NIH Planning Meeting for a Knockout Mouse Project

NIH Planning Meeting for a Knockout Mouse Project

Bethesda North Marriott
Rockville, Maryland
March 24-25, 2005

Chair: Geoff Duyk

Summary & Recommendations

Presentations at the workshop laid out the status of the current and planned mouse knockout resources:

• There are currently about 10,000 null mutants (ES cells or mice) that have been produced. Of these, there are slightly over 3200 targeted knockout mice that have limited availability because they are not in public repositories and therefore not readily available to the entire research community.
• The European Union Conditional Mouse Mutant Program (EUCOMM) plans to develop a resource of 20,000 conditional mutations in a 129Sv ES cell background.
• C57BL/6 is the preferred mouse strain for use in research; therefore efforts to establish more useful ES cell lines for generation of C57BL/6 knockouts would be valuable.
• Existing repositories have the capacity and capability to handle storage and distribution of additional resources.

Recommendations:

• The NIH's plan to construct a set of ES cells carrying null mutations with reporters, as recommended by the Banbury meeting, is important. The addition of 10,000 more nulls with a reporter will complete the resource so that there is a null for each gene in the mouse genome. Such a resource would be complementary to the set of conditional mutants that the EUCOMM project plans to generate.
• NIH should support continued efforts to develop the C57BL/6 ES cell lines as a system for generation of knockout mutants in an experimentally tractable genetic background.
• NIH should explore "repatriation" (collection) of existing ES cells and mice to repositories as an alternative to remaking the null mutants. However, issues such as quality control, relative expense, and potential IP restrictions must be carefully evaluated before a decision is made one way or the other.
• NIH should coordinate its efforts with other international programs that are developing the comprehensive mouse knockout resource.

Introduction

In the fall of 2003, a meeting was convened at the Banbury Center of the Cold Spring Harbor Laboratory to discuss construction of a public resource consisting of a comprehensive collection of mouse knockouts, i.e. a null mutation in every gene in the mouse genome. The meeting was international in scope and attended by scientists from both the public and private sectors. The issues discussed included the scope of the potential effort, type of mutations and the strain background that would be most desirable, the feasibility of producing such a collection in a cost-effective and highly efficient manner within a reasonable period of time, the data management requirements, the dissemination of data and materials, and the feasibility and costs of phenotyping the mutant mice generated in the project.

The 2003 CSHL Banbury meeting recommended the value of a knockout mouse project to generate a library of ES cells comprising a null allele, ideally conditional null for each gene in the genome. However, because they doubted that existing technologies were sufficiently robust to accomplish a genome-wide, high-throughput effort to produce mice with a conditional null mutation, the meeting attendees agreed that the project should use a mixed strategy employing both gene trapping and gene targeting technologies to produce a comprehensive collection of mutant ES cell lines representing null mutations for "all" mouse genes (~20,000) in the strain 129 background. The resource would be used to produce live mice (which would then be stored and distributed as frozen embryos and sperm). A graduated scheme, represented as a pyramid with three additional tiers, was recommended for the subsequent phenotyping phase of the project. Tier 1, or basic, phenotyping would be done on nearly all of the mice created by the project and the phenotyping would be done centrally. Tier 2 phenotyping would be carried out for only a subset of mice, although transcriptome analysis would be done centrally, and Tier 3 phenotyping would be specialized and performed by individual labs. It was envisioned that the project would international in scope.

Since the Banbury meeting, there has been a significant amount of activity around the world to organize mouse knockout efforts. Investigators in Europe are planning the European Conditional Mouse Mutagenesis Program (EUCOMM) that proposes to generate 20,000 mutations (12,000 conditional gene trap mutations and 8,000 conditional targeted mutations) in ES cells. The group proposes to archive and distribute the mutant ES cell lines centrally, along with existing Cre recombinase-expressing transgenic mouse lines. Additionally, they will establish 300 mutant mouse lines from the conditional mutants produced as part of the project and provide an online database for the dissemination of information about the resources. EUCOMM is a consortium of eleven laboratories and is led by groups at the GSF (German National Research Center for Environment and Health) in Munich, Germany and the Sanger Institute in Hinxton, UK. Other mouse knockout projects are under consideration in Canada, Australia, Japan and at the Sanger Institute.

