statistical Analysis
Data were collected and managed using REDCap electronic data capture tools hosted at A. C. Camargo Cancer Center, in São Paulo, Brazil. Considering the non-normal distribution of our data, all statistical tests employed in this study to examine the difference between categories and groups were non-parametric tests. The chi-square test or Fisher's exact test, the nonparametric Kendall tau-b correlation coefficient and the Spearman's rank correlation coefficient were used to examine differences in categorical variables, whereas the Kruskal-Wallis test was used to detect differences in continuous variables between groups of patients. However, to analyze the demographic and clinicopathological characteristics of the patients, the Person's Chi-Square test was used for these categorical variables. Qualitative data were described using relative frequencies. Overall survival (OS) was defined as the interval from the date of biopsy or surgical resection to death and OS curves were estimated using the Kaplan-Meier method. The Cox proportional hazards model was then used to analyze the association between OS rate and other covariances, and only parameters that presented P ≤ 0.02 in a univariate analysis were considered for multivariate analyses. We used the Statistical Package of Social Science (SPSS) version 18 for all statistical analyses. All tests with P<0.05 were deemed statistically significant and a Bonferroni correction was used when necessary.
RESULTS
Discovery Cohort: Differential Gene Expression Profiling of Primary PNENs Table 1 summarizes the clinical characteristics of patients, stratified by histological types. LCNEC and AC tended to be more frequent in female than male patients (3, 75.0% and 4, 80.0%). As for median age, a similar distribution was found across all histologic types. As expected, a history of tobacco smoking was more associated with SCLC when compared to LCNEC and carcinoid tumors, with statistical significance. Pathological stage showed a difference among the PNETs and PNECs histotypes. While all patients with carcinoid tumors were in an early stage of the disease, those with neuroendocrine carcinomas (SCLC and LCNEC) were in an advanced stage (P<0.01). Before surgical resection, patients with SCLC (3, Table 2 summarizes the distribution of the six genes studied in our cohort among histological types. Five of them -COL1A2, COL3A1, COL5A2, ITGA5, and ITGAVshowed a similar pattern of gene expression. Their expression increased as we moved from normal lung tissue to LCNEC and SCLC, where they were overexpressed, and decreased from normal lung tissue to carcinoid tumors, where they were downregulated (P<0.05). However, ITGB1 differed from the rest of the set and was overexpressed only in SCLC patients. In LCNEC and AC, its expression was lower than that of normal lung tissue and, in TC, expression was almost equal to that of normal lung tissue. The Figure 1 shows the box plots of COL and ITG gene expressions among PNENs histological subtypes.
Genes from the COL family seemed to play a particularly important role in several metrics. First, a significant association was found between COL genes (COL1A2, COL3A1, and COL5A2) and ITGA5, ITGAV, and ITGB1 (P<0.05) ( Table 3), but a correlation also emerged between COL and ITG expressions and disease progression profile. For instance, the overexpression of COL1A2, COL3A1,and COL5A2 in the TME was strongly associated with T stage, N stage, and pathological stage (P<0.05, in all three cases). M stage, in turn, was only significantly associated with COL3A1 (P=0.04). Similarly, the overexpression of ITGA5 and ITGAV were strongly associated with N stage, and pathological stage, (P<0.05), while ITGB1 was associated with T stage and N stage, (P=0.03 and P=0.00, respectively), as shown in Table 4 and Supplementary Figures 1 and 2. Functionally, the six fibrogenic genes (COL1A2, COL3A1, COL5A2, ITGA5, ITGAV and ITGB1) were involved in biological pathways associated with transmission of molecular signals to drive the activation of the resident fibroblasts into cancerassociated fibroblasts (CAFs) able to secrete and assembly the collagen fibers (cross-linking), and promote a desmoplastic reaction characterized by increased stiffness of the stroma for invasion and metastasis. In the Metascape analysis, the GO produced a list of top-level biological process and a heatmap of enriched terms related to the input genes, which included: "PID integrin 1 pathway", "extracellular matrix organization", "assembly of collagen fibrils and other multimeric structures", and "ECM-receptor interaction" (Supplementary Figure 3A). We then consulted the enrichment analysis in DisGeNET, where these six matricellular-associated genes seem to be involved in several diseases, as shown in Supplementary Figure 3B. Finally, Supplementary Figure 3C shows the PPI network formed in relation to biological processes; these enriched terms were produced using the following databases: BioGrid6, InWeb_IM7, OmniPath.
The PPI network with the proteins that are encoded by these genes using the STRING database included the 11 functional partners with the highest interaction score, namely COL1A1, COL1A2, COL3A1, COL5A2, LUM, ITGA5, ITGAV, ITGB1, ITGB3, ITGB6 and ITGB8. The edges represent proteinprotein associations that contribute to a shared function and involve only those proteins with a high edge score (confidence ≥ 0.9). These proteins' molecular organization can be visualized as a network of differentially connected nodes shown in Supplementary GSE1037 profile, as compared with normal samples, were determined using a GEO2R online analyzer (log FC>2 and adjusted P<0.05) and then compared to the data from our cohort. We observed a similar expression of collagens between the groups, mainly in the PNEC group, although the gene expression in our cohort was more expressive. However, some differences were detected in the expression of integrins. The expression of ITGA5 was inversely proportional between the groups: according to our data, this gene was overexpressed in PNECs and underexpressed in PNETs, whereas the GSE1037 data showed the opposite pattern. Similarly, while in our data ITGB1 was overexpressed mainly in PNECs, in GSE1037 it was mostly overexpressed in PNETs. A heatmap between fold changes was then established to show genes whose expression differed between our gene profile data (Supplementary Figure 5A) and the GSE1037 gene profile data (Supplementary Figure 5B).
COL and ITG Modulate Overall Survival and Risk of Death
A preliminary examination of the Kaplan-Meier survival curves generated in this study demonstrated that patients with advanced stage of disease (III/IV) had worse overall survival (OS) compared to those in early stage (I/II), 12 vs. 151 months (P=0.000). Thus, we coded the overall pathological stage as a single dummy variable with a value of 1 for stages I and II, and a value of 2 for stages III and IV.
The results of the Cox model analysis are found in Table 5.
Univariate Cox proportional hazards analyses showed that tobacco history, clinical stage, distant metastasis, and the expression of collagen and integrinwith exception to ITGAVwere significant predictors for OS and risk of death. These variables were accounted for in the multivariate analysis, in order to explore their independent predictive effect for OS.
The most effective survival model, under the conditions of the study, was controlled by tobacco history, distant metastasis, and COL5A2, ITGA5, and ITGB1 expression, all of them variables represented as co-dependent factors in the model. Distant metastasis was also presented as an independent factor of OS and risk of death (P=0.034). The chi-square including the covariates was 16.38 (P<0.01). In the Kaplan-Meier plots, the top curves represent the group with low expression of COL5A2, ITGA5, and ITGB1, whose median survival was quite long (112.78, 126.38, and 113.54 months, respectively). By contrast, patients with high expression of COL5A2, ITGA5, and ITGB1 (bottom curves) had a median survival time of just 16.91, 13.50, and 15.15 months, respectively (P<0.05, by log-rank test), as shown in Figure 2.
DISCUSSION
In the present study, we evaluated gene expression profiles using a set of twenty-four patients with surgically resected PNENs, including SCLC, and identified six fibrogenic genes: COL1A2, COL3A1, COL5A2, ITGA5, ITGAV, and ITGB1 up-regulated in PNECs and down-regulated in PNETs. We used two approaches where we first examined the six candidate fibrogenic genes using a whole-genome screen, and subsequently, validated the upregulation of these six genes in a similar independent validation cohort using in silico analysis. We additionally observed that low expression of three of these genes (COL5A2, ITGA5, and ITGB1) were significantly associated with metastatic-free and overall survival in PNENs. Our findings suggest that incorporation of collagen/integrin gene expression profile to routine genome-wide examination of biomarkers helps to predict metastasis in pulmonary neuroendocrine neoplasms and may be a promising tool to select and customize therapy. Although treatment options for metastatic PNENs have increased over the past decade, mortality and 5-year survival remain little altered for PNETs (19) and PNECs (20,21). Molecular studies identified somatic mutations, somatic copy numbers and pathway alterations in primary PNENs tumors (6)(7)(8); however, less is known regarding the effects of fibrogenic genes over the more lethal and therapeutically relevant stromaassociated fibrosis. Thus, studies that interrogate fibrogenic genes in tumors are critical to understanding the biology of invasion and metastasis in these tumors, the major cause of patient mortality.
The process of cancer cell invasion and metastasis undoubtedly comprises a series of complex, sequential stages, but among these the high collagen and integrins expression levels signalized by cancer-associated fibroblasts (CAFs) resulting in stroma-associated fibrosis is thought to be important because facilitates the migration of tumor cells and penetration of tumor by blood vessels (22)(23)(24)(25)(26). In order to understand the roles of fibrotic reaction in metastatic process, we explored the mRNA level of the different fibrillar collagens. We found that COL1A2, COL3A1, COL5A2, tumor mediators targeting of relevant structural components of the extracellular matrix (ECM) were activated to drive CAFs to synthetize fibrillar collagen creating an aberrant microenvironment (22). Convincing reports propose that a normal microenvironment avoids premalignant cells from developing into cancer, whereas an atypical or scarring repairassociated microenvironment can be tumor-promoting (27). The disruption in tissue homeostasis activates matrix fibroblasts into CAFs to synthetize collagen I, III and V (22), which in turn lead to a fibrotic repair of tumor stroma which is a major player in the development and progression of many cancers, including lung cancer, pancreas, breast, and hepatic carcinomas (28)(29)(30).
In the above scenario, fibrillar collagen types are the key actors in tumor stroma-associated fibrosis (also called desmoplasia), which is defined as a fibrotic state characterized by an excessive synthesis, deposition and remodeling of fibrillar a Univariate analysis was carried out without any adjustment in order to generate hazard ratios with confidence intervals for individual risk for each of the parameters on survival; b Multivariate analysis was carried out to analyze the effects of several risk parameters on survival; c HR, hazard ratio (b coefficient); d CI, confidence interval. Univariate and multivariate analysis employed a Cox proportional hazards model. Chi-square 16.38, P=0.006. † 8th International Association for the Study of Lung Cancer (14).
collagens surrounding the tumor (31,32). Collagen represents the most abundant ECM protein and collagen I, III and V deposition have been associated with increased desmoplasia leading to increased incidence of tumor formation and metastasis (33). A study found a gene expression signature that distinguished primary and metastatic adenocarcinomas and predicts the metastatic probability of these tumors; and a considerable proportion of the gene-expression signature described was composed of tumor COL gene expression, such as COL1A1 and COL1A2, both drivers of CAFs to synthesize collagen I fibers deposition resulting in stroma-associated fibrosis (34).
We show that, additionally to COL1A2, COL3A1 and COL5A2 were also up regulated in PNECs and downregulated in PNETs. Although collagen V is a minor constituent of the ECM compared to collagen I and III, collagen V is essential for fibrillogenesis, as its deletion leads to inability of collagen fibril assembly resulting in fibrotic stroma (35). Moreover, other studies observed that increased expression of collagen V individualizing malignant cells conferred an abnormal tumor stroma-associated fibrosis to local invasion and recurrence in malignant mesothelioma (36,37). In lung adenocarcinoma and breast cancer was observed that decreased expression of collagen V organized in an irregular texture of thin fibers involving large groups of malignant cells, facilitated invasion in a poor tumor stroma-associated fibrosis (36), also coinciding with previous work by Souza et al. (38). Hitherto, it was demonstrated that increased tumor stroma-associated fibrosis proportion in lung adenocarcinoma predicted a low risk of metastasis (39). According to the literature, this double edge sword of fibrosis in cancer due to collagen deposition is that this protein evokes multiple, and sometimes opposite, cellular responses, depending on the cell type (40). For instance, while collagen I and collagen V have been proven to represent an optimal substrate in fibrotic stroma for the attachment and growth of certain tumor cell types (41), this collagen species plays an antiadhesive and antiproliferative role in breast cancer cells (42). Furthermore, we showed a bimodal behavior of COL1A2, COL3A1 and COL5A2 up-regulated in PNECs and downregulated in PNETs. As PNENs are tumors with a limited stroma, the question is how the tumor cells migrate to gain access into vessels? We inferred that fibrotic genes in PNENs drives the migration of cells depending on collagen density in tumor stroma. In PNETs, downregulated fibrotic genes drive low collagen synthesis by CAFs, resulting in a loose fibrosis allowing tumor cells moving fast using pseudopodial protrusions as previously reported (43). In contrast, upregulation of fibrotic genes in PNECs increases deposition of fibrillar collagens making cell movements rely more extensively after collagenase cleavage of collagen fibers (44).
We have also found a strong quantitative relationship between COL1A2, COL3A1, COL5A2 and ITGA5, ITGAV and ITGB1 genes. As expected functionally, COL1A2, COL3A1, COL5A2, ITGA5, ITGAV, and ITGB1 genes were mostly involved in ECM remodeling, collagen fibers deposition, fibril harmonization, and cell adhesion. Elevated ITG signaling activates COL genes to deposit collagen proteins (fibers) in ECM creating a fibrotic stroma. Protein-protein interactions network showed 11 functional partners with the highest interaction score, namely COL1A1, COL1A2, COL3A1, COL5A2, LUM, ITGA5, ITGAV, ITGB1, ITGB3, ITGB6 and ITGB8. This signaling facilitates movement and migration of tumor cells between fibrotic stroma (45). Furthermore, increased stroma stiffness, a consequence of elevated expression of COL genes and collagen fibers deposition, causes activation of MAPKs and Rho-GTPases also via ITG signaling. These pathways are strong stimulators of fibrotic reaction for tumor migration, invasion, and metastasis (46). Integrin ITGB1 promotes cell invasion by sensitizing cancer cells to the changes in the fibrotic stroma (47), while Rho-GTPases are indispensable in the regulation of cell migration and control of multiple aspects of M phase and G1 progression of the cell cycle (46).
Recent studies have shown that several COL and ITG genes are increased in a variety of tumors, and have been associated with unfavorable outcomes. Among these, COL1A2, COL3A1, COL5A2, ITGA5, ITGAV, and ITGB1 have been especially noted (48)(49)(50)(51)(52)(53). Therefore, for all these reasons, we realize that COL1A2, COL3A1, COL5A2, ITGA5, ITGAV, and ITGB1 genes expression provide important predictive information about metastatic-free and overall survival in PNENs and our results now confirm the predictive importance of COL and ITG genes in PNENs. Whereas prior studies about lung cancer were able to show a significant relationship between COL5A2, ITGA5, ITGAV, and ITGB1 gene expression only in nonsmall cell lung cancer (54)(55)(56)(57)(58), our results suggest that COL5A2, ITGA5, and ITGB1 expression in PNENs, used as co-dependent variables, provide more information about the risk of metastasis and overall survival than does pathological stage. Moreover, the predictive value of COL5A2, ITGA5, and ITGB1 expression in PNENs persisted in the subset of patients with pathological stage I and II. In this context, we were able to identify two groups: patients with an expected low risk of metastasis and better overall survival versus patients with an expected high risk of metastasis and poor overall survival. Therefore, evaluation of the primary tumor for COL and ITG give us tools to guide the use of targeted therapy in patients expected to fail after surgical resection of PNENs.
In summary, the results presented herein provide important molecular evidence that collagen/integrin profiling are involved in the stroma-associated fibrosis to facilitate metastatic potential of PNENs. Specifically, our study indicates that collagen/integrin genes are up-regulated in PNECs and down-regulated in PNETs and thus potentially offer insight into novel therapeutic targets. Overall, these fibrogenic genes may represent partially an ECM 'remodeling' program to drive metastatic establishment. Preclinical studies are warranted, therapeutically, to select target up-regulated fibrogenic genes, mainly in SCLC, while maintaining proper ECM integrity in normal tissue.
DATA AVAILABILITY STATEMENT
The original contributions presented in the study are publicly available. This data can be found here: https://www.ncbi.nlm.nih. gov/geo/query/acc.cgi?acc=GSE181381.
ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics Committee for Research Project Analysis (CAPPesq) of the Hospital das Clıńicas, Faculty of Medicine of University of São Paulo. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements. | 2021-08-11T13:32:01.148Z | 2021-08-11T00:00:00.000 | {
"year": 2021,
"sha1": "9a01f7b3929603eccf651e361c7a19e9259ebf79",
"oa_license": "CCBY",
"oa_url": "https://www.frontiersin.org/articles/10.3389/fonc.2021.706141/pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "9a01f7b3929603eccf651e361c7a19e9259ebf79",
"s2fieldsofstudy": [
"Medicine",
"Biology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
233223391 | pes2o/s2orc | v3-fos-license | Fire-induced rock spalling as a mechanism of weathering responsible for flared slope and inselberg development
Inselbergs, such as Uluru in central Australia, are iconic landscape features of semi-arid and deeply denuded continental interiors. These islands of rock are commonly skirted by steep, overhanging cliffs (flared slopes) at ground level. The weathering processes responsible for formation of flared slopes and steep-sided inselbergs in flat, planated landscapes are enigmatic. One model emphasizes sub-surface weathering followed by denudation and excavation of saprolite to expose the unweathered bedrock while other models advocate slope development under subaerial conditions at ground level. We present a new hypothesis that identifies wildfire as a primary agent of flared slope development via fire-induced rock spalling around the periphery of inselbergs. Widespread fire-spalling following the 2019–2020 Australian fires illustrates that this is a common form of physical weathering in fire-prone environments but its effects are particularly evident in semi-arid regions where lateral fire-spalling dominates over fluvial and chemical weathering to create flared slopes and steep-sided inselbergs.
A ncient landscapes, such as central Australia, are characterised by deeply weathered bedrock and the development of well-preserved Cenozoic, multi-stage, and regolith profiles [1][2][3] . Isolated inselbergs, bornhardts, and mesas strewn across flat landscapes hint at a former higher relief [4][5][6][7] . Iconic examples of these steep sided inselbergs in Australia, include Uluru, Murphy Haystacks, Pildappa Rock, Burringurrah (Mount Augustus), and Katter Kich (Wave Rock) 4,5 , which rise steeply from the surrounding landscape ( Figs. 1 and 2) and in the case of Mt Augustus, may have been formed and exposed since at least the Jurassic 8 . The juncture between bare rock of the emergent inselberg and the surrounding plains covered by unconsolidated soil and sediment creates important sources of permanent water which promotes greater biodiversity in arid environments 9 . The peripheries of inselbergs are commonly the sites of rock shelters which represent important indigenous cultural sites that often host ancient rock art 10 , which in Australia, provides a record of the world's oldest living civilisation 11,12 . Whilst the history of occupation of these rock shelters is the focus of numerous studies, the geomorphological mechanisms responsible for flared slope formation in different rock types are poorly understood.
The main agents of bedrock weathering, erosion and sediment production in mountainous continental regions such as Europe, Scandinavia and North America are generally attributed to fluvioglacial processes or a combination of subaerial weathering and fluvial erosion in lower relief environments. Rates of vertical incision in active orogens vary between 1000 and 10,000 m Ma −1 (1-10 mm yr −1 ) 13 . However, flat, arid environments such as central Australia, are already denuded to local base levels across much of the planated landscape. This combined with very little rainfall, insignificant tectonic uplift, no recent volcanism and no glacial activity since the Permian or maybe the Cretaceous created flat landscapes that experience the slowest rates of erosion in the world 14 . Fission track studies [15][16][17] indicate relatively slow rates of 1-2 m Ma −1 (0.001 mm yr −1 ) of long-term landscape lowering throughout much of Australia during the Cenozoic. Cosmogenic studies reveal that climate is a major factor influencing rates of erosion with the lowest average rates of erosion of 1.5 m Ma −1 occurring in arid Australia whilst the highest rates, of 35 m Ma −1 are recorded from soil-mantled, spurs in humid temperate regions around the base of the SE escarpment 18 . Cosmogenic studies 14,19,20 of inselberg tops indicate rates of erosion of only~0.3-0.6 m Ma −1 (0.0003 mm yr −1 ) while rates around the periphery of inselbergs are an order of magnitude faster (3-3.9 m Ma −1 ) but highly variable 19 . Accordingly, the tops of inselbergs in Australia represent some of the most stable landscape features in the world. However, the development of steep, overhanging "flared slopes" around the periphery of the emergent rock hints at increased rates of weathering and the horizontal incision at or below ground level.
The term "flared slopes" is used to describe smooth, concave slopes at the junction between an emergent rock face and surrounding ground level 21 as seen at ground level around Uluru, Katter Kich, Murphy Haystacks, Pildappa Rock, Walga Rock and numerous other inselbergs in arid Australia ( Figs. 1 and 2). Previously, these concavities have been interpreted to be the result of subsurface, moisture-generated, weathering in the scarp foot beneath saprolite, sediment, or soil cover 22 . After prolonged subsurface weathering, the soft saprolite or unconsolidated sediment is evacuated during periods of landscape lowering to leave a concave slope in the bedrock. However, others 2 point out the obvious relationship between the ground surface and flared slope and suggest that the current ground surface acts as a local baselevel for the development of the overhanging slopes and only a thin veneer of soil covers the underlying, unweathered rock platform surface. Hence, they suggest a subaerial origin for flared slopes based primarily on the observation that so many of the rock platforms are coincident with the modern land surface and lack any evidence of weathered saprolite beneath the thin soil around the margins of the inselberg.
The steep, near-vertical slopes around the periphery of inselbergs and the development of overhanging flared slopes at ground level (Figs. 2 and 3) indicate that differential weathering processes are operating more rapidly laterally, around the flanks of inselbergs than vertically on the tops 22,23 . Many of these overhangs host important indigenous rock-art, which are the focus of effective preservation methods to reduce erosion 10,24 . Flared slopes are well developed but not restricted to arid Australia. Overhanging flared slopes are common around the edges of prominent granite domes in the New England and Lachlan orogens of eastern Australia (Fig. 4) and at the base of the sandstone escarpments in the Sydney Basin (Fig. 5). These areas were heavily affected by the intense "Black Summer" fires of 2019-2020 as well as previous fires in 2013 (Fig. 5d). Here, we show the role fire plays in physically weathering exposed rock surfaces using examples following the Black Summer fires and discuss how this might be responsible for flared slope development.
Results
Establishing the variables involved in rock weathering and fire behaviour is a key aspect of developing an accurate fire-induced rock spalling hypothesis. We expand on these variables by drawing on field observations and existing findings outlined below.
Mechanical weathering. The physical breakup and removal of rocks of varying hardness and degrees of weathering via mechanical weathering is the primary process that denudes and sculpts uplifted regions of Earth's surface. Sub-critical cracking describes the slow propagation of microfractures through a rock in low-stress, near-surface conditions as a result of thermal stress, ice wedging, mineral alteration (volumetric expansion) and biomechanical processes such as root growth 25 . Sheeting is characterised by thick (0.1-1 m) layers of rock peeling off exposed surfaces roughly parallel to the surface topography. There is debate as to whether sheeting is related to gradual unloading and release of stresses near the surface or a combination of other stresses 26,27 . The physical process of thermal expansion and contraction of rocks over thousands of years is responsible for the thinner, gradual flaking (exfoliation) of rock surfaces, which can be observed all over the surface of inselbergs in central Australia 5 and presumably the main process responsible for the slow rates of erosion at the tops of inselbergs 14 .
Fracture propagation is facilitated by the presence of water 28 , which helps to break chemical bonds leading to more fractured rock at shallow, superficial levels of the crust. Thus, rocks are generally more fractured in the superficial, near-surface environments than at deeper levels. Spontaneous rock-burst events were captured on video during a hot summer of 2014 in California when a granite dome at Twain Harte began explosively exfoliating 29 . Extreme thermal stresses associated with fire and lightning strikes are acknowledged as mechanisms of critical stress fracturing in rocks but generally considered to be a rare form of rapid and catastrophic mechanical weathering 25 . Our observations of rock surfaces following wildfires are that firerelated rock spalling is a commonly observed phenomenon wherever high-intensity fire has swept across rocky outcrops (for example, Figs. 5 and 6). We suggest that fire-spalling is a significant driving mechanism of physical weathering in arid, Wildfire temperatures. A detailed study of high-intensity wildfires in eucalypt forests of SW Australia 30 revealed that these fires burn at temperatures between 300°C at the tips of visible flames and up to a maximum of 1100°C near the flame base, while temperatures of up to 1330°C were recorded in Canadian crown fires 31 . Experimental fires conducted in jarrah forests of southwest Western Australia (Project VESTA) reveal that temperature correlates directly with the rate of spread, fire intensity, flame height and surface fuel bulk density 30 . This single case study measured the average flame-front residence time in eucalypt forest fuels of about 37 s. However, radiant heat and hot winds ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-22451-2 fanning out in front of the fire have the ability to pre-heat the rock surface and vegetation before and after the arrival of the fire front 31 particularly along cliff lines.
We report the first documented case of spalling in basalt from Mount Kaputar in northern N.S.W. (Fig. 4d). Basalt is a hightemperature volcanic rock with no quartz content. Fire-spalling was minimal across most of the outcrops and generally consisted of dislodged pyroxene phenocrysts. However, a few basalt outcrops adjacent to nearby fallen burnt logs were intensely scorched and displayed thin (1-4 mm) spalled flakes of basalt indicating that fire-spalling is not restricted entirely to quartz-rich lithologies. In mature eucalypt forests with large, woody fuels, termed 'down wood' 32 , fires can burn or smoulder for days, providing prolonged heat required for extensive spalling. Some cliff faces record distinct ghosted impressions of nearby tree trunks with the resultant spalling hollowing out the line and shape of a tree trunk in an otherwise flat, vertical rockface ( Fig. 5a -right-hand side). A discarded brown glass bottle adjacent to the basalt spalling had softened and undergone ductile collapse and partially melted. The glass had cooled slowly enough to avoid shattering indicating prolonged heating from the smouldering downward. This glass was collected and placed in a hightemperature oven where it was observed to become soft and malleable at 750°C and completely collapsed and started melting at 830°C indicating that this fire sustained ground surface temperatures of between 750 and 830°C next to the smouldering tree and fallen logs.
Fire-induced rock spalling. Fire is known to accelerate the rock flaking process 25,33-38 resulting in rock spalling 36,39 and shattering 38 . Conflagration leads to the rapid disintegration of the rock surface due to the differential expansion of the hot rock surface compared with the cooler interior. Fire-spalling can remove between 10 and 100% of the burnt rock surface in sheets between 5 and 50 mm thick 37 depending upon rock type and fire intensity. Detailed measurements of post-fire rock spalling after the Esperanza chaparral fire in California revealed that 7-55% of the granodiorite boulder surfaces were spalled to a depth of 11-24 mm 33 . They found that the thickest spalled sheets occurred around the flanks of the boulders and cautioned that, if sampled for cosmogenic dating, these freshly exposed, spalled surfaces would produce a significant underestimate of exposure age. These figures match our own observations of spalled granite following the fires in Cobargo, Moonbi and Thredbo N.S.W. (Fig. 4) in which granite boulders spalled sheets between 5 and 50 mm thick, while sandstones from the Blue Mountains spalled sheets between 5 and 22 mm thick (Fig. 5).
Quartz expands four times more than feldspar and twice as much as hornblende and shows a 3.76% volume expansion when heated from room temperature to 570°C 40 . Thus, quartz-rich rocks have a greater expansion potential and are more likely to spall. Experimental studies 41 show that rock elasticity reduces significantly at temperatures as low as 200°C, over a relatively short period of time. Goudie et al. 41 postulated that rock outcrops subject to intense fires would have an increased susceptibility to erosion via spalling and weathering. However, these findings have not been applied to broader landscape models or the formation of flared slopes around inselbergs.
Fire regimes. The potential rate of erosion due to fire-spalling at the base of inselbergs will be strongly influenced by fire severity and recurrence intervals, which vary greatly across Australia from 1-to 5-year recurrence intervals and <10,000 kW m −1 for tropical savannas of the north, to > 100-year intervals and >10,000 kW m −1 for tall, open forests of the cool, temperate south 42 . Accurately calculating the fire return period is difficult due to limited historical records but estimates for arid, spinifex-dominated regions such as the Tanami are in the order of every 7-9 years 43 . Analyses of satellite data between 1998 and 2004 revealed that 27% of arid Australia burnt at least once over that 6-year period 44 . Figure 1a shows the areas burnt in Australia since 2001. The surface area of the rock affected by spalling depends on the rock-type and severity of the fire. Fire severity is strongly determined by the bulk density 45 , height and proximity of the adjacent vegetation to rock surfaces and the surrounding slope gradient. All the examples of flared slopes shown in Figs. 1 and 2 reveal a close relationship between the height of the encroaching vegetation and the height of the concavity. Katter Kich, Pildappa Rock and Walga Rock form distinct embayments where the flared slopes are most pronounced, which appear to promote denser, taller vegetation growth and hence greater fuel loading and thus higher fire severity (Fig. 3).
The impermeable nature of inselbergs results in rapid and efficient water runoff from the bare-rock surface before draining into adjacent, thin soil profiles. This creates a "roof and gutter" effect around the periphery of many inselbergs which creates permanent water holes and shallow groundwater within easy reach of deep-rooted plants. Inselbergs create important geodiversity within otherwise flat landscapes and thus host important niche ecosystems that add to the overall biodiversity of desert regions 46 . Accessible groundwater around the fringes of the inselbergs encourages denser, taller vegetation at the interface between bare rock and unconsolidated surficial sediments which in turn increases the fuel load. Inselbergs are prominent topographic features in flat deserts that provide sources of permanent water, abundant flora and fauna and shelter.
Grassy plains and savannahs of central Australia are characterised by regular, low-intensity fires with fire recurrence intervals between 1 and 5 years 42 . However, where these fires encounter inselbergs they move into thicker, taller vegetation regimes with greater fuel loads (Figs. 2 and 3). Inselbergs are topographic highs within relatively flat landscapes and the slight increase in slope gradient around the inselberg will accelerate and intensify an approaching fire front. Steep slopes around the margins of inselbergs possibly act as chimneys, drawing in hot air from the surrounding plains and channelling them upwards. These factors possibly help to draw in fires from the surrounding Fig. 2 Flared slopes at ground level around Uluru. a The steep-sided inselberg of Uluru in central Australia. b, c Flared slope development at the foot of Uluru produces a remarkably uniform concave surface roughly 2 m high coincident with the level of the surrounding plain. Note the recently burnt trees and the height of the charcoaled trunks, which matches the height of the flared slopes, which themselves are blackened from this recent fire. d Fire-spalling on the arkose sandstone rock surface caused by a fire in 2012. e Sheeting and rockfall events produce large boulders at the foot of the inselberg which shield the inselberg from further fire-spalling and hence the lack of concave flared slope development behind the fallen boulders. f Rockfall boulders have themselves developed small caves on the side exposed to the vegetation and approaching fires indicating flared slope development occurred sub-aerially after the rockfall event. g Mala Puta Cave displays extensive flared slope and cave development at ground level. Mala Puta Cave is 2-3 m deep. In this example, the wet, black waterfall on the right highlights the difference in rates of fluvial incision compared to the rate of lateral erosion due to flared slope development. Overhanging caves develop on the right side due to the prevailing south-westerly winds in this region, which would control the direction from which most fire fronts would approach the rocks. plains into and around topographically high inselbergs where the intensity is enhanced at the base of inselbergs due to the denser vegetation and greater fuel load.
A fire-induced spalling weathering formula. Fire-spalling leads to physical weathering (erosion) and disintegration of exposed rock faces 36,37 as shown in Figs. 3-5. The degree and extent of spalling on different rock types and at varying temperatures and durations is less well understood and requires further experimental work 41 but essentially fire-spalling is a function of fire intensity (temperature), duration and rock type with quartzrich rocks having a greater propensity to expand and spall 40 .
We developed a simple fire-spalling erosion formula to estimate a long-term rate of fire-induced spalling that broadly considers the net result of fire-spalling in terms of the thickness (width) of the spalled flakes produced by a single fire event, the total surface area as a percentage of the exposed rock face affected by a single fire-spalling event, and the average fire recurrence interval for a given region. Together these variables can give some The formula for erosion due to fire-spalling.
where, E = rate of erosion due to fire-spalling (mm yr −1 ), W = average width (thickness) of spalled sheets (mm) for a single fire event. Dependent on rock type (quartz content and texture), rock strength, fire temperature and duration, A = area of rock surface affected by fire-spalling as a percentage (%) of total surface area. Dependent on temperature and duration of the fire, t = average fire recurrence interval (years). Determined from regional, historic fire records or palaeofire records for longer time periods. Dependent on vegetation, climatic regimes and land management practices.
Limitations: this equation applies to a near-vertical rock face at ground level which receives uniform heat radiation from a fire that burns right up to the rock face at ground level. The intensity of radiation will vary according to the dynamics of the fire front, fuel loading, vegetation type and slope gradient. Flame height is not critical to the overall rate of retreat of the cliff face because fire-spalling at the base of the cliff will gradually remove material supporting the cliff resulting in over-steepening at the base of the cliff and periodic sheeting and rockfalls as the overhanging cliff face becomes gravitationally unstable. The formula assumes that fire recurrence intervals have remained constant but we know from palaeofire records 47,48 that fire intensity and recurrence intervals are largely controlled by long-term climatic variations which affect vegetation types and thus fuel loads. Below, we give two end-member examples of long-term rates of spalling-related erosion for low and high-frequency fire regimes that may apply to temperate and arid environments, respectively. Example 1. Low intensity, irregular fire regime. In this scenario, the average fire against a cliff results in spalling and flaking of 10 mm sheets off~20% of the surface area at ground level during a single fire event. Fire recurrence interval is one event every 50 years. Example 2. High intensity, high-frequency fire regime. In this scenario, the average fire against a cliff results in spalling and flaking of~20 mm sheets (Fig. 5) off~80% of the surface area at ground level. Fire recurrence interval is one event every 5 years.
In an intensely fire-prone environment such as example 2 above, it may only take about 625 years of fire-induced spalling to weather out a 2 m deep flared slope at the base of a vertical rock face. The point at which undercutting due to fire-spalling would trigger massive sheeting of the unsupported, overhanging rock ledge and subsequent rockfall event is not well constrained but some of the flared slopes around Uluru and Walga Rock are at least 2-3 m deep (Fig. 2h).
Sediment production rates. If rates of erosion due to fire-spalling around the periphery of an inselberg are orders of magnitude greater than those across the top of the inselberg, then this has implications for mechanisms of sediment production in flat, arid environments like Central Australia.
Spalling of a 20 mm sheet from a 1 m 2 area of granite with a density of 2691 kg m −3 will yield 0.02 m 3 (53.82 kg) of rock. A flared slope around an inselberg such as Uluru with a circumference of~10,000 m and a height of 2 m, would produce 400 m 3 (1,076,400 kg) in a single event in which 100% of the 2 m high flared slope was spalled. Obviously, 100% spalling of the entire flared slope would never occur in a single event, so we use the long-term erosion rate based on fire recurrence intervals and average area spalled calculated in Eq. 1. This long-term estimate of sediment production from a single inselberg is compared with quantitative measurements of spalled granite surfaces sampled after the 2019-2020 fires in Cobargo on the south coast of N.S.W., Australia.
The formula for sediment production. This can be standardised to give a volume of rock spalled per year per square metre, which is the same as the erosion rate but in cubic metres per year. Given the density of the rock (granite = 2691 kg m −3 and compacted, meta-arkose sandstone (Uluru) are about the same) we can calculate the average mass of rock spalled each year. In the above scenario, it equals 8.61 kg per square metre per year.
The rate of background (non-fire related) sediment production (S BA ) from erosion of the surface area of an inselberg such as Uluru is equivalent to the surface area (~3,440,000 m 2 ) multiplied by the average denudation rate of~0.3-0.6 m/Ma (0.0003 mm yr −1 ) as established from cosmogenic studies. This equates to only 0.00081 kg per square metre per year. We estimate that fire-spalling on a 2 m high perimeter produces in the order of 64 times more sediment than the erosion of the entire surface of the inselberg due to background (non-fire related) processes.
Spalled granite material was collected from two locations following the 2019-2020 fires in the Cobargo region along the south coast of N.S.W. (Fig. 4) to assist in quantifying the amount of rock spalled from a single rock face. Spalled surface area can be estimated simply by measuring the maximum height and width of the spalled surface in the field. We also created a digital surface using photogrammetry MetaShapePro software to calculate a precise surface area of the spalled surface. All of the spalled material was weighed and a standard granite density of 2691 kg m −3 was used to determine total volume. Generally spalling occurs as thin (1-3 cm) sheets but occasionally includes large 20-30 cm thick slabs that substantially add to the overall weight of spalled material. Whilst complete spalling of a 2 cm sheet from one square metre of granite surface will produce 53.82 kg m −2 of rock, our two sites (Cobargo2 and 3A) produced 23.65 kg total (16.89 kg m −2 ) and 41.60 kg m −2 total (33.55 kg m −2 ), respectively, indicating an average spalling thickness of 0.63-1.25 cm although the spalled thickness was highly variable with spalling distinctly more prominent along sharp or protruding edges than on flat surfaces. Large logs or tree trunks have the potential to continue burning long after the fire front moves through and their presence near rock surfaces significantly increases the degree of spalling. The most intense spalling was observed at Moonbi Granite near Tamworth in northern N.S.W. where some granite boulders had 100% surface spalling up to 2 m above ground level and not one but several spalled sheets (5-20 cm total thickness) exfoliating off during a single, intense fire creating several hundred kilograms of spalled rock debris on the granite surface facing the fire front (Fig. 4). Likewise, lichen coated granites from Australia's most elevated alpine regions in the Snowy Mountains (Thredbo) displayed intense spalling but were covered in snow six months later. The fire recurrence interval for these alpine regions is probably in the order of one every 20-100 years thus the effects of fire-spalling are less pronounced than in arid regions and less evident than other forms of fluvial or chemical weathering that dominate in wetter climates. However, the abundant spalled surfaces shown in Fig. 4 reveal that large, intense fires such as the Black Summer fires of 2019-2020 will result in significant erosion and sediment production even in alpine environments.
