Servicio de Información Comunitario sobre Investigación y Desarrollo - CORDIS

FP7

TOXBANK Report Summary

Project reference: 267042
Funded under: FP7-HEALTH

Final Report Summary - TOXBANK (ToxBank – Supporting Integrated Data Analysis and Servicing of Alternative Testing Methods in Toxicology)

Executive Summary:
ToxBank established a dedicated web-based warehouse for toxicity data management and modelling, a „gold standards‟ compound database and repository of selected test compounds, and a reference resource for cells, cell lines and tissues of relevance for in vitro systemic toxicity research carried out across the FP7 HEALTH Alternative Testing Strategies SEURAT-1 program (http://www.seurat-1.eu/). The project developed infrastructure and service functions to create a sustainable predictive toxicology support resource going beyond the lifetime of the program. Specifically, based on an extensive requirements gathering from all SEURAT-1 consortia and an analysis of this data, a production version of the ToxBank data warehouse has been implemented that provides access to all experimental, processed data and protocols alongside relevant public information (https://services.toxbank.net/toxbank-ui/login). This includes the development of web-based interfaces for linking and uploading data, including raw and processed data, and model results. All steps of any experiments are linked to protocols describing the procedures. A web-based user interface for searching, browsing, and filtering the results has been implemented to provide access to all protocols and data across the cluster in a way that is sensitive of any intellectual property restrictions on access. Public access to the ToxBank data warehouse was established in Autumn 2014. Public data on reference compounds was incorporated into the warehouse supporting meta analysis and risk assessment being carried out on the SEURAT-1 case studies. A collaborative ToxBank Gold Compound database was established using the MediaWiki platform (http://wiki.toxbank.net/). It has been populated with a set of ca. 50 reference compounds, including information about chemical identities, adverse effects, toxicity mechanisms and therapeutic targets. These compounds were used as reference compounds by the SEURAT-1 cluster.
ToxBank Tutorial and Workshop proceedings have been captured and integrated into the ToxBank website (http://www.toxbank.net/). In 2015 ToxBank partners agreed to the continuation of ToxBank as a sustained OpenTox resource (http://www.opentox.net/) beyond the end of the project, including its extension and support of the EUToxRisk program (http://www.eu-toxrisk.eu/).

Project Context and Objectives:
The primary goal of the FP7-HEALTH-2010-Alternative-Testing HEALTH.2010.4.2-9 program (SEURAT-1) was the development of human safety assessment strategies which may be used to replace, reduce or refine (3Rs principle) repeated-dose systemic toxicity testing historically carried out in animals. The target of ToxBank under the SEURAT-1 project pillar was the establishment of a dedicated web-based Data Warehouse (DW) for toxicity data management and modelling, a “gold standards‟ compound database and repository of selected model compounds, and a bank for cells and cell lines, including stem cells, and tissues of relevance for in vitro toxicity testing.
The objectives were the following:
a) Establishment of a dedicated web-based data warehouse
The ToxBank Data Warehouse (TBDW) has established a centralised compilation of information and data for systemic toxicity. Links to relevant public databases are provided for data import. All projects under the SEURAT-1 program uploaded their raw and processed data into the TBDW as soon as these become available. Data generated were analysed and the outcome integrated whenever possible into computerised models capable of predicting repeated-dose toxicity. The TBDW has been organised so that it provides a sustainable source of information for toxicological research going beyond the life-time of the research projects, through sustainability and business planning carried on throughout the project, including discussions with industry groups such as Cosmetics Europe to meet their requirements
b) Establishment of a database of selected model compounds
The ToxBank Gold Compounds were selected to meet the highest quality standards. Chemicals in the database have been selected based on the scientific literature, including review of mechanistic publications, high-quality repeated dose toxicity in vivo data from animal studies, and adverse event and epidemiological data from humans. The database covers cosmetic ingredients, industrial chemicals, and pharmaceuticals that meet high-quality selection criteria for reference compounds. In addition, a list of selected model compounds, standard operating procedures for data quality control, processing and analyses have been provided. Whenever compounds are needed for training or validation purposes, they can be selected from the Database in correspondence with targeted mode of actions in systemic toxicity. An extensive information resource on reference compounds was created for public dissemination through the ToxBank wiki (http://wiki.toxbank.net/).
c) Establishment of a repository for the selected model compounds
The ToxBank Chemical Repository ensures the availability of test chemicals accompanied by analytical quality control procedures to the research projects on the program. We have investigated service models so that the Chemical Repository could be maintain beyond the end of the program to provide ongoing services to further toxicology research and validation programs.
d) Setting up of a cell and tissue bank for in vitro toxicity testing
An important service to European scientists is the formation of a bank of cells, cell lines (including stem cells and stem cell lines) and tissues to be used in the program projects and beyond the end of the program. A Biomaterials information resource was developed and incorporated into the ToxBank wiki.

Project Results:
WP1 Data Warehouse

T1.1 User requirements gathering
Approach to collecting requirements
The ToxBank consortium used a methodology referred to as contextual inquiry/design for the collection of user information, from visits and direct interviews with many SEURAT-1 partners, to use in developing the system requirements for the TBDW . An analysis of this data resulted in the user requirements for the TBDW. This included the need to develop a solution to manage, register, assign a status, comment, peer review, and sensitively share the diverse protocols being generated throughout the cluster as a high priority. Handling data presented a number of complex problems as a result of the diversity of the experiments being employed as well as the different workflows currently being adopted across the cluster. It was seen as essential that each step of an investigation be documented with a protocol and annotated with the resulting data, both the original results and any subsequent processed data. Providing information on the cells, reagents, and compounds was also highlighted as an important activity. It should be possible to search and download any protocols or investigation data.

To support the second phase development (integrated data analysis) ToxBank partners visited a further 10 SEURAT-1 partner sites and conducted interviews with ca. 20 investigators. The new requirements included the ability to precisely search for significant up or down regulated genes or proteins. In addition, chemical structure searching (exact, substructure and similarity) was added to support use cases such as read across. To understand how individual investigations, from a list of search results, were performed and what sort of data is available, a dashboard was implemented to summarize multiple investigations. This information allows users of ToxBank to understand both the experimental factors that were the basis for the investigations, but also the parameters and technologies used in producing the data. Data from the selected investigations can then be exported in a standardized way to enable combining data from different experiments. Since many tools were being used to perform data analysis and visualization across the cluster, it was decided not to replicate or build any of these specialized tools, but to enable their use through a variety of data export options. Specifically, linking the information to pathway tools is important to help understand the biological context.

