Pathology of the Laboratory Mouse

Abstract

The fifth in a series of European workshops for veterinary and human pathologists, “Pathology of the Laboratory Mouse: An International Workshop on Challenges for High Throughput Phenotyping,” was held in Bordeaux, France, from September 30 to October 1, 2010. In this report we outline the rationale for setting up this workshop series, summarize our experience, and suggest approaches for optimizing histopathology phenotyping for gene function discovery.

Over the past decade, large-scale chemical (N-ethyl-N-nitrosourea and ethylmethanesulphonate) mutagenesis projects (Acevedo-Arozena et al. 2008; Clark et al. 2004; Concepcion et al. 2004; Goldowitz et al. 2004; Munroe et al. 2000) inspired the development of high-throughput (HT), broad coverage ways to analyze mutant mouse phenotypes. These methods include, among others, traditional clinical pathology (hematology and blood chemistry), behavior, sensorimotor function, and morphological parameters together with physiological functions such as blood pressure and glucose tolerance (Brown et al. 2005; Gailus-Durner et al. 2009), collectively referred to as physiological phenotyping. Although physiological phenotyping methodologies have been refined and standardized over time, detailed histopathology (pathology phenotyping) has been implemented by some centers as part of a primary/primary-extended screen or included as a targeted study on cases triaged on the basis of findings in primary screening. One of the goals of this workshop was to assess whether there was internationally available expertise to perform pathology phenotyping (histopathology) on the scale required by future HT phenotyping programs, whether it was cost effective as a primary screen, and whether or not data generated to date justified pathology phenotyping in these initial workups. In addition to the mutagenesis projects, other large projects include the large-scale generation of genetically engineered embryonic stem (ES) cell lines (Collins et al. 2007) that are now being turned into living mice for comprehensive phenotyping by the International Mouse Phenotyping Project (IMPC; Abbott 2010), and massive aging programs in progress using “wild-type” inbred (Yuan et al. 2009) and collaborative cross mice (Chesler et al. 2008; Churchill et al. 2004); these issues and concerns remain the same for all studies on this scale. These large-scale programs will generate insights into normal and pathobiology and provide a huge resource for drug discovery by assisting in identifying targets and understanding the biology underlying off-target effects and toxicity.

All of the countries represented agreed that although there remains a shortage of pathologists (DVM or MD) or trainees with laboratory mouse expertise (Schofield et al. 2009), the current number of pathologists in the major centers, augmented by development of online data-sharing tools for image sharing and annotation between distributed experts, would make full pathology analysis feasible, at least for the expected caseload from the IMPC project. A major disincentive to young pathologists to specialize in genetically engineered mice (GEMs) is the lack of a viable career path in academia (Sundberg et al. 2004; Warren et al. 2009) or basic biomedical research, in part because of the increasing difficulty of obtaining research funds, even in established university positions, with junior pathologists opting instead for the more traditional and well-paid careers in industry (mainly in safety assessment toxicology and research). Lack of trainee awareness of academic/biomedical research-based career opportunities poses additional problems for recruitment. However, the large-scale projects now underway are showing welcome signs of stimulating renewed interest in the specialty, although the lag time in training means there is likely to remain a deficit of available mouse pathologists in the near future. One very useful function that large phenotyping centers could fulfill is that of providing internships and training courses, as they will have large amounts of material and should become centers of expertise, or at least a source of valuable materials for further study by others. One excellent example is the pathology program at the Centre for Modeling Human Disease (CMHD; Toronto, Canada), where trainees are being recruited. In 2004, the CMHD established a formal DVSc residency program in veterinary anatomic pathology in partnership with the Department of Pathology at the Ontario Veterinary College (OVC). This program provides post-DVM trainees with a funded residency that typically requires a minimum of a year and a half for graduate-level pathology coursework and anatomic veterinary pathology training (necropsy, biopsy, histopathology) at OVC followed by a minimum of two years within the CMHD’s Pathology Core, which provides a collaborative opportunity for a doctoral-level thesis project and focused anatomic pathology training on the spontaneous and induced diseases of the laboratory mouse.