Discussion
The development of vertical to overhanging flared slopes at ground level around the peripheries of inselbergs is direct observational evidence of the contrasting rates of vertical erosion operating slowly on the tops of inselbergs but much faster, lateral erosion at the base and around the periphery. We propose that exposure of rock surfaces to frequent, high-intensity wildfires over millennia creates a differential rate of lateral erosion at ground level around the periphery of inselbergs compared to slower vertical rates of erosion operating at the tops of the inselbergs. In arid areas where fluvial and chemical weathering processes are relatively slow-acting, fire-spalling results in the development of flared slopes and prominent inselbergs. Higher fire frequencies and intensities should accelerate lateral erosion at ground level, particularly along the thickly vegetated bases of inselbergs and escarpments. Even low-intensity fires such as the 2012 fire around Uluru resulted in minor spalling on flared slopes that were buffered from adjacent vegetation by a 2-5 m cleared walking path (Fig. 2). Fire-induced rock spalling and flared slope development undermines and potentially destabilises the outer shell of the inselberg at ground level, occasionally resulting in large-scale sheeting (~1-2 m thick sheets) and rock-fall events. Paradoxically, rockfall debris has the effect of temporarily shielding the inselberg from subsequent wildfire events resulting in no flared slope development on the inselberg surface behind the boulders, although the fallen boulders themselves are often subject to flared slope development on their exposed outer surface (Fig. 2e, f). We suggest that wildfire is largely responsible for the initial disintegration of rock in fire-prone environments and is an important process in the physical breakdown of rock and production of sediment, particularly in dry, flat continents such as Australia. We provide a simple fire-spalling erosion formula to predict the rates of fire-induced rock spalling for different fire regimes that can be applied to broader landscape models.
An inselberg such as Uluru has outcrop dimensions of roughly 2.5 × 1.5 km. The fire-spalling erosion Eq. 1 indicates that an intense fire regime with a fire recurrence interval of 5 years (Example 2), could result in a maximum, long-term, lateral erosion rates of about 3 mm yr −1 . Assuming that fire-spalling operates on all sides of the inselberg, this gives a combined rate of 6 mm yr −1 total. At this rate, the narrowest section of an inselberg the size of Uluru (1,500,000 mm) would be completely removed from the landscape in about 250,000 years. In reality, the rate of retreat would be much slower when we consider the shielding effect that rockfalls have in terms of temporarily protecting the inselberg and slowing the overall rates of rock spalling (Fig. 2g). Conversely, projected backwards, the periphery of an inselberg like Uluru may have been up to 200 m wider when humans first arrived in Australia some 65,000 years ago 49 . Twenty million years ago, Uluru was probably at the centre of a vast, undissected plateau stretching 120 km across from north to south and connected to other remnants of this ancient plateau, for example, Kata Tjuta (the Olgas) to the west and Attila (Mount Connor) to the east (Fig. 1).
We present a graph (Fig. 6c) that plots the rates of erosion due to fire-spalling (E (t) ) vs. fire recurrence intervals (t) to illustrate the formula in Eq. 1. Rock type and fire intensity control the thickness of spalled sheets with quartz-rich rocks such as granite and quartz sandstones displaying the most intense degrees of spalling (2-10 cm thick). In contrast, rocks such as basalt rarely spall and if so the spalled sheets are only a few millimetres thick and only occur as small patches and only in the most intense heat-affected rockfaces. Weathering of mafic rocks such as basalt is probably controlled primarily by the chemical breakdown of unstable, high-temperature minerals such as olivine, pyroxene and plagioclase rather than the physical process of spalling. The degree of spalling on a rock face is measured as a percentage of the total surface area and is controlled mostly by fire intensity and duration. The Moonbi Granite in Fig. 3 showed evidence of 3-4 sheets each about 2 cm thick being spalled off during one intense fire resulting in 100% surface spalling with a thickness of 5-20 cm. This was the most intensely spalled example we have encountered but given we were restricted to roadside outcrops we suspect there are more intense examples of spalling to be discovered. Whilst the relationships between rates of erosion with the thickness (W) and area (A) are linear, there is an inverse correlation with fire frequency (t) that creates a non-linear hyperbolic relationship (Fig. 6c). Dry, arid regions with short fire recurrence intervals of only 5-10 years (left side of the graph) will potentially experience increasingly higher rates of fire-spalling erosion than wet temperate regions with fire recurrence intervals of >50 years (right side). Examples of fire-spalling are added to provide context to the different rock types and climatic regimes across Australia.
In stark contrast, the rates of background, fluvial-related and/ or flaking-related erosion on top of an inselberg are an order of magnitude lower at about 0.3 m Ma −1 14 . At these slow rates, Uluru which sits 348 m above the surrounding plain, would take in the order of 1.16 billion years to be completely planated. This rate is obviously much slower than the overall rates of denudation for these regions and does not reflect the rate of scarp retreat and inselberg formation observed around rocks that are themselves are only a few hundreds of millions of years old, as for example, the heavily spalled Permian-Triassic granites in the New England Orogen of eastern Australia (Fig. 4).
Whilst fire-spalling was identified as a mechanism of physical weathering as early as 1927 36 it was largely regarded as an isolated, local phenomena. We recommend that fire-spalling needs to be incorporated into erosion and landscape evolution models that currently tend to focus almost entirely on fluvial, glacial, periglacial and erosive mass wasting processes operating within drainage basin catchments 50 . Wildfires are high-energy, episodic events that disintegrate rock and present a massive departure from the slow steady-state of water-based weathering and erosion operating during non-fire conditions. Palaeofire records in South Australia reveal evidence of episodic fire-related erosion events recorded in upland peat bogs of the Adelaide Hills 47 . This supports the notion that sediment mobilisation in vegetated fireprone regions is triggered by fire events that expose the soil directly to subsequent rainfall events leading to distinct pulses of hillslope erosion and sediment movement 51,52 . Statistical analysis of 223 fire records across Australia indicates climatic variations control fire regimes, with colder periods characterised by less burning and warmer intervals by more 48 . However, the records did not show any significant change to fire regimes with the arrival of humans and fire-stick farming techniques some 50-60 thousand years ago. This suggests that climate is the primary driver of fire regimes. We suggest that fire is not only important in terms of mobilising sediment reservoirs 52 through the temporary removal of binding vegetation cover, but also in generating new sediment, particularly around the peripheries of inselbergs and along escarpment fronts in hot, fire-prone continental interiors such as central Australia. Accurate long-term landscape models would need to adjust the fire recurrence interval according to climatic changes associated with glacial and interglacial cycles and possibly the more intense fire regimes associated with modern land management practices and a warming climate. Australia has just experienced an unprecedented fire season with over 18.6 million hectares burnt, mostly along eastern Australia 53 . The volume of rock fragments spalled from rock exposures after a mega-fire event of this severity represents a massive departure from "normal" background rates of bedrock erosion as well as subsequent soil/sediment mobilisation during rainfall events following the fires. To put this in perspective, if a conservative estimate of just 0.1% of the 18.6 million hectares of burnt areas across Australia in 2019-2020 were rocky outcrops affected by fire-spalling, on average 1 cm thick; then this would equate to over 5 million tonnes of spalled rock during the 2019-2020 fire season. If 1% of the burnt area spalled 2 cm of rock then the amount of sediment produced is closer to 100 million tonnes. Accurate estimates will vary depending upon the terrain, the outcrop, rock-type, fire intensity and whether the fires were in mountainous regions with abundant rocky outcrops.
We estimate that rates of lateral erosion due to fire-spalling in dry, fire-prone regions of central Australia may be up to 3.2 mm yr −1 which is not a lot less than the rates of vertical fluvial or glacial erosion (1-10 mm yr −1 ) measured in tectonically active mountains 13 and is four orders of magnitude (10,000 times) greater than the vertical rates of denudation (0.0003 mm yr −1 ) operating at the top of inselbergs. This differential erosion pattern is responsible for maintaining steep-sided inselbergs until complete planation. Fire-spalling acting on a 2 m high rock face around the periphery of an inselberg, such as Uluru, would appear to generate about 64 times more sediment than the slow rates of vertical denudation acting across the top of the entire inselberg. Fire-spalling only works laterally at ground level where there is sufficient fuel load to generate fires of enough severity to spall fresh rock. Fire will not have any significant effect on vertical denudation in a flat environment because soil and regolith insulate and protect the underlying rock from the effects of surface fires. Fire-spalling denudation and sediment production cease once the inselbergs are flattened however subsequent fires do remove vegetation and aid in the mobilisation of surface sediment via water or wind activity 52 .
The process of fire-spalling requires more quantitative studies including detailed photogrammetry and LIDAR surveys of flared slopes before and after fire events to establish the volume of mass wasting during fire events. Additional exposure age studies including cosmogenic exposure ages 14 and newly developed luminescence surface exposure dating 54,55 of exposed surfaces around the flanks of inselbergs might also help to establish the timing of past fire-spalling events and build more robust paleofire records. Likewise, lichenometry 56,57 , which is used to date Holocene glaciated terranes or landslides, might provide a method of calculating the age since the last fire-spalling event on a rock surface given that lichens and mosses are completely burnt and removed by fire-spalling at ground level but survive a few metres above the spalled rock surface (Fig. 4).
Recognition of fire-spalling as a major mechanism of weathering has relevance to the debate surrounding the formation of the steepened and flared margins of inselbergs. Two conflicting hypotheses for the origins of inselbergs have long been aired in multiple papers by and Twidale 21,61,62 and Twidale and Bourne 63,64 as well as in many other publications. King favoured inselberg formation by the parallel retreat of slopes in bedrock (pediplanation) over vast distances following initial valley incision. On the other hand, Twidale invoked a two-stage model of deep weathering of vulnerable rock and the stripping of the weathered material upon uplift, thereby exposing the weathering front and the more resistant unweathered bedrock as depicted in Fig. 6a. King noted the absence of deep weathering in some inselberg areas, as in the Valley of a Thousand Hills in Natal, South Africa, as well as incongruities between the depths of weathering and the heights of the inselbergs. Following this, Twidale introduced the notion of the episodic exposure of inselbergs through multiple stages of deep weathering and stripping, with the flared slopes representing the mere retouching of the outer flanks of the inselbergs. It was argued that hypotheses had been advanced and tested against the field evidence before arriving at the above theory, which was considered to be most likely to be correct. However, it was conceded that the conclusion could be modified if further evidence demonstrated its inadequacy 6 .
We consider that the role of fire-related weathering in the development of inselbergs and flared slopes may be a missing link in the previous theories proposed (Fig. 6b). Fire is a very effective mechanism of weathering, even of fresh bedrock, while the rates of weathering suggested are commensurate with the large-scale parallel retreat of slopes, and the modification and further development of the margins of inselbergs. Fire sculpting is an important agent of geomorphic change in any fire-prone environment but is particularly evident in hot, dry, non-glaciated and tectonically inert continental regions like Australia, where we hypothesise fire-spalling dominates over fluvial and chemical weathering to create flared slopes and steep-sided inselbergs. The role of fire-spalling requires consideration in the lateral erosion of inselbergs, scarp retreat, sediment production and landscape evolution models in fire-prone environments. The presence of flared slopes around the periphery of inselbergs and escarpments in arid environments where fluvial and chemical weathering processes are only operating slowly is a potential indicator of some of the most intensely, fire-affected regions on Earth and the role of fire as an agent of long-term weathering and erosion on a flammable planet.
Methods
This study is based on detailed field observations presented in the context of existing findings from published literature to develop a working hypothesis of fireinduced rock spalling. Field measurements of spalled flake widths and spalled surface areas were taken using an 8 m tape measure to the nearest mm and a portable luggage scale was used to measure the weight of spalled rock to the nearest 0.01 kg. All measurements are described in the text or annotated on the figures. Data sources are provided in captions.
Data availability
Data sharing not applicable to this article as no datasets were generated or analysed during the current study. Source data for Fig. 1 is provided with the paper. Locations (latitude and longitude WGS84) of new, un-named field sites pictured in Figs. 4 and 5 are provided in figure captions. | 2021-04-14T06:17:13.753Z | 2021-04-12T00:00:00.000 | {
"year": 2021,
"sha1": "4b4c4172974e5cc20f97f6ffcf114c61d61ab956",
"oa_license": "CCBY",
"oa_url": "https://www.nature.com/articles/s41467-021-22451-2.pdf",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "b78b147dd0f75147bd64f7ce645aaecf7a52a7d7",
"s2fieldsofstudy": [
"Environmental Science",
"Geology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
11717703 | pes2o/s2orc | v3-fos-license | Relation Classification via Recurrent Neural Network
Deep learning has gained much success in sentence-level relation classification. For example, convolutional neural networks (CNN) have delivered state-of-the-art performance without much effort on feature engineering as the conventional pattern-based methods. A key issue that has not been well addressed by the existing research is the lack of capability to learn %in modeling temporal features, especially long-distance dependency between nominal pairs. In this paper, we propose a novel framework based on recurrent neural networks (RNN) to tackle the problem, and present several modifications to enhance the model, including a max-pooling approach and a bi-directional architecture. Our experiment on the SemEval-2010 Task-8 dataset shows that the RNN model can deliver state-of-the-art performance on relation classification, and it is particularly capable of learning long-distance relation patterns. This makes it suitable for real-world applications where complicated expressions are often involved.
Introduction
This paper focuses on the task of sentence-level relation classification. Given a sentence X which contains a pair of nominals x, y , the goal of the task is to predict relation r ∈ R between the two nominals x and y, where R is a set of pre-defined relations (Hendrickx et al., 2009).
Conventional relation classification methods are mostly based on pattern matching, and an obvious disadvantage is that high-level features such as tags of part of speech (POS) and name entities are often involved. These high-level features require extra NLP modules that not only increase computational cost, but introduce additional errors. Also, manually designing patterns is always time-consuming with low coverage.
Recently, deep learning has made significant progress in natural language processing. Collobert et al. (2011) proposed a general framework which derives task-oriented features by learning from raw text data using convolutional neural networks (CNN). The idea of 'learning from scratch' is fundamentally different from the conventional methods which require careful and tedious feature engineering. Collobert et al. (2011) evaluated the learning-based approach on several NLP tasks including POS tagging, NER and semantic role labelling. Without any human-designed features, they obtained close to or even better performance than the state-of-the-art systems that involve complicated feature engineering.
A multitude of researches have been proposed to apply the deep learning methods and neural models to relation classification. Most representative progress was made by Zeng et al. (2014), who proposed a CNN-based approach that can deliver quite competitive results without any extra knowledge resource and NLP modules. And there are other valuable models such as MV-RNN (Socher et al., 2012), CNN(Nguyen and Grishman, 2015), FCM (Yu et al., 2014).
Despite the success obtained so far, most of the current deep learning approaches to relation learning and classification are weak in modeling temporal patterns. Note that the semantic meaning of a relation is formed in the context of the two target nominals, including the word sequence between them and a window of preceding and following words. Additionally, the relation is in fact 'directional', which means the order of the context words does matter. Therefore, relation learning is essentially a task of temporal sequence learning, and so should be modelled by a temporal model. Most of the current deep learning models men-tioned above are static models, and are potentially weak especially when learning long-distance relation patterns. For example, the CNN model can learn only local patterns, and so is hard to deal with patterns that is outside of the window of the convolutional filter.
In this paper, we propose a novel framework based on recurrent neural networks (RNN) to tackle the problem of long-distance pattern learning. Compared to other models such as CNN, RRN is a temporal model and is particularly good at modeling sequential data (Boden, 2002). The main framework is shown in Figure 1, which will be described in details in Section 3.
The main contributions of this paper are as follows: • Proposed an RNN-based framework to model long-distance relation patterns.
• Verified the proposed approach on the SemEval-2010 task-8 dataset and obtained state-of-the-art performance.
• Analyzed empirically the capability of the RNN-based approach in modeling longdistance patterns.
Related Work
As mentioned, the conventional approaches to relation classification are based on pattern matching, and can be categorized into feature-based methods (Kambhatla, 2004;Suchanek et al., 2006) and kernel-based methods . The former category relies on human-designed patterns and so require expert experience and time consuming, and the latter category suffers from data sparsity. Additionally, these methods rely on extra NLP tools to derive linguistic features.
To alleviate the difficulties in pattern design and also the lack of annotated data, distant supervision has drawn a lot of attention since 2009 (Mintz et al., 2009;Riedel et al., 2010;Hoffmann et al., 2011;Surdeanu et al., 2012). This technique combines resources of text data and knowledge graph, and uses the relations in the knowledge graph to discover patterns automatically from text data. However they still depend on NLP tools.
Our work follows the line of automatic feature learning by neural models, which is largely fostered by Collobert et al. (2011). A closely related work was proposed by Zeng et al. (2014), which employed CNN to learn patterns of realtions from raw text data and so is a pure feature learning approach. A potential problem of CNN is that this model can learn only local patterns, and so is not suitable for learning long-distance patterns in relation learning. Particularly, simply increasing the window size of the convolutional filters does not work: that will lose the strength of CNNs in modeling local or short-distance patterns. To tackle this problem, Nguyen and Grishman (2015) proposed a CNN model with multiple window sizes for filters, which allows learning patterns of different lengths. Although this method is promising, it involves much more computation, and tuning the window sizes is not trivial. The RNN-based approach solves the difficulty of CNN models in learning long-distance and variable-distance patterns in an elegant way.
Another related work is MV-RNN model proposed by Socher et al. (2012). The difference is that we based on different RNNs: the MV-RNN model is based on recursive NN while our work is based on recurrent NN, a temporal model. Additionally, MV-RNN relies on syntactic parsing, and our model uses only word vectors and so is more efficient especially in predicting process.
Finally, our work related to the FCM framework proposed recently (Yu et al., 2014). In principle, FCM decomposes sentences into substructures and factorizes semantic meaning into contributions from multiple annotations (e.g., POS, NER, dependency parse). It can be regarded as a general form of the MV-RNN and CNN models where the recursive hierarchy or max-pooling are replaced by a general composition function. Nevertheless, FCM is still a static model and shares the same disadvantage of CNN in modeling temporal data.
The advantage of the RNN model in learning sequential data is well-known and has been utilized in language modeling (Mikolov et al., 2010) and sequential labeling (Schuster and Paliwal, 1997). Compared to these studies, a significant difference of our model is that there are no predicting targets at each time step, and the supervision (relation label) is only available at the end of a sequence. This is similar to the semantic embedding model proposed by Palangi et al. (2015), though we have made several important modifications, as will be presented in the next section.
Recurrent Layer
Embedding Layer Pooling Layer Figure 1: The framework of the proposed model.
Model
As has been shown in Figure 1, the model proposed in this paper contains three components: (1) a word embedding layer that maps each word in a sentence into a low dimension word vector; (2) a bidirectional recurrent layer that models the word sequence and produces word-level features (representations); (3) a max pooling layer that merges word-level features from each time step (each word) into a sentence-level feature vector, by selecting the maximum value among all the word-level features for each dimension. The sentence-level feature vector is finally used for relation classification. These components will be presented in detail in this section.
Word embedding
The word embedding layer is the first component of the proposed model, which projects discrete word symbols to low-dimensional dense word vectors, so that the words can be modeled and processed by the following layers. Let x t ∈ {0, 1} |V | denote the one-hot representation of the t-th word v t , where |V | denotes the size of the vocabulary V . The embedding layer transfers x t to word vectors e t ∈ R D as follows: where W em ∈ R |D|×|V | is the projection matrix. Since x t is one-hot, W em in fact stores representations of all the words in V . Word embedding has been widely studied in the context of semantic learning. In this work, we first train word vectors using the word2vec tool 1 with a large amount of data that are in general domains, and then use these vectors to initialize (pre-train) the word embedding layer of our model. By this way, knowledge of general domains can be used. It has been shown that this pre-training improves model training, e.g., (Zeng et al., 2014;Yu et al., 2014).
Bi-directional network
The second component of our model is the recurrent layer, the key part for modeling sequential data and long-distance patterns. We start from a simple one-directional forward RNN. Given a sentence X = (x 1 , x 2 , ..., x T ), the words are projected into a sequence of word vectors, denoted by (e 1 , e 2 , ..., e T ) where T is the number of words. These word vectors are put to the recurrent layer step by step. For each step t, the network accepts the word vector e t and the output at the previous step h f w t−1 as the input, and produces the current output h f w t by a linear transform followed by a non-linear activation function, given by: where h f w t ∈ R M is the output of the RNN at the t-th step, which can be regarded as local segment-level features produced by the word segment (x 1 , ..., x t ). Note that M is the dimension of the feature vector, and are the model parameters. We have used the hyperbolic function tanh(·) as the non-linear transform, which can help back propagate the error more easily due to its symmetry (Glorot and Bengio, 2010).
A potential problem of the one-directional forward RNN is that the information of future words are not fully utilized when predicting the semantic meaning in the middle of a sentence. A possible solution is to use a bi-directional architecture that makes predictions based on both the past and future words, as has been seen in Figure. 1. This architecture has been demonstrated to work well in sequential labeling, e.g., (Schuster and Paliwal, 1997). With the bi-directional RNN architecture, the prediction at step t is obtained by simply adding the output of the forward RNN and the backward RNN, formulated as follows: where h bw t ∈ R M is the output of the backward RNN, which possesses the same dimension as h f w t defined by: are the parameters of the backward RNN. Note that the forward and backward RNNs are trained simultaneously, and so the addition is possible even without any parameter sharing between the two RNN structures.
Max-pooling
Sentence-level relation classification requires a single sentence-level feature vector to represent the entire sentence. In the CNN-based models, a pooling approach is often used (Zeng et al., 2014). With the RNN structure, since the semantic meaning of a sentence is learned word by word, the segment-level feature vector produced at the end of the sentence actually represents the entire sentence. This accumulation approach has been used in (Palangi et al., 2015) for sentence-level semantic embedding.
In practice, we found that the accumulation approach is not very suitable for relation learning because there are many long-distance patterns in the training data. Accumulation by recurrent connections tends to forget long-term information quickly, and the supervision at the end of the sentence is hard to be propagated to early steps in model training, due to the annoying problem of gradient vanishing (Bengio et al., 1994).
We therefore resort to the max-pooling approach as in CNN models. The argument is that the segment-level features, although not very strong in representing the entire sentence, can represent local patterns well. The semantic meaning of a sentence can be achieved by merging representations of the local patterns. The max-pooling is formulated as follows: where h is the sentence feature vector and i indexes feature dimensions. Note that we have chosen max-pooling rather than mean-pooling. The hypothesis is that only several key words (trigger) and the associated patterns are important for relation classification, and so max-pooling is more appropriate to promote the most informative patterns.
Model training
Training the model in Figure 1 involves optimizing the parameters θ = {W in , W f w , U f w , b f w , W bw , U bw , b bw }. The training objective is that, for a given sentence, the output feature vector h achieves the best performance on the task of relation classification. Here we use a simple logistic regression model as the classfier. Formally, this model predicts the posterior probability that an input sentence X involves a relationship r as follows: where σ(x) = e x i j e x j is the softmax function, and θ encodes the parameters of the RNN model.
Based on the logistic regression classifier, a natural objective function is the cross entropy between the predictions and the labels, given by: where n is the index of sentences in the training data, and X (n) and r (n) denote the n-th sentence and its relation label, respectively.
To train such a model, we follow the training method proposed by Collobert et al. (2011), and utilizes the stochastic gradient descent (SGD) algorithm. Specifically, the back propagation through time (BPTT) (Werbos, 1990) is employed to compute the gradients layer by layer, and the fan-in technique proposed by Plaut and Hinton (1987) is used to initialize the parameters. It was found that this initialization can locate the model parameters around the linear region of the activation function, which helps propagating the gradients back to early steps easier. Moreover, it also balances the learning speed for parameters in different layers (LeCun et al., 2012).
As has been discussed, pre-training the word embedding layer with word vectors trained from extra large amount corpus improves the performance. This approach has been employed in our experiments.
Position indicators
In relation learning, it is essential to let the algorithm know the target nominals. In the CNN-based approach, Zeng et al. (2014) appended a position feature vector to each word vector, i.e., the distance from the word to the two nominals. This has been found highly important to gain high classification accuracy. For RNN, since the model learns the entire word sequence, the relative positional information for each word can be obtained automatically in the forward or backward recursive propagation. It is therefore sufficient to annotate the target nominals in the word sequence, without necessity to change the input vectors.
We choose a simple method that uses four position indicators to specify the starting and ending of the nominals. The following is an example: "<e1> people </e1> have been moving back into <e2> downtown </e2>". Note that people and downtown are the two nominals with the relation 'Entity-Destination(e1,e2)', and <e1>, </e1>, <e2>, </e2> are the four position indicators which are regarded as single words in the training and testing process. The positionembedded sentences are then used as the input to train the RNN model. Compared to the position feature approach in the CNN model, the position indictor method is more straightforward.
Database and experimental setup
We use the dataset and evaluation framework provided by SemEval-2010 Task 8. There are 9 directional relations and an additional 'other' relation, resulting in 19 relation classes in total. Given a sentence and two target nominals, a prediction is counted as correct only when both the relation and its direction are correct. The performance is evaluated in terms of the F1 score defined by SemEval-2010 Task 8 (Hendrickx et al., 2009). Both the data and the evaluation tool are publicly available. 2 In order to compare with the work by Socher et al. (2012) and Zeng et al. (2014), we use the same word vectors proposed by Turian et al. (2010) (50dimensional) to initialize the embedding layer in the main experiments. Additionally, to compare with the work by Nguyen and Grishman (2015), additional experiments are also conducted with the word vectors pre-trained by Mikolov et al. (2013) which are 300-dimensional.
Because there is not an official development dataset, we tune the hyperparameters by 8-fold cross validation. Once the hypeparameters are optimized, all the training data are used to train the model with the best configuration. With Turian's 50-dimensional word vectors, the best dimension of the feature vector is M = 500, and with Mkolov's 300-dimensional word vectors, the best feature dimension is M = 700. The best learning rate is 0.01 in both the two conditions.
Results
Model F1 RNN 31.9 + max-pooling 67.5 + position indicators 76.9 + bidirection 80.0 very poor. This can be attributed to the lack of the position information of target nominals and the difficulty in RNN training. The max-pooling offers the most significant performance improvement, indicating that local patterns learned from neighbouring words are highly important for relation classification. The position indicators also produce highly significant improvement, which is not surprising as the model would be puzzled which pattern to learn without the positional information. The contribution of positional information has been demonstrated by Zeng et al. (2014), where the positional features lead to nearly 10 percentiles of F1 improvement, which is similar as the gain obtained in our model.
The second experiment compares various neural models by using the 50-dimensional word vectors. The results are presented in Table 2. It can be seen that the RNN model outperforms both the MV-RNN model proposed by (Socher et al., 2012) and the CNN model proposed by (Zeng et al., 2014). The most interesting observation is that the RNN model performs better than the MV-RNN model which uses syntactic parse as extra resources. This indicates that relation patterns can be effectively learned by RNNs from raw text, without any explicit linguistic knowledge.
Finally, we compare the RNN model with several representative models that achieve state-ofthe-art results in relation classification (see Table 3). The first model is based on SVMs and was proposed by Hendrickx et al. (2009). This model can represent the state-of-the-art patternbased system. All the other models are based on neural networks, which are MV-RNN (Socher et al., 2012), CNN (Zeng et al., 2014), CNN with multiple window sizes (Nguyen and Grishman, 2015), and FCM (Yu et al., 2014). Note that different authors use different features and extra knowledge, which makes it difficult to compare the results directly. Nevertheless, some interesting observations can be obtained. Firstly, the best performance obtained so far is 83.0, which was achieved by the FCM model using 300-dimensional word vectors plus dependency parse and name tags as extra knowledge. Among the learning-fromscratch models, i.e., no extra knowledge and NLP processing involved, the best performance (82.8) is achieved by the CNN model with multiple window sizes. Our RNN model with the same word vectors achieves a very similar result (82.5), and the network structure is simpler. In the next section, we will show that the RNN model possesses more potential in real application with complex long-distance relations.
Impact of long context
We have argued that a particular advantage of the RNN model compared to the CNN model is that it can deal with long-distance patterns more effectively. To verify this argument, we split the test dataset of SemEval-2010 task-8 into 7 subsets according to length of the context. Here the context is defined as words between the two nominals plus 3 words prior to the first nominal and 3 words after the second nominal, if they exist. To make the analysis more accurate, short clauses between two commas have removed from the context. Clearly, long contexts lead to long-distance patterns.
In order to compare performance of the RNN and CNN models, we tried to reproduce the CNNbased method proposed by Zeng et al. (2014). A little difference is that position indicators are used to specify the target nominals in the reproduction. This modification ensures that the two models learn the same input sequence with the same representation. The F1 results on the 7 subsets are reported in Figure 2. It can be seen that if the context length is small, the CNN and RNN models perform similar, whereas if the context length is large, the RNN model is clearly superior. This confirms that RNN is more suitable to learn longdistance patterns. Note that with both the two models, the best F1 results are obtained with a moderate length of contexts. This is understandable as too small context involves limited semantic information, while too large context leads to difficulties in pattern learning. Figure 2 shows that the RNN model significantly outperforms the CNN model, which is a little different from the results presented in Table 2, where the discrepancy between the two models is not so remarkable (78.9 vs. 80.0). This can be attributed to the small proportion of long contexts in test data.
Proportion of long context
To verify this conjecture, the distribution of the context lengths is calculated on the test dataset (SemEval-2010 task8). For comparison, another two datasets are also presented: the New York Time corpus with the entities and relations selected from a subset of Freebased recommended by Riedel et al. (2013); the KBP dataset with the modification proposed by Angeli et al. (2014).
The statistics are shown in Table 4. It can be observed that long contexts exist in all the three datasets. Particularly, the proportion of long contexts in the SemEval2010 dataset is rather small compared to the other two datasets. This suggests that the strength of the RNN model was not fully demonstrated by our experiments. With a more realistic dataset such as KBP, more advantages are expected to obtain with the RNN model.
Semantic accumulation
Another interesting analysis is to show how the 'semantic meaning' of a sentence is formed. First notice that with both the CNN and the RNN models, the sentence-level features are produced from local features (word-level for CNN and segmentlevel for RNN) by dimension-wise max-pooling.
To measure the contribution of a particular word or segment to the sentence-level semantic meaning, for each sentence, we count the number of dimensions that the local feature at each word step contributes to the output of the max-pooling. This number is divided by the number of total dimensions of the feature vector, resulting in a 'semantic contribution' over the word sequence. Figure 3 and Figure 4 show two examples of semantic contributions. In each figure, the results with both the CNN and RNN models are presented.
For the sentence in Figure 3, the correct relation is 'Instrument-Agency'. But CNN gives wrong answer 'Other'. It can be seen that CNN matches two patterns 'is able to' and 'magic', while the RNN matches the entire sequence between the two Dataset Context Length Proportion of ≤10 11 -15 ≥ 16 Long Context (≥ 11) SemEval-2010 task-8 (Hendrickx et al., 2009) 6658 3725 334 0.379 NYT+Freebase (Riedel et al., 2013) 22057 19369 3889 0.513 KBP (Angeli et al., 2014) 6618 11647 15546 0.804 Figure 3: Semantic distribution on words in the sentence "A <e1> witch </e1> is able to change events by using <e2> magic </e2> ." Figure 4: Semantic distribution on words in the sentence "Skype, a free software, allows a <e1> hookup </e1> of multiple computer <e2> users </e2> to join in an online conference call without incurring any telephone costs." nominals witch and magic, with the peak at 'by using'. Clearly, the pattern that the RNN model matches is more reasonable than that matched by the CNN model.
We highlight that RNN is a temporal model which accumulates the semantic meanings word by word, so the peak at 'by using' is actually the contribution of all the words after 'witch'. In contrast, CNN model learns only local patterns, therefore it splits the semantic meaning into two separate word segments.
Similar observation is obtained with second example shown in Figure 4. Again, the RNN model accumulates the semantic meaning of the sentence word by word, while the CNN model has to learn two local patterns and merge them together.
An interesting observation is that the RNNbased semantic distribution tends to be smoother than the one produced by the CNN model. In fact, we calculated the average variance on the semantic contribution of neighbouring words with all the sentences in the SemEval-2010 task-8 dataset, and found that the variance with the RNN model is 0.0017, while this number is 0.0025 with the CNN model. The smoother semantic distribution is certainly due to the temporal nature of the RNN model.
Conclusion
In this paper, we proposed a novel RNN-based approach for relation classification. Compared to other deep learning models such as MV-RNN and CNN, the RNN model can deal with long-distance patterns and so is particular suitable for learning relations within a long context. Several important modifications were proposed to improve the basic model, including a max-pooling feature aggregation, a position indicator approach to specify target nominals, and a bi-directional architecture to learn both the forward and backward contexts. Experimental results demonstrated that the RNNbased approach can achieve very competitive results, and for sentences with long-distance relations, the RNN model exhibits clear advantages. | 2015-08-05T11:03:46.000Z | 2015-08-05T00:00:00.000 | {
"year": 2015,
"sha1": "0f36511d47596b91b8d3db265e4ca7bafc0b15a4",
"oa_license": null,
"oa_url": null,
"oa_status": null,
"pdf_src": "Arxiv",
"pdf_hash": "c105040e7ef87494396f641c4d336a1bda905c98",
"s2fieldsofstudy": [
"Computer Science"
],
"extfieldsofstudy": [
"Computer Science"
]
} |
252866464 | pes2o/s2orc | v3-fos-license | Direct Interaction of Coronavirus Nonstructural Protein 3 with Melanoma Differentiation-Associated Gene 5 Modulates Type I Interferon Response during Coronavirus Infection
Coronavirus nonstructural protein 3 (nsp3) is a multi-functional protein, playing a critical role in viral replication and in regulating host antiviral innate immunity. In this study, we demonstrate that nsp3 from severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and avian coronavirus infectious bronchitis virus (IBV) directly interacts with melanoma differentiation-associated gene 5 (MDA5), rendering an inhibitory effect on the MDA5-mediated type I interferon (IFN) response. By the co-expression of MDA5 with wild-type and truncated nsp3 constructs, at least three interacting regions mapped to the papain-like protease (PLpro) domain and two other domains located at the N- and C-terminal regions were identified in SARS-CoV-2 nsp3. Furthermore, by introducing point mutations to the catalytic triad, the deubiquitylation activity of the PLpro domain from both SARS-CoV-2 and IBV nsp3 was shown to be responsible for the suppression of the MDA5-mediated type I IFN response. It was also demonstrated that both MDA5 and nsp3 were able to interact with ubiquitin and ubiquitinated proteins, contributing to the interaction between the two proteins. This study confirms the antagonistic role of nsp3 in the MDA5-mediated type I IFN signaling, highlighting the complex interaction between a multi-functional viral protein and the innate immune response.
Introduction
Coronavirus (CoV) is a main human and animal pathogen. The emergence of severe acute respiratory syndrome coronavirus (SARS-CoV) originated from bats in 2003, Middle East respiratory syndrome coronavirus (MERS-CoV) in 2012, and the current COVID-19 pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) calls for urgent development of effective intervention measures [1,2]. Understanding the molecular functions of individual CoV proteins related to viral replication and virus-host interactions, especially with host antiviral mechanisms, would be essential for the development of such tools.
Coronavirus encodes a single-stranded, positive-sense RNA genome of 27-32 kb [3]. Two-thirds of the virus genome translates into two large polyproteins, pp1a and pp1ab, which are cleaved by viral proteases into 15-16 non-structural proteins (nsp1-16). The papain-like protease (PLpro) encoded by nsp3 and 3CLpro encoded by nsp5 are two virusencoded proteases, cleaving the two polyproteins [4][5][6]. The mature nsps are assembled to form a multi-functional membrane-bound replicase complex (RTC), mediating the replication of the viral genome and initiating the transcription and translation [7]. One critical component is nsp3, a multi-functional protein containing a number of putative domains. So far, eight putative domains, including Ubl (ubiquitin-like domain1), the hypervariable region, X-domain (also called macrodomain), ubl2 (ubiquitin-like domain 2), the papain-like protease 2 (PLpro or PL2 pro ), the nsp3 ectodomain, Y and Y1 domains, are identified in almost all coronaviruses [8]. Except for the PLpro domain, the functionality of other domains is poorly characterized. The presence of these multi-functional domains in nsp3 makes it an attractive therapeutic target.
As the first line of host defense against pathogen invasion, the innate immune system can detect pathogen-associated molecular patterns (PAMPs) through pattern recognition receptors (PRR) and initiate a series of signal cascades to induce the expression of antiviral genes [9]. The antiviral response to cytosolic viral RNA is mediated by retinoic acid-inducible gene I (RIG-I)-like receptors (RLR). Melanoma differentiation-associated gene 5 (MDA5), a member of the RLRs family, consists of the N terminal tandem highly conservative caspase recruitment domains (CARDS), a central DExD/H box RNA helicase domain with an ATP binding motif, and a C-terminal domain (CTD) [10,11]. When MDA5 recognizes double-strand RNAs, the viral replication intermediates, protein phosphatase 1 (pp1) α/γ dephosphorylates the CARDs [12]. This promotes the binding and activation of mitochondrial antiviral-signaling protein (MAVS), leading to the phosphorylation of IRF3/IRF7 and the induction of type I interferons [13,14].
Avian coronavirus infectious bronchitis virus (IBV) infects chickens and causes respiratory symptoms and kidney and oviduct diseases [15]. Due to the absence of RIG-I in chicken [16], MDA5 would play a major role in recognizing IBV as well as other coronavirus RNA. In this study, we show that the full-length nsp3 from SARS-CoV-2 and IBV directly interacts with MDA5, forming complexes with MDA5, ubiquitin (Ub), and ubiquitinated proteins. The further functional characterization of these complex interactions demonstrated that the ubiquitylation of MDA5 enhances the MDA5-mediated IFN-β induction, whereas the deubiquitinating (DUB) activity of nsp3 PLpro demolishes this enhancement effect.
Antibodies, Reagents, Cell Culture and Virus Infection
The antibodies against β-actin, Flag, and HA were purchased from TansGen biotech (Beijing, China). The antibodies against MDA5 were purchased from proteintech (Wuhan, China). Antiserum against IBV N protein was prepared from rabbits immunized with bacterially expressed fusion proteins.
Vero-adapted IBV (p65) was obtained from the American Type Culture Collection (ATCC) and passaged in Vero cells 65 times, as previously described [17,18]. The rIBV-HA-3541 was rescued using the reverse genetic technique. The virus stocks were prepared by infecting monolayers of Vero cells with rIBVs at a multiplicity of infection (MOI) of 0.1, and the cell lysates were harvested when the complete fusion of the entire monolayer was observed. After repeated freezing and thawing three times, the total cell lysates were clarified by centrifugation at 5000× g at 4 • C for 10 min and stored at −80 • C.
In the infection experiments, the cells were washed with PBS before being infected with rIBVs and incubated in serum-free DMEM at an MOI of approximately 2. After 2 h of absorption, the cells were washed twice with PBS to remove the unbound viruses.
RNA Extraction and RT-PCR Analysis
The total RNA was extracted from the cells using TRIzol reagent (TansGen biotech, Beijing, China), and the first-strand cDNA was synthesized by reverse transcriptase M-MLV (Takara, Beijing, China) with specific primers. The relative transcriptional level of IFN-β was detected by fluorescent quantitative real-time PCR (qPCR), using specific primers based on the human IFN-β sequence.
The above cDNA (1 µL of the 20 µL RT reaction mixture) was used as templates and subjected to SYBR green PCR assays (TaKaRa, Beijing, China) at least three times. qPCR was performed using the CFX96 Touch Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA), as follows: initial denaturation at 95 • C for 3 min followed by 40 cycles at 95 • C for 15 s, 56 • C. for 30 s, and 72 • C for 30 s. A final melting curve analysis was performed from 65 • C to 95 • C at a rate of 0.1 • C/s (continuous acquisition). The results are expressed as the relative gene expression level with normalization to the expression level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The relative mRNA level was calculated using the 2 −∆∆Ct method.