T1.2 Design of system architecture
The TBDW architecture consists of a set of web services, providing access to protocols and data, a search service, and a Web GUI application, offering user-friendly access to the above functionality. ToxBank currently adopts the OpenTox framework design and incorporates the REpresentational State Transfer (REST) software architecture style, a formally defined common information model, based on the W3C Resource Description Framework (RDF) and authentication and authorisation based on OpenAM. All TBDW component and web service developments in this reporting period were based on these previously selected technologies.

T1.3 Incorporation of Systemic Toxicity Ontology
The ToxBank consortium created a keyword hierarchy that will be used in the TBDW. In addition to its use is facilitating collaborations, the keyword hierarchy is used to support searching, browsing and linking of resources within the warehouse. When information is uploaded into the TBDW, terms are selected from this hierarchy and linked to protocols and investigation datasets in the warehouse. The keyword hierarchy is currently organized into six main branches: biomaterials, investigative techniques, data and readouts, adverse events, modes-of-action, and gold compound standards. In addition, the TBDW continues to incorporate existing life-science related ontologies through the use of ISAcreator data entry tools that create ISA-tab compatible archives, where concepts (e.g. cells, experimental unit, and so on) in the data are linked to existing ontology terms, and the continued development and extension of a cross-SEURAT-1 keyword hierarchy.

T1.4 Implementation of the Data Warehouse through a suite of distributed data marts
To handle all of the data generated in the diverse investigations being performed across the cluster, as well as relevant public data, ToxBank adopted the ISA-TAB universal data format to represent the experiments, including the toxicity studies, any chemical analysis, and omics experiments. Data access and upload procedures are defined by the Investigation API. Data is uploaded in ISA-TAB format; data queries are performed with the SPARQL query language. REST operations are available for accessing individual investigations, studies, assays and data files. Work has been ongoing throughout the project on extending the data warehouse framework to support data uploading and processing.

T1.5 Implementation of Data Warehouse operations using web services
The warehouse has a series of operations for managing protocols, data, and searching. These operations have been implemented as a set of distributed web services that make use of existing OpenTox APIs. The specific services implemented are based on the needs of the cluster, as defined in the requirements described earlier. ToxBank’s REST resources are instances of the relevant RDF classes (e.g. https://services.toxbank.net/toxbank/protocol/SEURAT-Protocol-104-1 is an instance of the tb:Protocol class). ToxBank puts special emphasis on data confidentiality. The Authentication & Authorization infrastructure (AAI), in particular, builds upon what has already been developed and well tested in the OpenTox project and strives to further enhance it. Searching within the ToxBank system is provided as a separate web service that is deployable within an existing web container or as a stand-alone application. It was developed using Java and various open source technologies including Restlet and elasticsearch . The Search service is primarily accessed by the Protocol, Investigation, UI services and desktop applications such as ISAcreator . When protocol or data resources are uploaded, the corresponding service notifies the Search service that a new resource is available. The search service then retrieves the resource and makes it available for indexing.

To support the precise searching of omics and other data, the isa2rdf tool has been extended to support conversion of microarray, mass spectrometry and protein assignment data files to RDF/XML in addition to converting the ISA-Tab metadata. The RDF representation of the data files is based on an extension of the OpenTox Dataset RDF representation and each data file item is linked to the relevant sample, described by the ISA-Tab metadata. The conversion to RDF is performed transparently for the user, who uploads an ISA-Tab archive. The server preprocesses the archive, using ISA-Tab validation and isa2rdf and imports the triples generated into a triple store . Once all the information is available in the triple store, the relevant queries are defined as a set of predefined SPARQL queries and exposed as REST services in the general form of /investigation/sparql/{template_name}. The approach described allows a common data model for both metadata and data files, which is independent of a database technology.

To support searching by chemical structures, a dedicated ToxBank instance of OpenTox compliant web services from IdeaConsult was installed on a ToxBank server. The content is updated on demand through the OpenTox dataset API. An ISAcreator plugin allowing to query the chemical structure services was developed . It allows to search for chemical compounds (by identifier, similarity or substructure search) within ISAcreator and links the experiment metadata with the chemical structures in TBDW.

T1.6 Development of graphical user interface
Web-based graphical user interfaces have been designed, customised and implemented for loading the data into the TBDW as well as accessing the information and model results generated.

Data Entry
A series of forms-based user interfaces have been developed and/or customised for loading experimental data and related descriptions of experimental protocols. To collect investigation data in a consistent manner across the cluster, the ToxBank consortium selected to use ISAcreator, an open access tool (http://isatab.sourceforge.net/isahelp/ch03.html). ISAcreator provides a graphical user interface to create a consistently recorded series of data files that include the experimental design and information concerning the overall investigation, information on the experimental steps linked to both protocols as well as raw or processed data files.

Data Access & Analysis
The protocols and investigation housed in the TBDW are available from the web-based user interface. This GUI is a front-end user interface for the repository services defined by the ToxBank API. It is a standalone web application allowing users to log in, search and review existing protocols and investigations, and to upload new protocols and investigations. The interaction between the users’ web browsers and the ToxBank UI server relies on standard HTML/CSS/Javascript content, generated dynamically within a Java-based web application framework (the Play Framework).

A series of user interfaces to support chemical searching as well as searches on genes and proteins have been developed. The chemical structure search user interface allows structure queries to be specified in order to perform an exact, substructure or similarity search. The query molecule is defined as either a SMILES string, a MOL file or is drawn within an integrated structure drawing editor. The search queries all investigations with chemical structures that are defined as experimental factors in ISA-Tab. A biomaterials search window was developed that allows a ToxBank user to define a precise query for different biological materials. An additional option to display a dashboard for a selected set of investigations is also provided. The new dashboard is being developed to help understand a specific list of investigations in terms of what experimental factors were considered, what was the source of the biological material, as well as what protocols, technologies, and endpoints were used. From these investigations, it is possible to download and combine specific data to use in external bioinformatics, chemoinformatics, advanced data analysis or visualization software as well as data mining applications .