The United States and France are currently the leaders in training mouse pathologists, providing opportunities to subspecialize or focus on mouse research (especially for evaluation of GEM models), which should train twenty pathologists or more over the next five years (Sundberg et al. 2007; Sundberg et al. in press). The National Center for Research Resources and the National Institute on Aging (part of the National Institutes of Health in the United States) and Cancéropôle Grand Sud-Ouest (in France) have developed excellent collaborative pathology training and awareness programs in recent years, and the French government is exploring ways to develop a career path for the newly trained personnel to keep them in academic research. The proposal to establish dedicated posts in French veterinary schools or universities was warmly welcomed. There was strong support for the proposal to encourage other countries, particularly the United Kingdom, Germany, and Spain, to start similar training programs, and that an international shared residency or fellowship program be established to leverage the existing programs and resources.

On phenotypic analysis of GEM models, particularly in the context of the IMPC primary phenotyping pipeline (the sequential set of assays carried out in the phenotyping “clinics” [Brown et al. 2005]), we challenged the assembled group to debate whether pathology phenotyping was indeed necessary or could be replaced with other types of screens, or the opposite, whether pathology could replace some or all of the other phenotyping screens. The consensus from the attendees was that pathology is a uniquely valuable primary screen when carried out by experienced experts. The trend of some institutions to use inexperienced junior pathologists or even post-docs may not produce quality results, is likely to miss phenotypes, and could undermine users' trust in the value of pathology phenotyping. Although the IMPC data could provide invaluable training material and help increase the pool of pathologists with GEM expertise, there is no substitute for the trained expert in these high-throughput studies. When done well, carefully, and by experienced pathologists, pathology phenotyping is not only a powerful tool in itself, but by describing the structural basis and relevance of the tissue changes to the genotype, it also provides context for the various physiological phenotyping assays. The two approaches are complementary; histopathology will detect phenotypes not detected by physiological screens, will add value (e.g., structural basis of the phenotype) to physiological screens, and provides an integrative view based on whole-animal interpretation of findings. Pathology phenotyping is logistically feasible as part of the proposed IMPC primary baseline pipeline, and experience of phenotyping more than 4,500 mouse lines discussed at the meeting indicated that two full-time pathologists could readily examine twenty mutant mouse lines in a week using three animals of each sex from each line (120 necropsies).

Fiscal feasibility depends on expectations for a primary screen. As only one inbred strain is now being used for the IMPC program, the C57BL/6NTac, a minimum of two females and two males, ten slides per mouse, would provide a systemic overview of major lesions induced by a single gene mutation. This process follows well-established primary, initial screen phenotyping approaches used for mutant mice (Sundberg et al. 1998). At least one wild-type control mouse should be included with each group or on a regularly scheduled basis to screen for genetic drift with time, infectious diseases, or other untoward, unanticipated events that might affect the final interpretation. Disregarding the cost of obtaining live mice (generating the mice from ES cells and maintaining them in mouse rooms until they are fifteen weeks of age), the cost for a detailed pathology workup, per mouse, should run $300 or less, or $1,500 or less per line, with up to twenty lines processed per week. Additional younger, moribund mice should be processed before death, as these animals will most likely exhibit the most profound lesions. Cancer and some degenerative disease lesions are not detectable by physiological or behavioural assays at fifteen weeks of age, but early changes—for example, muscular dystrophies, polycystic kidneys, and pre-neoplasia—may be detected using histopathology, thereby extending the type of phenotypes detected without requiring aging. The cost of aging mice up to eighteen to twenty-four months of age is prohibitive as a first level screen, but maintaining mice up to twelve months of age seems more cost effective. Often, robust phenotypes occur only after fifteen weeks of age.