Plasmid Construction and Transfection
Human MDA5 was amplified from the total RNA of 293T cells by RT-PCR, using primer pair: TAGGGCGAATTCGGATCCatgtcgaatgggtattccacag and CGCCTC-GAGAAGCTTctaatcctcatcactaaataaacagcattc. The PCR product was inserted into vector pXJ40-Flag by homologous recombination. The SARS-CoV-2 nsp3 cDNA was synthesized (GENEWIZ) based on the SARS-CoV-2 nsp3 sequence (NCBI MT827843) and inserted into plasmid K1E-Hibit. The IBV nsp3 cDNA was reverse-transcribed and amplified from the total RNA extracted from the IBV-infected H1299 cells, and cloned into pXJ40 and pK1E vectors, respectively, by homologous recombination. All of the plasmids were verified by sequencing.
The HEK293T cells were plated to a 12-or 6-well plate and transfected with plasmid DNA when the cell density was about 70%, using the TransIntro EL reagent (Transgen Biotech, Beijing, China). According to the manufacturer's instruction, 1.6 µg plasmid DNA and 3.2 µL TransIntro EL were diluted with 100 µL Opti-MEM (Gibco) and incubated for 20 min. The cells were washed with PBS before being transfected and incubated in 900 µL Opti-MEM containing 10% FBS.
Western Blot Analysis
The cells were collected with a cell scraper (Corning) at a specified time point posttransfection. After centrifugation for 1 min at 16,000× g, the cell pellets were dissolved in 1 × RIPA buffer. After clarification, the protein concentration was determined using the BCA protein Assay Kit (Beyotime, Shanghai, China), according to the manufacturer's instructions. The cell lysates were mixed with Laemmli's sample buffer containing 100 mM dithiothreitol, boiled at 90 • C for 5 min, and centrifuged at 16,000× g for 5 min. Equal amounts of protein samples were subjected to sodium dodecyl sulphate-polyacrylamide gel electrophoresis and transferred to 0.2 µM BIORAD nitrocellulose membrane (BIORAD). After blocking the non-specific antibody binding with 5% skim milk in tris-buffer saline containing 0.1% Tween 20 (20 mM Tris-HCl pH 7.4, 150 mM NaCl), the membrane was incubated overnight with 1 µg/mL primary antibody at 4 • C. After washing, 1:2000 diluted anti-rat or anti-rabbit IgG antibody coupled with horseradish peroxidase (DAKO) was added and incubated at room temperature for 2 h. The proteins were detected using the chemiluminescence detection kit (Amersham Biosciences, Amersham, UK) and medical X-ray film (Fuji film, Tokyo, Japan) according to the manufacturer's instructions.
Co-Immunoprecipitation
The cells transfected with appropriate plasmid DNA or infected with the virus were lysed with 300 µL of RIPA buffer per well (6 well plate), 3 µL of protease inhibitor was added, mixed well, and placed on a table rotator at 4 • C for 15 min. After centrifugation at 10,600× g for 10 min, the supernatant was transferred to a fresh tube containing washed beads and added lysis buffer to keep the total volume to 1 mL, 10 µL protease inhibitors were added and incubated at 4 • C for 2 h. After centrifugation at 8300× g, 4 • C for 40 s, the supernatant was discarded, and the pellets were washed with 1 mL lysis buffer three times. The principates were then analyzed by Western blot.
Immunofluorescence
The H1299 cells cultured in 48-well plates to confluency were infected with rIBV-HA-3541 at MOI of~2. At 20 h post-infection, the cells were fixed with ice-cold 100% methanol for 15 min at −20 • C, rinsed 3 times with 1 × PBS for 5 min each. The cells were then incubated with blocking buffer at room temperature for 60 min, diluted primary antibody was added and incubated at 4 • C overnight, and rinsed 3 times with 1 × PBS for 5 min each. After incubation with diluted secondary antibody for 1-2 h at room temperature in the dark, the cell nuclei were stained by adding 10 µg/mL Hoechst 33342 and incubated at room temperature for 5-10 min, rinsed 3 times with 1 × PBS for 5 min each. Finally, the coverslips were mounted using the antifade reagent, and the cells were examined immediately or stored at 4 • C in the dark.
Efficient Expression of the Full-Length SARS-CoV-2 and IBV nsp3
Coronavirus nsp3 is a multi-domain protein, but the functional characterization of the protein was hampered by the lack of an expression system that can efficiently express the full-length nsp3 protein [19]. In this study, an efficient two-plasmid expression system was first developed based on the K1E phage RNA polymerase and a capping enzyme from the African swine fever virus [20]. The co-transfection of the two plasmids together with target genes cloned under the control of the KIE promoter would result in the cytoplasmic expression of the target genes. As shown in Figure 1, the full-length SARS-CoV-2 nsp3 with either an HA or a Hibit tag at the N-terminus (219.3 kDa) and the full-length IBV nsp3 with a Hibit or a FLAG tag (180.5 kDa) were highly efficiently expressed. This expression system was subsequently used to study the interactions between coronavirus nsp3 and host proteins. The HEK293T cells were transfected with pK1E (empty vector), pK1E-Hibit-IBV, SARS2-nsp3, pK1E-FLAG-IBV, and pK1E-HA-SARS2-nsp3, respectively. The cells were harvested at 24 h post-transfection and subjected to Western blot with indicated antibodies. Beta-actin was included as the loading control. Numbers on the left indicate protein sizes in kilodalton. The HEK293T cells were transfected with pK1E (empty vector), pK1E-Hibit-IBV, SARS2-nsp3, pK1E-FLAG-IBV, and pK1E-HA-SARS2-nsp3, respectively. The cells were harvested at 24 h post-transfection and subjected to Western blot with indicated antibodies. Beta-actin was included as the loading control. Numbers on the left indicate protein sizes in kilodalton.
Direct Interaction of SARS-CoV-2 and IBV nsp3 with MDA5
As MDA5 plays an essential role in anti-coronavirus innate immunity and was previously shown to interact with SARS-CoV-2 nsp3 in virus-infected cells [21], it would be interesting to investigate if the full-length SARS-CoV-2 nsp3 directly interacts with MDA5 in the absence of other viral proteins and viral RNA, and regulates the functions of MDA5. The direct interaction of SARS-CoV-2 nsp3 with MDA5 was first analyzed by co-expression of the two proteins as Hibit-tagged nsp3 (Hibit-SARS2-nsp3) and Flag-tagged MDA5 (Flag-MDA5) in 293T cells. The cells transfected with an empty vector plus one target protein (Hibit-SARS2-nsp3+Flag and Hibit+Flag-MDA5) were included as controls. Following immunoprecipitation with anti-Flag coated beads, the co-precipitated proteins were analyzed by Western blot using anti-Flag and anti-Hibit antibodies, respectively. As shown in Figure 2a, similar amounts of Hibit-SARS2-nsp3 were detected in cells either transfected with Hibit-SARS2-nsp3+Flag or co-transfected with Flag-MDA5 (Hibit-SARS2-nsp3+Flag-MDA5). In the immunoprecipitation assay, Hibit-SARS2-nsp3 was only detected when it was co-expressed with Flag-MDA5 ( Figure 2a). These results confirm the direct interaction between SARS-CoV-2 nsp3 and MDA5 and prompt to address if IBV nsp3 could also interact with MDA5. As shown in Figure 2b, Hibit-tagged IBV nsp3 could be efficiently co-precipitated with Flag-MDA5 when the two proteins were co-expressed in cells.
The interaction between IBV nsp3 and MDA5 was then studied in virus-infected cells. The immunofluorescent staining of the H1299 cells infected with rIBV-HA-3541 was carried out to further study the interaction between MDA5 and IBV nsp3. At 20 h postinfection, the cells were co-stained with anti-HA and anti-MDA5 antibodies and examined. The fluorescent images of MDA5 were detected in the cytoplasm in both infected and uninfected cells, and fluorescent images of the HA-tagged IBV nsp3 were mainly detected in the perinuclear region, the viral replication site, and in the rIBV-HA-3541-infected cells ( Figure 2d). Interestingly, in HA-tagged IBV nsp3-positive cells, significantly more MDA5 fluorescence overlapped with the nsp3 images was observed (Figure 2d), suggesting the enrichment of MDA5 in the viral replication site by the interaction between the two proteins. Taken together, these results verified the direct interaction between SARS-CoV-2 nsp3 and MDA5 in the absence of other viral proteins and RNA and further demonstrated the interaction between IBV nsp3 and MDA5 in IBV-infected cells as well as in cells overexpressing the two proteins. This interaction may position MDA5 to the viral RNA synthesis site to modulate the PAMP detection.
fected cells (Figure 2d). Interestingly, in HA-tagged IBV nsp3-positive cells, significantly more MDA5 fluorescence overlapped with the nsp3 images was observed (Figure 2d), suggesting the enrichment of MDA5 in the viral replication site by the interaction between the two proteins. Taken together, these results verified the direct interaction between SARS-CoV-2 nsp3 and MDA5 in the absence of other viral proteins and RNA and further demonstrated the interaction between IBV nsp3 and MDA5 in IBV-infected cells as well as in cells overexpressing the two proteins. This interaction may position MDA5 to the viral RNA synthesis site to modulate the PAMP detection.
Downregulation of the MDA5-Mediated IFN-β Induction by SARS-CoV-2 and IBV nsp3
The functional significance of the interaction between coronavirus nsp3 and MDA5 was then explored by checking the IFN-β production in the cells co-expressing the two proteins. In the cells co-transfected with same amounts of Hibit-SARS2-nsp3 and Flag-MDA5, efficient expression of Flag-MDA5 and Hibit-SARS-2 nsp3 was detected (Figure 4a). The overexpression of Flag-MDA5 induced 20-fold enhanced IFN-β transcription at 24 h post-transfection, compared with the control group transfected with the empty vector (Figure 4b). The IFN-β induction level was reduced by nearly half in cells co-transfected with MDA5 and SARS-CoV-2 nsp3 (Figure 4b). Similar results were also obtained in cells co-expressing Flag-MDA5 and Hibit-IBV-nsp3 (Figure 4a,b). These results demonstrate that the interaction between coronavirus nsp3 and MDA5 suppresses the MDA5-mediated IFN-β induction.
The domain(s) responsible for this suppressive effect was then investigated by coexpression of MDA5 with one of the four truncated SARS-CoV-2 constructs, nsp3-N1, nsp3-N2, nsp3-N3, and nsp3-C (see Figure 3a). The co-transfection of MDA5 with each of these constructs showed efficient expression of MDA5 and the truncated nsp3 products (Figure 4a,c), and the most significant reduction in the IFN-β production was observed in cells co-transfected with nsp3-N1 and MDA5 (Figure 4d). This suppressive effect was almost but not fully relieved by the deletion of the PLpro domain in nsp3-N2 (Figure 4d). Interestingly, significant enhancement of the IFN-β induction was detected in cells overex-pressing MDA5 with either nsp3-N3 or nsp3-C (Figure 4d), pointing to the possibility that these two regions may contain sequences that promote the MDA5-mediated IFN-β induction.
The domain(s) responsible for this suppressive effect was then investigated by coexpression of MDA5 with one of the four truncated SARS-CoV-2 constructs, nsp3-N1, nsp3-N2, nsp3-N3, and nsp3-C (see Figure 3a). The co-transfection of MDA5 with each of these constructs showed efficient expression of MDA5 and the truncated nsp3 products (Figure 4a,c), and the most significant reduction in the IFN-β production was observed in cells co-transfected with nsp3-N1 and MDA5 (Figure 4d). This suppressive effect was almost but not fully relieved by the deletion of the PLpro domain in nsp3-N2 (Figure 4d). Interestingly, significant enhancement of the IFN-β induction was detected in cells overexpressing MDA5 with either nsp3-N3 or nsp3-C (Figure 4d), pointing to the possibility that these two regions may contain sequences that promote the MDA5-mediated IFN-β induction. Significance levels were presented by the p-value (ns, non-significance, ***, p < 0.001).
Requirement of the Catalytic Activity of PLpro Domain in SARS-CoV-2 and IBV nsp3 for Suppressing the MDA5-Mediated IFN-β Induction
To further support the involvement of the PLpro domain in suppressing the MDA5mediated IFN-β induction, the catalytic triad of SARS-CoV-2 PLpro (Cys857, His1021, and Asp1032) and IBV PLpro (Cys601, His764, and Asp775) [22,23], were substituted with an Ala, respectively, and efficiently expressed (Figure 5a-c). Compared with wild-type nsp3, overexpression of all mutant constructs failed to suppress the MDA5-mediated IFN-β induction (Figure 5d,e), confirming that the catalytic activity of PLpro is essential for its role in the suppression of the MDA5-mediated IFN-β induction. It was also noted that the co-expression of MDA5 with either SARS-CoV-2 or IBV nsp3 did not result in cleavage of MDA5 in this and previous expression studies (Figure 5b,c), demonstrating that this suppressive effect is not caused by the nsp3-mediated proteolysis of the MDA5 molecule. Instead, the DUB activity of the PLpro domain would be responsible for this effect. , and IFN-β levels were determined by RT-qPCR after normalization with the GAPDH mRNA extracted from cells transfected with the empty vector. Significance levels were presented by the p-value (ns, non-significance, *, p < 0.05; **, p < 0.01; ***, p < 0.001). € Effects of co-expression of MDA5 and wild-type or mutant IBV nsp3 on the MDA5-mediated IFN-β induction. Total RNA was extracted from cells harvested in (e), and IFN-β levels were determined by RT-qPCR after normalization with the GAPDH mRNA extracted from cells transfected with the empty vector. Significance levels were presented by the p-value (ns, non-significance, *, p < 0.05; **, p < 0.01; ***, p < 0.001).
Enhancement of the MDA5-Mediated IFN-β Induction by Ubiquitylation
MDA5 and ubiquitin (Ub) were co-transfected into HEK293T cells to explore the possibility that the de-ubiquitination activity of coronavirus nsp3 may affect the ubiquitination of MDA5 and thus render a suppressive effect on the production of IFN-β. The co-expression of MDA5 with Ub significantly promoted the MDA5-mediated IFN-β induction, and this enhancement effect was suppressed by co-transfection with either SARS-CoV-2 or IBV nsp3 (Figure 6a,b).
Enhancement of the MDA5-Mediated IFN-β Induction by Ubiquitylation
MDA5 and ubiquitin (Ub) were co-transfected into HEK293T cells to explore the possibility that the de-ubiquitination activity of coronavirus nsp3 may affect the ubiquitination of MDA5 and thus render a suppressive effect on the production of IFN-β. The coexpression of MDA5 with Ub significantly promoted the MDA5-mediated IFN-β induction, and this enhancement effect was suppressed by co-transfection with either SARS-CoV-2 or IBV nsp3 (Figure 6a,b).
This possibility was further studied by transfection of cells with wild-type SARS-CoV-2 and IBV nsp3 as well as mutant nsp3 constructs (SARS2-nsp3-H1021A and IBV-nsp3-H764A). The expression of wild-type nsp3 from either IBV or SARS-CoV-2 significantly reduced the ubiquitination of MDA5 (Figure 6c), but over-expression of the two mutant nsp3 constructs failed to do so. These results confirm that the proteolytic activity of coronavirus nsp3 is essential for its DUB function, and this DUB activity inhibits MDA5 ubiquitination and suppresses the MDA5 signaling. Figure 6. Enhancement of the MDA5-mediated IFN-β induction by ubiquitylation. (a) Western blot analysis of the expression of Flag-MDA5 co-transfected with Myc-Ub and Hibit-tagged SARS-CoV-2 or IBV nsp3. HEK293T cells were transfected either with Flag-MDA5 alone or together with Myc-Ub, and Hibit-tagged SARS-CoV-2 or IBV nsp3. Cells were harvested at 16 and 24 h post-transfection (hpt), respectively, and subjected to Western blot analysis with anti-FLAG and anti-Hibit antibodies, respectively. Beta-actin was included as the loading control. Sizes of protein ladders in kDa are indicated on the left. (b) Effects of co-expression of MDA5 with Ub and SARS-CoV-2 or IBV nsp3 on the MDA5-mediated IFN-β induction. Total RNA was extracted from cells harvested in (a), and IFN-β levels were determined by RT-qPCR after normalization with the GAPDH mRNA extracted from cells transfected with the empty vector. Significance levels were presented by the p-value (ns, non-significance, *, p < 0.05;**, p < 0.01; ***, p < 0.001). (c) Co-immunoprecipitation of Ub with MDA5 and wild-type and mutant SARS-CoV-2 or IBV nsp3. HEK293T cells were transfected with Myc-Ub alone or together with Hibit-tagged wild-type and mutant SARS-CoV-2 or IBV nsp3, harvested at 24 h post-infection, total cell lysates prepared, and subjected to immunoprecipitation with anti-Myc beads. Total cell lysates and precipitates were analyzed by Western blot with anti-Myc, anti-MDA5, and anti-Hibit antibodies, respectively. Numbers on the left indicate protein sizes in kilodalton. This possibility was further studied by transfection of cells with wild-type SARS-CoV-2 and IBV nsp3 as well as mutant nsp3 constructs (SARS2-nsp3-H1021A and IBV-nsp3-H764A). The expression of wild-type nsp3 from either IBV or SARS-CoV-2 significantly reduced the ubiquitination of MDA5 (Figure 6c), but over-expression of the two mutant nsp3 constructs failed to do so. These results confirm that the proteolytic activity of coronavirus nsp3 is essential for its DUB function, and this DUB activity inhibits MDA5 ubiquitination and suppresses the MDA5 signaling.
Interaction of Ubiquitinated Proteins with nsp3 and MDA5
As ubiquitination and de-ubiquitylation of MDA5 and coronavirus nsp3 play essential regulatory roles in the MDA5-mediated innate immune response, we next set up to study if Ub forms complexes with nsp3 and MDA5 and the functional impact of these modifications on the interaction between nsp3 and MDA5. In cells overexpressing Ub, significantly more MDA5 ubiquitination was observed (Figure 7a). Immunoprecipitation with anti-Myc antibodies against the Myc-tagged Ub showed co-precipitation of MDA5 (Figure 7a). In the cells co-transfected with Myc-tagged Ub and HA-tagged SARS-CoV-2 nsp3, considerably more ubiquitinated proteins and MDA5 were co-precipitated with HA-tagged nsp3 by anti-HA antibodies (Figure 7b). Similarly, in cells co-transfected with Myc-tagged Ub and Flagtagged IBV nsp3, significantly more ubiquitinated proteins and MDA5 were co-precipitated with Flag-tagged IBV nsp3 by anti-Flag antibodies (Figure 7c). These results confirm the formation of complexes among coronavirus nsp3, MDA5 and ubiquitinated proteins.
Interaction of Ubiquitinated Proteins with nsp3 and MDA5
As ubiquitination and de-ubiquitylation of MDA5 and coronavirus nsp3 play essential regulatory roles in the MDA5-mediated innate immune response, we next set up to study if Ub forms complexes with nsp3 and MDA5 and the functional impact of these modifications on the interaction between nsp3 and MDA5. In cells overexpressing Ub, significantly more MDA5 ubiquitination was observed (Figure 7a). Immunoprecipitation with anti-Myc antibodies against the Myc-tagged Ub showed co-precipitation of MDA5 (Figure 7a). In the cells co-transfected with Myc-tagged Ub and HA-tagged SARS-CoV-2 nsp3, considerably more ubiquitinated proteins and MDA5 were co-precipitated with HA-tagged nsp3 by anti-HA antibodies (Figure 7b). Similarly, in cells co-transfected with Myc-tagged Ub and Flag-tagged IBV nsp3, significantly more ubiquitinated proteins and MDA5 were co-precipitated with Flag-tagged IBV nsp3 by anti-Flag antibodies ( Figure 7c). These results confirm the formation of complexes among coronavirus nsp3, MDA5 and ubiquitinated proteins.
Discussion
Coronavirus nsp3 is the largest mature nsp encoded by the genome, with an average molecular mass of about 200 kDa and containing multi-putative functional domains [24]. The multi-functional properties of this protein provide multiple research directions for the identification of novel antiviral targets. However, most current studies were carried out by infection with relevant viruses and by the overexpression of a certain nsp3 domain [24]. In this study, we established a two-plasmid system using K1E phage RNA polymerase and a related powerful promoter to efficiently express nsp3 from SARS-CoV-2 and
Discussion
Coronavirus nsp3 is the largest mature nsp encoded by the genome, with an average molecular mass of about 200 kDa and containing multi-putative functional domains [24]. The multi-functional properties of this protein provide multiple research directions for the identification of novel antiviral targets. However, most current studies were carried out by infection with relevant viruses and by the overexpression of a certain nsp3 domain [24].
In this study, we established a two-plasmid system using K1E phage RNA polymerase and a related powerful promoter to efficiently express nsp3 from SARS-CoV-2 and IBV. Using the expression system, we showed that nsp3 from both IBV and SARS-CoV-2 could physically interact with MDA5 in cells overexpressing nsp3 as well as in IBV-infected cells. This interaction would enrich MDA5 to the viral RNA synthesis site and modulate its PAMP detection, leading to the suppression of the MDA5-mediated IFN-β induction. This suppressive effect was further shown to be associated with the DUB activity of the PLpro domain from both IBV and SARS-CoV-2 nsp3.
Ubiquitination is one of the important post-translational modifications and an important step in the activation of PRRs. Regulating innate immune pathways through antagonizing ubiquitin and ubiquitin-like modifications by viral proteases is recognized as a common mechanism [25]. For example, the delivery of K63-linked polyubiqutin moiety to the CARDs of RIG-I can activate downstream signal transduction, and the K172 ubiquitination is critical for inducing type I IFN expression [26][27][28]. It was also reported that the binding of ISG15 to CARDs of MDA5 promotes the formation of a higher-order MDA5 assembly, triggering the activation of innate immunity against a range of viruses, including coronaviruses, flaviviruses, and picornaviruses [21].
Coronavirus nsp3 is implicated in cleaving proteinaceous post-translational modifications on host proteins as an evasion mechanism against host antiviral immune responses [12,29]. The DUB and deISGylating activities of nsp3 can remove Ub and ISG15 from cellular substrates in addition to cleavage of viral polyproteins to form a functional replicase complex [23]. SARS-CoV-2 nsp3 has been shown to recognize Lys48-linked polyubiquitin via S1 and S2 Ub-binding sites and remove it efficiently [30]. ISG15-dependent MDA5 activation is antagonized by the deISGylation of the SARS-CoV-2 PLpro region [31]. Furthermore, different coronavirus nsp3 may have different preferences for these substrates. For example, SARS-CoV nsp3 may predominantly target Ub chains, but SARS-CoV-2 nsp3 has a preference for cleaving ISG15 [31]. In this study, we show that SARS-CoV-2 and IBV nsp3 suppress the ubiquitination of MDA5 and the MDA5-mediated IFN induction with a similar efficiency.
The functions of other nsp3 domains, such as UBL1/2 and X-domain, in the regulation of IFN signaling are not fully understood. It has been reported that SARS-CoV nsp3 is antagonistic to the expression of type I IFN and does not only depend on the DUB activity [32], as the addition of inhibitors blocking DUB activity cannot completely eliminate this antagonism. This is consistent with our observations that PLpro played the strongest antagonistic role in MDA5-mediated IFN-β induction, but UBL1 and UBL2 also showed a certain inhibitory effect. Ubiquitination is also known to be closely related to NF-κB activation. Polyubiquitination of receptor-interacting protein (RIP), TNF receptorassociated factor 6 (TRAF6), and TNF receptor-associated factor 2 (TRAF2) activates these signaling intermediates, leading to the polyubiquitination of IκB [33]. It is likely that coronavirus nsp3 could not only inhibit MDA5 recognition of viral RNA but also affect the activation of other key factors in the type I IFN pathway. Further studies are required to further clarify these regulatory mechanisms.
The interaction between coronavirus nsp3 and MDA5 would lead to the deubiquitination of MDA5 and inhibition of the MDA5-mediated antiviral signal transduction, probably through interaction with the PLpro domain. In fact, the cotransfection of wild-type and mutant nsp3 PLpro domain from different coronaviruses confirmed the interaction between the PLpro domain and the MDA5-2CARD [21]. In addition to the interaction between the nsp3 PLpro domain and MDA5, at least two other domains, located at the N-and C-terminal regions, respectively, of nsp3, were found to be interacting with MDA5. Interestingly, co-expression of these two regions, each with MDA5, enhanced the MDA5-mediated IFN induction. It is currently unknown if these two regions can directly interact with MDA5 or through interaction with ubiquitination of the proteins.
In addition to evading the type I IFN pathway by nsp3, coronaviruses have evolved a diversity of other counter-measuring mechanisms. For example, Oh et al. demon-strated that SARS-CoV-2 N protein could inhibit the interaction between tripartite motif protein 25 (TRIM25) and RIG-I. N protein can also affect the expression of Tank-binding kinase 1 (TBK1), blocking the nuclear translocation of IRF3 [34]. Wang et al. found that SARS-CoV-2 nsp12 attenuated the IFN-β promoter activity induced by the components of the RIG-I/MDA5 pathway or Sendai virus (SeV) infection [35]. Zhu X has certified that over-expression of porcine deltacoronavirus (PDCoV) nsp5 inhibited the transduction of IFN signal through cleaving STAT2, an important component of the transcription factor complex of ISGF3 [36]. It was also reported that IBV infection disrupts the MDA5 signaling pathway by cleavage of the adaptor protein MAVS, which is essential for the activation of NF-κB and IRF3/7 downstream signaling pathways [37].
In conclusion, this study confirms that SARS-CoV-2 and IBV nsp3 functions as type I IFN antagonist by inhibiting the MDA5-mediated signal transduction pathway through multiple interactions with MDA5. Both nsp3 and MDA5 were shown to interact with Ub, and the DUB activity of nsp3 is essential for its IFN antagonistic function. This information would be of help in designing anti-coronaviral interventions. | 2022-10-13T15:40:36.776Z | 2022-10-01T00:00:00.000 | {
"year": 2022,
"sha1": "d10291d01ac240d1c121fac58b7cb4876ed0b113",
"oa_license": "CCBY",
"oa_url": "https://www.mdpi.com/1422-0067/23/19/11692/pdf?version=1665279454",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "d36bf6f03cebef152db073ca4e002b1d41c3cebb",
"s2fieldsofstudy": [
"Medicine",
"Biology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
219784011 | pes2o/s2orc | v3-fos-license | SUCROSE PULSING AND COLD STORAGE ON POST-STORAGE ATTRIBUTES OF CUT LILY FLOWERS IN DALAT, VIETNAM
The effects of sucrose pulsing, cold storage methods (dry and wet), and cold storage duration (one and two weeks) on the post-storage attributes of cut lilies in Dalat were studied. After harvest, the lilies were treated with sucrose pulsing solutions consisting of three concentrations of sucrose (0, 50, and 100 g/L) in combination with 0.2 mM silver thiosulphate (STS) and water as control for 24h at room temperature, then placed in wet and dry cold storage at 2.5 °C for one and two weeks. After each period of cold storage, the cut lilies were placed in water to evaluate the vase life, the time of bud opening, the bud opening rate, and water uptake. The results showed that when cut lilies were treated with 50 g/L sucrose in combination with STS 0.2 mM and one-week cold storage, the vase life increased to 16.8 days with better color, the time of bud opening slowed down to 5.8 days, buds fully opened (100%), and the water uptake remained constant at 91.5 ml.
INTRODUCTION
In Vietnam, like many other developing countries, the supply of cut flowers is very seasonal, while the lack of pre-treatment, refrigeration, and poor packaging leads to poor product quality and short vase life (Bich & Batt, 2011). Lily (Lillium) is the high economic value flower in Vietnam (Dang et al., 2016). Its vase life is an important poststorage attribute. The short vase life of lilies may be one of the most important reasons for the inability of florists to develop a suitable market (Nemati, Tehranifar, Esfandiari, & Rezaei, 2013). The main symptoms that shorten the vase life are abscission of buds, lack of bud opening, petal wilting, and leaf yellowing due to the sensitivity of cut lilies to ethylene. Cold storage of cut flowers during storage or shipment reduces the metabolism in the tissue, retards the respiration, transpiration, and multiplication of bacteria and fungi, and reduces ethylene production and action. Dry and wet storage methods are commonly used. In dry storage cut flowers are placed in a package and arranged in a carton. In wet storage cut flowers are held in a chemical solution (Rudnicki, Nowak, & Goszczynska, 1991). The cold storage temperature of cut lilies is about 2.5 o C (Prisa, Burchi, & van Doorn, 2013). However, a cold storage period often drastically increases the number of floral buds that fail to open, hastens tepal wilting, induces or increases leaf yellowing, and promotes bud abscission. Some degree of cold-storage stress is typically imposed on lilies. As in other species, the vase life of lilies depends on the cultivated varieties, as well as the duration of the handling and shipping (van Doorn & Han, 2011). Cold storage increases the sensitivity of flower buds to ethylene given after the cold period (Miller, 2014).
Many studies have shown that chemical treatment before cold storage (known as pulsing) to prevent infections and inhibit ethylene production and/or metabolism results in a prolonged storage and improved post storage vase life of cut flowers (Madhavi, 2007;Rudnicki et al., 1991;Song, Ko, Shin, Kwon, & Lee, 1995;Waithaka, Reid, & Dodge, 2001). Pulsing solutions for lilies are often composed of a mixture of chemicals, such as carbohydrates, plant growth regulators, ethylene inhibitors, biocides, and acidifiers, which may minimise the risk of physiological disorders that occur in cut flowers after harvest and cold storage, such as leaf chlorosis and blackening (Eason, 2002;Kim, Lee, & Suh, 2005). STS is used as a pretreatment to eliminate the effects of exogenous ethylene on cut flowers (Prisa et al., 2013;Waithaka et al., 2001). Sucrose is the major source of carbon for petal growth and the energy supply for cut flowers (Han, 2003). Cevallos and Reid (2001) studied the effect of dry and wet storage at different temperatures on the vase life of tulips, daffodils, irises, carnations, roses, and daisies. Their results showed that the vase life after wet storage at temperatures of 12.5 °C and greater was significantly higher than the vase life after dry storage at those temperatures for all the flowers studied. Irises and carnations only survived storage at 15 °C and 20 °C when stored in water. Han (2003) assessed the role of sugar in the vase solution of an oriental lily‚ 'Stargazer'. The addition of 2% sugar into the vase solution affected neither the longevity nor the size of the flowers, but significantly enhanced petal color. According to Nemati et al. (2013), the application of solutions containing nano-silver particles is recommended to improve postharvest of L. orientalis 'Bouquet'.
Specific postharvest data are lacking for cut flowers, and therefore farmers, traders, and retailers face problems to market and distribute high quality flowers to consumers (Rudnicki et al., 1991). Inadequate knowledge on proper preharvest and postharvest handling methods among the majority of farmers and sellers contributed largely to considerable postharvest losses in cut flowers (Weeraratne, Daundasekera, & Wijesundara, 2012). Miller (2014) asserted that there is a need to determine the effects of sucrose pulsing in cut stems and the effects of storage temperature and duration on a range of lily cultivars in different hybrid groups. Since then, an attempt was made to find out the effects of sucrose pulsing treatment combined with STS and the two cold storage methods (dry and wet) at 2.5 °C and two storage periods (one and two weeks) on the poststorage attributes of cut lilies in Dalat. This is in accordance with the aim of Dalat farmers and wholesalers to extend the vase life of the flowers after the expected time or storage of flowers in stock or transport.
Plant materials
This experiment was conducted at the Laboratory of Postharvest Technology, Dalat University, Lamdong, Vietnam. Three-month-old stems of hybrid lilies (Lilium longiflorum) were harvested on a morning in September. Lilies were cut when the first floral bud reached the earliest possible stage of maturity (a partly-colored stage on the outlet tepals), then were delivered to the laboratory and immediately prepared for the experiment.
Pulsing solutions and cold storage treatment
Stems were recut to 60 cm and the lowermost leaves were removed. Five flower buds were left on each stem, six replicate stems. Upon arrival in the lab, stems were held in water (control) or in pulsing solutions containing 0 g/L, 50 g/L, and 100 g/L sucrose at an ambient temperature of 23 o C for 24 hours, and STS at 0.2 mM was added in all pulsing solutions as a base solution. After pulsing, the stems were then placed in a cold room with an average temperature of 2.5 o C (Prisa et al., 2013) (range from 1.9 o C to 2.4 o C) for a period of one or two weeks. During cold storage, the stems for wet storage were placed in the water treatment and sucrose treatments in separate buckets; The remains were wrapped in plastic sleeves and kept in a cardboard box for dry cold storage. At the end of each one-or two-week period, lily stems were recut and placed in a vase containing water at room temperature to complete vase life.
Parameter analysis
Observations were made daily, at the same time of the day. Vase life of each stem was based on the time after one or two weeks of storage until it became fully chlorotic. Vase life was terminated when greater than 50% of the flowers were fading and wilting. The ratio of bud opening was calculated by dividing the number of opened buds with the total number of buds per stem during the vase life trial (Eason, 2002;Kim et al., 2005). The time of bud opening is when the first bud on a stem opened (Prisa et al., 2013). Water uptake was measured every three days as the holding solution volume decreased in the vases containing flowers (Hajizadeh, Farokhzad, & Chelan, 2012).
Statistical analysis
Data analysis was performed using the General Linear Model procedure (Ho, 2006) in SPSS (Statistical Package for the Social Sciences) software and data comparisons were performed according to Duncan's test (p ≤ 0.05).
Postharvest characteristics of cut lilies
The vase life of cut lilies and the bud opening day were found to be significantly influenced by the pulsing treatments (A) and cold storage duration (C) (p < 0.001). Data further revealed that cold storage methods (B) (dry and wet) did not have any significant effect on post-storage attributes of cut lilies. Statistical analysis performed on post-storage attributes also showed no interaction between solution pulsing and cold storage methods and duration A*B, A*C (except the bud opening day), B*C, and A*B*C (p < 0.05); Therefore, they were independently analysed using an ANOVA (analysis of variance) ( Table 1). The lily stems kept their quality, showing more than 10 days of vase life in all the evaluated treatments, for which the highest vase life was 16.8 days and the lowest was 13.3 days (Table 1). Vase life was significantly longer (up to one to two days) when cut lilies were pulsed with 0.2 mM STS and 50 g/L sucrose compared to water (Table 1). No differences in the lily vase life were observed between sucrose 50 g/L (16.8 days) and non-sugar (16.7 days) in combination with 0.2 mM STS in dry cold storage for one week. However, it is possible to see that the color of the lilies was different during experiment. At a concentration of 50 g/L sucrose with 0.2 mM STS, flower petals were darker in color (Figure 1). Waithaka et al. (2001) explained that pulse application of sucrose did not increase vase life or floret opening, but did overcome the deleterious effects of exogenous ethylene and anthocyanin content. Similarly, Han (2003) and Shiva, Chatterjee, and Bhattacharjee (2002) found that pulsing with sucrose along with STS increases the anthocyanin content, and thus the intensity of petal color. Many researchers showed that pre-treatment with sucrose and STS improves the vase life of cut flowers (Beura & Singh, 2001;Marandi, Hassani, Abdollahi, & Hanafi, 2011;Zhang & Leung, 2001). They suggest that sucrose and STS act similarly on soluble sugar changes and ethylene production associated with inhibiting flower senescence and improve the maintenance of high starch concentration in the floret during flower opening. STS may improve sucrose uptake and its subsequent hydrolysis (Marandi et al., 2011). It was found that sugar delays aging (Shahri, Tahir, Islam, & Ahmad, 2010). Sugar treatments at low concentrations cause flower buds to maintain proper form and improve the opening process. Pulsing cut carnations with STS and 10% sucrose inhibited the ethylene synthesis and improved the postharvest quality of cut flowers (Burzo, Dobrescu, Amariujei, & Stanica, 1995). Cut roses and chrysanthemums were pulsed with 3% sucrose, which extended the vase life by two days compared with the control (Mwangi, 2002;Song et al., 1995).
Figure 1. Difference between cut lilies pulsed with water and sucrose 50 g/L in combination with STS 0.2 mM
The lowest vase life occurred with water pulsing (control). The flowers only lasted 14.8 and 13.6 days in dry storage for one and two weeks, respectively and 14.1 and 13.3 days in wet storage for one and two weeks, respectively. The next lowest vase life occurred with sucrose pulsing at 100 g/L (15.3 and 14.2 days in dry storage and 14.5 days in wet storage for one and two weeks, respectively) ( Table 1). Leaf yellowing was also observed in the 100 g/L sucrose pulse treatment. This study shows that sugar treatment of lilies increases vase life, but that too much sugar can reduce vase life. Prisa et al. (2013) concluded that lily cv. Brindisi showed chilling injury when stored at 2.5 o C and that a Sucrose 0 g/L Sucrose 50 g/L pulse treatment with sugars at 100 g/L alleviated some of the chilling injury symptoms, but did not delay early tepal senescence and abscission, while in some experiments pulse treatments with sucrose aggravated and hastened leaf yellowing after cold storage. Many discussions about the role of sugar in solution found that sucrose pulsing had little or no effect on cold-induced hastening of tepal senescence and abscission. Vase sugar (glucose or sucrose) can increase leaf yellowing. Han (2003) found 'Stargazer' had more leaf yellowing in vase solutions containing higher concentrations of sucrose while cold storage accelerated leaf yellowing.
In this study, cold storage methods (dry and wet) were not found to affect the vase life of lilies (Table 1). This result is in agreement with the study by Cevallos and Reid (2001) who found that wet storage did not prevent the respiration-related loss of vase life during storage. Because the higher volume and weight of wet flowers will convert into higher costs of handling and transportation, this study clearly indicates that the industry should continue to transport flowers dry and minimize losses by paying closer attention to chemical pulsing and maintaining proper storage temperatures and relative humidity. Madhavi (2007) showed that pulsing treatments enhance the dry cold storage life of flowers, which can be used during export transit to overcome the delay during transport.
While dry storage and wet storage have no effect on vase life, cold storage duration has more effect on the vase life of lilies (Table 1). After one week of cold storage, non-sugar in combination with STS stored for one week extended the vase life of cut lilies by two to three days, compared to two weeks. Vase life of lilies pulsed with 50 g/L of sucrose and STS for one week was also significantly longer by one to two days than twoweek cold storage. However, there was no significant difference between the cold storage period for treatment with water and pulsing with 100 g/L of sucrose.
Regarding the day of bud opening, there was an interaction between sucrose pulsing and cold storage duration. It was determined that the treatment of cut lilies with a non sugar solution and STS and sugar solutions of 50 g/L and 100 g/L with STS for one week slowed down the time of bud opening, compared to two weeks of cold storage (Table 1). This is consistent with the study of Prisa et al. (2013) who found that bud opening during vase life was hastened by cold storage, and the more so the longer the period of cold storage. It is possible that the buds were physiologically more ready to open the longer the cold storage lasted.