In addition to these newly planned focused efforts, there are, of course, a large number of knockout mice that have been previously created in individual research laboratories, both public and private, around the world in the course of their specific research projects. The existence of these mice should be taken into account in planning the development of the proposed comprehensive resource, in order to avoid unnecessary redundancy. As a basis for doing this, the NIH has begun systematically to develop information about existing knockouts. The approach involved compiling data from the Mouse Genome Database (data curated from the published scientific literature), and gene trap sequence data submitted to the dbGSS section of GenBank, as well as conducting a new inventory of available but unreported knockout mice. The latter was done through an e-mail solicitation widely distributed to the research community. The results were collated by the MGD.

The information obtained to date indicates that mutations have been constructed in at least 8188 unique genes, by either gene-trap or targeted knockout methods. The information-gathering effort also indicated that only a small fraction of the mice or ES cells containing these constructs is publicly available and that investigators were willing to make available another small, but significant, number. One of the problems with this effort was that there was only a limited response from investigators outside of the US. Therefore it is almost certain that the current number is a significant underestimate of the number of unique genes that have been already been knocked out by the research community.

Since the Banbury meeting, the NIH has also been considering its role in the construction of a comprehensive mouse knockout and how to complement the efforts of the international community. The NIH Planning Meeting for a Knockout Mouse Project was organized to assess the status of research in the field eighteen months after the Banbury meeting, to obtain updates on current projects and to get input from the scientific community on a draft proposal for an NIH mouse knockout program

Meeting Summary

The agenda for the NIH Planning Meeting for a Knockout Mouse Project was designed to address a number of issues of importance to NIH planning effort:

1. How many knockout mutations, and of what type, would be appropriate for the NIH to contribute to a comprehensive international resource?
2. What ES cell strain background is preferred by the mouse research community?
3. What would be the optimal vector design? Should the knockouts be "simple" null deletions marked with a reporter gene or should they be conditional mutations?
4. What are the bioinformatics needs for the project?
5. Is current technology adequate to produce a comprehensive collection in a reasonable period of time at a reasonable cost, or is additional technology development needed? Are there other technologies on the horizon that could significantly reduce the value of a comprehensive set of knockout mutations before the project could be completed?
6. What are the repository needs for the project?

At the meeting, a series of panel discussions and break-out groups discussed each of these issues. The following is a summary of those discussions.

ES Cell Background

A panel of experts discussed the use of ES cells derived from inbred lines, particularly 129/Sv and C57BL/6, and of a 129/BL6 F1 hybrid cell line. The F1 line is very to easy handle and produces a high rate of germline mice. Despite the efficiency advantages, there was a strong recommendation that issues of mixed background genetics were of overwhelming importance. The influence of genetic mutation is most easily interpreted on a defined background. As a gene is bred onto a different genetic background, the role of modifier genes can be inferred. Moreover, a mixed genetic background makes breeding of congenic lines difficult, especially if it is into to a third genetic background. Thus the recommendation was to the use an ES cell line derived from an inbred strain to allow breeding to pure genetic inbred when needed.

There remained the question of which inbred strain, 129/Sv or C57BL/6, should be used. Strong reasons were presented for each alternative. In the end, a majority of those present recommended the use of C57BL/6, as the preferred line of choice for many groups including immunologists and neurobiologists.

There were several important caveats stated:

First, it was acknowledged that the attendees were not necessarily representative of the mouse genetics community as a whole and the choice of the community might be different. Second, it was also understood that there are still technical issues in handling the BL6 system that need to be solved before it can be introduced as the workhorse genetic background. For example, C57BL/6 ES cells are more fragile and have not been tested for high throughput use, as is the case of 129 ES lines that have been used in gene trapping efforts. Cryopreservation of sperm but not embryos is problematic for C57BL/6.Thus, there was also a strong recommendation to support further technology development research on the BL6 ES cell system, e.g. to optimize culture conditions for ES cells and to develop feeder independent BL/6 lines to reduce cost and increase efficiency.