T1.7 Integration of Tools for Data Analysis, Mining and Model Building
To support an integrated view of the derived or processed data generated from experiments across the SEURAT-1 cluster as well as outside the cluster, ToxBank uses preconfigured templates for assay metadata (as part of the ToxBank customised ISAcreator distribution) and proposed a standard file format for processed data. In this proposed standard, each type of experiment (e.g., transcriptomics, proteomics, and so on) has a different file format. The file containing this processed data is uploaded as part of an ISA-Tab archive (containing the experimental design, raw data, and links to the protocols) and can be used in ToxBank to support precise searching (e.g. identify all investigations where a specific gene has a fold change greater than 1.5) as well as a consistent integrated analysis of the data over the entire cluster. This standardization also supports effective integration with data analysis, data mining and model building applications.

T1.8 Data Warehouse Support Facilities
ToxBank has been supporting the preparation and upload of protocols and data into the data warehouse. To support the upload and use of the ToxBank data warehouse, a series of on-line tutorials have been generated. These include lectures on background material such as bioinformatics and tutorials to support the formatting of data, the upload of protocols, reports, and data as well as how to search, analyze and download information from the data warehouse. These tutorials are available through the toxbank.net website.

T1.9 Operation and Support of ToxBank Warehouse during the SEURAT-1 program
An interface to the COSMOS database via an API was developed and implemented in addition to approaches to integrate information on external resources with SEURAT-1 generated information. One of these projects was to integrate the ToxCast and Tox21 data with the SEURAT-1 data to support a meta-analysis of the combined information.

T1.10 Scientific Coordination of WP work activities and their interaction with other WPs and all other related program activities
A number of tools have been adopted to support this task: Mantis, ToxBank wiki web-pages, Google Docs and the Jenkins integration server. Weekly technology focused meetings were held to discuss the ToxBank technical progress and the development and supported through ToxBank wiki web-pages. The web-pages developed outlining the APIs are made available publicly for other developers to build interfaces using the OpenTox/ToxBank web services.

T1.11 Sustainability Planning for resources developed within WP
The TBDW has been organised to provide a sustainable service for toxicological research beyond the lifetime of the research projects and resources set-aside such that the public reference data will be available for at least five years after the completion of SEURAT-1 as an OpenTox resource.

WP2 Compound Database

T2.1 Establish Selection Criteria for ToxBank Gold Compounds
ToxBank was tasked to create a quality-controlled curated cheminformatics database for Gold Compound reference standards that can be used in the training and validation of in vitro assays and in silico models. Gold Compound selection criteria and standardized curation and operating procedures were established in the first half of year 1 in order to support the earliest possible creation of the database and its use in decision making on project and assay design across the SEURAT program. The SEURAT initiative addressed hepatic, cardiac, renal, neuronal, muscle, and skin toxicities, with the largest effort directed towards hepatotoxicity. The 6th Framework EU project, LIINTOP, identified the major hepatotoxicities of interest for development of in vitro testing methods. These included mitochondrial impairment, oxidative stress, apoptosis, and the lipid disorders of steatosis, cholestasis, and phospholipidosis (Gomez-Lechon et al. 2010). We therefore selected standards that are relevant to these pathologies, with the addition of specifically representing fibrosis with respect to cytotoxicity.

The SEURAT strategy encompasses “any substance”. Thus, the reference compounds must cover a breadth of chemical classes that include cosmetic ingredients, agricultural and industrial chemicals, and pollutants, among others. Therefore the core criterion for acceptance was established to be a promiscuous mechanism of toxicity. While “mechanism” may refer to several possible aspects of the overall mode of action of a toxicant, compound selection was based narrowly on the molecular initiating event for the adverse outcome pathway. Promiscuity refers to lack of structural specificity in ligand binding, where the specificity may relate either to ligand or receptor structure. The concept derives from the observation that small, hydrophobic ligands tend to have binding affinity for multiple different proteins, and conversely, receptors with large hydrophobic pockets tend to bind multiple diverse ligand structures. Promiscuity is clearly relevant to a strategy that must span a broad chemical space. The following were identified as basic promiscuous mechanisms of toxicity: alkylation, redox, membrane disruption, and binding to (nonselective) nuclear hormone receptors. Additional standards were selected to represent especially well-characterized initiating events for the targeted pathologies.

To the mechanism-related criteria were added general criteria such as stability, solubility, and availability. The selection criteria were initially published in the deliverable D2.1, and are available at the ToxBank Gold Compounds Wiki site: http://wiki.toxbank.net/w/index.php/Selection_Criteria.
Initial criteria also comprised non-idiosyncratic toxicity and relevance to repeated dose toxicity, but these criteria proved difficult to relate to objective standards of selection. There is no agreed frequency of occurrence that defines “non-idiosyncratic”, for example; and these criteria had only secondary impact.

T2.2 Mine ToxBank Data Warehouse and select ToxBank Gold Compounds
In the original ToxBank conception, identification of the Gold Compounds was to be accomplished via computational data mining. Once the SEURAT consortium was constituted, however, it became apparent that only a small number of reference compounds was required and that manual selection and extensive curation were preferred.

An initial set of 24 compounds was selected and published in 2012 along with information about chemical identities, adverse effects, toxicity mechanisms and therapeutic targets. This compound set completed the selection of hepatotoxin reference standards. The list was reported in D2.2 and formally submitted to the cluster and approved at the 2013 annual meeting.
In 2013, the full compound set was completed by the addition of the full list of renal-, cardio- and neurotoxin standards. The complete list of compounds is available at wiki.toxbank.net/wiki/CompoundSummaryTable. Data associated with the Gold Compounds, especially the hepatotoxins, comprised in-depth textual analysis of the literature on mechanism of action.

T2.3 Verify availability and stability of selected ToxBank Gold Compounds
Selection criteria for the chemical standards included an evaluation of the highest purity available on the market, the reliability of information and traceability, suitability for cell culture application, ease of access and shipment availability in the different countries involved, and lot size and price. Standard compound suppliers and product numbers were provided to ensure that all labs were using a common compound source. Information on the stability of the chemicals was collected from the literature, and additional calculated biodegradation and metabolism properties were provided as a component of a data table that was constructed for each compound.