An analysis of variance showed that the effects of sucrose pulsing, cold storage methods and duration on the open florets and their interaction were not statistically significant (Table 1). There was no significant difference between treatments on the rate of bud opening, both at one and two weeks of cold storage. These results confirm the findings of Han (2003) that sugar in vase pulsing does not overcome the increased bud blasting induced by cold storage, but it enables more flowers to open fully.
The water uptake
While water uptake of cut lilies after three days was found to be significantly influenced by the pulsing treatments (A), cold storage methods (B) (dry and wet), and duration (C) (one and two weeks) have a significant effect on water uptake after six and nine days. Statistical analysis performed on water uptake also showed an interaction between solution pulsing and cold storage methods A*B, but no interaction between pulsing and cold storage duration A*C, B*C (except after three days), and A*B*C (p < 0.05). Therefore, they were independently analysed using an ANOVA (Table 2). It is expected that water uptake will increase in solution pulsing compared to the water pulsing after cold storage. There was no significant differences in water uptake between solution pulsing and the control. The rates of water uptake from the 0 g/L, 50 g/L, and 100 g/L sucrose solutions and the control were 42 mL, 44 mL, 53 mL, and 48 mL per stem, respectively, in the first three days (data not shown). The water uptake then decreased to 30 mL, 31 mL, 33 mL, and 30 mL per stem in the next three days and finally to 27 mL, 28 mL, 33 mL, and 28 mL per stem in the last three days (data not shown). Although water uptake remained unaffected by individual cold storage methods and duration, the joint application of sucrose pulsing and cold storage methods (dried and wet) had an effect on the water uptake of lilies ( Table 2). The highest sucrose pulsing (100 g/L) in wet cold storage methods had the highest water uptake from about 57 mL/stem, 31 mL to 48.8 mL/stem and 35.8 mL to 44.2 mL/stem after three, six, and nine days, respectively, compared to other treatments. Similar results were observed in the treatment of 50 mL/stem and water in dry storage. The non-sugar solution with STS was significantly lower in water uptake from 38.3 mL to 40.2 mL/stem in the first three days. Water uptake of cut lilies pulsed with 50 g/L and 100 g/L sucrose and wet cold storage increased from 25 mL to 48.8 mL/stem in the next three-day (six days), compared to the non-sugar solution with or without STS.
After the last three days (nine days) in the vase solution, the water uptake did not vary significantly among treatments. Water intake was lower at nine days in all treatments, then was less from the 12 th day (data not shown). This parameter remained affected by cold storage methods, where wet storage had higher water uptake. Water uptake increases in lilies after two weeks cold storage in all treatments. This explains why the day of opening is earlier in cut lilies with two weeks of cold storage. De la Riva, Mazuela, Alvaro, and Urrestarazu (2009) attributed the prolongation of the vase life of cut flowers by sugar application to the increase in the uptake of water by the flowers. This is in conformity with the findings of many studies on rose, gladiolus, and dendrobium hybrid flowers summarised in studies by Asrar (2012) and Pal, Lahiji, and Roy (2003). It is suggested that the increase in water uptake with sucrose treatments could be due to the increase in the osmotic concentration of the florets and leaves (de la Riva et al., 2009). The flowers were handled in water (wet storage), there was continuous water uptake by the cut rose stems, which ultimately increased the turgidity of the flowers so that they were in a continuous developmental stage. Similar results were also obtained in gerberas and carnations where the water absorption increased and fresh weight decreased with an increase in the period of dry storage (Madhavi, 2007).
Adding sucrose to the pulsing solution improves vase life and petal color (Figure 2), but does not improve the opening, rate of opening and water uptake. This study chose pulsing with sugar at 50 g/L for the next experiment. Sugar can cause leaf yellowing, which is usually exacerbated by prior cold storage. These problems are easily reduced (with consequent improvements in flower longevity) by the use of gibberellic acid in the pulsing solution.
CONCLUSION
Both pulsing treatment and cold storage are identified as influencing the vase life of lilies, reinforcing the importance of pretreatment on lilies to improve their post-storage attributes as a means of postharvest risk mitigation. The recommended cold storage period, of up to one week in combination with a prior treatment with 50 g/L sucrose and 0.2 mM STS for 24h for dry cold storage before placing the flower stems in water, was found to extend the vase life up to two to three days compared to other solutions and two weeks of cold storage. The data suggest that sugar is an essential component of any effective postharvest solution for the treatment of lilies. The stems pulsed with 50 g/L of sugar in combination with STS and one week of cold storage had an average vase life two to three days longer than stems that were held in the same solution with two weeks of cold storage. Further analysis of lilies treated with gibberellic acid to delay the appearance of yellow leaves should be examined. | 2020-05-28T09:15:57.187Z | 2020-05-04T00:00:00.000 | {
"year": 2020,
"sha1": "0e2d77e1858e370c523184c0f6fde131a263127c",
"oa_license": "CCBYNC",
"oa_url": "http://tckh.dlu.edu.vn/index.php/tckhdhdl/article/download/575/318",
"oa_status": "GOLD",
"pdf_src": "Anansi",
"pdf_hash": "2c5548527ac7e2ff551b7ec9f5d04c75d7584127",
"s2fieldsofstudy": [
"Agricultural And Food Sciences"
],
"extfieldsofstudy": [
"Biology"
]
} |
17172245 | pes2o/s2orc | v3-fos-license | The biology of fear- and anxiety-related behaviors.
Anxiety is a psychological, physiological, and behavioral state induced in animals and humans by a threat to well-being or survival, either actual or potential. It is characterized by increased arousal, expectancy, autonomic and neuroendocrine activation, and specific behavior patterns. The function of these changes is to facilitate coping with an adverse or unexpected situation. Pathological anxiety interferes with the ability to cope successfully with life challenges. Vulnerability to psychopathology appears to be a consequence of predisposing factors (or traits), which result from numerous gene-environment interactions during development (particularly during the perinatal period) and experience (life events), in this review, the biology of fear and anxiety will be examined from systemic (brain-behavior relationships, neuronal circuitry, and functional neuroanatomy) and cellular/molecular (neurotransmitters, hormones, and other biochemical factors) points of view, with particular reference to animal models. These models have been instrumental in establishing the biological correlates of fear and anxiety, although the recent development of noninvasive investigation methods in humans, such as the various neuroimaging techniques, certainly opens new avenues of research in this field. Our current knowledge of the biological bases of fear and anxiety is already impressive, and further progress toward models or theories integrating contributions from the medical, biological, and psychological sciences can be expected.
Anxiety is a psychological, physiological, and behavioral state induced in animals and humans by a threat to wellbeing or survival, either actual or potential. It is characterized by increased arousal, expectancy, autonomic and neuroendocrine activation, and specific behavior patterns. The function of these changes is to facilitate coping with an adverse or unexpected situation. Pathological anxiety interferes with the ability to cope successfully with life challenges. Vulnerability to psychopathology appears to be a consequence of predisposing factors (or traits), which result from numerous gene-environment interactions during development (particularly during the perinatal period) and experience (life events). In this review, the biology of fear and anxiety will be examined from systemic (brain-behavior relationships, neuronal circuitry, and functional neuroanatomy) and cellular/molecular (neurotransmitters, hormones, and other biochemical factors) points of view, with particular reference to animal models. These models have been instrumental in establishing the biological correlates of fear and anxiety, although the recent development of noninvasive investigation methods in humans, such as the various neuroimaging techniques, certainly opens new avenues of research in this field. Our current knowledge of the biological bases of fear and anxiety is already impressive, and further progress toward models or theories integrating contributions from the medical, biological, and psychological sciences can be expected. n a book published in 1878 (Physiologie des passions), Charles Letourneau, who was contemporary with the French neuroanatomist Paul Broca, defined emotions as "passions of a short duration" and described a number of physiological signs and behavioral responses associated with strong emotions. 1 Emotions are "intimately linked with organic life," he said, and either result in an "abnormal excitation of the nervous network," which induces changes in heart rate and secretions, or interrupt "the normal relationship between the peripheral nervous system and the brain." Cerebral activity is focused on the source of the emotion; voluntary muscles may become paralyzed and sensory perceptions may be altered, including the feeling of physical pain. This first phase of the emotional response is followed by a reactive phase, where muscles come back into action, but the attention still remains highly focused on the emotional situation. With the knowledge of brain physiology and anatomy that was available at the end of the 19th century, hypotheses on the mechanisms possibly involved in emotions were of course limited. However, Letourneau assumed that "the strong cerebral excitation" that accompanies emotions probably only concerned "certain groups of conscious cells" in the brain and "must necessitate a considerable increase of blood flow in the cell regions involved." 1 He also mentioned that the intensity, the expression, and the pathological consequences of emotions were directly linked to "temperaments" (which he defined within the four classic Hippocratic categories).
I
It is amazing to see how Letourneau's views on emotions, more than a century ago, were in many ways premonitory. The fact that emotions are "intimately linked with organic life," his precise description of the sequence of the physiological and behavioral reactions that accompany a strong emotion, such as fear, the idea that emotions involve specific areas of the brain, and the theory that activation of these areas is associated with an increased blood flow have all been largely confirmed by modern neuroscience. The suggestion that temperament or personality traits influence the "affective style" and vulnerability to psychopathology is also an important aspect of our modern approach to anxiety and mood disorders. 2 For a long time, emotions were considered to be unique to human beings, and were studied mainly from a philosophical perspective. 3 Evolutionary theories and progress in brain and behavioral research, physiology, and psychology have progressively introduced the study of emotions into the field of biology, and understanding the mechanisms, functions, and evolutionary significance of emotional processes is becoming a major goal of modern neuroscience.
Three fundamental aspects of emotions
The modern era of emotion research probably started when it became obvious that emotions are not just "feelings" or mental states, but are accompanied by physiological and behavioral changes that are an integral part of them. This has progressively led to today's view of emotions being experienced or expressed at three different, but closely interrelated levels: the mental or psychological level, the (neuro)physiological level, and the behavioral level. These three complementary aspects are present in even the most basic emotions, such as fear. A detailed account of the many "theories of emotion" is beyond the scope of this review. However, a brief historical survey of the more biologically oriented ones may help to set some important conceptual issues. [3][4][5][6][7][8] One of the main questions addressed by earlier scientific theories of emotions was whether physiological changes precede the emotional experience, or if they are only a consequence of it. For James (1884) and Lange (1885), "[…] the bodily changes follow directly the perception of the existing fact, and […] our feelings of the same changes as they occur IS the emotion." In other words, according to the James-Lange theory of emotions, stimuli reaching the cerebral cortex induce visceral changes, which are then perceived as emotion. Bard (1915-1932) criticized this theory and proposed that the neurophysiological aspects of emotions are subcortical and involve the thalamus. 9 Stimuli from the environment activate the thalamus, which relays information to the cortex and viscera, and back again to the cortex to generate the "emotional state." Watson, the father of behaviorism, was also very critical of what he called the "introverted viewpoint" of James' theory. He considered that there were only three types of unlearned emotional responses, which he called "fear," "rage," and "love" for convenience, although he wanted to "[…] strip them out of all their old connotations." 10 These three emotional responses can be elicited by three sets of specific stimuli. Thus, a sudden noise or loss of physical support can induce an innate fear reaction, and restraint of bodily movements triggers rage. He also mentioned the fact that these emotional responses can be conditioned and that, although these reactions are usually accompanied by specific behaviors, "[…] visceral and glandular factors predominate." Papez's (1937) theory of emotions also had a physiological basis. For him, connections between the cerebral hemispheres and the hypothalamus, and between the cerebral hemispheres and the dorsal thalamus mediate emotions. He held the view that emotion implies behavior (expression) and feeling (experience, subjective aspects). Expression depends on the hypothalamus, and experience on the cortex. Although the "circuit of Papez" is still presented as "the emotional brain" in some handbooks, it is clear that many details of his original theory are now outdated. More recently, Schachter (1975) emphasized the importance of cognitive processes: bodily states are interpreted in a cognitive context and are modulated by experience. He also showed that the visceral response appears to be a necessary, although not sufficient, condition for the occurrence of emotion. The view that there is a limited set of emotions (eg, fear, anger, etc) with specific neurophysiological and neuroanatomical substrates that can be considered as "basic" and serve as the primitive building blocks from which the other, more complex emotions are built, was challenged as late as 1990. 11 However, Ekman has convincingly argued that there is now enough evidence of universals in expression and in physiology to suggest a biological basis for these elementary emotions. 12 Panksepp added to these arguments by stating that "genetically dictated brain systems that mediate affective-emotional processes do exist, even though there are bound to be semantic ambiguities in how we speak about these systems." 13
The biology of fear and anxiety
Fear versus anxiety: is there a difference?
The main function of fear and anxiety is to act as a signal of danger, threat, or motivational conflict, and to trigger appropriate adaptive responses. For some authors, fear and anxiety are undistinguishable, whereas others believe that they are distinct phenomena. Ethologists define fear as a motivational state aroused by specific stimuli that give rise to defensive behavior or escape. 14 Animals may learn to fear situations in which they have previously been exposed to pain or stress, and subsequently show avoidance behavior when they reencounter that situation. Young animals may show an innate fear reaction to sudden noise or disturbances in the environment, but rapidly become habituated to them. When they are used to a familiar environment, then a fear of novelty may develop. Ethologists have also made the important observation that fear is often mixed up with other aspects of motivation. Thus, conflict between fear and approach behavior may results in displacement activities (eg, self-grooming in rats). Such displacement activities may be the behavioral expression of an anxious state, but anxiety is a concept that is apparently not used by ethologists, perhaps because their definition of fear does in fact include all the more biological aspects of anxiety. Many authors, however, have argued that differences in their etiologies, response patterns, time courses, and intensities seem to justify a clear distinction between anxiety and fear. 15 Although both are alerting signals, they appear to prepare the body for different actions. Anxiety is a generalized response to an unknown threat or internal conflict, whereas fear is focused on known external danger. 15 It has been suggested that "[…] anxiety can only be understood by taking into account some of its cognitive aspects, particularly because a basic aspect of anxiety appears to be uncertainty. Also, it is reasonable to conclude that anxiety can be distinguished from fear in that the object of fear is 'real' or 'external' or 'known' or 'objective.' The origins of anxiety are unclear or uncertain […]." 3 Other authors pointed out that "[…] situations lacking in clear indications of situational contingencies or likely outcomes are associated with considerable stress. The uncertainty regarding these situations highlights a lack of control that contributes to feelings of anxiety and makes coping more difficult." 15 Barlow has described anxiety as "[…] a unique and coherent cognitive-affective structure within our defensive and motivational system […]. At the heart of this structure is a sense of uncontrollability focused largely on possible future threats, danger, or other upcoming potentially negative events, in contrast to fear, where the danger is present and imminent." 16 The fact that anxiety and fear are probably distinct emotional states does not exclude some overlap in underlying brain and behavioral mechanisms. In fact, anxiety may just be a more elaborate form of fear, which provides the individual with an increased capacity to adapt and plan for the future. 16 If this is the case, we can expect that part of the fear-mediating mechanisms elaborated during evolution to protect the individual from an immediate danger have been somehow "recycled" to develop the sophisticated systems required to protect us from more distant or virtual threats.
Defense and coping strategies
Fear or anxiety result in the expression of a range of adaptive or defensive behaviors, which are aimed at escaping from the source of danger or motivational conflict. These behaviors depend on the context and the repertoire of the species. Active coping strategies are
The biology of fear-and anxiety-related behaviors -Steimer
Dialogues in Clinical Neuroscience - Vol 4 . No. 3 . 2002 used when escape from threat is possible, and the autonomic changes associated with these active strategies are mediated predominantly by sympathetic activation (hypertension, tachycardia). This is the fight-or-flight response originally described by Cannon. 17 Passive coping strategies, such as immobilization or freezing, are usually elicited when threat is inescapable, and are usually characterized by autonomic inhibition (hypotension, bradychardia), and a more pronounced increase in the neuroendocrine response (activation of the hypothalamopituitary-adrenal axis and increased glucocorticoid secretion). This type of passive response was originally described by Engel and Schmale as a conservation-withdrawal strategy. 18 The concept of alternative (active/passive) strategies itself owes much to the work of Henry and coworkers. 19 Specific brain circuits appear to mediate distinct coping reactions to different types of stressors. 20,21 According to Panksepp, flight and other active coping behaviors are unconditional responses to proximate threat, whereas passive coping strategies, such as freezing, are conditioned responses to distal stimuli predictive of danger. These two strategies have distinct and successive roles, and are modulated by the (cognitive) apprehension of the environment and probability of success, eg, whether or not there is a route of escape. Thus, when an animal faces a predator, freezing is preferentially activated when the source of known danger is still far away. When danger gets closer, and the stimulus passes through some critical "psychometric" distance, it becomes a true unconditional stimulus and a flight pattern is activated. 22 Defensive behaviors have been studied in a large number of species, 23 and it has recently been shown that human defensive behaviors to threat scenarios are not unlike those seen in nonhuman mammals. 24 The importance of risk assessment in making a proper decision about the best strategy to be used in a particular context has been emphasized. 25 It should be underlined, however, that the choice between an active or passive defense strategy does not entirely depend on contextual clues. Individual differences in coping styles do exist and may also influence this choice. In a given situation, some individuals may react actively ("proactive" style), whereas other individuals may react in a more passive way ("reactive" style). These coping styles are characterized by consistent behavioral and neuroendocrine patterns, and may explain individual differences in vulnerability to stressinduced diseases. 26 Differences in coping styles have also been found between various strains of mice, 27 or between genetically selected rat lines, 28 which suggests that they have a genetic basis. The capacity to cope successfully with life challenges, whether innate or acquired, is probably a primary determinant of resistance to stress-induced diseases. 29,30 Normal versus pathological anxiety Although anxiety is a natural adaptive reaction, it can become pathological and interfere with the ability to cope successfully with various challenges and/or stressful events, and even alter body condition (eg, formation of gastric ulcers).
In 1926, following a major flooding disaster in Leningrad, Pavlov reported a state of "chronic inhibition" and learning impairment in the dogs that had been successfully trained for conditioned responses in his laboratory and had directly experienced the flood. 31 This observation (which may be one of the first laboratory-based accounts of the symptoms of posttraumatic stress disorder) and other experiments were the basis for his later studies on "experimental neuroses" in dogs. Pavlov discovered large differences in dogs' individual susceptibility to psychopathology, and attributed these differences to "nervous types." He described four types analogous to the four temperaments of Hippocrates, which, according to him, resulted from the combination of three factors: the "strength" of the nervous system (its degree of resistance to excitation or inhibition), the equilibrium between excitation and inhibition processes, and the capacity to shift from inhibition to excitation and vice versa. 32 Although Pavlov's typology is outdated, it is now recognized that increased vulnerability to anxiety and its disorders is associated with particular traits or endophenotypes, ie, traits that may be intermediate in the chain of causality from genes to disease. 33 These traits may be innate or acquired during development or through experience. Barlow has defined three interacting sets of vulnerability factors for the development of human anxiety disorders in humans: (i) a generalized biological vulnerability, mainly of genetic origin; (ii) a generalized psychological vulnerability, resulting in particular from early life experiences; and (iii) a specific psychological vulnerability, focused on particular events or circumstances. 16 The lat-
S t a t e o f t h e a r t
ter set is probably implicated in the development of specific anxiety disorders (as opposed to generalized anxiety disorders), ie, social phobia, obsessive-compulsive and panic disorders, and specific phobias. Increased anxiety in animal models, as a trait, can be attributed to at least two sets of factors: (i) a genetic predisposition, essentially linked to the expression of genes that are involved in the various neurochemical mechanisms underlying fear and anxiety; and (ii) the influence of environmental factors. These environmental factors can interact with the expression of the relevant genes during early development and determine the functional properties of the neural and biochemical systems involved in coping with stressful events. They can also modulate the learning processes that occur at a later stage, when the individual is confronted with various life events, and determine the capacity to cope successfully with aversive or threatening situations in adulthood. These predisposing factors, either innate or acquired, determine individual "affective styles" 2,34 or coping strategies, 26 which are thought to play an important role in vulnerability to psychopathology.
Animal models
Some of the neurobiological mechanisms underlying anxiety may already be present in very simple organisms, such as the snail Aplysia, which can show forms of learning akin to anticipatory and chronic anxiety. 35 However, most animal models of anxiety are based on the use of mammalian species, particularly rats and mice. [36][37][38][39][40][41][42] These models fall into two broad categories. In the first one, animals are confronted with situations that generate an anxious state (state anxiety models). This state of anxiety can be either conditioned (eg, conditioned fear, avoidance, and punishment-induced conflict tests) or unconditioned (eg, aversive and ethological conflict tests). In the second category, the models are concerned with trait or "pathological" anxiety: genetic manipulations (transgenic or "knockout" animals) or selective breeding creates lines of rats or mice that permanently express an increased or decreased level of anxiety.
Functional neuroanatomy
As already suspected by Letourneau and others, emotional experience and the associated behavioral responses are likely to activate specific circuits in the brain. The search for the neuroanatomical substrates of fear and anxiety has been a successful field of research over the last decades. For a long time, it was assumed that emotions, including fear and anxiety, were almost exclusively generated or processed in a "primitive" part of the brain, ie, the limbic system ("the emotional brain"). The view that emotions and cognitions are separate functions of the brain and must therefore have different underlying neuroanatomical substrates is probably responsible for this simplification. As pointed out by LeDoux in a recent review, 43 modern research with the most advanced neuroimaging technologies still uses this dichotomic approach to higher brain functions as a post hoc explanation: "When a so-called emotional task is used, and a limbic area is activated, the activation is explained by reference to the fact that limbic areas mediate emotions. And when a limbic area is activated in a cognitive task, it is often assumed that there must have been some emotional undertone to the task." However, neuroanatomical and behavioral data obtained during the last decades clearly indicate that this dichotomy between cognitive and emotional processes is obsolete.
The locus ceruleus and arousal
Autonomic activation and increased arousal are among the earlier psychophysiological responses observed in a state of fear or anxiety. Since the immediate consequences of autonomic activation (eg, tachycardia) are perhaps the most readily perceived when experiencing a state of fear or anxiety, it has been proposed that the ascending noradrenergic system originating from the locus ceruleus (LC) is the core around which feelings of anxiety are organized. 44 The LC contains a large proportion of the noradrenaline (NA) cell bodies found in the brain and it is a key brain stem region involved in arousal ( Figure 1). It is highly responsive to alerting/stressful stimuli. In rats, cats, and monkeys, increased LC neuronal firing rate is associated with alertness, selective attention to meaningful and/or novel stimuli, and vigilance. The meaning, as well as the intensity of stimuli, seems to be an important factor in LC response. In cats, confrontation with a novel, but nonthreatening stimulus, such as a mouse, does not cause a specific increase in LC firing, whereas confrontation with a threatening stimulus (eg, a dog) causes a marked increase in LC firing. Thus, novelty by itself is not sufficient to activate the LC/NA system, but stimuli that signal reward, as those that signal danger, may activate the system. 45 Recent data suggest that a phasic mode of LC activity may promote focused or selective attention, whereas a tonic mode may produce a state of high behavioral flexibility or scanning attentiveness. 46 Figure 1. A schematic view of major brain circuits involved in fear and anxiety. External auditory, visual, olfactory, or somatosensory stimuli are relayed by the thalamus to the amygdala and cortex. The basolateral complex (BLA) of the amygdala is the input side of the system, which also receives contextual information from the hippocampal formation (entorhinal cortex, hippocampus, and ventral subiculum). After intra-amygdala processing of the emotional stimuli, the central nucleus of the amygdala (CeA), on the output side, activates the locus ceruleus (LC) and central and peripheral noradrenaline systems (via corticotropin-releasing factor [CRF] neurons), and the hypothalamus (paraventricular nucleus [PVN] and lateral hypothalamus [LH]). The bed nucleus of the stria terminalis (BNST, part of the "extended amygdala") is also a control center for the neuroendocrine system by integrating information originating from both the hippocampus and the amygdala. In addition, the CeA directly activates various midbrain regions or nuclei responsible for different aspects of the fear/anxiety response: freezing or escape (periaqueductal gray [PAG]), increased respiratory rate (parabrachial nucleus [PBN]), startle (caudal reticulopontine nucleus of the reticular formation [RPC]), and the dorsal motor nucleus of the vagus (DMN) in the medulla, which (together with the lateral hypothalamus) is responsible for the increase in heart rate and blood pressure associated with emotional events. The prefrontal cortex (PFC) processes more elaborate ("cognitive") information; it modulates the physiological, neuroendocrine, and behavioral responses (via the amygdala), and it is also involved in the extinction of fear-and anxiety-related conditional responses. ACTH, adrenocorticotropic hormone; ANS, autonomous nervous system; BP, blood pressure; GABA, γ-aminobutyric acid; Glu, glutamate; NA, noradrenaline (neurotransmitter) or nucleus ambiguus (structure); NTS, nucleus tractus solitarius.
tating the stress response associated with increased anxiety ( Figure 1). However, although 6-hydroxydopamine lesions of the LC in rats affect the HPA axis response to acute stress, they do not appear to substantially affect its response to chronic stress. 47 Noradrenergic LC neurons also project to the amygdala (mainly to the central nucleus of the amygdala [CeA]), the prefrontal cortex (PFC), the bed nucleus of the stria terminalis (BNST), the hippocampus, the periaqueductal gray (PAG), the hypothalamus, the thalamus, and the nucleus tractus solitarius (NTS), which are all areas involved in the fear/anxiety response (Figure 1). The LC is in turn innervated by areas such as the amygdala (which processes fear-related stimuli) and other areas receiving visceral stimuli relayed by the NTS. The LC is therefore in a key position to integrate both external sensory and internal visceral stimuli and influence stress-and fear-related neuroanatomical structures, including cortical areas. 48
The septohippocampal system and behavioral inhibition
The inhibition of ongoing behaviors is the first behavioral manifestation of an anxious or fearful state. In the 1970s, Gray suggested that vulnerability to anxiety is associated with individual differences in the activity of a septohippocampal behavioral inhibition system (BIS). According to Gray, this is one of the three major emotional systems, which also include the behavioral approach system (BAS) and the fight/flight system (F/FLS). 49,50 The primary function of the BIS is to compare actual with expected stimuli. If there is a discrepancy between the actual and expected stimuli (ie, "novelty" or "uncertainty"), or if the predicted stimuli are aversive, the BIS is activated, arousal and attention to novel environmental stimuli is increased, and ongoing behaviors are inhibited. Thus, according to Gray, anticipatory anxiety reflects a central state mediated by BIS activation, which is elicited by threats of punishment or failure, and by novelty or uncertainty. 51 The central role of behavioral inhibition in generating an anxious state has also been pointed out by Laborit. 52 Anxiety is associated with the "alarm reaction," as defined in Selye's original description of the stress response (or general adaptation syndrome). 53 According to Laborit, anxiety appears when one realizes that a proper adaptive action is not possible, ie, that there is loss of control over the situation, and it depends on the activation of the HPA axis.
Panksepp has argued that the activities of the ascending NA systems and the descending BIS are not causally related to the affective experience of fear and anxiety. 22 They may be correlated, supportive, or permissive systems for establishing brain states that participate in the many brain readjustments accompanying fear. These systems certainly participate in the genesis of fear and anxiety behaviors: the NA system is involved in the initial alarm reaction, whereas freezing promoted by septohippocampal inhibition may help regulate the intensity and duration of fear. However, according to Panksepp, the amygdala-central gray axis plays an essential role in creating the emotional state associated with fear and anxiety. 22
The amygdala-hypothalamus-central gray axis and fear
In all mammalian species, there are three distinct sites in the brain where electrical stimulation will provoke a full fear response: the lateral and central zones of the amygdala, the anterior and medial hypothalamus, and specific areas of the PAG. A circuit coursing from the lateral and central nuclei of the amygdala, throughout the ventralanterior and medial hypothalamic areas, down to the mesencephalic PAG, may constitute the executive system for fear, since freezing, as well as flight behavior and the autonomic indices of fear (eg, increased heart rate and eliminative behavior) can be evoked along the whole trajectory of this system. 41 In rats, stepwise increases in the electrical stimulation of the dorsolateral periaqueductal gray (dlPAG) produce alertness, then freezing and finally escape, replicating the sequence of natural defensive reactions when exposed to threat. Recent data suggest that dlPAG stimulation produces freezing independently of any contextual fear conditioning, whereas stimulation of the ventral periaqueductal gray (vPAG) appears to be critical to the expression of conditioned fear. 54 Because electrical or pharmacological stimulation of PAG produces a range of fear-related responses similar to those seen in a panic attack, this area be could be directly implicated in panic disorder. 55,56 The amygdala and fear conditioning The elegant studies carried out by LeDoux, based on a simple fear conditioning paradigm in rats, have emphasized the primary role of the amygdala in controlling 60 based mainly on the behavioral models of freezing and fear-potentiated startle in rats 61 have been worked out in detail. In LeDoux's model, the amygdala and thalamic pathways are responsible for the primary appraisal of threat by allowing a rapid, automatic analysis of potentially dangerous stimuli. Additional brain structures, including the hippocampus and cortical pathways, provide more information on the situational context and relevant stimulus characteristics (Figure 1). Thus, the amygdala plays a central role by integrating rapid, direct thalamic inputs, eg, visual information, with more detailed information, eg, cortical integration of sensory information, originating from longer and slower neuronal pathways. 43 Activation of the amygdala by threatening stimuli then influences cognitive processes, perception, selective attention, and explicit memory.
The cognitive representation of fear may preferentially involve the left amygdala, as shown by recent functional magnetic resonance imaging (fMRI) studies. 62 Interestingly, a sex difference in amygdala activation during the perception of facial affect has recently been reported. 63 Amygdala activation (measured by fMRI) differed for men and women depending on the valence of the expression: happy faces produced greater right than left amygdala activation for males, but not for females. Both sexes showed greater left amygdala activation for fearful faces. These data suggest that the left amygdala may be more involved in the representation of negative affect. The role of the various amygdala nuclei in fear conditioning is now well established, notably by lesion studies. 43,59,60,64 In rats, the central and medial nuclei of the amygdala are important in mediating conditioned aversive states, but conditioned freezing may be mediated independently. 65 Thus, different types of fear-conditioned behavior may be mediated by separate nuclei within the amygdala. 66 The amygdala plays a pivotal role in coordinating the behavioral, neuroendocrine, and prefrontal cortical monoamine responses to psychological stress in rats. In a fear-conditioning paradigm, pretraining amygdala lesions blocked freezing behavior, ultrasonic vocalizations, adrenocortical activation, and dopaminergic metabolic activation in the medial prefrontal cortex (mPFC). Posttraining lesions blocked mPFC dopamine, serotonin (5-hydroxytryptamine [5-HT]), and NA activation and stress-induced freezing and defecation, and greatly attenuated adrenocortical activation. 67
The amygdala and positive reinforcement and attention
The role of the amygdala is not limited to fear-conditioning and the processing of aversive stimuli. Studies in rats using food-motivated associative learning indicate that the basolateral amygdala may be involved in the acquisition and representation of positive reinforcement values (possibly through its connections with the ventral striatal dopamine systems and the orbitofrontal cortex). 68 Therefore, the amygdala is probably a key structure for the integration of behavior in conflicting situations, when both potentially rewarding and aversive stimuli are present. Recent studies indicate that the human amygdala can also process both positively and negatively valenced stimuli. 69 Recent studies also indicate that the CeA may contribute to attentional function in conditioning, by way of its influence on basal forebrain cholinergic systems and on the dorsolateral striatum. 68
The amygdala and social behavior and phobia
The amygdala may play an important role in regulating social behavior. Thus, in adult macaque monkeys, selective bilateral lesions of the amygdala result in a lack of fear response to inanimate objects and a "socially uninhibited" pattern of behavior. 70 The amygdala may function as a protective "brake" during evaluation of a potential threat, and it has been suggested that social anxiety may involve a dysregulation or hyperactivity of the amygdala evaluative process. 70 Studies in rats also suggest that the basolateral nucleus of the amygdala may play a crucial role in the consolidation of information that leads to the formation of a specific phobia. 71
The extended amygdala (BNST) and anxiety
Although the amygdala is clearly involved in conditioned fear, its role in anxiety is less evident, because it is often difficult to specify the stimuli that triggers anxiety. 72,73 Thus, lesions of the rat amygdala that suppressed fearelicited startle or freezing behavior did not affect measures of anxiety in the elevated plus-maze and shockprobe-burying tests, two classic tests of anxiety for rodents. 74 Moreover, diazepam was effective in these tests, even in amygdala-lesioned rats, suggesting that the anxiolytic effects of benzodiazepines are not necessarily mediated by the amygdala. 75 Recent studies in primates also suggest that the amygdala is involved in mediating some acute unconditioned fear responses in rhesus monkeys, but that it is unlikely to be a key structure regarding the dispositional behavioral and physiological characteristics of the anxious temperament. 76 The BNST is considered to be part of the extended amygdala. 77 It appears to be a center for the integration of information originating from the amygdala and the hippocampus (Figure 1), and is clearly involved in the modulation of the neuroendocrine stress response. 78,79 Activation of the BNST, notably by corticotropin-releasing factor (CRF), may be more specific for anxiety than fear. Studies in rats with the startle reflex suggest that explicit cues such as light, tone, or touch activate the amygdala, which then activates hypothalamic and brainstem target areas involved in the expression of fear, whereas less specific (or more complex) stimuli of longer duration, such as exposure to a threatening environment or intraventricular administration of CRF, may preferentially involve the BNST. 73
The PFC and the control of emotional responses
The primary roles of the PFC appear to be the analysis of complex stimuli or situations and the control of emotional responses.
In a revised version of his original BIS model, Gray postulated that the PFC may modulate septohippocampal activity, and that lesions to this area would impair the processing of vital information for the subicular comparator, and subsequently affect behavioral inhibition and anticipatory anxiety. 51 He also suggested that the role of cortical structures in anxiety was probably more prominent in primates, based on the increased anatomical relationship between the septohippocampal system and the prefrontal and cingulate cortices observed in monkeys. Recent studies in humans and primates have largely confirmed Gray's hypothesis, and it is now clear that the various subdivisions of the human PFC (dorsolateral, ventromedial, and orbital sectors) have specific roles in representing affect in the absence of immediate rewards or punishments and in controlling emotional responses. 80,81 There appear to be important functional differences between the left and right sides within each of these sectors. Earlier studies on patients with unilateral brain lesions have already emphasized the role of cerebral lateralization in emotional information processing. 82 More recently, brain electrical activity measures and positron emission tomography (PET) studies have indicated that negative affect and anxiety are associated with increased activation of the right PFC; moreover, individual differences in baseline levels of asymmetric activation in the PFC may be associated with individual differences in affective styles and vulnerability to mood and anxiety disorders. 81 There is also increasing evidence that the PFC plays an important role in controlling anxiety and the associated stress response in rats, and that cerebral laterality is an important feature of the PFC system. Thus, in a recent study right, but not left, lesions of the ventral medial PFC were shown to have anxiolytic effects, and were also more effective in suppressing the neuroendocrine and autonomic stress response. 83
Neurochemical correlates
A large number of neurotransmitters, peptides, hormones, and other neuromodulators have been implicated in fear and anxiety. We shall only discuss a few representative examples.
The noradrenergic system
Several preclinical studies have shown that stress and anxiety cause a marked increase in NA release in several rat brain regions, including the hypothalamus, the amygdala, and the LC. 84 In agreement with these data, yohimbine, an α 2 -adrenergic receptor antagonist that increases NA release in the brain, has been shown to have anxiogenic effects in rats. 84 However, pharmacological experiments involving the administration of various α 2A -receptor agonists or antagonists in several animal models of anxiety are
The biology of fear-and anxiety-related behaviors -Steimer Dialogues in Clinical Neuroscience -Vol 4 . No. 3 . 2002
inconsistent, perhaps due to their interaction with other monoaminergic receptors. 85 In a recent study, local administration into the LC region of an antisense oligodeoxynucleotide (AS-ODN) corresponding to the α 2A -receptor mRNA was shown to have an anxiolytic effect, 85 but another study has also shown that genetic knockout of the α 2A -receptor in mice resulted in a more anxious phenotype than that of the corresponding C57BL/6 wild type. 86 The role of the various NA receptor subtypes in mediating NA action on fear-and anxiety-related behaviors is therefore not settled. The precise location of the receptor subtypes within the complex circuitry mediating fear and anxiety responses is probably critical.
The serotonergic system
Data on the role of 5-HT in anxiety are conflicting: there is no agreement whether 5-HT enhances or, conversely, decreases anxiety. Thus, a 5-HT 2C agonist such as mchlorophenylpiperazine (mCPP) has anxiogenic effects in humans and may induce panic attacks, obsessions, and other neuropsychiatric symptoms, whereas selective 5-HT reuptake inhibitors (SSRIs) and 5-HT 1A or 5-HT 3 receptor-selective drugs can have antianxiety effects in certain anxiety disorders and animal models. 87 On the basis of data obtained from animal models, Graeff et al have proposed a "dual 5-HT fear hypothesis" postulating that 5-HT may enhance conditioned fear in the amygdala, while inhibiting innate fear in the dorsal PAG. 88 The ascending 5-HT pathway originating from the dorsal raphe nucleus (DRN) and innervating the amygdala and frontal cortex facilitates conditioned fear, while the DRN-periventricular pathway innervating the periventricular and PAG matter inhibits inborn fight/flight reactions to impending danger, pain, or asphyxia. 89 The same authors have also proposed that the pathway connecting the median raphe nucleus (MRN) to the hippocampus may promote resistance to chronic, unavoidable stress by facilitating hippocampal 5-HT 1A transmission. 89 These results demonstrate that it is not possible to conclude about an "anxiogenic" or "anxiolytic" role for 5-HT (or, for that matter, of any other neurotransmitter, peptide, or hormone) without considering its site of action in the brain and/or the receptor subtype implicated.
