October 15, 2014 (Vol. 34, No. 18)

Animal models, such as rodents, have been indispensable to progress in the life sciences. Mice, in particular, have played an important role in human research because they share many genes and molecular pathways with people. Ninety-nine percent of mouse genes have homologs in humans. Researchers all over the globe use mice to study diseases such cancer, diabetes, and cardiovascular disease, among others.

GEN recently interviewed seven experts to learn just what makes a good mouse model for biopharmaceutical research and to find out for which applications and diseases better models are most needed. These experts are Joe Cornicelli, Ph.D., scientific advisor at Charles River Discovery services; Jean-Pierre Wery, Ph.D., president of Crown Bioscience; Philip Damiani, Ph.D., global head of embryologic sciences and transgenic support at Harlan Laboratories; Sheryl Wildt, global manager of genetic quality and breeding at Harlan Laboratories; Cathleen Lutz, Ph.D., director of the Jackson Laboratory Repository; Xiaoxia Cui, Ph.D., vice president of R&D at Sage Labs; and Benjamin August, Ph.D., director of research models and services at Taconic.


Scientists use mouse models to study numerous disorders that afflict humans, including cancer, arthritis, diabetes, and cardiovascular and neurological diseases. [mgkuijpers/Fotolia.com]

GEN: What characterizes mouse models that have demonstrated clear benefits in drug discovery and clinical research? In other words, what makes a mouse model truly useful?

Dr. Cornicelli: The acid test for any animal model is its translatability to the clinic.  Truly useful animal models of human disease are those that resemble some aspect of the outward appearance—arthritic joints in a model of rheumatoid arthritis or elevated blood pressure in a model of hypertension. These models have the ability to predict the course of the disease, and they share common mechanisms and pathways with respect to etiology and pathogenesis.

Few, if any, animal models mimic human diseases. However, many models have been used to validate drug targets. In addition to predicting disease activity and therapeutic efficacy for drug candidates, a useful model would also predict adverse events associated with the treatment.

Dr. Wery: The most important property of animal models (including mouse) is their predictive power, i.e., the ability to use the model to gain useful information regarding how the clinical candidate will behave in human patients with respect to certain properties such as efficacy, toxicity, ADME, etc. Other properties can make the models more useful and practical such as cost, timeliness, and availability.

A recent example of a new generation of mouse models would be the Patient Derived Xenograft models, which are created by implanting a fresh human tumor in an immune-deficient mouse. They have been shown to be extremely predictive on how a molecule will interact with human tumors.

Dr. Damiani: Many mouse models have been developed to extrapolate or mimic the effects of human health aliments and diseases. However, some models are not always effective as preclinical models. For example, a model may fail to properly metabolize drugs developed for treatment. The development of humanized mouse models has been extremely useful in both basic and applied human disease research. Humanized mouse models are capable of containing human tissue/tumors or multilineage human hematopoiesis, and now represent an in vivo human model for both pathological and physiological disease states. These models can be instrumental in further ascertaining the effect of new drugs to combat human diseases as well as provide better models for basic research and discovery.

Dr. Lutz: There are many criteria that I consider critical in evaluating the usefulness of a model for translational research: First, the construct design of the genetically engineered mouse should match, as closely as possible, the disease variants in humans. Second, the model should phenotypically recapitulate one or more of the primary deficits associated with the human disease. In-depth characterization of models is a must and rigorous preclinical experimental design and application of best practices are necessary to reliably inform clinical trials. Finally, it is important to remember that mice are not humans and humans aren’t mice. Ultimately a good mouse model requires a good researcher who uses a model to inform but not over translate preclinical data.

Dr. Cui: When used properly, the mouse is a powerful model system for fields such as functional genomics and preclinical drug discovery. First, the mouse genome is highly similar to that of the human on both DNA and transcription levels, providing biological relevance. Secondly, mice are small so they are cheaper to maintain and take a smaller drug dose than larger animals. Mice also have a short gestation time and lifespan and good-sized litters. As a result, the mouse is an economical model system. Finally, the mouse has been the most malleable to precise genetic manipulations and has become the predominant model system for the past couple of decades.