T2.4 Establish Standard Operating Procedures for data quality control, acceptance, processing and analyses of selected ToxBank Gold Compounds
The compound selection strategy evolved in consultation across the SEURAT-1 cluster. Each project team, the Scientific Expert Panel (SEP), and Colipa (now Cosmetics Europe) provided representatives at a kick-off meeting in Cascais, Portugal in February, 2011. An advisory Gold Compound Working Group with 18 members was assembled from the attendees at the Cascais meeting, and an evaluation team of 10 scientists was assembled from the SEP, industry, and academic labs to serve as the working Gold Compound Selection Team to evaluate specific compounds for acceptance. As a matter of process, it was agreed that compounds recommended as standards require unanimous agreement by the evaluation team and would be submitted to the working group for review and comment before being accepted as Gold Compound standards.

The selection strategy that emerged from the Cascais meeting was defined with respect to adverse events such as steatosis and cholestasis. The subsequent evolution to an MOA-based approach was endorsed by the SEP and developed in a series of monthly teleconferences with the Gold Compound Working Group starting in early August 2011. Explicit consideration of repeated dose toxicity was initiated at a meeting of experts organized by COACH in Ispra and a subsequent meeting with the SEURAT cardiotoxicity team in Cologne in November of 2011.

The final compound list was formally submitted to the Scientific Expert Panel and accepted in February of 2013. Compound selection efforts culminated in 2014 with the publication of a peer-reviewed review of the hepatotoxin reference compounds. This review assimilated information from the ToxBank Wiki data tables for individual compounds into an integrated discussion of mechanisms of toxicity and provided for peer review of the strategy and results.

T2.5 Build ToxBank Gold Compound Database including data import services and maintenance for SEURAT-1 program data passing curation evaluation and acceptance criteria
--And--
T2.6 Populate ToxBank Gold Compound Database with existing high-quality repeated-dose toxicity data from animal and human studies
Although it was originally envisaged that the Gold Compound Database would comprise a relational database, it was agreed with the Gold Compound Working Group in 2011 that a textual discussion of the rationale for selection of the compounds and their relevance to assay validation was preferred. Accordingly, a standard table of acceptance criteria was established and the data supporting compound acceptance was published as the ToxBank Gold Compound Wiki (http://wiki.toxbank.net). The wiki format was selected because it supported incremental, rapid publication of compound data tables as they were developed and approved by the working group. This wiki ultimately became the primary repository for an extensively curated compilation of reference information on each Gold Compound.

Data associated with the Gold Compounds that was generated internally by the SEURAT cluster was then collected within the ToxBank Data Warehouse as described under the final report for WP1. Details about data import, maintenance, support, and curation for the Warehouse were reported in Deliverables 1.1 and 1.2. The core principle was to adopt existing standards and solutions, in particular OpenTox infrastructure and semantic web technologies (W3C Resource Description Framework - RDF3) to create an underlying database structure and then to provide data import and expose the data and functionality through web services. Data integration efforts were enhanced by establishing a database structure based on the ISA-Tab standard for uploading new biological data for to the Warehouse. This standard was proposed and adopted for use across the cluster. Finally, links from the Warehouse to external data sources were established as described below under T2.7.

T2.7 Establish procedures and analytical tools for selection of compounds for in vitro R&D programs, QSAR model building and training, integrated testing strategies, and model development and validation towards regulatory acceptance under REACH and the Cosmetics Directive
In the original ToxBank conception, identification of Gold Compounds was to be accomplished via computational data mining. Once the SEURAT consortium was constituted, however, it became apparent that only a small number of reference compounds was required and manual selection and curation were preferred. Therefore, the ToxBank data mining and analytical capabilities were redirected to two outcomes.

For the first outcome, efforts concentrated on supplementing the literature data in the Gold Compound Wiki with computed properties for the compounds. These properties were primarily physical properties such as solubility and predictions of elements of pharmacokinetics and can be accessed at the wiki (http://wiki.toxbank.net).

For the second outcome, efforts concentrated on integrating the ToxBank Gold Compound Warehouse with public databases and demonstrating proof of principle for data mining the linked data sources. These capabilities were then applied to developing methods for computationally relating chemical structure to mechanism of action in order support data mining efforts that address this future demand.

In order to support development of data mining tools, the Gold Compounds were associated with external mineable database collections in 2013, including the well-known TG-Gates and assay data from PubChem as well as the COSMOS database of the SEURAT cluster. The WP2 team then demonstrated application of analytical methods for Read Across, enriched meta-analysis of multiple omics and functional data, background knowledge from GO ontologies and Kegg pathways, and pathway visualization for SEURAT-1 Gold Compounds.

In 2014, links were extended to the Munro (2006) database of LO(A)EL and NO(A)EL data and HESS (Hazard Evaluation Support System) database from the OECD QSAR Toolbox and the EPA’s Integrated Risk Information System (IRIS) database. Tools were then created to mine these databases to identify compounds chemically similar to the Gold Compounds, retrieve biological data for these compounds, and then predict biological targets for new compounds of interest, for example, in Read Across predictions and Adverse Outcome Pathway predictions. These tools and proof-of-principle for their employment are exemplified by analysis of ToxCAST/Tox21 Phase II data for potential hepatotoxicity and analysis of the Munro, HESS and IRIS database for prediction of hepatotoxicity, prediction of repeated dose toxicity, and prediction of developmental and reproductive toxicity (see D2.4 and D2.5 reports).

T2.8 Define and enforce procedures for project-based and long-term ToxBank Gold Compound data submission, curation and maintenance, and support and training for ToxBank Gold users
As stated above, the primary relational database for the Gold Compounds is the ToxBank Warehouse, which was created and supported under WP1 and stores experimental data produced across the SEURAT cluster. Details of data IO processes, maintenance, and support are provided under the WP1 final report. Data submission was originally supported via the ISA-TAB standard for representation of final, curated data, and support for submission of preliminary, unformatted data was added in 2015.