Many phenotypes/lesions—for example, development defects in selected organs including the normally tiny lymph nodes, retinal lesions, brain lesions, and subclinical degenerative renal lesions—will be discovered only by using histopathology. This limitation will have a significant impact on phenotype-driven user data needs, and there is the risk that without the integrative aspect of histopathology, a mutant may be allocated to a misleading phenotypic group. For example, in a physiologically detected azotemia, is the underlying cause pre-renal, renal, or post-renal? Is a gait dysfunction a result of vestibular defects, cerebellar lesions, or even lymphadenopathy? The additional contextual information helps potential users to better select a model on which to work and would reduce wasted effort.

Embryonic and perinatal lethality represents further categories of mutant mouse phenotypes. These classes of phenotypes can account for a significant proportion of phenotypes generated in the large-scale programs such as the IMPC, and, in particular, embryonic lethalities require a specialized pathology-based screen. Although such a screen can be guided or assisted by specialized imaging, stand-alone histopathology approaches are well established (Ward and Devor-Henneman 2000). It has been noted that there are few pathologists skilled in evaluation of mouse embryos and extra-embryonic membranes, highlighting yet another specialty training requirement. Integration of imaging, histopathology, and other methods optimizes the ability to discover the causes of embryonic mortality.

Although general, service-based anatomic pathologists can be used for IMPC-type screening, some admitted to missing phenotypes outside of their area of specialty or acknowledged that they may recognize a lesion but not be certain of its significance and need further consultation. A distributive model was proposed in which experts in each area or organ system would review the primary slides or provide secondary review expertise. This process could take advantage of the worldwide community of mouse pathology expertise. In-house pathology demands might therefore decrease, or each center would develop its own group of specialists. The practical limitations of depending on a distribution network of organ-specific experts in a high-throughput primary baseline screen keeping pace with physiological phenotyping needs to be integrated into this model.

With the advent and now general availability of whole-slide imaging, all or selected slides could be scanned and centrally stored. Slides could then be analyzed by specialists using the Internet without the need of having expertise in each field at each center or slides generated one organ at a time. Storage of primary image data could be done at major national centers. It was estimated that assuming 17,000 mutant lines are phenotyped by the IMPC (taking into account lines in which the mutations are unobtainable or early lethal and no material is available), about 600 TB of computer storage would be needed to archive whole-slide images of twenty slides per line (ten per mouse, of each sex) containing forty tissues at an average file size of 2 GB. This represents a significant but soluble challenge.

Standardization of necropsy methods and tissue trimming is needed between all large-scale phenotyping operations. The number of blocks, slides, and tissues taken varied considerably between pathologists, with up to forty tissues taken in a total of six to twenty-four blocks. Fixatives also varied, but most laboratories use neutral buffered 10% formalin. Others use 4% formaldehyde in phosphate buffered saline, Bouin’s solution, or Fekete’s solution (37–40% formaldehyde + glacial acetic acid in 70% alcohol). Fekete’s solution was shown to be one of the best general purpose fixatives for immunohistochemistry by at least one major laboratory (Mikaelian et al. 2004). It would be advantageous if a minimal tissue set and standardized fixation and trimming protocol were agreed upon to enable comparison of results between laboratories.

Informatics, nomenclature, and the use of data and image sharing tools should be standardized and specific for histopathology. Such tools for histopathology data acquisition already exist (Sundberg et al. 2009; Sundberg et al. 2008). Making databases function between centers would enable rapid dissemination of the data, both summative (diagnoses) and primary (images), to the community and would also facilitate computational analysis. Integrating images with annotation from these projects into public databases, such as Pathbase and the Mouse Tumor Biology Database, provides additional online training opportunities as well as reference materials (Begley et al. in press; Schofield et al. 2010). Data integration requires standardized nomenclature and ontologies, in this case MPATH (Schofield et al. 2010), and, for example, the recent international classification of rodent lesions (INHAND; Renne et al. 2009) is a valuable contribution toward this goal.

High-throughput phenotyping is an ambitious undertaking with the potential for extremely high rewards for science. For the first time, achieving these ambitions depends on concerted, long-term international cooperation and coordination on a scale and detail not previously attempted. Technology has made this goal possible, but only the scientific community can make it work.