Indirect evidence that the anxiolytic action of 5-HT is mediated by the 5-HT 1A receptor has been obtained by three independent groups who have reported an "anxious" phenotype in 5-HT 1A receptor knockout mice compared with corresponding wild-type mice, using three different genetic backgrounds. 90 Depending on this background, the null mutation may be associated with changes in GABAergic transmission. 91 More recently, it has been shown that 5-HT 1A receptor knockouts display an "anxious-like" phenotype not only at the behavioral, but also at the autonomic response level. 92 This seems to provide a strong argument in favor of an important role of 5-HT 1A receptor gene expression for anxiety-related behaviors. In contrast, 5-HT 1B receptor knockout mice were found to be more aggressive, more reactive, and less anxious than their wild-type counterparts, suggesting that this receptor may also modulate 5-HT action on defense mechanisms. 93 Serotonin transporter (5-HTT) knockout mice (5-HTT-/-) have also been produced, and shown to display elevated anxiety in various behavioral tests, and an increased stress response (adenocorticotropic hormone [ACTH] secretion) following a mild stress, which was also observed to a lesser degree in the 5-HTT+/-heterozygotes. 94 The GABAergic system γ-Aminobutyric acid (GABA) is the most abundant inhibitory neurotransmitter in the brain. The GABA Abenzodiazepine receptor is an important target for several anxiolytic drugs and may therefore play an important role in anxiety-related disorders. 95 Several GABA A receptor subtypes have been described. 96,97 The diazepam-sensitive α 2 -GABA A subtype appears to be specifically involved in anxiolysis. 96 This subtype is largely expressed in the hippocampus, the amygdala, and the striatum. 98 Two mouse lines were generated with a knockin point mutation on the α 2 or α 3 subunit, which rendered them insensitive to diazepam. The anxiolytic action of diazepam was suppressed in mice with the α 2 (H101R) point mutation, but not in those with the α 3 (H126R) point mutation. 99 Heterozygous γ2-knockout mice (γ2+/-) have been generated (the homozygous mutation is not viable). 98 These mice show enhanced reactivity to natural aversive stimuli, increased passive avoidance responses, and a deficit
The biology of fear-and anxiety-related behaviors -Steimer
Dialogues in Clinical Neuroscience - Vol 4 . No. 3 . 2002 in ambiguous cue discrimination. 100 They have been proposed as a model for trait anxiety characterized by harm avoidance behavior and explicit memory bias for threat cues (enhanced sensitivity to negative associations).
In contrast to the anxiolytic action of benzodiazepinelike compounds, inverse agonists of the GABA/benzodiazepine receptor such as the β-carbolines are well known to be anxiogenic. Recently, intrahippocampal injections of a novel inverse agonist (RY024) have been shown to produce a fear response (freezing) and to interfere with fear-conditioning in rats. 101
The neurosteroids
The neurosteroids are a novel, interesting class of neuromodulators synthesized in the brain directly from cholesterol. 102 They appear to act essentially via an allosteric modulation of the GABA A receptor, although other receptors may also be involved. 102,103 As early as 1987, Majewska suggested that neurosteroids could play an important role in mood regulation. 104 Several studies have shown that positive allosteric modulators (which potentiate GABA action), such as progesterone and allopregnanolone, have anxiolytic effects in various animal models. 103 Neurosteroid synthesis is regulated by a peripheral benzodiazepine receptor (PBR) located on the outer mitochondrial membrane, 105 and part of the anxiolytic effects of benzodiazepine could in fact involve increased neurosteroid synthesis. Compounds with a selective affinity for the PBR, such as FGIN-1-27, have shown an anxiolytic action in rats. 106 Neurosteroids are currently attracting a lot of interest because of their potential role as natural, endogenous anxiolytics.
Hormones of the HPA axis
Hormones of the HPA axis, such as cortisol, or corticosterone (in rodents),ACTH, and CRF are usually increased in a state of fear and anxiety.They also appear to modulate the response to threatening events.
Corticotropin-releasing factor
Intracerebral administration of CRF has been shown to elicit anxious-like behavior in rats. 107 More recent preclinical studies suggest that CRF and its receptors play a pivotal, integrative role in the stress response and anxiety-related behaviors. 108,109 There are two major CRF sys-tems in the brain: the neuroendocrine system in the PVN, and another system with CRF cells located in the amygdala (CeA) and BNST, which would be more directly related to the physiological and behavioral responses associated with fear and anxiety.Whereas glucocorticoids restrain CRF production in the PVN (the neuroendocrine negative feedback loop), they appear to increase CRF expression in the amygdala and BNST, thus promoting fear-and anxiety-related behavior. 110 CRF neurons originating from the amygdala project onto the LC ( Figure 1) and contribute to increased arousal in fear and anxiety states. 111 In a rat model, a full postsynaptic CRF agonist, CRF , increased arousal at low dosage and had an anxiogenic action at higher doses. 112 This suggests that progressively increasing levels of CRF in the brain may ensure the transition from the initial state of increased arousal to the anxious state of expectancy in stressful situations. Transgenic mice overexpressing CRF show a behavioral and neuroendocrine profile consistent with an increased level of stress and anxiety, including elevated plasma ACTH and corticosterone levels, and generally exhibit the same behavioral changes as those observed in mice following exogenous CRF administration. [113][114][115] Recent data indicate a desensitization of postsynaptic, but not presynaptic 5-HT 1A receptors in mice overproducing CRF. 116 Another line of transgenic mice overexpressing CRF (CRH-OE(2122)) has shown a reduced startle reactivity, habituation, and prepulse inhibition. 117 Deletion of the CRF gene (CRF-KO mice) results in chronic glucocorticoid insufficiency, and this may cause severe developmental problems. 114,118 Despite an impaired stress-induced activation of the HPA axis, the behavioral stress responses do not appear to be markedly affected in CRF-deficient mice, suggesting that other CRF-like molecules may be implicated in the behavioral effects mediated by CRF receptors. 114,118-120 CRF-KO mice also display normal startle-and fear-conditioned responses. 120
CRF receptors and CRF binding protein
Deletion of the genes coding for CRF receptors 1 (CRF-R1) or 2 (CRF-R2) have more profound behavioral effects. 114,115,[121][122][123][124] CRF-R1-deficient mice display decreased anxiety and an impaired stress response, 125 whereas deletion of the CRF-R2 gene has the reverse effect in males (but not in females): anxiety is increased in Crhr2-/-. 126 These data suggest that CRF-R1 mediates the anxio-genic effects of CRF, whereas CRF-R2 may be involved in anxiolysis. Recently, mice deficient in both CRF-R1 and CRF-R2 receptors have been generated. 127 These double mutants display altered anxiety-related behavior and an impaired HPA axis response to stress. Interestingly, the effects on anxiety are again sex-dependent: females show a decreased anxiety similar to that observed in Crhr1-/-mutants, whereas the genotype has no effect on male anxiety-related behaviors. These studies have also demonstrated a novel role of the mother's genotype on the development of pup anxiety: pups born to a heterozygous or mutant mother display significantly more anxiety, regardless of that pup's genotype. 127 The CRF binding protein (CRF-BP) may play an important modulatory role in CRF action. 128 Interesting data consistent with a modulatory action of CRF-BP have recently been obtained with transgenic and knockout models: transgenic males overexpressing CRF-BP tend to show less anxiety, whereas the behavior of CRF-BPdeficient mice was consistent with increased anxiety. 129
Corticosteroids
Corticosteroids effects on anxiety-related behaviors may be mediated by both genomic and nongenomic mechanisms (control of neuronal excitability). Hippocampal corticosteroid receptors play an important role in the termination of the acute stress response. 130 Studies with a model of posttraumatic stress disorder in rats suggest an alteration of the mineralocorticoid receptor (MR) vs glucocorticoid receptor (GR) balance, as measured by the expression of mRNA levels in the hippocampus, during the recovery phase following acute stress: the MR/GR ratio was decreased, but only in animals with an enhanced fast feedback. 131 Recent data also suggest that, at low circulating levels, corticosteroids exert a permissive action (via MRs) on acute freezing behavior and other acute fear-related behaviors. At higher levels, corticosteroids enhance acquisition, conditioning, and consolidation of an inescapable stressful experience, as well as processes underlying fear potentiation, via GR-dependent mechanisms. 132 Mice with targeted mutation of the MR and GR receptors display altered anxiety-related behaviors. 133
Other peptides, neurotransmitters, and hormones
Several peptides, such as cholecystokinin (CCK), neuropeptide Y (NPY), tachykinins (substance P, neuro-kinins A and B), and natriuretic peptides (atrial natriuretic peptide or C-type natriuretic peptide) may play important roles in fear-and anxiety-related behaviors. 134 CCK may be particularly relevant for panic disorders, 135,136 and may influence cognitive processes. 137 Excitatory amino acids (EAA), such as glutamate, are also important. In rats, microinjections of EAA into the dorsolateral PAG induce a flight reaction. Part of the effects mediated by N-methyl-D-aspartate (NMDA) receptors may involve nitric oxide (NO). Nitric oxide synthase (NOS) inhibitors injected in the dorsolateral PAG have been shown to have anxiolytic effects, and psychological stress (restraint) induced an increased expression of neuronal NOS in the same area and in other areas related to defense mechanisms, suggesting that NO may participate in these defensive responses. 138 We have also shown that anticipatory anxiety can lead to a decreased secretion of luteinizing hormone (LH) and testosterone in young, healthy male subjects. 139
Genetic and environmental factors
Individual differences in sensitivity to threat or stress, and particular coping or affective styles appear to be critical predisposing factors for anxiety-related disorders. Genetic and environmental factors have been implicated, and how these factors interact during development is one of the major questions addressed by recent clinical and fundamental research.
Genetic determinants
A genetic basis for anxiety-related behaviors is now clearly established, notably through several family, twin, and adoption studies. In mice, targeted gene mutations have shown that modifying the expression of particular genes can have a profound effect on anxiety-related behavioral phenotypes. 39,140 Some examples were mentioned in the preceding section. Natural variations in trait anxiety, or emotionality, in inbred rat and mouse strains are being extensively studied. 27,39,[141][142][143][144][145][146] Some of these strains show differences in sensitivity to anxiolytic agents such as diazepam. 147,148 Crossbreeding of inbred rodents strains has shown the quantitative nature of many anxiety-related traits. 149,150 The quantitative trait locus (QTL) method is based on a comparison between the allelic frequency of DNA S t a t e o f t h e a r t markers and quantitative behavioral traits. 146,150 It has been used to assess gene effects on fear, emotionality, and anxiety-related behaviors in mice from various genetic backgrounds. 140,151 Loci on mouse chromosomes 1, 4, and 15 were found to operate in four tests of anxiety, whereas loci on chromosomes 7, 12, 14, 18, and X influenced only a subset of behavioral measures. 152 A QTL influencing anxiety has also been found recently on rat chromosome 5. 153 Selective breeding of mice and rats has also been used to create lines that show extreme behavioral characteristics within the range of the normal population. 140 Various selection criteria can be used, which may not be directly related to anxiety. Thus, rat lines initially selected for their good versus poor performance in two-way, active avoidance were subsequently shown to differ in trait anxiety, or emotionality. For instance, the Roman high-(RHA/Verh) and low-(RLA/Verh) avoidance rat lines display clear differences in emotionality and anxiety-related behaviors. 28,154 The more anxious (RLA/Verh) rats display increased neuroendocrine and autonomic reactivity to mild stressors. 28,155,156 Differences in vasopressin, oxytocin, and CRF action at the level of the amygdala, 156,157 dopaminergic and GABAergic neurotransmission, 158 basal vasopressin mRNA expression in the hypothalamic PVN, 159 and 5-HTT levels in the frontal cortex and hippocampus 160 have been reported. We have shown an increased capacity (enzymatic activities) for the production of progesterone-derived, anxiolytic neurosteroids in the frontal cortex and BNST of RHA/Verh rats, which may explain in part the differences in emotional reactivity of these two lines. 28 These two rat lines also differ in their respective coping styles and response to novelty, 154,155 and this model may therefore prove useful for studying the interaction between anxiety and defense mechanisms. Recently, two Wistar rat lines have been selected and bred for high anxiety-related behavior (HAB) or low anxiety-related behavior (LAB) on the elevated plusmaze, a classical test for anxiety in rodents. 149 The neuroendocrine, physiological, and behavioral characteristics of these two lines are being extensively studied, and show some similarities, but also differences, as compared to the Roman rat lines. [161][162][163][164][165][166][167] Further comparison between lines such as the RHA/RLA and HAB/LAB rats, which have been selected on different behavioral criteria (avoidance versus anxiety in the elevated plus-maze test), but show a similar, anxiety-related behavioral phe-notype, may be extremely fruitful to delineate brain mechanisms underlying specific aspects of anxiety disorders.
Environmental influences
The role of environmental influences in the etiology of anxiety is also well established. 15 Early adverse experience is a major developmental risk factor for psychopathology. [168][169][170] Prenatal stress in animal models has been shown to permanently alter brain morphology, anxiety-related behavior, coping, and regulation of the HPA axis in adulthood. 171 Naturally occurring variations in maternal care can also alter the regulation of genes controlling the behavioral and neuroendocrine responses to stress, as well as hippocampal synaptic development. These effects are responsible for stable, individual differences in stress reactivity, as well as the maternal behavior of female offspring. 172 They could constitute the basis of a nongenetic mechanism for the transmission of individual differences in stress reactivity and coping styles across generations.
In 1958, Levine reported that rats handled for the first 21 days of life exhibit reduced fearfulness compared with nonhandled controls. Since then, several studies have shown the beneficial effects of neonatal handling and a progressive habituation to stress on adults' stress responses and anxiety-related behaviors. Neonatal handling can even reverse the behavioral abnormalities induced by prenatal stress. 173 These effects appear to be mediated essentially by the CRF/HPA axis system, 174,175 although the serotonergic and catecholaminergic systems could be also involved. 176,177 A study has shown that neonatal handling increases the expression of the peripheral benzodiazepine receptor (PBR), which has been implicated in the synthesis of endogenous, natural anxiolytic agents such as the neurosteroids, in rat adrenals, kidney, and gonads. 178 It is likely that increased adrenal production of naturally anxiolytic compounds such as allopregnanolone contributed to the decrease in anxiety reported in this study. Sex differences in the effects of neonatal handling have been recently reported: neonatal handling may provide males with a greater capacity to actively face chronic stressors. 179 Recent data indicate that neonatal handling can also affect memory processes involved in contextual fear conditioning. 180 In the Roman rat lines, neonatal handling has been shown to alter the behavioral phenotype of the more anxious RLA/Verh rats so that, in adulthood, they behave in the same way as their nonhandled, hypoemotional RHA/Verh counterparts. Females were found to be more sensitive than males to the positive influences of early stimulation. 181 The effects of neonatal handling on RLA/Verh rats were not limited to behavioral stress responses and coping behaviors, but were accompanied by a concomitant decrease in stress-induced ACTH, corticosterone, and prolactin release, indicating that the neurochemical substrates underlying these responses were also permanently affected by early experience. 182,183 This and other examples indicate that the developmental processes that determine individual sensitivity to stressors, or emotionality, and coping behaviors involve complex interactions between genetic and environmental factors, and that anxiety-related phenotypes cannot be predicted on the sole basis of a genetic predisposition or early adverse experience.
Conclusions
The biological bases of fear and anxiety are now recognized, and the major brain structures and neuronal circuits involved in emotional information processing and behavior are delineated. Emotional and cognitive processes cannot be dissociated, even when considering such a basic emotion as fear. The cognitive apprehension of events and situations is critically involved in emotional experiences and also influences coping strategies or defense mechanisms. This is reflected in the important role now attributed to the PFC in controlling emotional behavior in humans and animals. Molecular biology techniques, such as those used to create transgenic and knockout mice, have been successful in exploring the role of various neurotransmitters, peptides, hormones, and their receptors in mediating the appraisal of stressful stimuli, information processing through the various neuronal circuits, and the physiological responses and behaviors associated with fear and anxiety.
It is now clear that individual differences in affective or coping styles, which are also observed in nonhuman species, are directly associated with vulnerability to psychopathology. Studying these individual differences, including sex-related differences, in humans and in animal models will give interesting clues about the brain mechanisms of emotional behavior. Finally, the study of genetic predisposition and environmental influences, particularly during early development, in determining vulnerability traits and anxietyprone endophenotypes is certainly becoming one of the major, and perhaps most promising, domains of contemporary research with respect to our understanding of the etiology of anxiety and mood disorders. ❏ The author would like to express his gratitude to the Swiss National Science Foundation for supporting work on the Roman rat lines in his laboratory (grant 32-51187-97).
S t a t e o f t h e a r t 244
La biología de las conductas relacionadas con el miedo y la ansiedad La | 2014-10-01T00:00:00.000Z | 2002-09-01T00:00:00.000 | {
"year": 2002,
"sha1": "1d5235ca42b0e9b91afe9e56593056a883b685e9",
"oa_license": "CCBYNCND",
"oa_url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3181681",
"oa_status": "GREEN",
"pdf_src": "PubMedCentral",
"pdf_hash": "1d5235ca42b0e9b91afe9e56593056a883b685e9",
"s2fieldsofstudy": [
"Psychology",
"Biology"
],
"extfieldsofstudy": [
"Medicine",
"Psychology"
]
} |
214750396 | pes2o/s2orc | v3-fos-license | Classical homocystinuria: A common inborn error of metabolism? An epidemiological study based on genetic databases
Abstract Background Biallelic pathogenic variants in CBS gene cause the most common form of homocystinuria, the classical homocystinuria (HCU). The worldwide prevalence of HCU is estimated to be 0.82:100,000 [95% CI, 0.39–1.73:100,000] according to clinical records and 1.09:100,000 [95% CI, 0.34–3.55:100,000] by neonatal screening. In this study, we aimed to estimate the minimal worldwide incidence of HCU. Methods The 25 most common pathogenic alleles of HCU were identified through a literature review. The incidence of HCU was estimated based on the frequency of these common pathogenic alleles in a large genomic database (gnomAD). Results The minimum worldwide incidence of HCU was estimated to be ~0.38:100,000, and the incidence was higher in Europeans non‐Finnish (~0.72:100,000) and Latin Americans (~0.45:100,000) and lower in Africans (~0.20:100,000) and Asians (~0.02:100,000). Conclusion Our data are in accordance with the only published metanalysis on this topic. To our surprise, the observed incidence of HCU in Europeans was much lower than those described in articles exploring small populations from northern Europe but was similar to the incidence described on the basis of neonatal screening programs. In our opinion, this large dataset analyzed and its population coverage gave us greater precision in the estimation of incidence.
cysteine levels. The main clinical complications in untreated HCU patients are found in the eyes, skeleton, central nervous system, and vascular system (Morris et al., 2017).
More than 200 pathogenic variants have been described in the CBS gene, and most of these are rare and private variants (Stenson et al., 2017). However, the four most prevalent mutations (p.Ile278Thr, p.Thr191Met, p.Gly307Ser, and p.Trp323Ter) represent half of all HCU alleles reported worldwide (Kraus, 2019). Rare metabolic monogenic diseases like HCU are usually characterized by allelic heterogeneity and show a broad spectrum of clinical expressivity (e.g., for some diseases, even the penetrance is not 100%). Besides that, in the absence of newborn screening programs, their diagnosis is usually delayed. Those factors together contribute for making the estimating of the incidence/prevalence of those diseases a real challenge. In the case of HCU, this is even more complicated, since there is not a good biomarker for newborn screening of the milder forms (the "responsive to pyridoxine" patients).
The worldwide prevalence of HCU based on the number of known patients is estimated to be between 0.29 and 1:100,000 individuals (Moorthie, Cameron, Sagoo, Bonham, & Burton, 2014;S.H., H.L., & J.P., 2001). Moorthie et al., 2014 performed a systematic review and meta-analysis to estimate the prevalence of HCU and found a worldwide prevalence based on diagnosis of symptomatic individuals of 0.82:100,000 [95% CI, 0.39-1.73:100,000], while that based on neonatal screening by MS/MS was 1.01:100,000 [95% CI, 0.34-3.55:100,000] newborns (NBs). We like to point out that the study of Moorthie et al., 2014 included the Qatari population, with an extremely high prevalence of HCU of 55:100,000 which may introduce an overestimation of the worldwide prevalence.
Another strategy to estimate the incidence of HCU is via determining the frequency of carriers for pathogenic alleles in the CBS gene and next use it to calculate the expected number of patients with HCU via the Hardy-Weinberg (HW) equation. The first researchers to use this approach were Gaustadnes, Ingerslev, & Rütiger, 1999, who screened 500 consecutive Danish NBs for the c.833T>C mutation and estimated the incidence of HCU to be at least 4.8:100,000. Linnebank et al., 2001 also conducted screening for the c.833T>C mutation in 200 healthy unrelated German controls and calculated the frequency of homozygosity for this mutation to be 5.6:100,000 individuals. In Norway, Refsum, Fredriksen, Meyer, Ueland, & Kase, 2004 determined the presence of six specific mutations of the CBS gene in 1,133 NB blood samples randomly selected from ~12,000 samples, and they calculated an HCU prevalence of ~15.6:100,000. Janosík et al., 2009 estimated the frequency of HCU in the Czech Republic via determining the presence of the c.1105C>T mutation in 600 NB blood spots, and they calculated the birth prevalence for HCU to be at least 2.5:100,000.
Thus, there is an about a 6-fold unexplained discrepancy between the number of known patients with HCU and that calculated on the basis of the number of carriers detected via genetic analyses of relatively small populations in Northwest Europe. There is no clear explanation for this discrepancy, but it could be due to the low penetrance or expressivity of some genotypes or to underdiagnoses. This discrepancy triggered us to obtain a more reliable estimate of the minimal worldwide incidence of HCU, using the data available in relevant large genomic databases.
| METHODS
We determined the 25 most common pathogenic variants in HCU patients via a literature review using the key terms "CBS mutation" and "Classical homocystinuria" in PubMed (www.ncbi.nlm.nih.gov/pubmed) and by examining references cited in related papers. Publications that contained molecular data in HCU patients were selected and used in the analysis, the search resulted in the inclusion of forty papers, containing 1,026 independent alleles from 25 countries. Since several of these studies described only a few patients, which could lead to overestimation of the frequency of a specific allele, only data from those countries in which at least ten alleles (five non-related patients) were used in the analyses totaling 1,014 alleles (Table 1).
Based on the 25 most frequent variants of the literature review, we conducted searches to determine the prevalence of these variants in the general population in two relevant genomic databases: gnomAD v2.1.1 (Lek et al., 2016, last accessed October 2019 and ABraOM (Naslavsky et al., 2017, last accessed July 2019). The first database includes worldwide data from 141,456 unrelated individuals sequenced as part of various disease-specific and population genetic studies and it is possible to access different subgroups in which individuals can overlap (e.g., one individual could be included in more than one subgroup): controls, non-cancer, non-neuro, and non-TOPMed. The individuals are clustered according to their genetic determination of ancestry. For example, individuals residing in the USA or Brazil may be clustered as European, African or Asian according to their genetic background. The second database, ABraOM, uses data from 609 healthy elderly individuals who were selected using a standardized sampling process from the city of São Paulo, Brazil; nearly 10% of the Brazilian population is located in this city, making it reasonably representative of the country.
The estimated incidence of HCU was calculated based on the assumption that HW equilibrium exists; thus, the frequencies are ''p'' for the wild-type allele and ''q'' for the pathogenic allele. The different allele frequencies for each pathogenic variant were summed.
| RESULTS
HCU patients from all selected studies were grouped according to their country of origin, and allelic frequencies were calculated for each variant in each country. The 25 most frequent variants of the CBS gene are described in Table 1.
| Most common variants
The five most common pathogenic CBS variants identified in our literature review (46% of alleles) were p.Arg336Cys, p.Ile278Thr, p.Gly307Ser, p.Thr191Met, and p.Trp323Ter. The countries where these pathogenic variants are most common are highlighted in Figure 1 and Table 1.
| p.Arg336Cys
In this study, p.Arg336Cys presented an overall allele frequency among HCU patients of 14% (149 alleles), and it was by far the most common variant in Qatar (97% of alleles). p.Arg336Cys was found in 15% of HCU Saudi Arabian patients but in no more than 10% of cases in European and Asian patients.
In the gnomAD, this variant was found only in non-Finnish Europeans, and it was present in 0.004% of alleles in this population. Patients homozygous for p.Arg336Cys are usually unresponsive to treatment with pyridoxine, and untreated patients present a severe clinical phenotype with involvement of the eyes, bones and vascular and central nervous systems.
| p.Ile278Thr
Our data showed an allele frequency among HCU patients of 13% (133 alleles) for p.Ile278Thr, which is the most widely dispersed variant in the world. The p.Ile278Thr was the most common pathogenic variant reported in the USA, Brazil, France, Italy, Germany, the Netherlands, the Czech Republic, Slovakia, Poland, Denmark, England, and Israel. Figure 2 illustrates the presence and frequency of this pathogenic variant around the world in HCU patients.
| p.Gly307Ser
A reported allele frequency of 10% (108 alleles) was found for p.Gly307Ser in HCU patients from the USA, Europe, Israel, Australia, and Qatar. p.Gly307Ser was the most common reported pathogenic variant in Ireland (66%) and Australia (22%).
According to gnomAD this pathogenic allele was present in Europeans (0.03% of all alleles) and Africans (0.008%) ( Table 2). Patients homozygous for p.Gly307Ser are usually non-responsive to treatment with pyridoxine and present a severe clinical phenotype.
| p.Thr191Met
p.Thr191Met presented an allele frequency among HCU patients of 8% (82 alleles) and was the most common pathogenic variant reported in countries of the Iberian Peninsula and in their former colonies in Latin America. The highest frequencies of this variant among HCU patients were found in Spain (44% of the alleles), Portugal (23%), Colombia (73%), and Venezuela (20%).
Data from gnomAD indicated the presence of the variant in Latin Americans (0.038%), but it was not identified in ABraOM. Patients who are p.Thr191Met homozygous are usually non-responsive to pyridoxine and present a moderate to severe clinical phenotype.
| p.Trp323Ter
The overall allele frequency among HCU patients of p.Trp323Ter was 2% (25 alleles). This variant was reported in patients from Saudi Arabia (77% of alleles) and northeast Brazil (6% of alleles). Interestingly, according to gnomAD data this variant is very rare and found only in one allele among Asians. This variant was not observed in ABraOM, which analyzed persons from São Paulo, Brazil. Patients homozygous for p.Trp323Ter are usually non-responsive to treatment with pyridoxine and present a moderate to severe clinical phenotype.
| HCU worldwide incidence
In the genetic database gnomAD, we found 304 individuals who were carriers for any of 20 of the 25 most frequent pathogenic alleles of the CBS gene, yielding an estimated HCU incidence (i.e., homozygosity or compound heterozygosity) of ~0.38:100,000 (95% CI, 0.29-0.39:100,000) individuals. When we analyzed only the control group of gnomAD, the estimated incidence was ~0.22:100,000; if we analyze only the non-neuro subgroup the incidence is ~0.42:100,000 in (Table 2). No homozygous individuals, for the 25 most frequent pathogenic variants, were found in this database.
| HCU incidence in southeastern Brazil
In the ABraOM database, we found only two of the 25 variants analyzed (p.Ile278Thr and p.Ala114Val). A total of 12 individuals carrying either of the pathogenic variants were included in this database (11 carriers of p.Ile278Thr and one of p.Ala114Val), yielding an estimated incidence of HCU of ~9.7:100,000 individuals. No homozygous individuals were found.
| DISCUSSION
Knowledge of the genetic background of HCU in different populations is generally poor and even contradictory. This omission hampers proper patient genetic counseling and appropriate genetic testing. Knowledge of the prevalent pathogenic variants and their frequencies will support decision making within national screening programs. Furthermore, there is an approximately 6-fold discrepancy between the number of known patients with HCU and the estimate calculated on the basis of the number of heterozygotes detected via genetic analyses of relatively small populations.
In this study, we used the results of published articles to characterize the worldwide mutational profiles of HCU patients. Next, we used the 25 most common published pathogenic variants (Table 1) to determine the corresponding allele frequencies in genomic databases and to calculate the incidence of HCU in different ancestralities (Table 2). Interestingly, the frequencies of these 25 most commonly pathogenic variants reported in various countries are in line with the data from genomic databases; for instance, p.Ile278Thr was very common in different ancestries, p.Thr257Met and p.Ala114Val were described in patients with HCU in different continents and found in several ancestralities and p.Thr191Met was found in Latin Americans. Europeans seems to be the group with the greatest allelic diversity, which leads us to hypothesize that dispersion of these pathogenic alleles occurred during the colonization period of America and Africa. In addition to the much lower incidence of HCU, we observed a distinct pattern of mutations in Asia and Russia, where 75% of alleles differ from the 10 most common pathogenic variants worldwide.
Based on biochemical neonatal screening data obtained by the measurement of Met in dried blood spots (DBSs), Naughten, Yap, & Mayne, 1998 reported HCU incidences of 0.77:100,000 NBs in Germany, 0.8:100,000 NBs in England and 0.34:100,000 NBs in the USA and higher incidences of 1.5:100,000 NBs in Ireland and 1.8:100,000 NBs in Italy. According to Mathias & Bickel, 1986, the incidence in Germany was 0.32:100,000 NBs based on biochemical neonatal screening data of almost 1 million individuals. Biochemical neonatal screening of 820,797 individuals in New South Wales, Australia, around the 1960s revealed 14 cases of HCU, resulting in an incidence of 1.72:100,000 NBs (Wilcken & Turner, 1978).
In Asian countries, a much lower HCU incidence is observed. In Japan, an extremely low incidence of 0.11:100,000 NBs was observed despite an effective biochemical screening program . National biochemical neonatal screening performed in the Philippines between 1996 and 2001 identified no HCU patients among 176,548 samples (Padilla, 2003). In Taiwan, 5 million individuals were subjected to biochemical neonatal screening for HCU, and only 3 were diagnosed with the disease. In sharp contrast, an extremely high frequency of HCU of 416:100,000 individuals was found on an island inhabited by an Austronesian Taiwanese Tao tribe (Lu et al., 2012). Kaur, Das, & Verma, 1994 investigated 2,560 high-risk patients with strong suspicion of an inborn error of metabolism in northern India, and the most commonly found disorder was HCU (0.6%).
Qatar is the country with the highest incidence of HCU in the world due to a founder effect of p.Arg336Cys. This pathogenic variant rate is very frequent in three tribes of the Qatari population, and consanguineous marriages even enhance the high incidence of HCU. Initially the incidence of HCU was estimated to be ~33:100,000 individuals (El-Said et al., 2006). However, after the implementation of neonatal screening through the detection of tHcy and Met combined with genetic screening, the estimated incidence of HCU in the Qatari population increased to 55:100,000 NBs (Gan-Schreier et al., 2010).
Newborn screening is being carried out in countries with high incidences of HCU, such as Ireland and Qatar Zschocke et al., 2009). For this purpose, tHcy is measured in DBSs with a dedicated method in Qatar. However, all other newborn screening programs measure Met in DBSs, which results in a high proportion of false negatives, particularly for pyridoxine-responsive forms of HCU, because these patients seem not to develop hypermethioninemia in the first days of life, so due to the limitation of newborn screening method this patients are likely not diagnosed (Bowron, Barton, Scott, & Stansbie, 2005;McHugh et al., 2011;Peterschmitt, Simmons, & Levy, 1999). Countries in Latin America have no neonatal screening program for HCU; however, based on our estimation of HCU incidences in these populations of ~0.45:100,000 we like to advocate to introduce newborn screening for HCU in Latin American countries. Countries such as Japan and the USA have neonatal screening programs for HCU, even though the incidences in these countries are lower than (Japan) or similar to (USA) those estimated in Latin America.
Among CBS mutations, p.Ile278Thr is geographically the most widespread. Studying the emergence and dispersal of this mutation, Vyletal et al., 2007 reported that haplotype c. [833C; 844_845ins68] is very common in sub-Saharan Africa (up to 40% of control chromosomes), less frequent throughout Europe and America (5%-10% of control chromosomes), and rare in Asia (0.16%-2.5% of control chromosomes). It was concluded that the p.Ile278Thr variant occurred repeatedly and independently in the recent history of the European population. Interestingly and confusing is the haplotype c.[833C; 844_845ins68] on the CBS gene, which contains the c.833T>C. But this haplotype is considered non-pathogenic since c.844_845ins68 creates an alternative splice site that rescues the wildtype CBS sequence from the mutated allele, resulting in normal CBS enzyme activity and normal Hcy concentrations (Kluijtmans et al., 1997). We have to take into account that in the Brazilian database the presence of variant c.844_845ins68 has not been described, but since the frequency of p.Ile278Thr is very high (0.9%), we cannot rule out the possibility that some of these individuals may carry the non-pathogenic haplotype. In the gnomAD, we assumed that the individuals heterozygous for p.Ile278Thr (0.08% of the total sample) had the isolated pathogenic variant, since the frequency of individuals with the c.844_845ins68 variant was approximately 12%.
The prevalence of HCU varies dramatically between regions from 416:100,000 on Orchid Island and 55:100,000 in Qatar to less than one in one million in the Taiwanese Han population (Gan-Schreier et al., 2010;Lu et al., 2012). In this study, we used the genetic database gnomAD to determine the frequency of CBS heterozygotes and next calculated the worldwide incidence of HCU, which was found to be approximately 0.38:100,000 individuals. Stratifying populations by ancestry, the highest incidence of HCU was found in Europeans and Latin Americans. A much lower incidence was found in Asians. The incidences in these various regions were more or less in line with those found through neonatal biochemical screening. For instance, in Europeans, an incidence of 0.72:100,000 individuals was calculated on the basis of the number of heterozygotes in gnomAD versus 0.77:100,000 according to neonatal screening, and in Asian, the corresponding values were 0.02:100,000 versus 0.07:100,000 individuals .
Another remarkable finding is that the incidence calculated in this study for Europeans of approximately ~0.72:100,000 individuals is much lower (approximately 6 to 7 times) than those described in four different studies. At least 4.8:100,000 live births in Denmark, 5.6:100,000 in Germany, 2.5:100,000 in Czech Republic ~15.6:100,000 in Norway. We have no explanation for this discrepancy except that the numbers of studied individuals were relatively small (200 to maximal 1,133 individuals) and that publication bias may have played a role. The Europeans sample size of gno-mAD is approximately 115 times larger than the sample sizes of these studies and should therefore provide a much more precise incidence rate.
We estimated the number of HCU patients using HW equilibrium. The HW principle presents limitations because it analyzes allele frequencies and genotype counts in successive generations and predicts that in a random mating population of infinite size, allele and genotype frequencies should remain constant from one generation to the next. Factors that may disrupt the HW equilibrium included mutation rate, natural selection, migration, population structure (nonrandom marriage and/or consanguinity) and nonrandom selection of the samples studied (Piel et al., 2016;Waples, 2015). Overall, we assume that these limitations do not substantially affect the numbers we calculated.
The genomic database gnomAD consists of 282,912 alleles and includes data from consortia such as 1,000 genomes, GO-ESP and TOPMed and provides sequence data from unrelated individuals from various disease-specific populations included in genetic studies. Therefore, our frequency analysis is based on diverse populations from various countries and ethnicities clustered according to their genetic determination of ancestry. Although the calculated frequency of HCU is relatively low, we consider it to be rather precise estimation of incidence because of the large number of individuals and the different genetic backgrounds included.
Genetic data provided by databases present limitations because of the heterogeneous inclusion criteria of the original studies such as age or selection based on diseases. Although gnomAD is the largest public genetic database to our knowledge, it should be taken into account that individuals in this database are clustered according to their genetically determined ancestry and not according to the country or continent where they reside. Approximately 45% of individuals are classified as exhibiting European ancestry, so gnomAD only partially reflects global genetic diversity.
Another possible limitation of this study was the method used to define the 25 most common pathogenic variants among HCU patients. We have considered studying all variants described in the CBS gene in the gnomAD population, but the filtering of truly pathogenic variants is rather poor and inaccurate, and we kept in mind that the low penetrance or expressivity of some genotypes, could lead to a falsely increased incidence result. So, an extensive review of the literature was performed to identify the most common pathogenic variants in HCU patients, but some relatively common variants may still have been missed. Each study presents its own methodology for the inclusion and diagnosis of patients. In many countries, indicated in white in Figure 1, there are no reported HCU patients. No alleles were found in the genomic databases for five of the twenty-five pathogenic variants analyzed (p.Leu101Pro, p.Cy-s165Tyr, p.Ala226Thr, p.Lys441Ter, and p.Lys523Serfs). Our study included three pathogenic variants that are known to be responsive to pyridoxine (p.Arg266Lys, p.Ala226Thr, and p.Ile278Thr), which accounted for an important percentage of the alleles present in this population (~20%). Pyridoxine-responsive patients are known to present with a milder clinical phenotype and presenting later in life or may even have no symptoms at all. | 2020-04-02T09:32:31.096Z | 2020-03-30T00:00:00.000 | {
"year": 2020,
"sha1": "0f17554b5773f0262c985dc5219c89bd92cb232d",
"oa_license": "CCBY",
"oa_url": "https://onlinelibrary.wiley.com/doi/pdfdirect/10.1002/mgg3.1214",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "6ec51bf5c5d12ec64c5ac7aab6e4dfce099d0053",
"s2fieldsofstudy": [
"Medicine"
],
"extfieldsofstudy": [
"Medicine"
]
} |
245421882 | pes2o/s2orc | v3-fos-license | Personal therapeutic approach in Gestalt therapists working with clients suffering from medically unexplained psychosomatic symptoms
Treatment specificity and adherence to treatment manuals represent essential components of the medical model in psychotherapy. The model assumes that psychotherapists who work with the same type of clients and who identify with the same theoretical approach work very similarly. This study illustrates the shortcomings of that assumption and explores how therapists’ individuality forms and shapes their unique approaches that resonate with their own personalities, inclinations, and worldviews. Semi-structured interviews with eight Gestalt therapists working with clients who experienced medically unexplained physical symptoms were analysed using the grounded theory method. Considerable differences were found among the therapists within four domains of the personal therapeutic approach, namely Case Conceptualization, Therapeutic Task, Therapist’s Position, and Alternative Strategy. However, regardless of the differences, all the therapists endeavoured, either implicitly or explicitly, to convey to the clients what they considered to be healthy functioning. There is considerable diversity in the way therapists work even when they subscribe to the same psychotherapeutic approach and work with the same type of clients. The exploration of psychotherapists’ usual strategies, as well as the alternative strategies they use when their usual strategies do not work, appears helpful for capturing their personal therapeutic approaches.
Introduction
Psychotherapists tend to develop, either deliberately or unintentionally, individualized approaches that resonate with their personality, inclinations, and worldviews (Fernández-Alvarez, Gómez, & García, 2015;. These personal approaches are created from a blend of one's theoretical orientation, natural skills and abilities, personal history and experiences, accumulated professional experience, and their own experience in therapy (Maruniaková, Řiháček, & Roubal, 2017;Maruniaková & Řiháček, 2018). Indeed, it has been shown that the proportion of variance explained by differences among therapists is much higher than the proportion explained by differences among treatments (Baldwin & Imel, 2013). Therapists' individual approaches seem inseparable from their personalities and cannot be fully standardized.
Such a point of view contrasts with the medical model in psychotherapy (Wampold & Imel, 2015). Assessing psychotherapists' adherence to a treatment manual became an integral part of randomized clinical trial methodology (Waltz, Addis, Koerner, & Jacobson, 1993).