T2.9 Operation and Support of ToxBank Gold Compound Database during the SEURAT-1 program
Since the primary Gold Compound data repository under WP2 was the Gold Compound Wiki, the primary effort was expended in collating the data that supported selection of compounds as reference standards, and building the wiki. Subsequent maintenance and support requirements were minimal. The successful support of the wiki is exemplified by the following graph of the number of hits on the wiki across 2014 and 2015.

Support requirements for the Gold Compound Data Warehouse for experimental data was extensive, however, and exemplified by a series of detailed video tutorials which are reported under D2.5. Further details are available from the final report for WP1.

A summary of the ToxBank gold compounds and associated datasets is provided in Table 1. In addition to SEURAT-1 datasets, discussions with the US EPA during the program resulted in the additional generation of ToxCast and Tox21 assay datasets, which were incorporated into ToxBank.
WP3 Compound Repository

The main objectives within WP3 were the establishment of a physical repository of the test chemicals used, and the detailed characterisation of the chemical structure and relevant physico-chemical properties (including predictions on stability and binding properties) of the chemical used. We developed analytical methods for critical unstable test compounds, and the stability measurement under experimental conditions of doxorubicin, tamoxifen, amiodarone, bosentan, chloropromazine, valproic acid (and its liver metabolites), methotrexate and piperonyl butoxide. Particular attention was given to the measurement of the actual concentration of the substances within the in vitro assays. The measurement was made by LC-MS/MS in order to determine their stability during in vitro testing, at different time points. Furthermore, activities during the whole duration of the project aimed to direct the development of standard operating procedures (SOPs) for the test protocols and education on Good Chemical and Cell Culture Practice (see WP4).

A physical repository for test chemicals used within SEURAT-1 was established at Mario Negri Institute, Italy. Analytical methods were developed for doxorubicin, tamoxifen, amiodarone, bosentan, chloropromazine, valproic acid and its metabolites, methotrexate and pyperonil butoxide. Extraction procedures from cell culture medium were optimized for tamoxifen, amiodarone, bosentan, chloropromazine, valproic acid and its liver metabolites, methotrexate and pyperonil butoxide in order to measure their concentration during in vitro testing at different time points. Protocol guidelines were developed and uploaded in the ToxBank Data Warehouse providing definition, information on content, advices for the compilation, uploading and sharing within SEURAT-1 of research protocols and standard operation procedures.

Task 3.1. Establishment of a physical collection of test chemicals used within the SEURAT projects
The prime objective achieved within WP3 was the establishment of a physical repository for test chemical to be used within SEURAT-1. It addressed the availability and detailed characterization of the chemicals used within SEURAT-1 and was strictly related to the Data Warehouse established within ToxBank. According to the statement in the Description of Work the “Compound Repository” was established as a dedicated freezer at the partner institute in Milano (Mario Negri Institute), who held the responsibility for its maintenance and the associated chemical analysis facilities.

Task 3.2. Definition of a quality assurance framework for data inclusion on test chemicals to be used within SEURAT projects
A quality assurance framework for data inclusion on test chemicals to be used within SEURAT-1 was defined. This work was strictly related to the extensive assessment performed in WP2 for the selection of Gold Compounds for testing, since data on physico-chemical properties, purity and stability were gathered for the evaluation of the suitability of the standards in term of physico-chemical behaviour in test conditions.

The collected properties were those required for the characterization of the Gold Compounds proposed within WP2: testing within the US EPA ToxCast/Tox21; structure and isomeric form; stability to storage, light, freeze thaw; buffers and water solubility; dimethylsulphoxide solubility; binding to plasticware; availability commercially with highest purity; volatility.

The sources which were considered reliable and used were: scientific literature, official databases, material certification, information sheets and specification provided by the standard’s suppliers and US EPA. For data confirmation, multiple sources were considered and compared contemporaneously, whenever available. For particular properties, such as water solubility and volatility, if data were not available, they were calculated through commercially and freely available software for modelling, as planned in the Description of Work. The used software programs were ACD Labs (Advanced Chemistry Development, Inc., Canada) for the calculation of solubility and Episuite (EPA, USA) for the calculation of vapour pressure for volatility evaluation. Quality principles were established and followed in order to guarantee the reliability and completeness of the information collected.

All the data were reported in a standardized form (scientific unit, conditions, etc.) along with specification on how the data were generated (testing parameters and conditions, calculation methods) and with references.

Task 3.3. Development and distribution to the participants of the SEURAT projects of procedures for chemical handling during in vitro testing
The procedures for chemical handling during in vitro testing are strictly related to the specific test conditions (e.g. solvent, pH, temperature, duration, reactive conditions, other factors) and then to Standard Operating Procedures (SOPs) which describe how an experiment is performed. SOPs were developed within SEURAT-1 and published and shared through the ToxBank Data Warehouse. During the first year of the project the Data Warehouse was constructed within the ToxBank project and the SOPs were developed and applied within the cluster projects. Instructions on the correct handling of chemicals was transmitted through the indications reported in the “Stability” part of the “Physical properties” and in the “Storage” part of the “Recommended product and source” section on the wiki, where there is available specific information on the stability to light, storage, freeze-thaw, pH sensibility, etc.

Task 3.4. Data Warehouse entry of relevant information on test chemicals including source, purity, structure, isomers, and tautomers
The entry of the relevant physico-chemical information on already selected test chemicals in the Data Warehouse was achieved through the organization of these properties and those evaluated within WP2 in a wiki resource which will be linked to the Data Warehouse (ToxBankWiki, http://wiki.toxbank.net). The data are reported in the “Physical properties” section of each compound’s table in the wiki. For each chemical all the properties collected for the evaluation of the criteria defined in WP2 are reported, including source, structure, purity and specification of the isomeric form.

Task 3.5. Calculation of stability of test chemicals
The calculation of stability for the Gold Compounds accepted has been made and all the results were reported in the deliverable D3.2 “Entry of structures, physico-chemical, stability and binding properties of test chemicals”, due for M24. The calculation of biodegradation and metabolism properties was performed with the following software: Episuite (EPA, USA), Topkat (Accelrys, USA), MetabolExpert (CompuDrug, USA). These programs allow the estimation of several properties related to stability: atmospheric oxidation, Henry’s law constant, melting point, boiling point, aerobic and anaerobic biodegradability, aqueous hydrolysis rate constant and half-lives (only for particular classes of chemicals), metabolites which may be formed in humans, animals or through photodegradation. These results contributed to the complete characterization of the Gold Compounds and to the evaluation of possible problems of stability in specific test conditions.