Vector Design

Given that the European and other efforts will be focused on conditional mutations, the meeting recommended that the NIH should consider providing a complementary resource of benchmark reporter-marked null deletions. The "simple" deletions would provide an opportunity to generate baseline data that could then provide the context for the interpretation of the data from the more complicated conditional mutations, and the deletion/conditional heterozygote would be a particularly valuable experimental reagent.

In designing an NIH program, the mutation-generation strategies should be fully validated, have a benchmarked cost per allele, and result in mice with a high germ-line transmission frequency. Any strategy employed must ensure that the target vector resource remains useful in the future. The attendees also ratified the Banbury meeting's conclusion that the mutations to be incorporated in the comprehensive resource (whether existing or newly generated) must be fully validated as to null/conditional/reporter status.

There was general agreement that the NIH should not undertake the creation of a complete resource but, because of the large number of international efforts already underway, the NIH effort must be cognizant of and complementary to the already active efforts in Europe, the UK, Canada, and elsewhere. Thus, the attendees strongly recommended that an NIH gene targeting-based program must be closely coordinated with international efforts to ensure that appropriate choices of genes to target are made. However, it was also considered crucial that there be an agreement among all of the groups participating in an international mouse knockout project that the resource generated be available to all users around the world.

Finally, whatever strategy is chosen, BAC recombineering-ready libraries of targeting vectors that can be reengineered for any future genetic alteration of locus are a vital first-line resource.

Beyond the considerations of vector design and mutation strategy, it was acknowledged that baseline gene expression and phenotype information would make the resource more valuable. However, it was recognized that this would also increase the cost significantly and, as a result, such characterization was not included as a recommendation for the initial phase of the mouse knockout project at the NIH.

Bioinformatics

The following information will be needed for a knockout mouse project and accumulating the information will require bioinformatics solutions:

• Gene Model List: A canonical list of genes in the mouse genome, tied to the accessioned genome and version number will be required for a targeting strategy for the knockout project. Similarly, the results of the efforts to knock out genes by trapping should be tracked using sequence from the knockout event.
• Prioritization of Gene Models: Tools are needed to prioritize the gene models to be targeted in the KOMP. This prioritization will need to be coordinated with international efforts such as EUCOMM, IGTC and existing targeted knockouts. Another decision that will have to be made will be whether to implement a regional process (systematic by chromosome) of assignments or a functional process (based on gene families, biological pathways or systems, or requests).
• Inventory of Existing Knockouts: The NIH's effort to catalogue existing gene traps and targets is based on sequence data from the International Gene trap Consortium deposited in dbGSS, published information from MGD, and an inventory of unpublished knockouts. Further efforts are needed to get more information from groups (in Europe, Japan, Canada and Australia) that were not reached by the current inventory. If an effort were made to repatriate individual lab-generated knockouts based on community agreed standards (see below), informatics tracking tools will be needed to assess the cost-effectiveness of that approach, as opposed to remaking the mutants.
• Tracking Database: Each knockout group will have its own laboratory management database and it was agreed that a new, uniform lab management database for all participating laboratories was not necessary. However, a single project tracking center will be needed to track and consolidate project-wide results, information about the status of assigned knockouts (allocation of gene regions, vector and ES cell availability), and results (molecular event on graphical browser, where to get the vector, cell lines, or mouse reagent resources). This project tracking center will likely also need to point to the participating laboratory database for further information and individual knockouts.
• Project-wide Quality Control Standards: The participating groups will need to develop a set of common standards and then provide sufficient information to characterize the molecular event from each knockout with high degree of confidence (sequencing should be required).
• Annotation of the Mouse Genome: While this is not an explicit component of the mouse knockout project, annotation of the mouse genome is an issue of major interest to the scientific community. The Sanger's VEGA annotation project is interested in providing detailed annotation of the genome, but is not currently funded for the entire mouse genome.