Technique is regarded as the active ingredient of treatment and, therefore, a group of psychotherapists applying the same therapeutic approach are expected to think and behave in a very homogeneous manner throughout the therapeutic process. Another assumption embedded in the medical model is treatment specificity (Wampold & Imel, 2015). Various disorders are assumed to have different causes. Since treatments are expected to specifically address these causes, distinct techniques are prescribed for different disorders and conditions. This also naturally leads to the standardization of treatments and to a demand for psychotherapists to follow these specific guidelines.
In the present study, we challenge the above-described assumptions of the medical model using the concept of a personal therapeutic approach (PTA) Maruniaková, Řiháček, & Roubal, 2017;Maruniaková & Řiháček, 2018) as it naturally appears in daily psychotherapy practice. We start from the position of seeing psychotherapy as a primarily interpersonal process in which the technique cannot be separated from the relationship, and the therapist participates in the process with his or her whole personality (e.g., Orlinsky & Rønnestad, 2005). We are equally critical of the assumption that psychotherapists, if properly trained, are, in fact, interchangeable (Elliott, 1998;Orlinsky & Ronnestad, 2005).
We also question the concept of treatment specificity for different kinds of mental health problems. Categorical concepts of psychopathology are being revisited and adjusted in favour of more general psychopathology factors (Caspi et al., 2014;Hopwood et al., 2019) and the same therapeutic change mechanisms have been identified across various mental health issues (Wampold, 2015). From this point of view, we may expect that change mechanisms targeted by different therapists would be applicable across diagnostic categories, even if they work with a specific group of clients. Consequently, treatment strategies would differ across psychotherapists because they would be influenced by the therapists' personal preferences rather than the diagnostic category in question.
In the present study, we deliberately chose a setting where a rather uniform approach to treatment would be expected: the psychotherapists were all trained in the same psychotherapy approach (Gestalt therapy), and their clients all had the same kind of clinical symptomatology (medically unexplained somatic symptoms, MUPS). Under these conditions, we explored how therapists' individuality forms and shapes their own unique psychotherapy approaches.
The term MUPS is being used for a somatic symptom experience that lacks any medical explanation (Brown, 2007). These clients, after a series of unsuccessful medical consultations, are often referred to psychotherapy and are considered difficult to treat (Heijmans et al., 2011;Luca, 2011). This study is a part of a larger project that included the analysis of clinical strategies used in the treatment of MUPS by psychotherapists of various theoretical orien-tations. In a previous study, we analysed data from the whole sample (Čevelíček et al., 2019). For the purpose of this study, however, we selected only those for whom Gestalt therapy was their primary theoretical orientation. This approach was chosen simply because it was the most represented one in the sample.
Change mechanisms emphasized in humanistic and experiential psychotherapies for the treatment of MUPS include validating clients' experience of their somatic symptoms, increasing bodily and emotional awareness and regulation, helping clients understand how their symptoms are grounded in their relational and cognitive patterns, and nonjudgmentally accepting the presence of symptoms while focusing on client resources and abilities, rather than symptoms and deficits (Řiháček & Čevelíček, 2019). Several change mechanisms used in Gestalt psychotherapy have received empirical support when examined across different psychotherapeutic approaches. These mechanisms include increasing symptom acceptance, which represents the willingness to experience unwanted emotions, thoughts, and bodily sensations with the ability to act despite symptoms; development of coping strategies that can build on clients' strengths; instilling positive expectations from the treatment; helping clients engage in pleasurable activities and reducing their fear of symptoms; helping them sense more control over symptoms; and fostering working alliance (Pourová et al., 2020).
In the present study, while keeping both the self-identified psychotherapeutic approach and the target client population constant, we aimed to explore the inter-individual differences in therapists' PTA. Clients with MUPS are considered difficult to treat and we can therefore expect that a significant portion of cases may be viewed as unsuccessful by the psychotherapists (Heijmans et al., 2011). Therefore, to explore diverse strategies within psychotherapists' PTAs, we asked them about both successful and unsuccessful cases. This was motivated by the intention to explore the alternative strategies they used when they perceived their usual strategy as unsuccessful. Through a qualitative analysis of interviews with therapists, we sought to answer two complementary research questions. The first one was focused on the uniqueness of each therapist's personal approach: In what ways do Gestalt therapists differ from each other when working with MUPS clients? The second one, on the other hand, aimed to identify common patterns behind the heterogeneity: What are the common principles that, on a meta-level, govern the functioning of the individualized therapist's personal approach? With regard to their professional background and education, two were psychiatrists and six were psychologists. All the participants had finished graduate-level academic training in their respective professions and then finished a certified psychotherapy training with a minimum of 750 training hours. Most of them had completed additional shorter trainings, two were trained in body-psychotherapy, two in systemic and family therapy, one in hypnosis and CBT, one in relaxation and imagination techniques, one in EFT, and one in supervision. Gestalt therapy is one of the humanistic psychotherapies and is based on a phenomenological exploration of the processes of experiencing as they emerge in the therapeutic meeting (Francesetti, Alcaro, & Settanni, 2020). Gestalt therapy, as an existential and relational approach, focuses on what the client and the therapist experience and how they make meaning of the co-created therapeutic situation. Symptoms, including MUPS, are seen as products of the creative self and display human uniqueness (Perls, Hefferline, & Goodman, 1951). Psychopathology represents an originally unique creative adjustment in a difficult situation (Roubal, Francesetti & Gecele, 2017). However, once the situation is resolved, the adjustment no longer serves the client's needs and restricts their ability to have satisfying contact with their environment. In this sense, symptoms can be seen as a plea for a kind of contact that would render the symptoms themselves unnecessary (Sichera, 2001), which is explored in the hereand-now relationship with the therapist.
Participants
Gestalt therapy, in general, focuses on the mutual bodily processes of affective co-regulation between the client and the therapist (Francesetti, 2019;Jacobs & Hycner, 2009;Philippson, 2009), and also focuses its approach on psychosomatic symptoms (Iaculo, 1997;Kepner, 1993;Nemirinskiy, 2013). It does not deal with the human body per se but with contact in the body/environment field (Реrls, Hefferline, & Goodman, 1951). Psychosomatic symptoms can be perceived as a retroflected form of contact with the world (Nemirinskiy, 2013) which is actualized in the therapy situation, allowing for the exploration of the relational aspects of psychosomatic symptoms. However, Gestalt therapy as a humanistic approach does not primarily focus on the treatment of psychosomatic symptoms. Rather, it aims to support the human growth of the clients in the unique contexts of their lives. Despite the internationally widespread network of practitioners, rich theoretical developments, as well as a growing amount of research literature (Barber, 2006;Brownell, 2008Brownell, , 2016Strümpfel & Goldman, 2001;Roubal et al., 2016;Strümpfel, 2006), specific research on MUPS is vastly missing in the Gestalt therapeutic tradition, and Gestalt therapy relies on studies of other humanistic and experiential approaches.
Recruitment.
As an invitation to participate in the research, an email with a short summary of the research goals and the interview description was sent to 71 Czech psychotherapists working with clients suffering from MUPS. We used the following inclusion criteria: i) completed psychotherapy training; ii) at least five years of experience working with clients suffering from MUPS; and iii) willing to talk with researchers about one successful and one unsuccessful case of psychotherapy with clients suffering from MUPS. Thirty-one of them were willing to participate and met the inclusion criteria and eight self- Data collection. From February to May 2018, two research assistants conducted the interviews. The duration of the interviews ranged from 1.5 to 2 hours. We used a semi-structured interview schedule with each therapist to explore one successful and one unsuccessful psychotherapy case of a client with MUPS. We excluded cases with other serious mental health problems or severe addiction. While no specific time limit was used to exclude cases, participants were instructed to select recent cases so that they could recall the treatment process in detail. The interview schedule covered three themes: i) descriptive information about the client (somatic symptoms; other somatic and mental health problems; age; sex; education; occupation; family situation; and treatment motivation); ii) the client's understanding of their problems and the psychotherapist's conceptualization of the client's problems; and iii) the psychotherapy process in detail (see Appendix for the interview schedule). All interviews were audio-recorded and transcribed for analysis.
There were several reasons for this strategy. First, by focusing on specific cases, we aimed to eschew the psychotherapists' theorizing about the treatment of MUPS in general and explore their genuine experience. Second, the choice of one successful and one unsuccessful case allowed for the exploration of strategies the psychotherapists used in different scenarios. Third, from an idiographic perspective, this approach should help explore psychotherapists' styles when working with specific clients and allow for detailed descriptions of those styles.
Data analysis. The data was analysed by the first two authors using grounded theory methodology (Glaser & Strauss, 1967;Charmaz, 2006). The analysis proceeded in the following steps: i) Using open coding procedures, each interview was analysed by one of the first two authors as a separate case in order to investigate the personal approach of each therapist. The emerging concepts were then provisionally interconnected into a working theoretical model for each case separately. Each of the first two authors analysed a half of the interviews; ii) Each case-specific model was audited and further developed by the other of the first two authors (i.e., the one who did not analyze the particular case in Step 1). Then both authors discussed and consolidated each model via a teleconference. Afterwards, the case-specific models were validated by the last two authors; iii) Using the constant comparison method and theoretical coding procedures, the case-specific models were compared to each other, and, within this process, a paradigmatic model was developed to facilitate inter-case comparisons. The paradigmatic model consisted of four domains, namely Case conceptualization; Therapeutic task; Therapist's position; and Alternative strategy. Each case-specific model was then reformulated using this paradigmatic model to allow for more systematic comparisons among cases. The paradigmatic model was validated by the last two authors; iv) To capture the uniqueness of each case, we strived to develop a title for each that would express its essence in a more holistic manner; v) On a more abstract level, we searched for common patterns across the whole sample related to the functioning of the individualized therapist's personal approach; vi) Due to the space limitations of this article, we chose four cases that best represented the heterogeneity of the personal therapeutic approaches and present them in the Results section.
Trustworthiness of the results was reinforced by: i) Every case being analysed by two researchers, where one of them conducted the primary analysis and the other served as an auditor; and ii) Using the principle of consensuality (Hill, 2012) as a means of securing the validity of the analytic procedure -at each step of the analysis, findings were personally discussed until a consensus between all four authors was reached that the created concepts were grounded in the data, understandable, and relevant.
Results
As a means for the description of the participants' PTA, we developed four domains. These domains emerged from the data and included: i) Case conceptualization of how a therapist understands the client's difficulties; ii) Therapeutic task deduced from the conceptualization; iii) Position from which the therapist approaches their clients; and iv) Alternative strategy used when the main strategy does not work.
We use these domains to structure the vignettes of the therapists. Each vignette is first characterized by a title in the form of a statement paraphrasing the strategy typical for the particular psychotherapist. Due to space limitations, we selected four cases that best represent the heterogeneity of the sample. The therapists' names used in the Results section are fictional to preserve participant anonymity (see Table 1 for a more schematic depiction of the differences among the four therapists). Some contrasts that may not be apparent from the narrative vignettes are more readily visible in Table 1. For example, while Magda and Veronika are characterized by a similar approach to Conceptualization (MUPS are associated with unexpressed emotions), they differ considerably in the Task and Position domains: Veronika uses theoretical models to teach the client and offer coping strategies, whilst Magda relies on corrective relational experience provided by an accepting non-expert therapist.
Vincent: I lead the client to change her life
Vincent (Therapist 4) is a 53-year-old male psychiatrist with 19 years of psychotherapy practice.
Case conceptualization. According to Vincent, symptoms signalize that the client is not living a healthy lifestyle. MUPS are caused by the client being over- whelmed by the number of tasks imposed upon him, which is also related to his personal perfectionist nature and his inability to perceive the existential dimension of life. MUPS are caused specifically by existential anxiety, which the client is unaware of and which he manages in a dysfunctional way oriented only to 'material criteria': In this case, the most primitive issues mattered: to be as rich as possible so the family has a good life. Nothing was really important; the meaning of life or the universe were not of any interest to him. So maybe this is why he had pain in his back and belly (...) He experienced great stress and, in fact, existential anxiety which he was not aware of… (...) For such a person, financial safety is never enough, he repeatedly succumbs to anxiety, because the reason for it lies somewhere else.
Being goal-oriented cannot provide an answer to existential anxiety, the cause of MUPS. Therefore, it is important to understand the client's existential values and to direct him towards a lifestyle that allows for their fulfilment. Therapeutic task. The task of psychotherapy is to understand what change in lifestyle is being indicated by the somatic symptoms. Vincent offers a psychological perspective on MUPS and assesses whether the client can 'accept' this perspective. The client needs to 'get caught up' in therapy and 'start working on it himself'. He needs to 'be ready to change his life and himself'. Vincent aims to help the client 'broaden his awareness': 'Draw attention to what the patient experiences, such as somatic symptoms (...). Broadening of awareness, increasing self-support, increasing competencies in the sense that he himself can decide more about his life.' Therapist's position. Symptoms have specific causes that need to be addressed in psychotherapy. This causal way of thinking probably reflects Vincent's medical education and practice. There is a clear hierarchy in the psychotherapeutic relationship. Vincent takes responsibility for the client, especially at the beginning of therapy. From the position of a kind and respectful authority, Vincent legitimizes the client's troubling experiences: 'This was probably important for her. (...) That, unlike her, I did not see what was happening to her as something horrible. (...) She simply got scared. (...) She just needed some support and (...) she needed to hear that nothing truly horrible was happening, and that it would be all right again.' Alternative strategy. Depending on the nature of the achieved change, Vincent distinguishes between 'relief' and 'recovery': 'He left with at least some small relief from his symptoms, but this is not a successful therapy for me. Success would be if he would recognize how he is creating the symptoms himself and what they tell him, and if he started to change his life accordingly.' When it is not possible to achieve the desired change in therapy, Vincent respects this as the actual limit of the client: 'He could sense that it would lead to some changes which (...) he would not like.' In such cases, Vincent turns to the medical approach: he educates the client and offers him medications and herbs. He calms the client and relies on the 'magic of the physician's reassurance': 'She was afraid she would not be able to walk normally again. I told her that was nonsense and that she will be alright. So, I used my role of a physician.'
Magda: I cradle the client patiently
Magda (Therapist 7) is a 54-year-old female clinical psychologist with 28 years of psychotherapy practice.
Case conceptualization. Magda understands the symptoms as an expression of 'frozen emotional processes'' that the client is unaware of. They are connected to situations in which actual emotions could not be experienced or expressed: 'Either great sorrow and loss or often some anger or hurt or trauma that could not be expressed. (...) So, it stays with the person and expresses itself through the body.' Through focusing consciously on experiencing a somatic symptom in relation to a certain demanding situation, therapy 'brings to life what was numbed'. Magda recognizes a sequence of processes leading to MUPS: The client who experienced anger with her husband turned it against herself, which then led to the development of MUPS. In therapy, Magda works in the opposite direction, starting from the end: By deepening the awareness associated with MUPS, the client re-experienced aggression, accepted the reality of her relationship with her husband, re-experienced sorrow, and in a safe therapeutic environment, she allowed herself to express her emotions. The MUPS often remain, but the client 'is dealing with them in a much more realistic way' in the broader context of her life conditions. Therapeutic task. Magda focuses on creating a supportive therapeutic relationship. Experiencing the therapist's support leads to building the client's self-support, which, in turn, leads to change. The change comes with a modified approach to one's own body. Until now, the body was 'causing trouble'. Magda's task is to help the client find kindness in relation to her body, to experience it and understand it. The client 'would like her body to function, [she perceives her body] as an object, as a computer. (...) So, [she needs] to become kinder [to herself].' Magda challenges the client's somatic explanation of MUPS using an explorative and confrontative approach pointing to obvious contradictions between the client's explanation and observable behavior here and now in the therapy situation. Magda helps the client understand the dynamics of emotional processes related to MUPS in everyday situations.
Therapist's position. Magda takes the 'not-knowing approach', respecting that somatic symptoms are initially understandable neither for the client, nor for the therapist: Magda must not approach the client from an expert position. Such a non-expert approach, which differs from that experienced with medical doctors, provides the client with a corrective experience. Only then can the client be ready to accept the psychological explanation of MUPS. Magda patiently trusts the slow, self-healing process of therapy and respects 'the wisdom of resistances' and respects client's limited capacity, which stops the client from making excessively large changes: 'Whenever she would fully realize fully in a given moment that she would have to do something [very radical in her relationship with her husband], then a part of her inner dialogue would stop her [as if she would say to herself:] wait, maybe it [MUPS] is just a somatic issue anyway.' Alternative strategy. When the change cannot be achieved through her basic strategy, Magda does not change her approach, she just intensifies her original strategy: 'it is simply necessary to be even more patient'. Magda needs to conceal her frustration from the client, especially at the beginning of the therapy. The therapist's task is to accept her own helplessness regarding MUPS and wait with respect and curiosity. Supporting the creation of a safe and trusting therapeutic relationship is crucial because even if the client is unable to make a visible change, the therapy works on the level of the 'emotional field', the shared here-and-now. In this implicit way, Magda can support emotional change in the client through a corrective relational experience.
Irena: I let the client learn from a strong woman
Irena (Therapist 6) is a 42-year-old female psychologist with 18 years of psychotherapy practice.
Case conceptualization. According to Irena, the client needs to learn to physically sense her body. Then she can more sensitively monitor her bodily experiences in difficult life situations and consider her physical symptoms in a broader relational context, 'so that she becomes aware of the situation before the physical pain comes.' Irena attributes a symbolic meaning to MUPS: 'My first hypothesis certainly was that when the patient brought up pain in the pelvis (...), then it was some sexual or female issue.' Irena tries to convey this understanding to the client: 'Her skin started to hurt so that she would be protected against intimacy. (...) I told her that.' Therapeutic task. Irena's task is to redirect the client back to her feelings, to 'go deep' to 'get to inner values'. The basic method is to build sensitivity to feelings that are experienced bodily: 'I repeatedly turned her back to her feelings, to experiencing her body, simply to monitor what was changing.' In addition, Irena believes that clients needlessly limit their understanding of their situation by focusing solely on their actual problems. Instead, Irena strives to 'broaden the context for understanding' and actively directs the client's attention to resources that are available to them in demanding situations: 'I constantly try to somehow remind her of her resources, inner or outer ones, which are at her disposal. So, I repeat this to her again and again.' Therapist's position. Irena takes the position of an expert, who -from a professional distance -assesses both the difficulties and the resources of the client: 'With some clients (...), it is clear to me from the beginning who has the resources and who does not. Then I set my therapeutic goals accordingly.' Irena 'leads' the client to become aware of her emotions and demands that the client 'work hard between the sessions'. She advises the client in a kind way, from her position as a strong, yet authentic female authority: 'Sometimes I say: '(...) you need to exaggerate [emotional expression] a few times to learn how to deal with it.' At the same time, I say: 'your husband will definitely survive it.'' Irena leads the client from a power position: 'When it is needed, I pull her. I do not motivate her; I simply say directly what needs to be done'. However, her authority is a very humane one, she uses herself as a model for explaining how MUPS work: '[I told her:] 'Where I would scream, you experience it bodily'… And we made fun of that (...) [because] laughing strengthens the therapeutic alliance. (...) I sit there and laugh and say: 'I am so proud of you!'' Irena is clearly personally involved: 'I invested quite some effort in her and I still do. I love her, I care for her.' Alternative strategy. When the above-mentioned strategy does not work, Irena resigns from pursuing change by searching for insights. She stops trying to produce changes directly and resorts to using techniques and consultancy: 'Mostly relaxation and imagination, these are a kind of rescue'. She directs the client: 'I do not motivate her, I rather tell her directly that she needs to keep visiting the rehabilitation practitioner regularly.' She also offers specific ideas for solving the client's problems, 'because she lacked the potential for discovering it herself.' Irena offers simple human support and an opportunity to share: 'The client does not have the resources for change, so I am a surrogate friend to her […] so that she is not totally alone in it.' Irena lets herself 'be used (...) as a kind of anxiolytic'; she even perceives her position as 'a statefunded emotional prostitute'.
Veronika: I teach the client to use emotions properly
Veronika (Therapist 8) is a 41-year-old female psychologist with 17 years of psychotherapy practice.
Case conceptualization. Veronika understands symptoms as something caused by restraining emotions, especially aggression: 'The whole process of her symptoms was based on her holding back her aggression.' Veronika uses theoretical models to explain to the client how her symptoms are associated with emotions: '[I explained to her] how it is and why it is happening this way. (...): 'This is what is written [as a theory] and it actually seems to me that this principle also applies to your case.'' In psychotherapy, change occurs when the client learns 'to experience emotions and to be kind towards herself'. Therapeutic task. Veronika explains the psychological context and offers a coping strategy. 'I analysed the situation [with her husband] for her, I was helping her to see his motivations.' She relies on the theoretical frame that she learned in her psychotherapy training, and she tries to share it with the client: 'I helped her to understand how her disorder works in relation to her personality from a psychological standpoint. (...) And I offered her a strategy for how to get out of it.' Veronika demands personal involvement from the client. If the client does not accept the concept suggested to her, the therapist understands it as resistance and tries to eliminate it: 'I worked with the resistance for a long time, I tried to bring it into the therapy (...), but she always somehow went around it. So I ultimately ended [the therapy] and referred her to a colleague, who she refused, however.' Therapist's position. Veronika perceives herself as the initiator of the change. She requires the client to actively take over the responsibility for change: 'I will not try to persuade someone who is objectively not willing, not trying, and not aiming anywhere.' Veronika uses her knowledge and skills to teach the client. At the same time, she offers kind patience: 'I trusted in my knowledge, theoretical background (…). And also in the patience and kindness I felt towards that woman.' Alternative strategy. Veronika is very clear about her basic strategy, and when it does not work, she feels helpless. In such cases, she at least offers relaxation training. However, she continues to require the client's active personal involvement in what the therapist is offering to them. If this does not work, Veronika ends the therapy. 'We really worked on her resistance. (...) [But] she was not even willing to accept the fact that she does not want [to collaborate]. (...) So, as a logical consequence, we ended the therapy.' At the same time, however, Veronika is respectful to the client as a person, she acknowledges her limitations and does not accuse her of being unable to make a change. She also admits limitations on the side of the therapist and the therapy setting.
General patterns
In the final phase of our analysis, we adopted a more general perspective from which a crucial common pattern could be recognized: All the therapists were trying to convey their fundamental beliefs to their clients. Although the individual beliefs might have differed, we could see how the therapists were trying to pass on to their clients what they considered as healthy functioning. In the successful cases reported by the therapists, the clients seemed to have adopted these fundamental beliefs of their therapists and started to make changes in their lives accordingly.
At the same time, we could see that each of the therapists used their own unique approach to convey their beliefs. The difference was in the relational setting with the client, which varied on the scale between two extreme poles: EXPERT LEADERSHIP ←→ JOINT EXPLORATION On one side, there is an explicit position of a directive teacher who knows what is healthy and offers expert leadership: I want to teach the client what I have learned to be useful. On the other hand, there is the position of an exploring companion: Together, we search for a meaningful way of understanding, which will provide the client with guidelines. From this position, the teaching was an implicit dialogical process in which the therapists were 'seducing' their clients into adopt the therapists' beliefs about healthy functioning.
However, the therapists also reflected on unsuccessful cases in which the clients did not accept the therapists' beliefs on healthy functioning. This point is especially important regarding clients with MUPS who often refuse to accept the psychological aspects of their somatic problems. They also tend to come to psychotherapy with low, or sometimes even negative, expectations after a long series of unsuccessful treatments with different specialists on somatic disorders. For both of these reasons, psychotherapists often reach the limits of their approach when working with MUPS clients.
In such cases, the therapists were pushed to modify the relational setting with the client.
We found that the therapists' flexibility was very limited in this regard. While on the level of specific therapeutic interventions, the therapists seemed to show high flexibility, on a more general level, they adhered to their favourite collaboration setting. In several cases, they had one more alternative to which they could switch, but no more than that. Accordingly, we distinguished four types of collaboration setting modifications the therapists used in unsuccessful cases: 1. Expert leadership was modified to Joint exploration.
For example, Irena (I let the client learn from a strong woman) changed the Therapist's position from that of expert assessment and advice to offering simple human support and an opportunity to share. The Case conceptualization changed from 'The client needs to understand her physical symptoms in a broader relational context', to 'It is important that she is not totally alone in it.' The Tasks changed accordingly from 'Turning the client toward her bodily experienced feelings' to 'Becoming a substitutional friend for the client'. 2. Joint exploration was modified to Expert leadership.
For example, Vincent (I lead the client to change her life) changed the Therapist's position from 'broadening awareness' in a safe and trusting therapeutic relationship to educating the client and offering medicines and herbs. The Case conceptualization changed from 'Symptoms signalize that the client is not living healthily, he is omitting the existential dimension of life', to 'Enabling the client to experience at least some small relief from his symptoms'. The Tasks changed accordingly from 'Raising competencies to decide more about one's own life by broadening awareness' to 'Calming the client down by using the magic of the doctor's reassurance'. 3. Joint exploration was modified to even more intense Joint exploration. For example, Magda (I cradle the client patiently), who patiently trusts the slow selfhealing process in the supportive therapeutic relationship, intensifies this approach by being even more patient, forcing the therapist to process their own frustration. The Case conceptualization ('Symptoms are an expression of frozen emotional processes') and the Therapist's position (Providing the client with a corrective relational experience) remained the same, but the Tasks became less ambitious and demanding (Being even more patient and concealing frustration from the client). 4. Expert leadership was modified to even more intense Expert leadership. For example, Veronika (I teach the client to use emotions properly) who, as the initiator of the change, explained the psychological context and offered a coping strategy, stressed this expert Therapist's position even more and became a relaxation trainer for the client. Case conceptualization (Symptoms as being caused by restraining emotions) remained the same, but the Tasks (Motivating the client to learn a relaxation technique) became less ambitious and more specific. The way the therapists structure the relational setting with the client, and how they change it in unsuccessful cases, illustrated the personal therapeutic approach not only on the level of working strategies, but also from a more complex point of view. It enabled us to summarize each therapist's working strategies into one coherent picture.
Discussion
The present study is the first to explore PTA in Gestalt psychotherapists who reflected on their clinical work with clients suffering from MUPS. The first question of this study was 'In what ways do Gestalt therapists differ from each other when working with MUPS clients?' The question assumed that Gestalt psychotherapists would differ substantially in their work, as their PTA would overshadow the fact that they all used the same theoretical model and worked with the same group of clients (Baldwin & Imel, 2013;. The results of our study show this assumption plausible, as shown by the differences identified among psychotherapists. Our results complement the already existing models of therapist´s individualized approaches (Fernández-Alvarez, Gómez, & García, 2015; with a new way of capturing the individual difference between therapists by focusing on therapist´s understanding of the client's difficulties; on a therapist's task deduced from the conceptualization; on a relational position from which the therapist approaches their clients; and on the alternative strategy used when the treatment did not progress in the desired direction. The psychotherapists' clinical work represents creative adaptations of the therapeutic style with a specific group of clients, showing an individualized approach that is complementary to the usage of predefined treatment manuals (Wampold & Imel, 2015). The need for a flexible use of techniques that aims to find a helpful focus of treatment for clients has been mentioned in other recent studies of professionals who focus on the treatment of people with MUPS (Balabanovic & Hayton, 2019;Cevelicek et al., 2020). These clients are considered difficult to treat (Heijmans et al., 2011;Luca, 2012), a position supported by observably mediocre therapeutic results (Kleinstäuber et al., 2011;van Dessel et al., 2014), leaving a high desire to explore different approaches to engage them in treatment. A flexible usage of personal therapeutic styles, similar to the one demonstrated in the present study, might improve psychotherapy outcomes and client engagement, specifically when client characteristics such as psychotherapy preferences, stage of change, reactance level, coping style, and attachment style are accounted for (Norcross & Wampold, 2018). The overview of different personal therapeutic styles that the therapists in our study used in the treatment of clients with MUPS challenges the assumptions of treatment manuals adherence and of treatment specificity (Wampold & Imel, 2015), and may serve as an inspiration for other practitioners. However, it is important to note that not just any personal therapeutic style can be presented as effective. A personal style must make use of effective mechanisms of change grounded in a clearly defined theoretical model to be effective.
Since psychotherapists in the studied sample had considerable experience in the treatment of people with MUPS, multiple mechanisms of change were present in their work. For instance, even though the psychotherapists used different positions from which they approached clients in the relational field that unfolded during sessions (Francesetti, 2019), both the psychotherapist who used the position of a health-care authority and the therapist who assumed a not-knowing and non-hierarchical position used their positions to legitimize clients' symptoms and enhance their self-compassion. Similarly, the effective change mechanism of increasing clients' emotional awareness and regulation (Pourová et al., 2020) was invoked by both the psychotherapist teaching clients emotional management and redirecting clients back to their feelings.
The second question of the present study was 'What are the common principles that, on a meta-level, govern the functioning of the therapist's personal approach?' PTA represents a system with its own inner logic of organization and principles of development according to the core aspects of PTA : It is selective in an idiosyncratic way; created from metabolized theories and techniques; responsive to the context of a thera- peutic situation and evolves throughout a therapist's career. In our study, the role of the underlying beliefs was found crucial for each therapist´s PTA. All the psychotherapists applied their views of healthy functioning and the changes they perceived as beneficial across clients and situations rather uniformly. Similarly, when eminent psychotherapists with different theoretical orientations were asked about their usage of beliefs in psychotherapy (Williams & Levitt, 2007), they expressed that beliefs governed their emotional reactions to clients and, in turn, the psychotherapists used their emotional reactions to monitor client progress. However, the present study is the first to our knowledge that illustrates how therapists use their beliefs to determine the desired direction of change in clients with MUPS. Since therapists' beliefs are intertwined with their perception of a desirable treatment outcome, it is not surprising that they associate clients' adoption of their beliefs with outcomes (Jadaszewski, 2017;Kelly, 1990). This is relevant in clients with MUPS who are perceived by professionals as often having views and characteristics that thwart what therapists consider to be healthy and useful in psychotherapy (Balabanovic & Hayton, 2019;Heijmans et al., 2011;Luca, 2011). The effort to 'convert' clients' beliefs was even more apparent in this study because the psychotherapists perceived clients with MUPS as less open to change. Because suggestion is an effective factor in psychotherapy, 'converting' clients' beliefs about healthy functioning is considered a useful therapeutic strategy (Cuijpers, Reijnders, & Huibers, 2019). Indeed, some psychotherapy approaches intentionally aim to change specific beliefs held by clients because they hypothesize dysfunctional beliefs to be the causal or maintaining factors in people suffering from MUPS (e.g., Salkovskis et al., 2016). The theoretical and empirical base of therapist's beliefs about healthy functioning supports their orientation in a complex situation. That said, the need for some flexibility in beliefs about healthy functioning might be an underrepresented theme in treatment and training.
Study limitations
Asking therapists about two of their recent psychotherapies with clients suffering from MUPS, one successful and one unsuccessful, could have influenced the variety of strategies we were able to capture. The therapists might use different positions, tasks, and conceptualizations with different clients. However, the aim of the present study was to capture the therapists' unique styles from an idiographic perspective, which was still possible within the chosen design. Although the chosen interview process could limit the range of captured strategies, it was successful in capturing the psychotherapists' beliefs connected to therapeutic success and healthy functioning in clients with MUPS.
We relied on the therapists' own descriptions of and reflections on their therapeutic work. These may not fully correspond to the therapists' actual behavior that would be reported by an external observer. While there is research showing that practitioners can accurately recall types of techniques they used (Castonguay et al., 2017), other studies suggest that the reliability of self-report is limited (Santa Ana et al., 2008). The present study relied on a high degree of analytic inference from the researchers, i.e. it addressed themes that the respondents did not talk about explicitly. For instance, most of them did not explicitly formulate the positions they used in contact with clients. When a high degree of inference is used, data analysis may more often lead to alternative interpretations by different researchers. We attempted to reduce the risk of bias by including multiple researchers in the analysis process and by using the principles of consensuality (Hill, 2012). Admittedly, the results of the study are influenced by our focus on the uniqueness of each therapist's PTA. Researchers who would endeavour to find commonalities among the therapists in the first place would probably come to different conclusions.
The focus on Gestalt therapists might be perceived as reducing the generalizability of findings for practitioners of other therapeutic orientations. However, we believe that our results point to common factors (Pourová et al., 2020; rather than to the specifics of the Gestalt therapeutic approach in clinical practice and work with psychosomatic clients specifically (Nemirinskiy, 2013). We hypothesize that similar variability could be found in other therapeutic orientations. However, this hypothesis must be tested in future studies.
Conclusions and implication for training and practice
The study revealed considerable diversity in the way psychotherapists work, even when they share the same approach and the same type of clients. The specific features of each personal therapeutic approach can be characterized by the particular therapist's Case conceptualization, Therapeutic task, and the Therapist's position. However, to understand a therapist's flexibility, this framework needs to be supplemented with explorations of the alternative strategies a therapist uses when working with an unsuccessful case.
Our finding that the therapists' unique style of working with clients suffering from MUPS resulted from their beliefs about general healthy functioning has implications for psychotherapy training. Psychotherapy trainees are usually taught specific theories of change that they should apply to clients, and they also tend to adopt their trainers' beliefs about healthy functioning. In the initial stage of development, trainees tend to apply these theories to clients (Maruniaková, Řiháček, & Roubal, 2017; Maruniaková & Řiháček, 2018), later discovering that many clients do not change as expected. Exploring the perspective of implicit theories of change, beliefs, and values, as they translate to interventions, therapeutic relationships, and psychotherapists' view of treatment success, could help broaden trainees' concepts of healthy functioning. The flexibility they thus develop might reduce their initial and unnecessary disappointments and protect clients from being pushed into 'boxes' into which they cannot fit. | 2021-12-23T16:13:32.979Z | 2021-12-20T00:00:00.000 | {
"year": 2021,
"sha1": "c2cd3300145df97f6a4cd669cfa6ec7a0211502f",
"oa_license": "CCBYNC",
"oa_url": "https://www.researchinpsychotherapy.org/index.php/rpsy/article/download/535/517",
"oa_status": "GOLD",
"pdf_src": "PubMedCentral",
"pdf_hash": "fdfeafe89d5f77b95f782c81614725da08f2fd72",
"s2fieldsofstudy": [
"Psychology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
235466763 | pes2o/s2orc | v3-fos-license | Evaluation of a new body-focused group therapy versus a guided self-help group program for adults with psychogenic non-epileptic seizures (PNES): a pilot randomized controlled feasibility study
Objective Psychogenic non-epileptic seizures (PNES), a common phenomenon in neurological settings, are regarded as a paroxysmal type of functional neurological disorder (FND). In a substantial proportion, PNES are disabling with poor long-term outcomes and high economic costs. Despite the clinical and financial consequences of PNES, there is still a lack of controlled clinical trials on the treatment of this challenging disorder. The study aims to evaluate the feasibility and collect first evidence of the efficacy of a group based-intervention in PNES-patients. Methods A pilot randomized controlled feasibility study with a parallel-group design was performed in adult outpatients with PNES to evaluate a new body-focused group therapy (CORDIS) versus guided self-help groups. Self-assessment of dissociation (Dissociation Experience Scale—DES-20) and seizure severity (Liverpool Seizure Severity Scale—LSSS) were assessed two weeks before and two weeks after the treatment intervention and also six months after treatment as primary outcome parameters. Results A total of 53 patients were recruited from a specialized outpatient clinic, and out of those, 29 patients completed either the body-focused group therapy program (n = 15) or a guided self-help group (SHG) therapy (n = 14). When analyzing the ITT sample (n = 22 CORDIS group, n = 20 SHG), both groups showed an effect on seizure severity and level of dissociation. In the per protocol sample (n = 13 CORDIS group, n = 12 SHG), CORDIS was superior to the self-help group for reducing seizure severity 6 months after the treatment. Significance CORDIS is a newly developed body-focused group therapy program for adults with PNES. Further studies should include a multicentric design with a higher number of participants.
Introduction
Psychogenic non-epileptic seizures (PNES) are paroxysmal episodes characterized by the loss of voluntary control over body functions that are usually intentional. Non-epileptic seizures resemble epileptic seizures but are not related to abnormal electrical activity in the brain [1]. The prevalence of PNES in the general population has been estimated at 2-50/100.000 [2]. In general neurology outpatient clinics, PNES account for 2% of new referrals [2]. In tertiary Epilepsy units, the proportion of PNES among all patients is around 30% [3]. The aetiology of PNES is still subject to research. Theoretical psychological models assume that PNES occur in response to distressing stimuli when 1 3 alternative coping mechanisms are inadequate or have been overwhelmed [1]. The occurrence of PNES is often associated with psychological trauma and posttraumatic stress disorder [4]. Current treatment recommendations rely on psychological interventions after a thorough workup. Antiepileptic drugs are neither indicated nor helpful in PNES [5]. Although caring for PNES patients thus usually demands an interdisciplinary approach, they are often solely treated either in neurological or in psychiatric/psychosomatic settings, leading to disadvantages in treatment quality.
Despite the high frequency of PNES in clinical settings, there are surprisingly few studies of high methodological quality focusing on guideline-oriented treatment options for PNES. Three previous pilot treatment trials have focused on cognitive behavioural therapy (CBT) programs for PNES patients [7][8][9]. The first two monocentric studies showed that a structured CBT treatment may reduce the seizure frequency in PNES [7,8]. There is one multicentric study from 2020 involving 368 patients [9]. This study revealed that a structured CBT treatment plus standard medical care was superior to only standard medical care regarding the outcome of the burden of somatic symptoms, bothersomeness of seizures, the general quality of life, work and social life functioning, and overall psychological distress [9]. Based on those pilot trials in the English-speaking area, we developed a body-focused group therapy program in German. Its content is presented in detail in a previous publication [10], including cognitive-behavioural aspects that are already established in the existing English treatment programs [7][8][9]. A new aspect is that it also involves a body-focused psychotherapy approach with symptom-specific exercises. This body-focused approach refers to the German guidelines for the treatment of functional body symptoms. In those guidelines, the involvement of body-focused psychotherapy is recommended due to the body-focused symptomatology of the disease [6].
In this study, we performed a randomized controlled feasibility trial to evaluate the effects on seizure severity and dissociation level of our treatment program CORDIS for PNES patients. We compared these effects with those of a guided self-help group (SHG), an effective therapy option for psychological diseases with expected nonspecific treatment effects [10]. We hypothesized that our program is superior to an SHG in decreasing seizure severity and level of dissociation in patients with PNES.
Trial design
The primary objective was to estimate the effect of the manualized group treatment program 'CORDIS' on seizure frequency compared to that of SHG in a parallelgroup study design. After the inclusion of the first three patients, we changed our primary outcome parameter from "seizure frequency" to "seizure severity" and the "level of dissociation", based on the results of psychometric assessment tests. The reason for this change of study design were difficulties with assessing seizure frequencies include high variability (multiple seizures per day versus one seizure per month), lack of valid scales and instruments, and different definitions of when to count symptoms like seizures. This resulted in high discrepancies between the seizure documentation by patients and the actual interpretation of the seizure situation from the evaluating physician.