Task 3.6. Measurement and calculation of binding properties of test chemicals
The calculations on binding properties were performed on the accepted Gold Compounds and the results was reported according to the Description of Work in deliverable D3.2 “Entry of structures, physico-chemical, stability and binding properties of test chemicals”, due for M24.

Task 3.7. Development and distribution to the participants of the other SEURAT projects of a template for preparing in vitro Standard Operating Procedure (SOP), education on GCCP, and continuous revision of SOPs
Partner IRFMN established contacts with the EC JRC, which is formalizing and disseminating the template for alternative methods within SEURAT-1 and also within a broader initiative of the EC. Partner Leadscope has been also actively involved in this task in order to harmonize the procedures used by SEURAT-1 partners and upload in the Data Warehouse within ToxBank.

Telephone conference meetings and informal discussion with partner HPA-UKSCB led to development of a two tiered approach to protocol capture development of the terminology of “Research Protocol” for laboratory protocols provided by other SEURAT-1 partners and “SOP” where the protocol is produced in a standard form with a minimum data set and also supported by appropriate controls and qualifying data to demonstrate effectiveness of the SOP.

Partner HPA-UKSCB worked closely with Partner JRC in the first six months to establish a review of quality control issues and potential markers and tests which was published and selected as the journal editors’ selection when published in the January 2012 edition of Expert Opinion On Drug Metabolism and Toxicology (F. Pistollato, S. Bremer-Hoffmann, L. Healy, L. Young and G. Stacey. (2012) Standardisation of pluripotent stem cell cultures for toxicity testing, Expert Opinion On Drug Metabolism and Toxicology, 8(2):239-57. Epub 2012 Jan 17). Partner HPA-UKSCB gave a presentation on the principles of Good Cell Culture Practice as part of the main SEURAT-1 scientific programme and provided a training session on stem cell culture for ToxBank partners at the SEURAT-1 first scientific meeting in Cascais, March 2011.

At the SEURAT-1 February 2012 scientific meeting Partner HPA UKSCB led a workshop on Quality Control for Stem Cells, which has been reported to COACH and is being developed as a cross-SEURAT-1 cluster activity. In addition, HPA-UKSCB has collaborated with the Karolinska Institute ToxBank partner to provide a Good Cell Culture Practice training session at the SEURAT-1 “Summer School” at IBET in Oeiras, Portugal, in June 2012. Partner HPA-UKSCB in collaboration with partner Leadscope, developed guidelines for preparing research protocols and Standard Operating Procedures, which is available to all SEURAT-1 partners in the ToxBank Data Warehouse.

Task 3.8. Development of analytical methods for critical, unstable test chemicals and
The development of analytical methods for critical, unstable test chemicals and the following measurement of stability properties of test chemicals were the tasks planned to start in a second part of the project course, in particular from months 19 and 25 respectively. Experiments were performed to determine the actual concentration of tamoxifen, amiodarone, bosentan, chloropromazine, valproic acid and its liver metabolites, methotrexate and piperonyl butoxide present in cell culture medium after different treatment times and in controls. The analyses were performed to support the request by a cluster project partner from NOTOX project. For this reason a specific analytical method based on acetonitrile extraction followed by centrifugation was developed and optimised for the extraction of each chemical from cell culture medium and the subsequent determination was performed by electrospray ionization liquid chromatography tandem mass spectrometry. Additionally, a chromatographic based method for the quantification of doxorubicin was developed on tandem mass spectrometry.

Task 3.9. Measurement of stability properties of test chemicals
The developed LC-MS/MS analytical methods were applied to the determination of the actual concentration of tamoxifen, amiodarone, bosentan, chloropromazine, valproic acid and its liver metabolites, methotrexate and piperonyl butoxide in cell medium samples from acute and long term toxicity studies on HepaRG cells. Analysis was performed both on medium incubated with cells and without cells to estimate the stability of the drug in cell medium under experimental conditions and to evaluate cellular uptake. Furthermore the stock solutions of the drug in cell medium maintained in the fridge have been analyzed. Valproic acid concentration was also measured in samples from different cultivation conditions: 2D and 3D HepaRG cultures; methotrexate and piperonyl butoxide was measured in samples from 3D HepaRG cultures. The measured concentration was compared with the nominal concentration; the partial loss of the active compound was related to different processes such as adsorption to containers, degradation, evaporation and absorption into cells.

Cells were exposed to the drugs dissolved in the culture medium for acute and long term (28 days) toxicity experiments, in which the medium was renewed every 48 h. Cells were treated with concentration in the micromolar range except for valproic acid which was tested in millimolar concentrations. After 48 h of incubation at 37°C, supernatants were collected in triplicates into 0.5 ml plastic reaction tubes and centrifuged for 10 minutes at 13k rpm (4°C). They were then transferred into glass vials with micro inlets and frozen at -20°C for storage. Before shipment from the NOTOX partner to ToxBank partner IRFMN, the samples were further thawed and aliquots were prepared. The samples for time points for days 6, 14, 20 and 28 were analyzed for chemical quantification, except for methotrexate and piperonyl butoxide where the experiment was a long term (21 days) toxicity experiment on 3D HepaRG cultures. Supernatant and cells were transferred in ice cold plastic tube and then the separation of supernatants from cells was obtained by sedimentation and consecutive relocation of supernatant. The calculated limits of detections for the developed analytical method were 0.001 ng/µl for tamoxifen, 1 nM for amiodarone and bosentan, 8 nM for chlorpromazine, 0.2 µM for valproic acid, 0.021 µM for piperonyl butoxide and 0.0042 µM for methotrexate.