Dr. August: The usefulness of a mouse model in this context should be judged by the predictive value of research results obtained with it. In other words, is the phenotype and the reaction to compounds indicative of the clinical situation? At the same time, it is critical to know the limitations of a given model, as they usually reproduce certain aspects of a medical condition but not all. Another issue to be taken into consideration is the timing of the condition in the model and the available experimental window. Is the onset and progression of the model comparable to the human situation, so that, e.g., preventive treatment can be assessed? These three parameters—predictive value, limitations, and timing—need to be looked at.

GEN: Which areas of life science research are most in need of new and improved models and why?

Dr. Cornicelli: Behavioral and developmental disorders, e.g., autism spectrum disorders (ASDs) have been difficult to recapitulate in animals, in part because identifying the triggers in humans has been so elusive. Data have only recently begun to emerge linking candidate genes with an increased risk for developing an ASD, spurring development of novel animal models. On another front, the ability to grow human tumors as patient-derived xenografts has advanced the potential of personalized medicine. The availability of mice with fully humanized immune systems will advance this field in ways that weren’t possible before. Those same models can be useful in studying autoimmune regulation and diseases and provide a stable platform for evaluating the utility of human stem cell therapies.

Dr. Wery: The life science area that would benefit the most from new and improved models with greater predictive accuracy is oncology drug development. There are still many unmet medical needs in oncology. Despite the fact that our fundamental understanding of cancer biology has greatly improved, our ability to take a promising preclinical cancer drug candidate and execute a successful clinical development is poor. Some reviews place the success rate of cancer clinical trials at about 5%, probably the lowest among all therapeutic areas. New predictive models could have a tremendous impact on our ability to bring lifesaving medicines to patients.

Ms. Wildt: Two areas come to mind: ADME/Tox testing and obesity. There is a need for more humanized or knockout mouse models to further assess the absorption, distribution, metabolism, excretion, and toxicity of newly developed drugs for preclinical trials. Unlike in vitro assays, these in vivo models can clearly show important relationships to drug dose, concentration, toxicity, and possible drug–drug interactions in humans.

The prevalence of obesity has risen dramatically over the past 30 years. At least 30 genes have been identified that correlate to obesity disposition. Monogenic obesity mouse models do not mimic human obesity as well as polygenic models. Only one polygenic model, the C57BL/6 (DIO), is commonly used for obesity studies. New models that better mimic human obesity are needed for research.

Dr. Lutz: There are thousands of rare and orphan diseases that individually affect only a small group, but taken together cause disease for a large segment of our population. Funding for research into these diseases is limited, which has slowed the pace of model development and drug discovery.

The increasing availability of genome sequencing is leading to the discovery of more and more rare—or even unique—disease-causing genetic mutations. At the same time, CRISPR/Cas technology and other advances have made it cheaper, easier, and faster to engineer these human mutations in a mouse model. The challenge will be putting this technology to good use by creating rigorously constructed models; moving them quickly through rigorous characterization; and advancing them promptly to preclinical trials.

Dr. Cui: New and improved models are likely to have the most impact on preclinical drug discovery. Mouse models are used for absorption, metabolism, efficacy, and toxicity tests of any new drug. The better a model recapitulates a human disease, not only on the genetic level but also histologically and phenotypically, the better the chance to achieve more accurate predictions on clinical outcome and to help reduce drug failure rates in expensive clinical trials.

Dr. August: I would say neurological disorders are most in need of new and improved models. While there are some rodent models available for several of these diseases, including Alzheimer’s, Parkinson’s and ALS, they come with considerable limitations. But regardless of research area, many complex, multifactorial conditions would benefit from additional models. Unfortunately, those are generally difficult to develop, with a high risk of failure in the end. An emerging research area that could greatly benefit from the development of mouse models is research into microbiota. This is a new and hot topic that does not have many models available and so far mostly relies on clinical correlations for publications.

Previous articleAgilent Eyes Cutting 300 Jobs after Exiting NMR
Next articleProteogenomics: Integrating Proteomics and Genomics to Unlock Biological Function