Technology Development

The issue of evolving technologies that might obviate the need for a mouse knockout resource within the next few years was discussed. It was agreed that siRNA-based technologies for selectively knocking out gene function are likely to be a good complement to genetic knockout techniques, but are not going to replace them. Additionally, there are no conditional transcription technologies robust enough at present to compete with the recombinase-mediated conditional approaches utilized in mouse knockout systems. There are new inducible technologies that may be useful but at present the data are too preliminary to assess as they have not been validated in mice. Additionally, many of these usually include a transgenic system expressed with a Knockout. Thus the KO resource is still a valued resource.

The group agreed that the availability of primary targeting vectors would be extremely useful in facilitating the secondary modifications of alleles. There is a need to engineer additional mouse strains to support conditional strategies and no need to preconfigure ES cells for this effort. There are a variety of histochemical and fluorescent protein markers that can be expressed in transgenics or knock-ins and are available to assess the activity of integrases and recombinases, and the group felt that technology is moving rapidly and many new tools are appearing. There was consensus that the ideal reporter or recombinase expression system has not been validated. There was no strong feeling among the participants that any new approach is ready to be incorporated into mouse knockout projects or that there is a limitation of availability of existing tools.

To make the conditional knockout resource maximally useful, the loxP/Cre-based approaches are favored, but data presented at the meeting raised a concern about secondary chromosomal damage as a consequence of Cre activity. Some preliminary sites suggested that Cre induction can induce deletions that can be as large at 20Mbases, presumably due to recombination between cryptic loxP sites in the mouse genome. If the Cre-lox system is to be used, good Cre driver strains are critical to the maximal usefulness of a conditional resource, and this is an element of the system in which the private sector has a particular interest. However, no proposal for developing a specific set of strains was discussed at the meeting.

Repositories

As a public resource, it will be essential for public repositories to handle the products produced by the knockout mouse project. The following points were made in the discussion of this issue:

• Repository efforts should be centralized. This will improve efficiency and ensure the quality of the materials distributed to the user community.
• New repositories are not needed to handle the products of a knockout mouse project; existing repositories have the capacity to scale up to accommodate the output of the proposed project.
  " The repositories must have expertise in handling, storing and distributing ES cells, embryos, sperm and mice and must have good standards for quality control for each type of product.
• The repositories must also have capability to create mice from ES cells.
• The repositories will need to provide user access to all information on the strains they maintain.
• There are currently bottlenecks in distribution that need to be addressed.
• Users will need to share the cost of access to materials in the repository.
• Repositories will need to put in place plans to replenish, expand and extend the knockout project's materials in the repository.
• Repositories will need to Integrate with other international resources (the International Federation of Mouse Repositories, the Mouse Genome Database, The International Mouse Strain Repository, the National Center for Biotechnology Information, Ensembl, and the University of California, Santa Cruz).

Feedback to the NIH

Considerable progress in producing knockout constructs and mice has been made since Banbury. The NIH's plan to construct a set of ES cells carrying null mutations with reporters, as recommended by the Banbury meeting, is important. The addition of 10,000 more nulls with a reporter will complete the resource so that there is a null for each gene in the mouse genome. Such a resource would be complementary to the set of conditional mutants that the EUCOMM project plans to generate. However, there is considerable interest in continued efforts to develop the C57BL/6 ES cell lines as a system for generation of knockout mutants in an experimentally tractable genetic background. Private sector scientists are interested in "humanized" mice, that is mice in which a human gene has been "knocked in" in to replace the corresponding mouse gene. They urged NIH to consider this element in its planning.

While generation of new mutants to complete the knockout resource is important, participants urged NIH to explore "repatriation" (collection) of existing ES cells and mice to repositories as an alternative to remaking the null mutants. It was noted that there are, issues such as quality control, relative expense, and potential IP restrictions that must be carefully evaluated before a decision is made one way or the other. Overall, there was agreement that whatever initiative NIH puts forward, it will need to coordinate its efforts with other international programs that are also participating in developing the comprehensive mouse knockout resource.

Finally, a warning that a complete catalog of mouse knockouts will totally alter the sociology of the mouse research community landscape. In that regard, it will be a disruptive innovation, and will create the need for an extensive educational/re-training platform. Such education programs will facilitate the wide-spread distribution of mice, advance research, and minimize duplicative efforts.