Results showed that the analyzed compounds had different stability in the experimental conditions examined. Amiodarone was detected only in the acute toxicity test samples, when the tested concentration was higher than 10 µM, the measured concentrations were 1000 times lower than expected. The critical stability of amiodarone is probably related to its low solubility and binding properties towards plastic materials. Chlorpromazine was measured 3 to 27 times lower than the expected concentrations in the cell medium incubated for the long term toxicity experiments without cells. Bosentan measured in cell medium incubated for the long-term toxicity experiments without cells was about 50-70% of the expected concentration. Valproic acid was measured nearly at the expected concentrations. Methotrexate level was about 50-70% and the piperonyl butoxide was calculated 2-5% due to solubility problems and binding properties toward plastics. The protocols for extraction and LC-MS/MS analysis of the drugs have been uploaded on the ToxBank Data Warehouse and have been linked to the ISA-tab reporting the results on their concentration in cell medium samples.

Task 3.10. Scientific Coordination of WP work activities and their interaction with other WPs and all other related SEURAT program activities
The WP3 activities were strictly related to the selection of the Gold Compounds performed within WP2 and their characterization in terms of physical chemical properties. Collaboration with WP1 was achieved for the definition of the structure of the ISA-tab reporting data on mass spectrometry analysis.
The interaction with the other cluster projects was mainly achieved through the reported work performed in collaboration with the Saarland University, partner in the NOTOX project.

Task 3.11. Sustainability Planning for resources developed within WP
The repository of the substances will be maintained for 5 years. This is not demanding and critical, since it simply requires to keep the freezer on. The chemicals will be possibly available for other projects, such as, in particular, EU-ToxRisk. Similarly, the methodologies for the analyses of the chemicals will be available within others projects.

Significant Results WP3
• LC-MS/MS methods developed for doxorubicin, tamoxifen, amiodarone, bosentan, chlorpromazine, valproic acid and its metabolites, methotrexate and piperonyl butoxide.
• Determination of the actual concentration of amiodarone, bosentan, chlorpromazine, tamoxifen, valproic acid and its metabolites, methotrexate and piperonyl butoxide in cell medium samples from acute and long term toxicity studies on HepaRG cells (overall more than 1,000 samples measured).
• Protocol guidelines developed and uploaded in the Data Warehouse

Statement on the use of resources within WP3
The resources used, in terms of person months, to achieve the objectives within WP3 during the first part of the project were consistent with what was planned. In the last few months of SEURAT-1 a much higher amount of samples has been delivered, and with a sense of responsibility were carried out with anextensive amount of additional work, using personnel resources higher than expected in this final part of the project. For the evaluation of stem cells, input has been provided by the HPA-UKSCB partner staff in several cases at no cost to the project.

WP 4 Cell and Tissue Bank

T4.1 Determine Materials Requirements
The Biomaterials requirements for ToxBank were gathered as a cross-partnership collaboration following an initial scoping meeting (Milan, 2012) and subsequently through a combination of email, phone calls, web-based surveying and participation in the ToxBank Knowledge Café laboratory visits on user requirements (See WP5 D5.1). The developing materials requirements (D4.1 Report on Materials Requirements in Systemic Toxicity) were further updated and coordinated at ToxBank partner meetings organised around the SEURAT-1 consortia scientific meetings. This included fundamental criteria for acceptable characteristics of hPSC lines (D4.2 General quality and regulatory criteria for establishment and dissemination of hPSCs), which were established in a collaboration with the SEURAT-1 partner consortium Scr&Tox (Pistollato et al., 2012). Initiatives coordinating output from relevant national, regional or European activities (e.g. UK Stem Cells for Safer Medicine, California CIRM, International Stem Cell Banking Initiative) were also engaged in this process. Information requirements on Biomaterials were fed into WP5 (see Deliverable 5.1).

T4.2 Establish suppliers (biomaterials) virtual network and registry with user friendly information access for toxicologists
This goal was addressed by reviewing the general resources available in Europe (D4.3 Summary of existing European biobanks and biorepositories) and identifying those with special relevance to in vitro systemic toxicology and the detailed user requirements identified under T4.1. Based on information provided by suppliers and users, an evaluation procedure (T4.2, T4.3) was be set up to produce a web-based registry of suppliers that builds on existing registries focused on the central EC-funded database of hPSC lines hESCreg (D4.5 First version of a virtual suppliers network, D4.6 Points to consider in gaining access to human tissues and cell lines and D4.7 Inventory and map of European suppliers: materials, resources, facilities and standards.

T4.3 Define a quality assurance framework
This activity was aimed to identify key quality criteria on which to base evaluation of suppliers (see T4.4, D4.4 Evaluation process for suppliers of hPSCs) and in addition provide advice on best practice for SEURAT-1 partners and ultimately to the broader toxicology field. These were based on key user criteria established in the earlier process of establishing biomaterials requirements (see D4.1, D4.2).

An additional and important development for this task was an active collaboration with the SEURAT-1 consortium Scr&Tox and the ToxBank WP4 lead coordinating the SEURAT-1 “Stem Cell Group” incorporating input from ECVAM. These interactions led to joint poster presentations at SEURAT-1 annual scientific meetings, peer reviewed publications (Pistollato et al 2012; Stacey et al., in press) and reports and guidance documents now available on the public ToxBank wiki. These dealt with accessing suitable human tissues and cell lines (D4.2 and D4.6) and key quality criteria for establishment and use of hPSC lines under Good Cell Culture Practice (D4.2). D4.6 identified critical issues to be addressed to enable compliance with legal, ethical and commercial requirements for use of stem cell lines in the SEURAT-1 programme and ongoing industry utility and involved the completion of an ethics review questionnaire developed in collaboration with partners in Scr&Tox.

T4.4 Establish an inventory of suitable cell culture passaging facilities and procedures
An evaluation process was created to invite feedback from suppliers of stem cell lines and other key reagents, indicating how they meet the quality criteria established for biomaterials in T4.2 and T4.3. This was developed to provide researchers with the ability to evaluate suppliers based on a standardised assessment based on information gathered using a standard questionnaire.