Further, we established an active control-group design based on the decision framework for randomized trials of behavioural interventions in psychiatry [11]. All participants provided informed consent for research participation. All data were collected and interpreted at Charité University Medicine Berlin. The study was approved by the Ethics Committee of Charité University Medicine Berlin (EA1/185/16). Registration as a randomized clinical trial was prospectively initialized in the German registry for clinical studies on 06/04/2018 (DRKS, study number DRKS00014251).
Power calculation
For calculating the case number, we referred to the monocentric study by Goldstein et al. in which a comparable study design was used for an individual intervention [8]. We calculated Cohen's F [25], which resulted in an effect size of 0.36. On the basis of this effect size, a total sample number of N = 18 patients was calculated to identify a treatment difference between the two treatment arms with a probability of 80% and a two-sided significance level of 0.05. It has to be noted that the highly significant difference between the study groups in the 2010 study is likely to be partially due to a bias in time and attention since the intervention group with 10 therapy sessions received significantly more attention than the control group (treatment as usual). In our study plans, we attempted to avoid this bias by providing an SHG within the same time frame. Accordingly, we expected a lower effect than in the 2010 study and corrected the improvement in the seizure frequency in the control group from 1.25 to 7 per month. With these corrected raw data, a Cohen's F of 0.2 was calculated. On the basis of this assumed effect size, a total sample number of N = 52 patients was required to detect a mean difference between the two treatment arms with a probability of 80% and a two-sided significance level of 0.05. With an expected dropout rate of 15%, 60 patients had to be recruited (30 patients per group).
Participants
We recruited patients from a specialized outpatient clinic for PNES patients at Charité University Berlin, Department of Psychosomatic Medicine. Adults (> 18 years) that met the following inclusion criteria: PNES diagnosed by an experienced epileptologist via 24-h continous videoelectroencephalography, seizure videos and/or description of the seizures from relatives as well as ongoing seizures for more than six months with at least one seizure every 2 months and at least one seizure 4 weeks prior to the start of the intervention. Exclusion criteria were comorbidity of epilepsy, current psychotherapeutic treatment, psychosis, substance abuse, acute suicidality, insufficient language skills, and inability to complete the questionnaires (e.g. blindness).
Randomization and dropout
We performed computer-generated block randomization using the program Randomizer (www. rando mizer. org). Randomization and enrollment of patients were performed by different members of our study group. All participants received detailed information about the study and written consent was obtained.
Demographic and clinical characteristics
All participants completed questionnaires to capture demographic data, psychiatric and somatic pre-existing conditions, age, gender and level of education. In addition, the Childhood Trauma Questionnaire (CTQ) was used to determine early-life traumatization. The CTQ is one of the most commonly-used and well-validated measures for early-life traumatic events [22].
Outcome parameters
We measured potential treatment effects at baseline two weeks before the treatment program started (pre-treatment), 14 days after treatment (post-treatment) and six months and two weeks after treatment (follow-up). We defined the level of dissociation and seizure severity as primary outcome parameters. Secondary outcome parameters were depressive symptoms and somatoform symptoms [17].
Primary outcome parameters
Level of dissociation We measured the level of dissociation with the Dissociation Experience Scale-20 (DES-20). The DES-20 (German version: FDS-20) is a 20-item scale rated in percentage from 0 to 100, which measures the frequency of dissociative experiences [13]. The total DES-20 score is the mean of all item responses. The scale considers a wide range of dissociative symptoms but focuses on those regarding awareness and consciousness. The German version showed good internal consistency [14].
Seizure severity
We measured seizure severity with the Liverpool Seizure Severity Scale (LSSS). The LSSS is a 20-item self-report questionnaire to assess the severity of seizures within the past four weeks. As a supplement question to the LSSS, we added a question about the seizure frequency in the past year. The LSSS is scored from 0 to 100, with higher scores reflecting greater seizure severity. The LSSS showed good internal consistency (α = 0.72-0.96) in patients with epilepsy [15] and has been used in patients with PNES [16].
Secondary outcome parameters
Depressive symptoms We measured depressive symptoms with the the Patient Health Questionnaire (PHQ-9). The PHQ-9 is a self-report instrument and screens for depressive symptoms. It consists of 9 items with a total summation score of 27. Spitzer et al. published the reliability and validity studies [18].
Somatoform symptoms Somatoforms symptoms were measured with the Patient Health Questionnaire for somatoform symptoms (PHQ-15). It consists of 15 items, with a total summation score of 30. The PHQ-15 has been shown to have good internal consistency (α = 0.8) [19].
Body-focused group therapy program CORDIS
The manualized group treatment program CORDIS was developed to fill the gap of specific treatment options for PNES patients in Germany. The name "CORDIS" stands for CORpus (referring to bodyfocused therapy) and DISsociation. It was developed by three authors with expertise in epileptology, neuropsychosomatics and psychotherapy and is partly based on existing programs for English natives [7,8]. It consists of a combination of measures for psychoeducation, emotion-regulation and body perception. CORDIS consists of 10 weekly sessions of 90 min each. Each session is structured with repetitive elements such as greeting rituals, discussion of the symptom diaries and guided performance of the group exercises. Each session contains a detailed plan for every therapy session, see Table 1. In our study, the treatment was performed by two experienced group therapists who received training in how to use the CORDIS manual. Adherence to the manualized intervention strategy was supported through regular supervision provided by the authors of the manual.
Self-help group (control intervention)
The guided SHG consisted of 10 weekly sessions of 90 min each. Each session was introduced by a trained student. In the first group session, the student led a round of introductions. Then, the student introduced a general topic with advice on how to discuss the issue (e.g. mind mapping). The topics of discussion provided by the students were as follows: anxiety of seizures, feelings of embarrassment regarding seizures, loss of control, taking control over seizures, helpful proxies, seizures and the family, triggers for seizures, pessimism or optimism and finding helpful resources to deal with seizures. The last session consisted of a summary from the previous sessions and a farewell round without any specific topic.
Publications
We reported interim results of our study only from the intervention group in an earlier publication [10]. Based on the patient collective of our study some other studies with a cross-sectional design were performed and published [28][29][30].
Statistical analysis
We analyzed all data using SPSS, Version 24.0 (IBM Corp. Armonk, NY). Normality in distribution and potential outliers of all outcome scores were assessed with histograms. Demographic characteristics between intervention groups were compared using univariate analysis of variance (ANOVA) for continuous variables and χ 2 tests for dichotomous variables. Missing data in our Intention-to-treat (ITT) sample were imputed with the last observation carried forward (LOCF) approach. Due to high dropout rate in the ITT sample we also conducted all analyses in the per protocol sample (patients completing pre and post-intervention assessments, n = 25). Follow-up data was calculated based on the per protocol sample due to extremely high dropout numbers. For main analyses, we calculated change scores of pre-treatment versus post-treatment (two weeks after completion of the treatment intervention). For these we also calculated Bias Corrected accelerated 95% Confidence Intervals (BCa 95% CI) based on 1000 bootstrap samples. We conducted separate ANCOVAS with change scores as outcome variables and intervention (Cordis vs SHG) as group variable, controlling for pre-treatment scores. We also calculated the number of responders (25% reduction of outcome score) in both intervention groups. Follow-up assessment scores were compared with a repeated measures ANOVA with three time points (pretreatment, post-treatment, follow-up) as within subject factor and intervention group as between-subject factor.
Participant flow
For details of recruitment and dropout see also Fig. 1 (CON-SORT flow diagram). Sixty-seven patients were screened for eligibility between May 2018 and December 2018. Five patients declined to participate for unknown reasons and nine patients did not meet the inclusion criteria (four patients started psychotherapy, the diagnosis of two patients revealed unclear seizure etiology, and three patients did not have any more seizures for more than three months). The remaining 53 patients were recruited and randomized 1:1 to one of the two treatment arms (n = 27, group treatment program (CORDIS); n = 26, SHG). After randomization, 11 patients withdrew their informed consent and refused to participate in further visits or follow-ups, thus pre-treatment assessments could not be obtained. Reasons for withdrawal were spontaneous symptom reduction (CORDIS group, n = 2), discontent with the treatment plan (long waiting time) (CORDIS group, n = 3) and disappointment of being put in the 'wrong' group (SHG, n = 6). Our ITT sample, therefore, consisted of all patients, that were randomized and provided pre-treatment data (n = 42) with 22 patients in the CORDIS group and 20 patients in the control group (SHG).
Thirteen patients interrupted the study; seven patients from the treatment group (three patients needed hospital treatment and the remaining four left for unknown reasons) Fig. 1 Consort flow diagram of the RCT study (LOCF last information carried forward, GTP group treatment program, SHG self-help group, n number) and six patients from SHG (three patients started individual psychotherapy during the study and the remaining three left for unknown reasons). Twenty-nine participants completed either the treatment program (n = 15) or SHG (n = 14), but only 25 patients completed the questionnaires and could therefore be included in the per-protocol analysis (n = 13 CORDIS group, n = 12 SHG).
Demographic and clinical characteristics
The CORDIS and control group did not differ significantly in terms of demographic and clinical characteristics (see Table 2). Regarding seizure frequency 14 days prior to the beginning of the intervention from all 42 patients, n = 24 patients stated to have seizures every day, n = 15 patients had 3-5 seizures per week, two patients had one seizure per week and one patient had only one seizure in the last two weeks before assessment.
Primary outcome parameters
All primary and secondary outcome parameters were calculated based on the ITT sample (n = 22 CORDIS, n = 20 SHG) in consent to the "last information carried forward"method. All outcome variables (means, SD) for all time points are shown in Table 3.
Secondary outcomes
Comparing pre and post-intervention scores with respect to depressive symptoms, PHQ-9 yielded no significant results (change scores PHQ-9 were 0.52 (SD 2.86) for CORDIS group vs. No harms or unintended effects were observed, neither in the intervention nor in the control group.
Discussion
The present study is the first to evaluate a symptom-specific treatment program for German patients with PNES in a randomized clinical trial design. It features a new therapeutic aspect by including body-focused therapy exercises next to well-established cognitive-behavioural treatment methods. Furthermore, it is the first study that compares the effects of psychotherapeutic treatment for PNES with an active control treatment (guided SHG).
Patients who participated in the CORDIS treatment program and SHG showed improved dissociation levels and reduced seizure severity after the respective intervention (ITT sample analysis). However, there was no significant difference in the efficacy between either of the interventions, but the effect size showed a small to medium effect. After the six-month follow-up testing, the CORDIS treatment was superior to the active control intervention regarding the outcome parameter 'seizure severity' (per protocol analysis). Our study adds evidence to the currently small spectrum of treatment options for PNES patients and points toward the notion that body focussed interventions may be effective in PNES. According to Röhricht et al., body psychotherapy is characterized by the central guiding principle that the body remains the focus of the therapeutic work. This counts especially for diseases with body-focused symptomatology, as given in PNES [12]. PNES patients might get irritated if therapists confront them directly with explanations concerning the psychological genesis of their disease. It can be confusing that the seizures in PNES are caused by psychological reasons because patients tend to experience them clearly as a body dysfunction. Therefore, CORDIS tries not to directly address psychological theoretical explanation models unless the patient brings them up [12]. This theoretical approach in combination with repeated exercises helping to improve body perception might play a role in the positive long-term effects of our treatment program, as shown in other studies focusing on body psychotherapy for medically unexplained symptoms [27].
In contrast to earlier studies we used seizure severity and not seizure frequency as outcome parameter. This may reduce comparability, however, there is growing evidence suggesting that seizure frequency alone may not be the most important determinant of quality of life in patients with PNES [9,16].
Interestingly, the only available multicentric study on the efficacy of a treatment program for PNES patients showed that seizure frequency in PNES patients did not improve significantly in a CBT-treated group, compared with that in the standard medical care group, however somatic symptom burden in general did [9].
In a study from 2005, the usefulness of seizure frequency as an outcome parameter had been already questioned [24]. Difficulties with assessing seizure frequencies include high variability, lack of valid scales and instruments and different definitions of when to count symptoms as seizures resulting in discrepancies between the seizure documentation by patients in a diary and the actual interpretation of the seizure situation from the evaluating physician (epileptologist/study physician). Furthermore, a lot of patients experience strong derealisation or amnestic states making subjective ratings difficult to rely on. We, therefore, chose seizure severity, measured with a standardized instrument of epileptic seizures (LSSS) as the outcome parameter in our study. However, we are aware of the shortcomings of using a scale for epilepsy in patients with PNES.
Another difference to earlier studies is that we used an active control group. First, we believe that using an active control group is a strength of our study because it tests for the so-called nonspecific treatment factors. Several studies have shown that common (nonspecific) factors cause an average placebo response of around 40% in psychotherapy. This condition is comparable to drug RCT studies under similar conditions when primary outcome efficacy measures are patient-reported outcomes [18][19][20]. These nonspecific effects might play a role in the very high significant effects of previously evaluated treatment programs for PNES patients [7,8].
Gold et al. describe a framework in which they offer an algorithm that helps construct clinical psychiatric studies. Considering this framework for our study condition, an active and nonspecifically effective control group, especially for patients with a high risk inherent to the condition, is recommended, no matter if there are already effective treatment options [11]. Therefore, we chose the study design with an active control group mainly for ethical reasons, which is in line with the decision framework from Gold et al. [11]. Generally, it can be discussed whether SHGs offer a cost-effective and widely available alternative to treat PNES patients. SHGs could be most valuable in rural areas where trained therapists are less available. These findings are also consistent with those of previous studies in other areas of behavioural health [21], pointing towards self-help approaches as a veritable alternative to therapist-delivered treatment [10].
There are several limitations to our study: The small sample size, mainly due to high drop out rates, led to insufficient power for small effects to reach statistical significance.
Nevertheless, our dropout rate is comparable to the dropout rates in previous studies with the same design and disease [7,8]. However, we had a remarkably high dropout rate (study withdrawal) before the intervention groups started. From initially 67 screened patients, only 42 started the intervention in one of the two study arms. This high withdrawal rate can be partly explained by the non-blinded design, which triggered a 'nocebo'-effect in patients randomized in the control group arm. Six patients did not show up to the control group intervention because of the disappointment of being randomized to the 'wrong group'. For further studies, we recommend an improved communication, for example, neutral labelling of the study groups at recruitment, starting from a position of equipoise of the two treatment arms. This might help to reduce early withdrawal.
Previous studies have discussed the specific challenges of group psychotherapy interventions because the influence of the so-called 'proxies' (placebo effects evoked by family members or other patients who participate in psychotherapy groups) becomes even more virulent in group settings [23]. Of course, positive placebo effects are expected to rise when offering a group intervention, making our study design even more challenging. Further studies could change the study design by offering eHealth interventions, consisting of chatroom-based or blog-based group interventions with psychoeducational content. However, this might lower the effectiveness of the verum intervention because of weakened effects by proxies (e.g. other participants).
Regarding our CORDIS treatment program, some social aspects could improve the positive effect of the treatment compared with that in the active control group. We did not specifically include relatives in our treatment program, which could be a promising option due to the extreme involvement of the social environment of PNES patients [23].
In summary, CORDIS is a promising therapeutic tool for the treatment of PNES patients. In our German sample, the effects of CORDIS were most notable on the outcome of seizure severity 6 months after the treatment, compared with the effects of an active SHG. Our study widens the still small spectrum of clinical research projects regarding the treatment options for PNES patients. To further investigate the efficacy of our treatment program, a study with a higher statistical power (multicentric study) and an improved study design would be promising. | 2021-06-18T13:50:14.566Z | 2021-06-18T00:00:00.000 | {
"year": 2021,
"sha1": "fdba1ede647995652aee5a18b9067d0085b572e8",
"oa_license": "CCBY",
"oa_url": "https://link.springer.com/content/pdf/10.1007/s00415-021-10652-0.pdf",
"oa_status": "HYBRID",
"pdf_src": "Springer",
"pdf_hash": "fdba1ede647995652aee5a18b9067d0085b572e8",
"s2fieldsofstudy": [
"Medicine",
"Psychology"
],
"extfieldsofstudy": [
"Medicine"
]
} |
250280098 | pes2o/s2orc | v3-fos-license | Generalised model-independent characterisation of strong gravitational lenses VIII. automated multi-band feature detection to constrain local lens properties
As established in previous papers of this series, observables in highly distorted and magnified multiple images caused by the strong gravitational lensing effect can be used to constrain the distorting properties of the gravitational lens at the image positions. If the background source is extended and contains substructure, like star forming regions, which is resolved in multiple images, all substructure that can be matched across a minimum of three multiple images can be used to infer the local distorting properties of the lens. In this work, we replace the manual feature selection by an automated feature extraction based on SExtractor for Python and show its superior performance. Despite its aimed development to improve our lens reconstruction, it can be employed in any other approach, as well. Valuable insights on the definition of an `image position' in the presence of noise are gained from our calibration tests. Applying it to observations of a five-image configuration in galaxy cluster CL0024 and the triple-image configuration containing Hamilton's object, we determine local lens properties for multiple wavebands separately. Within current confidence bounds, all of them are consistent with each other, corroborating the wavelength-independence of strong lensing and offering a tool to detect deviations caused by micro-lensing and dust in further examples.
INTRODUCTION
Highly dense mass distributions like galaxy clusters or galaxies can act as strong gravitational lenses, deflecting the light of background objects into multiple images. Previously established approaches to reconstruct the deflecting mass density profiles, e.g. Lenstool (Jullo et al. 2007), Grale (Liesenborgs et al. 2010), Glee (Grillo et al. 2015), or the lighttraces-mass-approach of Zitrin et al. (2009) to name a few galaxy-cluster reconstruction tools or, e.g. the galaxy-scale reconstruction algorithms developed in Keeton (2001) and Saha & Williams (2004), use global mass density profiles that are fit to the observables of the multiple images, mainly the positions of their centres of light.
The method developed in this paper series, mainly in paper I (Wagner 2017), paper II (Wagner & Tessore 2018), and paper III (Wagner 2018a), principally differs from this methodology. It does not fit a global mass density model to all multiple images together, but directly uses them to constrain local properties of the gravitational lens at their positions and in their close proximity. These local lens properties, ratios of scaled mass densities, reduced shear components and approximations to the critical curves, can be directly inferred from the equations of the gravitational lensing formalism without additionally imposing a mass density profile, may it be of parametric type like used in Lenstool or consisting of a basis function set as in Grale. Thus, the local lens properties yield all data-based information about the lens that the model-based approaches have in common. Using these local lens properties, we can also back-project the multiple images into the source plane to obtain a reconstruction of the source morphology. The latter can be determined up to an overall scaling factor and is free of any bias that may be caused during the back-projection when using a mass density profile model. As detailed in paper I, II, and III of this series, the extended surface brightness profiles of multiple images deliver the most constraining observables to apply this method. Paper IV (Wagner 2018b) and VI (Wagner 2019b) of the series deal with the mathematical derivation and the physical interpretation of the degeneracies of the strong gravitational lensing formalism. They show which transformations can be applied to the equations of the lensing formalism leaving all observables in the multiple images invariant. In paper VII Wagner (2022) the degeneracies between local lens properties, in particular higher-order terms like flexion, and intrinsic source characteristics, such as ellipticity, are investigated further. Since paper IV and VI investigated lineof-sight contributions to the degeneracies as well, we used a supernovae ensemble to set up observation-based cosmic distances that are least model-dependent in paper V (Wagner & Meyer 2018). These are usually needed to determine the distances between us as observers, the lens, and the source. A summary of the theory and some applications of the method highlighting its usefulness to track dark matter can be found in Wagner (2019a).
This Paper VIII aims to set up a robust, automated image processing pipeline in order to identify the peculiarities in surface brightness profiles that are used as observables to constrain local lens properties as detailed in paper I and II. Due to the increasing amount and quality of multiple-image observations, several use-cases for our approach have already been discovered. On galaxy-cluster scale, we applied it to the five-image configuration of a blue spiral background galaxy in CL0024 in and recently to the tripleimage configuration of a star-forming galaxy in a newly detected galaxy cluster as detailed in Griffiths et al. (2021). For these two cases, we identified matching brightness features in multiple images manually by eye. Only in Wagner & Williams (2020), we could use pre-determined brightness features which were provided by a standard analysis tool of radio interferometer data. With many more possible application cases being discovered, we now address the issue how to extract and employ quantitatively robust features that can be detected automatically using image processing algorithms for given spatial distributions of light, like optical filter bands or radio band data after the conversion into the spatial domain.
Emphasis is put on an automated, quantitatively measurable feature extraction and not on an automated featurematching process across multiple images because the former is independent of any strong gravitational lens modelling and therefore allows to infer local lens properties devoid of assumptions on the global deflecting mass density. As detailed below, this is also the reason why the feature extraction can be based on a multi-purpose, standard astrophysical image processing tool. Subsequently matching these features automatically would require the use of additional assumptions or models as in lens reconstruction algorithms mentioned above. Thus, local lens-model-independent properties cannot be obtained in this way, which is the reason why we only aim for an automated feature extraction.
For the sake of self-consistency, Section 2 briefly summarises the approach to determine local lens properties out of substructure in extended surface brightness profiles of multiple images in the theoretical framework of single-plane strong gravitational lensing. In Section 3, we detail our data processing pipeline from the data used to the extraction of our features of interest which are to be inserted into the equations introduced in Section 2. As many well-established collaborations have relied on SExtractor (Bertin & Arnouts 1996) as the standard software to generate catalogues of multiple-image observables, like Postman et al. (2012) or Lotz et al. (2017), we also base our data processing on it. Yet, instead of using SExtractor directly, we employ the routines provided by the 'Python library for Source Extraction and Photometry', sep, (Barbary 2016) because recent developments in cosmology and astronomy increasingly rely on Python as the standard programming language.
After the introduction of the robust feature extraction for our approach, Section 4 shows the resulting local lens properties based on these improved features for the two previously mentioned multiple-image configurations and compares them to the results obtained with the features identified by eye in and Griffiths et al. (2021). Beyond the comparison of performance between manual and automated feature extraction, we also investigate the dependence of the local lens properties on the filter band that is used to extract the features. As gravitational lensing effects are wavelength-independent, analysing the same configuration over multiple filter bands allows us to investigate biases, for instance due to dust extinction. Subsequently, Section 5 summarises the results and discusses the advantages of the new robust feature extraction for our lens reconstruction approach and others. Section 6 concludes the paper discussing further usages to robustly detect deviations from wavelength-independent strong gravitational lensing in further multiple-image configurations.
THEORETICAL BACKGROUND
A general overview of gravitational lensing in arbitrary spacetimes can be found in Fleury et al. (2021) or any standard text book on gravitational lenses, for instance, Schneider et al. (1992). The approach discussed here is based on the standard single-lens-plane strong gravitational lensing formalism in a concordance Λ-Cold-Dark-Matter cosmological background.
For consistency across all papers of this series, we denote two-dimensional positions in the lens plane by x. The angular diameter distance to the lens plane is D d corresponding to its redshift z d . Analogously, we denote two-dimensional positions in the source plane by y, the angular diameter distance is Ds corresponding to zs. The distance between lens and source plane is denoted by D ds . Multiple images are indexed by i, j = 1, 2, 3, ... and positions of brightness features by µ = 1, 2, 3, ... .
Summary of the formalism to extract local lens properties
For self-consistency, we briefly summarise the approach of Wagner & Tessore (2018) and to match features in multiple images onto each other in order to extract local lens properties. The linearised lens equation around a point xi,0 in multiple image i = 1, 2, 3, ... maps vectors around xi,0 into vectors in the source plane around the source point y 0 , which is the common source position of the multiple image points xi,0. Thus, using the distortion matrices A(xi,0), we can state meaning that the lens mapping back-projects vectors between corresponding brightness features µ = 1, 2, ... in multiple images, here image i and image j, onto the same vector in the source plane (see also Fig. 1). The distortion matrix usually contains the convergence κ and the shear components γ1 and γ2 at the position xi,0. The former is a measure of the local scaled mass density which scales the source properties by an overall factor. The latter two represent the local leading-order distorting strength of the lens. Yet, as we are interested in properties that can be directly linked to observables, we write Ai ≡ A(xi,0) in terms of the reduced shear components g1 and g2 as Since the vectors in the source plane cannot be observed, we first identify corresponding features in the multiple images to set up the vectors (xi,µ − xi,0) for i = 1, 2, 3, .... Subsequently, we use the last part of Eq. (1) to set up a system of equations linking the vectors in different multiple images and their unknown distortion matrices, which is then solved for the ratios of convergences between multiple images and the reduced shear components at all multiple image positions with the magnitude of the shear of image i given by if we have at least three multiple images each containing two non-parallel vectors that can be matched. Fig. 1 (3) and (4) using Eq.
(2), we also determine the ratios of magnification ratios which can be compared to observed flux ratios between the multiple images for a consistency check. Differences between Eq. (6) determined by the brightness feature locations and the observed flux ratios can hint at additional micro-lensing of individual images or dust extinction due to intervening gas clouds along the line of sight.
Only ratios of the characteristic local lens properties κ, γ1, γ2 can be determined because all observables are angular distances on the celestial sphere (see Wagner (2018b) and Wagner (2019b) that these quantities are not subject to any of the common strong-lensing degeneracies and that they are indeed the maximum information retrievable from observables without imposing additional model assumptions). (6). As the graphics shows, a linear transformation T between two images close to corresponding multiple-image positions x 1,0 , x 2,0 can be determined from the brightness feature vectors (white arrows). This T corresponds to the product of the distortion matrices of these multiple images, enabling ptmatch to solve for the local lens properties.
To be converted to physical distances, they require an overall size scale, like a physical distance, to be fixed. In standard approaches to gravitational lens reconstructions, the angular diameter distances based on a chosen cosmological model or on observations, like set up for our approach in Wagner & Meyer (2018), are used.
In principle, Eq. (3) can contain any pair of convergences, but, for the sake of efficiency, the implementation we use, called ptmatch 1 , fixes a so-called 'reference image' with respect to which all ratios of convergences are determined. As in all related works, we set the reference image to be called image 1. ptmatch also takes into account uncertainties in the brightness feature positions and yields confidence bounds on the local lens properties of Eqs. (3) and (4) as sampled from the full likelihood distribution of the f and g values for the given observed positions. Details about the confidence bounds can also be found in Wagner & Tessore (2018) and .
Robust features
Based on the strong lensing formalism, the most reasonable robust features are maxima in the brightness distribution of the multiple images because maximum intensities in the surface brightness profile of the source object are mapped into maximum intensities in the surface brightness profiles of the multiple images. This statement relies on the conservation of photons and can be proven as follows: Assume the intensity at a point y 0 in the source object is I(y 0 ). If this is an intensity maximum, then ∇yI(y 0 ) = 0 holds at y 0 , meaning, we first take the gradient ∇y with respect to y in the source plane and then evaluate it at y 0 . Now, let x0 be a corresponding image point to y 0 in the lens plane with intensity I(x0), like, for instance, x1,0 to y 0 in Fig. 1. If the surface brightness is conserved, then I(x0) = I(y 0 ). Deriving both sides with respect to x implies ∇xI(x0) = ∇xI(y 0 ) = A(x0)∇yI(y 0 ) = 0 , where we transformed the gradient from the lens to the source plane first and then used Eq. (7) in the last step. Although the photon conservation is an idealised assumption that may not strictly hold for observations, cross checks can be made when jointly analysing data from several filter bands and using Eq. Since Eq. (8) also implies that intensity minima in the source plane are mapped to intensity minima in the lens plane, one could also think of using them as 'darkness features'. Some features used in Griffiths et al. (2021) actually relied on this argument and yielded a robust and reasonable ptmatch result. However, the darkness features may remain rather an exception than becoming standard because the negligible contaminations due to stray light of neighbouring foreground objects and the very low and stable background noise level as observed for the triple-image configuration in Griffiths et al. (2021) has only been rarely found so far.
ROBUST FEATURE EXTRACTION BY IMAGE PROCESSING
As motivated in Section 1, our image processing pipeline to extract the intensity maxima as robust features, detailed in Section 2.2, relies on the 'Python library for Source Extraction and Photometry', sep, and consists of the usual steps of background subtraction (see Section 3.1) and subsequent feature detection and extraction (see Section 3.2). We systematically analyse the optimum parameter ranges for both subroutines to maximise the number and stability of detected intensity maxima for the star-forming regions of the multiple-image configuration in CL0024 discussed in in the 475W filter band. We will show in this and the following section that training and calibrating on a single, very clear example case in one filter band, generalises well to further observations of other wavelengths because all steps are based on sound physical and imageprocessing principles. Limitations, as also noted by other automated lens reconstruction pipelines like Shajib et al. (2019), are summarised in Section 6. To accelerate the image processing, we choose a 128 × 128 pixel box to contain the individual gravitationally lensed images, including an area of 64 × 64 pixels devoid of any image pixels and further unrelated high-intensity signals for the local background estimation. Fig. 2 shows the boxes (with areas for background estimation marked by white squares) for all five images of our example used in this section. No pixels were excluded from the analysis pipeline, meaning that we did not make use of any masking functions due to the negligible number of corrupt pixels. As detailed below, these choices are applicable to many other multiple images as observed with current telescopes, in particular the Hubble Space Telescope (HST).
Background subtraction
The background subtraction routine of sep is based on the same mixture of κ σ clipping and mode estimation as SExtractor, documented in Bertin & Arnouts (1996). As such, it could estimate a varying background over an extended region observed on the sky even in the presence of foreground signals and stray light. However, run-time is increased when applying this routine to an extended region. Furthermore, the robustness of the resulting background estimate underneath the foreground pixel has to be checked, see, for instance, Melchior et al. (2018) for a more detailed discussion. We therefore choose to estimate a constant mean background value, µ bg and its root-mean-square value, σI, from a sufficiently large, separate background patch close to the foreground pixels of the gravitationally lensed images (see white squares in Fig. 2).
Using a patch size of 64 × 64 (the sep default value) approximately covers the same area spanned by the pixel of a gravitationally lensed image at HST resolution, so that the resulting σI obtained from this size of background patch yields an adequate measure of the noise level per pixel to robustly detect intensity maxima as detailed in Section 3.2. Using smaller or larger patch sizes can lead to a reduced amount of detected features or a less efficient detection, depending on whether the respective σI is larger or smaller than the optimum value 2 . Accounting for potentially occurring biases in the background estimation, we define the sufficient (µ bg , σI) pair when the maximum intensity positions of all star-forming regions at significance levels larger or equal than 3σI above µ bg are correctly identified. Choosing the background patch size too small or too large particularly affects the quality of feature detection in multiple images in the central part of galaxy clusters, as image 5 in Fig. 2 demonstrates. Too small a background patch size will set µ bg as too high, while too large a background patch will increase σI when extending the patch out to less luminous areas in the galaxy cluster. As a consequence, the features with low signal to noise in these images will remain undetected.
Next, we determine a suitable value for the width and height of the median filtering which is typically used to suppress bright foreground signals like stars. As expected by construction of background patches, different values of the filter width and height do not change the detected number of features (see Section 3.2 for details on the detection and extraction step), so that we keep the default value of three pixel.
Thus, the parameter settings for the sep background estimate routine are set to the default values having corroborated their usefulness for our case. After applying this routine to all background patches shown in Fig. 2, we retrieve the globalback and globalrms return values. They correspond to µ bg and σI, respectively, so that we then can subtract the mean background estimate from the entire 128 × 128 box to continue with the feature identification in the multiple images.
Feature extraction
To identify and extract the maximum intensity values of the star-forming regions, we continue with the sep routine to extract these features. The latter requires a backgroundsubtracted array of data as input. We insert the resulting 128 × 128 pixel box obtained after the steps detailed in Section 3.1.
To separate foreground from background objects, sep employs a simple thresholding at each pixel in which the signal strength nI is user-defined and (µ bg , σI) are determined by the routine of Section 3.1. Contiguous areas of pixels above t are subsequently grouped into foreground objects. (The significance of detection in terms of nIσI above background thus marks the pixel-wise signal strength and should not be confused with the integrated signal strength for the whole object based on all of its pixels.) The resulting collection of detected objects and their properties can be visualised as shown for the example of image 1 in Fig. 3 for nI = 5. Increasing nI from 3 in steps of 1 to the value at which no objects are detected anymore, we track the size of the detected objects in the box surrounding image 1 of Fig. 2 as plotted in Fig. 4. The plot shows that the star-forming regions of interest can be extracted in this way because many of them are identified as objects up to a much larger threshold than set by the minimum requirement of nI = 3. To avoid spurious detections, the minimum amount of required pixels to be counted as an object, minarea, is kept at the default of five pixel. While the star-forming regions of interest cover a much larger area at nI = 3-5, the next step of our robust feature extraction goes to the limit of higher thresholds and smaller areas. Therefore, we do not choose to set minarea to a higher value.
As star-forming regions may lie closely together or be even merged into elongated wingy objects, we need to switch the clean parameter in the sep extraction routine off to as- sure that all star-forming regions can be detected and are not pruned as spurious detection in the wings of a brighter one.
To further maximise the number of features that can be matched across all multiple images, we tune the deblending parameter to split a multi-modal intensity distribution into separate objects. Tracking the creation and annihilation of objects is easiest in a 'persistence diagram', as shown in Fig. 5 for the example of image 2 of Fig. 2. The left-hand side tracks the x1-coordinate of all objects for increasing threshold, the right-hand side the x2-coordinate. Objects below five pixel area simply cease to exist, while new ones can be created at higher thresholds when large contiguous areas split into separate parts. Fig. 5 is plotted setting the minimum contrast ratio used for object deblending, deblend cont, to its default value 0.005. Hence, the minimum flux ratio to split a fainter object with lower flux from a brighter one with larger flux is 0.005. The positions of all objects detected in this way remain stable over all thresholds and only one object is newly created above nI = 3. In contrast, we increase deblend cont to 0.1 in Fig. 6 and observe that detected object positions become less stable. Most severely the objects with coordinates (33, 95) and (33, 101) are affected, which, due to their mutual proximity, are not reliably separated anymore. Testing deblend cont values from 0.000 to 1.000, we find most stable object separation for the default value, without introducing spurious artefacts at values below 0.005. As expected, increasing deblend cont above 0.005, causes objects to mix and the number of detected objects to decrease because faint features disappear and features that lie close together merge.
Thus, the only change in the default setting of the sep extraction routine is to set the clean parameter off, all other pre-defined defaults are suitable for our application.
In this parameter configuration, we extract all objects for all images in Fig. 2 for nI ≥ 3 and plot their persistence diagrams to determine which of these objects are stable features that can be matched across all multiple images. First, the longer an object persists in the diagram over increasing thresholds, the higher the probability it is a suitable feature. Second, objects showing horizontal lines in the persistence diagrams are also more likely to be suitable features because their centre of light coordinates do not vary, for instance, due to a large amount of noise. Third, the smaller the distance between the centre of light position, calculated as the first moment of an object's brightness distribution, and the peak position of maximum intensity, the more likely it is that the object under analysis is a suitable and not very noisy feature. Fourth, we would expect star-forming regions to be of approximately circular shape which are distorted into ellipses to first order by the gravitational lensing effect. Assuming that these features are far enough away from the critical curve not to be extremely distorted in the shape of arcs (see Wagner (2022) for a detailed analysis in galaxy-cluster scale gravitational lenses), the semi-major and semi-minor axes of the second order moment of an object's brightness distribu-tion also contribute to evaluate the usefulness of an object. Highly elongated arcs can be excluded to be star-forming regions and only included in the feature set if such arcs are clearly identified across several multiple images as a more complex feature. Fig. 7 shows the detected objects and their properties for image 2 of Fig. 2 at nI = 3 (left) and nI = 5 (right). We note that the many very faint objects that do not correspond to suitable features do not persist to nI = 5 and are therefore easily excluded by the first criterion of persistence. As for the second criterion, we already saw the mostly horizontal lines in the persistence diagrams of Fig. 5. The largest change in position of the centre of light for the most unstable feature (15, 106) of Fig. 5 is found to be 2 pixels, which is comparable to the most unstable features in the other multiple images and thereby sets the range of tolerable non-zero slopes for the second criterion. Comparing the centre of light (magenta dots) and maximum intensity (blue dots) positions within each of the objects passing the first two criteria in Fig. 7, we note that the extended background structure of the galaxy and the proximity of individual features to each other cause the two positions to have distances of the order of 2 pixels at nI ≤ 5. These distances are reduced below one pixel for increasing nI > 5. For the neighbouring features which overlap at low nI, reducing the deblend cont may reveal the additional features visible by eye in the wings of the ellipses at larger nI. Yet, as noted before, this change in parameter value also increases the amount of spurious, persistent features consisting of bright, locally restricted noise. The axes ratios of the semi-minor to semi-major axis of the quadrupole moments are also tracked over the increasing thresholds and remained greater than or equal to 0.15, thus giving an order of magnitude to identify star-forming regions as contained in the five images in CL0024.
In summary, running the default configuration of the sep extraction routine and switching clean off, the detected objects are analysed for their persistence at least existing to nI ≤ 3, stability of the centre of light position with only 2 pixel variations over all nI, the distance between the latter and the maximum intensity position of up to 1 pixel at the highest nI, and an only slightly elliptical quadrupole moment of the intensity distribution with axis ratios larger than 0.15. After determining the objects fulfiling these criteria, their properties at the highest nI at which they can still be detected are used as stable features in the next steps 3 .
Calibration of ptmatch
After the automated extraction of robust features using the criteria in Section 3.2, we manually match the maximum number of corresponding features found across all images. For the example in Fig. 2, we find four features fulfiling the criteria of Section 3.2 that can be matched across all images. They correspond to the features 1, 2, 4, and 5 identified by eye in .
These features can now be employed to run ptmatch and retrieve the local lens properties as detailed in Section 2. In the idealised setting depicted in Section 2.2, the peak positions as found by our sep pipeline would be inserted into ptmatch. However, as Section 3.2 indicated, an intensity maximum may be generated by noise, so that it is not the corresponding feature to the respective intensity peaks in other multiple images. Other lens reconstruction approaches which determine the 'positions of multiple images' are not always clear about their precise choice of coordinates, it also depends on the image pre-processing. Caminha et al. (2016), for instance, clearly use the peak position and thus assume all noise was eliminated. On the scale of star-forming regions as features, Wagner (2022) showed that any physically caused offset between the peak and the centre of light, for instance due to higher-order lensing effects like flexion, is unlikely to occur and be resolved. This corroborates our third selection criterion for robust features of Section 3.2.
The persistence analysis introduced in Section 3.2 allows to evaluate the quality of the chosen features to be matched. Tracking the peak position with respect to the centre of light position in each object over increasing intensity thresholds, peaks created by noise and peaks representing genuine intensity maxima can be distinguished. A second example, in addition to the one in Fig. 7, is the object detection in the central image 5 in CL0024 as observed in the F555W filter band shown in Fig. 8. This image is located in a low signal-to-noise environment. Therefore, at the highest threshold, nI = 4, that identifies the object in the centre of Fig. 8 (left), the peak position (yellow cross) is still 3.2 pixel away from the centre of light position (magenta dot). To investigate the cause of the distance, we manually increase the threshold to nI = 4.6, at which the object is still detected. As can be read off the comparison between Fig. 8 (left) and (right), the peak position does not shift closer to the centre of light along the connection between the peak and the centre of light, but completely changes its position with respect to the centre of light. Thus, this unstable peak should not be employed as a feature to be used in ptmatch. According to the feature selection criteria of Section 3.2, the object is discarded due to the distance between the peak and the centre of light exceeding one pixel.
Yet, cases may arise in which such objects are erroneously employed as features or have to be considered for the lack of qualitatively better ones. To test the impact of these inferior features on the ptmatch results, Fig. 9 shows the local lens properties determined for image 5 of Fig. 8 (further details on this and other filter bands are detailed in Section 4.1). The measurement uncertainties for the positions, σ1 and σ2 for the x1-and x2-coordinates respectively, were set to one pixel in all cases. Then, ptmatch was run on all multiple images for the F555W filter band to determine the local lens properties of Eqs. 3 and 4. Leaving all feature coordinates in all other multiple images at the stable peak positions, different choices of positions as detailed in Fig. 9 are used for the features in image 5 4 . Comparing the f and g values for the four different choices, employing the noisy peak position leads to deviations in the inferred local lens properties beyond consistencies within the 68% confidence bounds marked by the error bars. The comparison of the two position choices shows that local lens properties as inferred from positions that vary on the scale of one pixel are consistent with each other and that the χ 2 red is not the only measure to be monitored when evaluating the overall quality of fit, as it does not indicate whether the chosen position for a feature is physically meaningful.
Having investigated the impact of the coordinate choices on the inferred local lens properties, we continue to apply our pipeline for automated feature extraction based on sep and subsequent inference of local lens properties by ptmatch to multiple wavebands for CL0024 and the galaxycluster containing Hamilton's Object. Unless noted otherwise, we apply the feature selection criteria of Section 3.2, choose the stable peak positions, and run ptmatch in its standard configuration with measurement uncertainties of σ1 = σ2 = 1 px, motivated by the typical distances between the peak and the centre of light position. Figure 8. Identification of peak positions generated by noise for the example of image 5 in CL0024 observed in the F555W filter band, thresholded at n I = 4.0 (left) and n I = 4.6 (right). The central object's peak position (yellow cross) at the last detectable threshold with n I = 4 is several pixel away from the centre of light position (magenta dot). Increasing the threshold to n I = 4.6, the peak position shifts closer to the centre of light. Thus the peak is no genuine maximum intensity value but an artefact due to the low signal-to-noise environment. Figure 9. Impact of different feature position choices on inferred local lens properties f , g 1 , and g 2 for image 5 of Fig. 8. A: results using peak points at n I = 4 (χ 2 red = 1.013), B: results using peak points at n I = 4.6 (χ 2 red = 0.459), C: results using peak points at n I = 4 for all features but the noisy one which is replaced by the centre of light (χ 2 red = 0.623), D: results using centre of light positions at n I = 4 for all features (χ 2 red = 0.478). Plots show the mean local lens property (black mark) with its 68% confidence bounds (error bars) and the most likely local lens property (red mark) as obtained from ptmatch.
APPLICATIONS
We now apply our robust feature extraction pipeline to the two multiple-image configurations previously analysed in and Griffiths et al. (2021) and extend the manual feature extraction to multiple wavelengths in the case of CL0024 (detailed in Section 4.1) and to a feature extraction from individual filter bands (instead of a joint picture) for the cluster containing Hamilton's Object (detailed in Section 4.2). Subsequently, we run ptmatch on the manually matched features and obtain the local lensing properties, Eqs. 3, 4, and 6 for all filter bands. Tables 4 and 5 summarise all local lens properties obtained for both cases.
Five-image configuration in CL0024
The first example is the five-image configuration of a blue spiral galaxy at zs = 1.675 behind the galaxy cluster CL0024+1654, abbreviated as CL0024, at z d = 0.39, which we already investigated in using the approach of Section 2.1. Here, we re-analyse this example using the automatically extracted features to demonstrate that an objective feature extraction yields similar local lens properties as features identified by eye. The standardised feature extraction also allows us to compare the local lens properties obtained in several wavelength filter bands to investigate the wavelength independence of strong gravitational lensing. The observations used here were made by HST in November 2004 (PI: Ford) using filters F435W, F475W, F555W, F625W, F775W and F850LP with exposure times 6435s, 5072s, 5072s, 7756s, 5072s, and 8164s, respectively. Each waveband has the same spatial resolution, but different signal-to-noise ratios, as the background estimation reveals. The feature extraction as outlined in Section 3 is performed for all filter bands. A manual cross-check confirms that it leads to a stable detection of features 1, 2, 4, and 5, the Table 1, mean (black dot), 68% confidence intervals (error bars), and most likely value (red cross). star-forming regions used as manually identified features, in . Out of the 24 required feature identifications, the only noisy peak is the one in image 5 in F555W, shown in Fig. 8, which is replaced by the centre of light, as detailed in Section 3.3, because the feature is vital to span a larger area (see on the impact of the image area on the precision of the local lens properties). Our manual follow-up analysis showed that the true stable peak lies close to the centre of light and therefore, the local lens properties as obtained from the automated feature extraction for the centre of light are in agreement with the one from the true peak within the confidence bounds in this case. Features 3 and 6 of were found in larger images with a higher signal-to-noise ratio such as images 1 and 4, but could not reliably be detected in images with lower signal to noise such as image 5.
In Figs. 10-14, we compare all local lens properties for each image over the six filter bands and with the local lens properties that are obtained for the same four manually identified features and all six manually identified features of . To shorten notations, Table 1 lists the abbreviations used as labels on the abscissa of the plots and the χ 2 red values for the best-fit obtained by ptmatch. Comparing the χ 2 red values between the manual feature identification and the automated one, we find that, except for F775W and F850LP, the matching is improved and the χ 2 red values reduced from a biased fitting to over-fitting. The F475W filter band shows the lowest χ 2 red values, potentially due to its role as calibration example. Yet, Figs. 10-14 show no signs of specialty with respect to other filter bands. Confidence bounds for image 2 and image 4 are even the second largest for F475W, only surpassed by those in F775W. Section 4.2 will show that similarly low χ 2 red are obtained in a different setting as well (see Table 2) to support the claim that F475W yields typical results. Thus, the feature extraction generalises as expected to other filter bands. Yet, the increased χ 2 red values for F775W and F850LP and the increased confidence bounds particularly observed for F775W show that the automated detection reaches its limits for those filter bands. The emission profile of the blue background galaxy fades into the background light for these wavelengths. This effect is clearly visible in Fig. 11 for image 2, which requires a very precise feature identification because it is located close to an isocontour of κ = 1, as detailed in .
As also observed from Fig. 10-14, all local lens properties are in agreement within the 68% confidence bounds. For further details, Table 4 contains all values. Variations in the value and the size of the confidence bounds can be observed, most significantly for images 2 and 4 in F475W and F775W as noted above. Altogether no significant disagreement is found within the current precision. We can thus conclude that no deviations from a wavelength independence of strong lensing can be found in this multiple-image configuration.
As a cross-check, we determine the magnification ratios of Eq. 6. The results are plotted in App. A for completeness. As expected, these graphs also show a high degree of agreement and thus yield the same conclusion. Compared to the f and g values, their confidence bounds are smaller (due to the implementation of ptmatch), such that they are even a better test for constant local lens properties over all wavelengths. To investigate potential biases of the local lens properties, for instance, by dust extinction or micro-lensing, a comparison to the measured flux ratios is required. This can be easily performed for multiple images like Hamilton's Object, detailed below in Section 4.2. Its main flux is contained in a central bright feature and can thus be determined by a standard flux measurement implemented in SExtractor or sep. For this image configuration in CL0024, however, such a flux measurement is hard to set up because it should contain the flux collected from the area spanned by all features. This flux cannot be calculated with the standard astrometry and photometry programs because they only focus on individual objects like the star-forming regions but not arbitrarily shaped areas. So we leave such a comparison, which is beyond the scope of this work, for future investigations.
Three-image configuration with Hamilton's Object
Our second example is the triple-image configuration containing Hamilton's object as first analysed in Griffiths et al. (2021). The background source is a star-forming galaxy at zs = 0.8200 multiply-imaged by the galaxy cluster RM J223013.1-080853.1, J2230 for short, at z d = 0.526. Up to our knowledge, Hamilton's Object is the first fold configuration with almost perfectly aligned images orthogonal to their straddling critical curve which also has an extended counterimage clearly showing the same substructure, as spectro- scopically identified in Griffiths et al. (2021). Fig. 15 gives an overview of the three images. For this case, your focus lies in first determining robust features for each individual filter band from the automated pipeline outlined in Section 3, which could not be achieved manually. To identify matching features by eye, we had to create a coloured image as a combination of the F606W, F814W, and F110W filter bands. Assembling this coloured image was an image processing task on its own, as detailed in Griffiths et al. (2021). The F606W and F814W filter bands with a resolution of 1.4 × 10 −5 degrees per pixel observed by ACS/WFC had to be downsized and aligned with the F110W filter band with a resolution of 2.5 × 10 −5 degrees per pixel observed by WFC3/IR. Thus, this triple-image configuration can also be used to study the robustness of our feature extraction and the inferred local lens properties for the same data at different resolutions. Our second goal here is therefore to compare the local lens properties inferred for F606W and F814W and their downsized versions, abbreviated as F606W d and F814W d , respectively.
The three images denoted as A, B, and C in Griffiths et al. (2021) will be referred to as images 2, 1, and 3 respec- Figure 16. Detection of an additional robust feature in image 1 in J2230. Matching feature 4 in image 1, two choices (green crosses) with peak-to-peak separation of 6 pixels show up. The plot shows the detected objects in images 1 in F606W waveband at a n I = 3, notation as in Fig. 3. tively because we now use image 1 (image B in Griffiths et al. (2021)) as a reference image for ptmatch to obtain more robust confidence bounds for the local lens properties and χ 2 red closer to 1. The observations used in this paper of Hamilton's Object were made by HST with the filter bands F606W and F814W taken in 2015 and the band F110W taken in 2016 (PI: Ebeling for both). They had exposure times of 1200s, 1200s, and 705.88s, respectively.
Running our feature extraction pipeline as detailed in Section 3 for all filter-band configurations listed in Table 2 5 , features 1, 3, 4, and 7 of Griffiths et al. (2021) are robustly identified without any noisy peaks. Hence, the automated feature extraction succeeds in identifying features across different filter bands with different signal-to-noise ratios and different resolutions, which was not possible by eye.
Further investigating all identified features, we note an interesting detection in the F606W, F814W, their downsized versions, and the coloured image as shown in Fig. 16 in the F606W filter band. While all other images clearly only have one object detection that can be considered as feature 4, two possible objects show up in the above mentioned filterband configurations. As the two almost vertically aligned features, called 4t and 4b (top feature and bottom feature, respectively), are not visible in the F110W filter band image, one may interpret them as a split of feature 4 into two parts due to the increased resolution in the F606W and F814W filter bands. Yet, the downsized versions also show feature 4t and 4b, so that this explanation can be excluded. With the same argument, it can be rejected that image 1 being the largest image of all images may show more details of the background source because it is subject to a stronger gravitational lensing effect. Thus, two interpretations remain. The second additional feature can have a narrow emission profile restricting its visibility to the optical wavelengths. Alternatively, given that the two observations happened about 5 The automated detection of features in the coloured image is performed on its conversion into a grey-valued image by means of the Python Imaging Library (PIL) function 'convert("L")'. Table 2 using feature 4t (black lines) or 4b (blue lines) as feature 4 for image 1 (see Fig. 16). Notation analogous to Fig. 10.
a year apart, the second feature can be a transient microlensing event similar to the ones made by Kelly et al. (2018) and Kaurov et al. (2019), explained by a two-scale lens modelling of the global galaxy-cluster and the local micro-lensing scale in Diego et al. (2018).
Comparing intensities and relative positions, feature 4t is considered the best matching one and will be used in all following plots. Yet, assuming we use feature 4b instead, we can perform a similar test for robustness of ptmatch as for image 5 in CL0024 detailed in Section 3.3. Fig. 17 shows a comparison of all local lens properties in image 1 obtained over all filter-band configurations of Table 2 when using feature 4t (black lines) or 4b (blue lines) as feature 4. The χ 2 red values for the ptmatch fits are of the same order of magnitude for both configurations. The plots for images 2 and 3 look similar to the g1-value plot in Fig. 17 and are thus insensitive to the feature exchange. They are plotted in Appendix B. In contrast, the g2 values in image 1 show a shift to smaller g2 values, which is most prominent in the F606W and F814W filter bands. The unchanged values in all local lens properties except for the shift in the g2 value of image 1 show that ptmatch only changes those local lens properties that are directly affected by a change in the relevant feature coordinates in the respective image. Larger |g2| values are expected because feature 4b is farther away from the bulge of other features and thus contributes to increase the amplitude of the shear in g2 direction. Comparing these results to the ones obtained by eye with four and seven features, there is a trend for the seven-feature configuration to agree with the local lens properties obtained using feature 4b, indicating that the remaining features also favour larger |g2| values. In addition, the error bars are slightly smaller when using feature 4b (see also the plots in Appendix B that show less stable local lens properties for the downsized version of F606W and the colour image). Hence, the correct matching cannot be determined with certainty.
Using feature 4t, the local lens properties for all multiple images are shown in Figs. 18, 19, and 20. (Due to different random realisations for the confidence bounds, the error bars compared to Fig. 17 can differ.) Table 5 summarises all values in detail. As can be read off Table 2, all χ 2 red values remain well under 1 and are in the regime of over-fitting. For this case, over-fitting is not surprising, particularly because image 1 and image 2 are more or less reflected copies of each Table 2, analogous to Fig. 10. other, such that not all degrees of freedom offered by the linear transformation encoded by the local lens properties may be required for the mapping.
Based on some very large error bars that are observed for the downsized versions of F606W and F814W and the colour image compared to the very robust results for the original data, it is possible that the additional processing of these observations introduces biases in the positions of the features. Therefore, any additional image pre-processing, for instance, denoising, should be handled with care in order not to bias the positions of the multiple images and its features. We also find that the confidence bounds for the local lens properties inferred in the F606W and F814W filter bands are all smaller, ranging from 12% to 67% of the ones obtained by manual feature extraction (at the worse Table 3. Flux ratios F 2 and F 3 between image 2 and 3 to image 1 in J2230 for the central bulge as given in Table 1 in Griffiths et al. (2021) for the filter bands used in our analyses. resolution of the F110W filter band). They are even smaller than the confidence bounds obtained using seven manually extracted features. Both results together corroborate that the resolution of the observation and its pre-processing has a non-negligible impact on the precision of the inferred local lens properties. At last, we plot the magnification ratios of images 2 and 3 to image 1, Eq. 6, inferred by ptmatch in Fig. 21. Compared to the five-image configuration in CL0024, the confidence bounds on J2 and J3 are reduced for F606W and F814W, contributing to a stronger variation between the different configurations. As we see in Figs. 17 and 18, the latter are mainly caused by the variations in image 1 over the different filter-band configurations. All magnification ratios agree within their 68% confidence bounds so that we find no hint for biases of the wavelength independence of strong gravitational lensing.
From the measured magnitudes of the images contained in Table 1 of Griffiths et al. (2021), we obtain the flux rations for the three filter bands analysed in Table 3. While there is some flux in feature 7 and also contributions from the other features, the central bulge was found to be the major source of emission in Griffiths et al. (2021) and its observed flux was found in overall agreement with the magnification ratios as determined by ptmatch. The same is found for the magnification ratios obtained when using our automated feature extraction. Yet, all magnification ratios are systematically lower than the observed flux ratios except for J3 for F814W. This may be a hint of micro-lensing in image 1, supported by the detection of feature 4b. Assuming (locally) brighter than expected fluxes due to micro-lensing that are not incorporated in the ptmatch analysis, the flux ratios would be lower than the ones inferred from ptmatch. Alternatively, it is equally well possible that the parts outside the centre of the galaxy can explain the under-predicted magnification ratios. To follow up on that hypothesis, the same problem with the flux determination from a set of non-contiguous bright features as noted in Section 4.1 arises and is left for future work.
DISCUSSION AND SUMMARY OF RESULTS
In summary, this part of the paper series focuses on the identification of robust features within multiple images instead of the local lens reconstruction method as the previous parts. As such, it yields valuable information for other lens reconstruction approaches as well.
We showed that standard techniques exploiting the physical principles of the data acquisition, for instance implemented in SExtractor, are still adequate to set up an effi- Table 4: Synopsis of local lens properties, Ji, fi, gi,1, and gi,2 of the five-image configuration in CL0024 for all filter-band configurations of Table 1. Results obtained by eye differ from those in because ptmatch was re-run for our analyses, but results are based on the same input data and therefore statistically consistent. Figure 21. Magnification ratio J , Eq. 6, for image 2 (left) and image 3 (right) with respect to image 1 in J2230 for all filter-band configurations of Table 2. Notation as in Fig. 10.
cient feature extraction. A calibration on one representative example case scales well across different filter bands, resolutions, and signal-to-noise environments. Enhanced deblending and noise reduction, like Melchior et al. (2018), may increase the number of detected features in crowded and noisy environments. However, it comes at a higher computational and run-time cost and, using surface-density brightness profiles and noise models to deblend and denoise objects, it may also bias the feature extraction. Thus, it should only be used if the simple attempt does not yield the required outcome. For the two multiple image configurations analysed here, the cost-saving method found less features than manual identification, but detected the salient ones that led to comparable or even smaller sizes of confidence bounds for the inferred local lens properties.
Concerning the feature positions, we can combine the findings of Wagner (2022) that higher-order lensing effects like flexion can mostly be neglected for galaxy-scale multiple images in galaxy-cluster scale lenses with the findings of Section 3.3 that robust features have peak positions within 1-2 pixel distance to their centre of light. Consequently, using the peak position as the theoretically motivated coordinates (see Section 2.2) or using the centre of light as the coordinates robust against noise should not lead to strongly deviating results in the lens reconstruction. As detailed in Section 3.3, the choice which one is better suited depends on the signal-to-noise ratio of the observation under analysis and the desired lens reconstruction precision.
Depending on the pre-processing of the data, biases in the positions can be introduced which do not increase the χ 2 red . This implies that the latter should not be taken as the sole goodness-of-fit measure to evaluate whether the lens reconstruction is physically meaningful. As an example detailed in Section 4.2, the local lens properties obtained from the images in the downsized versions of F606W and F814W have a χ 2 red < 1, but larger confidence bounds than other local lens properties.
Setting up persistence diagrams to track detected objects over increasing significance levels increases the runtime and computational costs compared to a single thresholding. At the same time, extracting only features fulfiling the persistence criteria outlined in Section 3, noisy peak positions and a manually undetected feature could both be successfully identified, alleviating or at least detecting potential biases caused by the pre-processing. Specifically for our approach to reconstruct local lens properties using ptmatch (or, the analytic approximation to the critical curve detailed in Wagner (2017)), the automated feature extraction allows for a standardised evaluation by defining quantitative selection criteria for features to be matched. It even succeeded in finding a sufficient amount of features when a manual approach was not successful.
While inconsistencies can show up for an incorrect matching between features across multiple images clearly pointing at the matching error, for instance shown in Wagner & Williams (2020), ambiguities to match available features can also lead to viable results for all matching possibilities. Section 4.2 details one such case for the two possible feature 4 in image 1 of Hamilton's Object. Further examples for ambiguous feature matchings are Abell 3827, see, for instance, Chen et al. (2020), in which the present matching is incompatible to standard leading-order lensing theory similar to the case in Wagner & Williams (2020) or the matching of features in the giant arc in WHL0137-08 according to Welch et al. (2022). They only show one possible matching and this only maps parts of the giant arc onto each other. Since the resulting lens reconstructions can differ for different matchings, conclusions about properties of (dark) matter should be treated with caution in such cases. For ptmatch we showed that translations of feature coordinates are only related to the corresponding local lens property constrained by these coordinates. Hence, coordinate changes and entailed changes in local lens properties are directly linked and the impact of the former on the latter is intuitively understandable.
CONCLUSION
This part of the paper series focused on the development of an automated feature extraction approach to robustly and quantitatively identify bright, elliptical intensity features like star-forming regions in multiple images of strong gravitationally lensed background galaxies. Our approach is based on publicly available software, sep, adapted and calibrated to our task as detailed in Section 3. The technical benefits for our lens reconstruction approach and others are summarised in Section 5, which also gives recommendations for the detection and the usage of brightness features to infer (local) lens properties based on the results found for two representative multiple image configurations detailed in Section 4.
Most importantly, we found that, for brightness features within multiple images or for featureless multiple images generated by a galaxy-cluster scale strong gravitational lens, the fact that higher-order lensing effects like flexion are negligible in most cases (Wagner (2022)), puts the maximum intensity position and the centre of light only 1-2 pixel apart from each other. As a consequence, basing lens reconstructions on either of them has a small impact on the lens reconstruction. For our lens reconstruction approach, using either of them leads to indistinguishable local lens properties within their confidence bounds. Furthermore, we showed that any change in the feature positions directly and understandably shifts the corresponding local lens property of the respective multiple image. The impact for other lens reconstruction approaches may not be so simple to single out.
How well the feature extraction performs in further cases remains to be investigated. Yet, as we saw in our manual cross-checks for the two examples considered in this paper, the persistence diagrams that our feature extraction is based on yield the necessary information on occurring anomalies to exclude these cases. In 48 robust feature detections that our examples required, only one needed a manual follow-up due to the importance of that feature which was located in a low signal-to-noise regime. With the results we found, even that case can be automatically treated in the future.
The much more problematic issue is the automation of the feature matching which currently seems impossible in principle to avoid making any a priori assumption about the lens when matching the features. In that sense, every lens may be unhappy in its own way and require manual adjustment in their feature matching but feature extraction, as shown here, is easily automated.
As the automated feature extraction can improve the precision of the inferred local lens properties, maximally over 30% decrease in the confidence bounds for the cases considered here, it may become possible in the future to detect deviations from the wavelength independence of strong gravitational lensing. As estimated in Er & Mao (2014), plasma lensing effects may only shift image positions of the order of tens of milli-arcseconds, so that its impact is negligible at the current resolution of telescopes and particularly at the observation wavelengths considered here. A more likely impact is caused by the dust attenuation affecting the overall observed flux of a multiple image in the blue optical wavelengths and micro-lensing due to granular matter densities close to the critical curve as simulated, for instance, in Diego et al. (2018).
By construction of our approach, the inferred magnification ratios have the smallest confidence bounds and can thus be compared to the measured flux ratios between pairs of multiple images over multiple filter bands. These comparisons allow us to investigate the influence of dust or (transient) micro-lensing on top of the galaxy-cluster scale strong gravitational lensing generating the multiple images under analysis. For the five-image configuration in CL0024 and the triple-image configuration in J2230, all local lens properties were in agreement with each other for all filter bands analysed. A rough comparison to the measured flux ratios for the triple-image configuration in J2230 also showed an overall agreement. While model-based lens reconstructions require a global lens mass reconstruction for the entire area on the sky where the multiple images are located to determine the magnification ratios of the multiple images of interest, our approach constrains them without making any additional assumption about the global lens-mass density profile and therefore requires much less run-time and resources to arrive at the same result. Tests for anomalous flux ratios of multiple images are available in about a second and can be used to inform lens model reconstructions about additional attenuation or micro-lenses or be used as a standalone to scan a lot of multiple image configurations for flux ratio anomalies that may not be explainable in terms of these two known contaminations. For anomalous flux ratios in individual filter bands, deviations from cold dark matter or non-linear lensing effects accumulating deflections along the line of sight can be considered, while deviations between filter bands are expected to be caused by baryonic effects. Table 2 using feature 4t (black lines) and 4b (blue lines) as feature 4 in image 1.
This paper has been typeset from a T E X/L A T E X file prepared by the author. | 2022-07-06T01:15:46.351Z | 2022-07-04T00:00:00.000 | {
"year": 2022,
"sha1": "a1bf1a1cbd7fc5349c637f0ec7d5b1f3dafa89f5",
"oa_license": null,
"oa_url": "http://arxiv.org/pdf/2207.01630",
"oa_status": "GREEN",
"pdf_src": "Arxiv",
"pdf_hash": "a1bf1a1cbd7fc5349c637f0ec7d5b1f3dafa89f5",
"s2fieldsofstudy": [
"Physics"
],
"extfieldsofstudy": [
"Physics"
]
} |
216696013 | pes2o/s2orc | v3-fos-license | A new subspecies of Ottoman viper, Montivipera xanthina (Gray, 1849), (Squamata: Viperidae) from Geyik Mountains, Mediterranean Turkey
A new Ottoman viper subspecies, M. xanthina varoli n. subsp., is described from the higher altitudes of Gündoğmuş (Antalya). The new subspecies differs from the other M. xanthina populations by pholidosis; higher number of intercanthals and lower number of subcaudalia. Also, the whiteness between windings or spots on dorsum pattern were observed in new subspecies, similar to the southern populations. Furthermore, the spots on the ventrals became denser in the mid-body and forms darker colorization at the end of body of the males and the tail tips are yellowish-orange or light orange on both sexes.
Introduction
The Ottoman viper, Montivipera xanthina was firstly described by Gray in 1849 as Daboia xanthina, from Xanthos ruins in Muğla. The specimens collected from Jaffa (Israel) by Müller (1878) included in Vipera genus which was described by Laurenti (1768) and defined as Vipera xanthina. In later years, Boettger (1880Boettger ( , 1888 included specimens from Xanthos in Vipera genus as V. xanthina. Later, the nomenclature was used in systematic studies regarding the species (Werner 1898(Werner , 1914Bodenheimer 1944;Mertens 1952Mertens , 1967Mertens et al. 1967;Baran 1976;Nilson & Sundberg 1981;Nilson & Andren 1986;Nilson et al. 1999) for many years. Werner (1898Werner ( , 1914 first described the new species from Turkey (Southern Taurus, Bolkardağı) and Lebanon as Vipera bornmuelleri, then considered that Eskişehir and İstanbul specimens in addition to Bolkar Mountains (Bolkar dağları) ones as a subspecies of M. xanthina (V. x. bornmuelleri). Bodenheimer (1944) also made evaluations without mentioning the subspecies, and he stated that M. xanthina is found in Western and Central Anatolia, and the one specimen from Giyi (Alanya) on Taurus mountains was included as V. aspis balcanica. However, Mertens (1952) pointed out that the locality of that specimen must be İstanbul and Ecologica Montenegrina 22: 214-225 (2019) This journal is available online at: www.biotaxa.org/em that specimen is defined as V. x. xanthina. Nilson & Andren (1986) stated that with dark skin color and high numbers of ventralia, that specimen corresponded to north population of xanthina.
In detailed systematic studies on both M. xanthina specimens (Mertens 1967;Baran 1976) and the specimens previously identified as V. bornmuelleri (Werner 1898) showed that Turkish specimens were different from Lebanon-distributed V. bornmuellleri. Thus, the nominated subspecies M. x. xanthina was distributed in central, western and southern Anatolia, whereas the subspecies M. x. raddei was in Eastern Anatolia, around Kars. Also, Mertens et al. (1967) defined a new species as Vipera latifii from Iran and mentioned its similarities to raddei.
Based on high level of morphological similarities both subspecies which are previously identified in M. xanthina's distribution area (M. x. xanthina, M. x. raddei and M. x. palaestinae) and M. latifii which were considered as different species before, were evaluated as the Vipera xanthina complex by Nilson & Sundberg (1981). In the same study, venom proteins of the xanthina, raddei and paestinae were compared and, suggested to be accepted as species as first described. In subsequent years, a comprehensive monograph on Vipera xanthina complex was made by Nilson & Andren (1986) and two groups were defined based on the head scalation: xanthina group (V. xanthina, V. bulgardaghica, V. bornmuelleri and V. wagneri) and raddei group (V. latifi, V. albicornuta, V. raddei). In the same study, M. xanthina was evaluated by considering the morphological characters and important intra-species variations were determined and divided into two groups as the northern and southern populations within the distribution area in Western Anatolia without any taxonomic separation.
Taxonomic position of V. xanthina complex were re-evaluated by Nilson et al. (1999) and new subgenus Montivipera was introduced and not only xanthina but also bornmuelleri, bulgardaghica, albizona, wagneri, raddei, albicornuta, latifii and kurdistanica were included in this subgenus. Lenk et al. (2001) supported this new subgenus and in following years Montivipera was considered as a genus by Joger & Nilson (2005). In addition, in Rajabizadeh et al. (2014) they described a new viper species, Montivipera kuhrangica from Central Zagros Mountains belonging to M. raddei complex. This species has a unique color-pattern, high count of head scales and highly projected supraoculars comparing to other taxa of the M. raddei complex.
A study on latest approaches in taxonomy and phylogeny of Near and Middle East vipers by Stümpel & Joger (2009) stated that M. xanthina (s. str.) populations in Western Anatolia has a genetic diversity which was not taxonomically described. Later, by using morphological methods, Cattaneo (2014) described Chios (Aegean Sea) population as M. x. nilsoni and Leros (near Bodrum, Turkey) population as M.
x. dianae. Molecular phylogeny and variation of genus Montivipera is described by detailed study of Stümpel et al. (2016) and four lineages; Aegean and Greece lineage and Lycia and Taurus were determined in "xanthina" group. Even though they are geographically close, it is stated that specimens from Greece and Thracian part of Turkey do not share the same clade. In parallel with Stümpel et al. (2016), Cattaneo (2017) determined morphological differences in Greek Thrace population and described the subspecies M. x. occidentalis. In the Stümpel et al. (2016), all known taxa (except Chios and Leros island populations) were included in the analysis, and the genetic data stated that M. xanthina generate a cryptic species complex with three or four new taxa. However, it is also stated that phenotypical distinction (taxonomical distinction) was not possible due to lack of material and morphological characteristics might be helpful for determining these taxa. The aim of this study is to determine the morphological characteristics of Montivipera xanthina specimens collected from the Geyik Mountains (Gündoğmuş/Antalya), which reveal the differences with the other populations and describe the new population as a new subspecies. (Figure 1a, b). Except for the two male specimens, the remaining specimen found was crushed by a car on the road. Specimens were immediately photographed for color-pattern analysis when they were captured and preserved in 96% ethyl alcohol for further morphological analysis in the laboratory.
Materials and Methods
All specimens have been deposited in the Collection of the Molecular Zootaxonomy Laboratory of Çanakkale Onsekiz Mart University, Turkey. Morphological measurements and the evaluation of pholidosis characteristics were performed according to Nilson & Andren (1986). All the measurements were taken with a digital caliper with an accuracy of 0.01 mm and the characters were measured (Table 1): total body length (TBL), snouth-venth length (SVL), tail length (TL), head length (HL), head width (HW), rostrale width (RW), rostrale high (RH). Other meristic and pholidotic characters; number of ventrals according to Dowling (1951) (V), number of preventrals (PreV), number of subcaudals (SubC), number of anterior dorsal scale row (aDSR), number of midbody dorsal scale row (mDSR), number of posterior dorsal scale row (pDSR), number of apical plates (Ap), interocular row of scales (anterior part) (aInto), number of scales between canthals, supranasal and apicals-intercanthals (IntC), number of all scales between supraoculars (IntSup), number of canthal plates (right) (Canr), number of canthal plates (left) (Canl), first circumocular row (right) (1cirR), first circumocular row (left) (1cirL), second circumocular row (right) (2cirR), second circumocular row (left) (2cirL), shortest distance between supralabials and eye-subocular (Suboc), supralabials (right) (supR), supralabials (left) (supL), sublabials (right) (subR), sublabials (left) (subL); Index calculations were performed according to Baran, 1976: RW/RH x 100 (Rostrale index), TBL/TL x 100 (Tail index). All the measurements and index values were compared with literature (Baran, 1976;Başoğlu & Baran 1980;Nilson & Andren 1986;1992;Stümpel et al. 2016). Diagnosis: This subspecies has similar characteristics with those of the southern populations according to Nilson & Andren (1986). The whiteness between spots and zig-zag windings in the dorsum is more prominent in the new subspecies than in the southern population, and females have darker greyish-brown skin color than males. Whiteness between blackish spots can be also more prominent in male specimens compared to only adult female specimen. In three male specimens, there was almost no spotting and colorization under the head. The spots on ventrals became denser in the mid-body and forms darker colorization at the end of body. Also, just as stated in Nilson & Andren (1986), in three of the four evaluated specimens, they have neck spots united with the dorsal band. The tail tips of all four specimens are yellowish-orange or light orange while tail tip is yellow for other M. xanthina populations according to Nilson & Andren (1986 Figure 2). This specimen was chosen as holotype since the larger male specimen's tip of tail was damaged.
Dark brownish-black spots are present on dorsum on the gray ground-colour. The bands that begins at the back of the eyes have darker brownish-black colorization. Blackish-brown spots are present on the dorsum and occasionally these spots were circled blackish scales. The spots on the dorsum might be separate but they are connected especially at the end of the tails. At the connection region of spots or zig-zag windings on dorsum, a lighter white colorization is present. Large spots that are close the back of head and the first dorsal spot closest to head are not united. A small extension from dorsal spots reaches between head spots. Two small spots are present on the head close to the eyes. At the lateral sides of the body, between the two spots and zig-zag windings, there are brown spots which form a thin line that reaches to ventral. The head and back scales are carinated (Figure 2, 3A).
In male specimens, the dorsum ground-colour is brownish-grey. The spots and zig-zag windings are blackish brown, and darker scales are present around it. These spots can be connected or separated. At the connection points of dorsal spots, there is a lighter white colorization. Two large oblique spots on the backsides of the head are united with back pattern at the neck and form a "Y" shape. Following this Y shape, the spots with the shape of quadrangle, circular or half-moon shaped form zig-zag bands especially in male specimens and turn into a straight line on the tail. On the lateral side, the spots with the same color as dorsal spots form a line towards ventralia. The ventral is yellowish white, and under head it is almost spotless, towards the tail the spots continuously increase but there are almost no spots, or fewer spots under the tail. Less prominent little dark spots make a line on each side of the ventrals. Anal plate has less spots, however between midbody to anal plate spot the density is very high. Towards the tip of the tail, yellowish-orange colorization is present, and it is especially prominent below the tail.
The only female specimen has the brownish-grey dorsal background, and dark black spots and, darker scales around the spots. At the connection points of dorsal spots and zig-zag windings, a lighter whitish colorization can be observed. Dorsal spots and zig-zag windings continues after midbody in a separated manner and forms a line at the tip of tail. Large spots that are close the back of the head and neck are united with the first dorsal spot close to the neck and forms a "Y" shape ( Figure 4A).
Ventral background color is yellowish white and dark spots are present in the sides of ventrals close to the dorsal scales. In some ventrals, the middle sections also have dark spots. The tail has a prominent colorization with yellowish-orange color and very few spots under the tail ( Figure 4B).
Habitat and Distribution:
The specimens were collected in rocky areas with subalpine herbaceous plant vegetation on mountain slope on Mühür Mt. on western of Geyik Mountains, Gündoğmuş/Antalya, Turkey ( Figure 5). Euphorbia nicaeensis is common, and Verbascum sp. is also observed in the area. In addition, Astragalus sp. and Juniperus oxycedrus are also seen. There are steppe areas in the higher parts, and moist areas in the regions close to the valley floor. Other sympatrically living reptiles are; Testudo graeca, Anatololacerta pelasgiana, Stellagama stellio, Eirenis modestus, Platyceps najadum and newly described critically endangered viper subspecies Vipera anatolica senliki. Also, Pelophylax bedriagae and Ablepharus budaki were given as sympatric species in Göçmen et al. (2017). Derivatio nominis: The newly described subspecies were named in honor of Prof. Dr. Cemal Varol TOK who made valuable contributions to Turkish Herpetofauna and also the word "var'ol" is a verb that means "to exist, to stay alive, to be in existence".
Discussion
The new subspecies are similar to descriptions in Baran (1976) and Başoğlu & Baran (1980), however in these studies, the Turkish specimens were considered without a taxonomic distinction. In the previous studies, researchers did not collect samples from the localities where our new subspecies live for the morphological analysis. The one specimen from Giyi (Alanya) on Taurus mountains which was included as V. a. balcanica by Bodenheimer (1944), reevaluated by Mertens (1952) and described as V. x. xanthina, but it was pointed out that the locality of the specimen must be Istanbul. This opinion was also supported by Nilson & Andren (1986) and they noted that the dark background color and high number of ventralia correspond to the northern population of M. xanthina.
As is mentioned in the diagnosis, it is obvious that the new subspecies varoli is distinct from other populations of the Vipera xanthina group by color-pattern and meristic characteristics ( Table 2). New subspecies varoli is similar to the southern populations since females have 27-32 subcaudalia, however since subcaudalia of males of southern populations 30-35, in new subspecies' males are not closely less than 30, the new subspecies has a higher number of intercanthals (scales between canthals, apicals and supranasals) and the slightly white coloration between dorsal spots or zig-zag windings that are mentioned by Nilson & Andren (1986) xanthina for the southern populations are more distinct and also have different in spot or zigzag windings shape (Table 2). Furthermore, the latest molecular studies pointed out that M. xanthina specimens have to be comprehensively evaluated based on their morphology and it is distributed as a cryptic species complex with three or four new taxa by Stümpel et al. (2016). Specimens evaluated in this study were collected from around Geyik Mt as stated in Stümpel et al. (2016) and they belong to Taurus group of xanthina clade since they have the color-pattern showed in that study. Three of four of the specimens showed neck spots united with the dorsal band whereas the ratio of this property is 56% for the southern population in Nilson & Andren (1986).
The new subspecies differs from M. b. bulgardaghica which has a distribution on the eastern part and included in nominate subspecies by Stümpel et al. (2016) by following characteristics: dorsal pattern, [According to Nilson & Andren (1986) dorsum pattern has a dorsal band that is inclined towards the right side or partly round or rectangular with dark spots], apicalia is 2 (3 in bulgardaghica), supralabialia is 10 on both sides (9-9 in bulgardaghica), ventralia is 157 in males and 160 in the female (bulgardaghica has lower numbers: 150-154 in males, 145 in females). In conclusion, we believe that the examined species should be considered as a separate taxon based on color-pattern and pholidolial features differing from known M. xanthina populations. | 2020-04-23T09:12:57.358Z | 2019-10-15T00:00:00.000 | {
"year": 2019,
"sha1": "c60a7271abdaba3aefcd1000919831994e7f0aca",
"oa_license": "CCBY",
"oa_url": "https://doi.org/10.37828/em.2019.22.17",
"oa_status": "GOLD",
"pdf_src": "ScienceParsePlus",
"pdf_hash": "83f31f4b9e0325e67e419b66be0588629735a227",
"s2fieldsofstudy": [
"Biology",
"Environmental Science"
],
"extfieldsofstudy": [
"Geography"
]
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.