Securing satisfactory information from suppliers was a significant challenge and less than 50% of those suppliers contacted provided a response, or were able to provide sufficient information to use in the evaluation. All suppliers providing satisfactory responses were listed in a registry with details from their websites and their authorised response to the evaluation process (D4.3 & D4.5). These details included any quality standards or regulations under which they are inspected or to which they proposed to be compliant. A formal review and a registry of these suppliers was compiled and published (D4.5). These criteria were then utilised in the final registry of suppliers and a map with hyperlinks has also been developed for international suppliers of stem cell lines (D4.7). Whilst not identified as a formal commitment in T4.2, this activity has been extended to address the availability of biomaterials from outside Europe to identify international players in the biomaterials field with a special focus on human pluripotent stem cell lines (D4.9 Directory of suppliers of Materials for in vitro toxicology). The Deliverable Report 4.9 also contains the template questionnaire which was published on the ToxBank wiki, a questionnaire researchers can use for ongoing selection of biomaterials.

T4.5 Develop ongoing operations beyond the period of the project
Ongoing supply of Materials under the proposed framework could be delivered through a number of business models. A business case and service model were developed based on an options appraisal which concluded that a sustainable online Cell and Tissue Bank for provision of access to research materials and reagents for in vitro systemic toxicology was possible (D4.8 Options appraisal for coordinated supply of hPSCs via directory of suppliers). An estimate for the investment required to achieve this was also included in this report.

The WP4 activity has developed a significant virtual network engaging a range of stakeholder groups and an outline for a supply network for Materials for European in vitro systemic toxicology now available through the ToxBank wiki (D4.7, D4.9). These resources can be used to coordinate procurement of suitable research biomaterials and reagents for in vitro systemic toxicology procedures using stem cell lines which has ready capacity for expansion as a primary access conduit for users via a directory of suppliers (D4.9).

T4.6 Scientific Coordination of WP work activities and their interaction with other WPs and all other related 4.2.9 program activities
This task involved coordination of all the work activities and resources associated with the ToxBank Cell and Tissue Bank (achievement of implementation goals, requirements analysis, software development, delivery of services) including interaction with other SEURAT-1 projects and coordination of cross-project working groups.

ToxBank WP4 lead (partner NIBSC) also coordinated the SEURAT-1 “Stem Cell Group” which gathered SEURAT-1 partners together to discuss key issues of quality control for stem cell lines, formulated suitability criteria for cell lines used in SEURAT-1 workplans (D4.1) and published a consensus, with a number of SEURAT-1 partners, on quality assurance of hPSC lines and the development of stem cell-based toxicology assays (Pistollato et al., 2012; Stacey et al., in press).

T4.7 Sustainability Planning for resources developed within WP4
The current on-line facility is accessible at the ToxBank website (www.toxbank.net/) where the resources from WP4 described above are located with other elements of ToxBank. This system has a direct link with hPSCreg (www.hpscreg.eu) for access to scientific and ethics data on stem cell lines. ToxBank partners HPA-UKSCB and DC are also now partners in the IMI EBiSC project (European bank of iPSC lines), where DC leads the development of the information management system, which in turn enables ongoing coordination and interoperability with ToxBank features and resources.

WP5 Dissemination

The results obtained within te ToxBank project were presented in several national and international congresses and also published in peer-reviewed international scientific journals. A significant dissemination of resource information was provided through the website, tutorials, virtual seminars, workshops and working groups. More information about this activity is provided in Section 4.2.

WP6 Management
In order to plan and the check the status of the scientific and administrative work, the consortium regularly organized face-to-face meeting (at least one per year) and virtual meetings (usually every week).

Maintenance of ToxBank services and security updates at project end
At project-end the security of the ToxBank services were brought into state of the art condition. Secure communication between ToxBank web services and web clients are handled by HTTPS technology. HTTPS is a communication over HyperText Transfer Protocol within an encrypted connection by SSL or TLS. SSL, and its successor TLS, are cryptographic protocols to provide secure communication over the Internet. In the current server setup outdated and unsecure SSL protocols in version 1 to 3 are blocked and only newer TLS protocols 1.0, 1.1 and 1.2 are in use. TLS do support a large number of "cipher suites". Cipher suites are collections of symmetric and asymmetric encryption algorithms used by hosts to establish secure communication. Some offer better level of security than others. At the start of the connection between a web client and a web service the client sends a list of possible ciphers to the webserver. The webserver replies with the cipher suite that it has selected from this list. The ToxBank data warehouse server allows the use of a collection of 'latest' cypher suites that offers a good balance between security and the possiblity to connect with a variety of current web clients. The web server’s configration was adapted and tested with the free online SSL test service from Qallys SSL Labs at https://www.ssllabs.com. The web services operating systems were upgraded to newest versions (Debian 8 with support until May 2018/May 2020 LTS) to ensure secure and "easy to update" operating system over the end of the project.

Potential Impact:
We have uploaded numerous protocols, reports and datasets from the SEURAT-1 program activities (see Annex I and II). We have agreed as a consortium to maintain all ToxBank reference information as a public OpenTox resource for a minimum of 5 years beyond the end of the project, providing the scientific community access to the results of SEURAT at no charge for access. OpenTox was initially an FP7 project, but importantly has developed as a community around open resources and standards, and in 2015 was formed as an international member-based non-profit organisation (http://www.opentox.net/the-opentox-association).

During 2016 we will in collaboration with COACH and the EC JRC, continue to develop and extend case studies around the ToxBank resources. Part of the consortium (DC, IRFMN) will also be engaged in the new EUToxRisk program forming a bridge between SEURAT results and their extensions. We will continue to communicate with SEURAT-1 coordinators and data owners throughout the final reporting and review process, and beyond. Data Quality has been assured by continuing to support the best practices on data preparation as reported extensively in previous ToxBank reports and tutorials (www.toxbank.net).

To promote wider access, public registration and access to ToxBank is supported, and will continue to be supported throughout the next five years. We also plan to further develop the ToxBank resource as an important key reference resource interoperating with the growing set of OpenTox resources on data, algorithms, modelling, and analysis and visualization components.
As data owners provide public access to their datasets and results, we will promote their availability to users registered on ToxBank, and more broadly to the scientific and general public.

List of Websites:
Public website: http://toxbank.net/
Subdomains:
• ToxBank Wiki: http://wiki.toxbank.net/wiki/
• ToxBank Data and Analysis services: https://services.toxbank.net/

Related information

Documents and Publications

Contact

Benfenati, Emilio (Laboratory Head)
Tel.: +390239014420
Fax: +390239014735
E-mail
Record Number: 182090 / Last updated on: 2016-05-17
Information source: SESAM