Efficient use of genetically modified animal models can accelerate the progression of candidate therapeutics through your drug development pipeline with more predictive in vivo results.
In this guide, learn how genetically engineered organisms can better support all four phases of drug discovery research.
Target discovery is the start of the drug development pipeline, where the goal is to identify genes that can be modulated to affect the clinical manifestations and outcome of specific diseases.
High value target candidates can be defined as genes that, when modulated, affect disease progression without having severe harmful effects on the physiology of the organism.
Traditionally, potential target genes were identified by forward genetic screenings performed using in vitro systems or relatively inexpensive animal models such as zebrafish or Drosophila Melanogaster1,2. More recently, mice have been used in genetic screening experiments to identify genes linked to specific physiological and pathological pathways3,4. The major advantage of using the mouse for genetic screenings is the possibility to directly validate results in well-established genetic disease models5.
Advances in genome sequencing opened a new approach to the identification of potential drug targets. Genome Wide Association Studies (GWAS), performed by sequencing the exome or even the full genome of patients suffering from specific syndromes, are extensively exploited to identify novel target genes.
The major appeal of this approach is that GWAS data are directly derived from the clinic, providing the drug development pipeline with potentially high value drug targets.
Since GWAS are exclusively based on clinical data, they do not rely on the use of animal models. However, once potential target genes have been identified, animal models are still required to confirm that their modulation can affect the disease without resulting in severe effects on the organism6,7,8.
The main goals of target validation are to explore whether deletion (or over-expression) of a target gene might be detrimental to the organism, and to confirm that its modulation alleviates the phenotypes associated with a specific pathology.
The most common models used in target validation are represented by knockout (KO) mice. KO models provide a very high degree of specificity and, if designed correctly, are extremely informative on the potentially toxic effects of inactivating defined genes.
One issue with constitutive knockout mice is that the absence of a gene in embryonic development might have very different effects than in the adult animal. The use of conditional KO alleles coupled with inducible Cre recombinases can address this issue. By regulating Cre recombinase activity, it is possible to induce genetic deletion of a target gene in adulthood and model drug treatment more predictively.
Sometimes, gene deletion is not the preferred approach for modeling the effects of a specific drug. Many potential drug target proteins have multiple functional domains which are required for separate functions.
Therapeutic molecules usually interact specifically with distinct domains, inhibiting only some functions of their target proteins. Deletion of entire genes, however, results in the loss of all of their associated functions, and therefore does not necessarily mirror the effects of specific drugs.
One example of how to model inactivation of a single activity in a multifunctional protein is described in Vaillancourt et al.10. In order to evaluate if inhibition of phosphatidylinositol-4-Kinase IIIα (PI4KIII) kinase activity without affecting its scaffolding functions could be used as a strategy to target the hepatitis C virus, the authors exploited a mouse model carrying a conditional KI allele expressing a protein bearing a single amino acid substitution that inactivates the kinase domain. Using this tool, the authors show that specific inactivation of PI4KIII kinase function in the adult mouse leads to massive necrosis in the digestive tract, suggesting that its inhibition is not a viable path for the development of therapeutic molecules.
In addition to testing potential negative effects on the physiology of the organism, a major goal of target validation experiments is to verify that modulation of target gene candidates ameliorates the pathology in models of human disease.
A remarkable example of target validation in the mouse has been described by the group of Paul Greengard at the Rockefeller University11. In this article, the authors describe the use of an in vivo RNA interference approach to modulate the activity of gamma-secretase activating protein (GSAP), a potential target for the development of drugs for the treatment of Alzheimer's disease.
The use of an shRNA model provided two main advantages: its genetic simplicity, since the entire shRNA system is coded by a single allele12; and the reduction, but not-complete elimination, of GSAP activity, mimicking the dynamics of a drug treatment more closely than genetic deletion.
After a gene is validated as a target for drug development, the next step is to select molecules able to interfere with its function. The use of combinatorial chemistry, paired with high-throughput functional assays, allows efficient synthesis of potential therapeutic molecules.
Due to the relatively high costs of developing and maintaining transgenic lines, genetically modified animals are not typically used in the initial identification of active molecules. Once a compound has been selected, however, in vivo experiments are crucial before investing more resources to progress it to the next phase of the drug development pipeline.
The in vivo stability and tissue distribution of selected compounds need to be addressed before they can be considered clinical candidates. Testing of compound stability in rodents has always been a challenge, since the major pathways related to drug metabolism and distribution are not well conserved between species.
One of the major physiological pathways used by organisms to eliminate drugs from their bodies is to metabolize them into secondary compounds lacking the therapeutic activity. Members of the cytochrome P450 family are central in the activity of drug metabolism, both in human and mouse. This multigenic family shows profound evolutionary divergences within mammals, making it impossible to directly compare their activity in different species. To overcome the limitations of murine models in drug stability testing, one approach has been to replace mouse genes involved in drug metabolism with their human counterparts13,14. For example, using a genetically humanized mouse model, Zhang et al. demonstrated in vivo the effect of co-administration of a specific cytochrome inhibitor with the antiviral drug, accurately predicting the clinical outcome of this therapeutic strategy15.
Preclinical testing is a crucial step in the drug development pipeline, when candidate molecules are tested in vivo to assess safety and efficacy. The appropriate use of genetically engineered models can help preclinical researchers address many of the concerns raised by the introduction of a new chemical reagent into the drug market.
One major issue is the potential carcinogenic activity of candidate drugs. Regulatory agencies are very strict on the assessment of this aspect of drug safety.
The transgenic rasH2™ mouse model, developed by Tatsuji Nomura at CIEA in Japan16, is an indispensable tool for laboratories involved in the evaluation of compound genotoxicity. This model is significantly more susceptible to genotoxic carcinogens than wild-type mice, supporting increased sensitivity in safety testing and a reduction in both animal test cohort size and project timelines17.
Testing the efficacy of potential therapeutic compounds is another major challenge for preclinical scientists. Drug candidates developed to interfere with the function of human proteins may not comparably interact with their murine counterparts, making it challenging or even impossible to perform in vivo tests on wild-type rodents.
One strategy to address these species-specific differences is to humanize the mouse, replacing a mouse gene with its human homologue. This method was used by scientists at Eli Lilly to test the in vivo efficacy of DETQ, a novel dopamine receptor D1 potentiator developed to treat Parkinson's disease symptoms18. Since the mouse receptor is impervious to DETQ binding, researchers developed a humanized mouse model where the endogenous gene was replaced with its human homologue.
Using this model, the scientific team at Eli Lilly successfully tested DETQ in vivo, confirming its efficacy and lack of major side effects. These experiments contributed to the advancement of the molecule to phase I clinical testing.
A complex challenge to efficacy testing of drugs in vivo is posed by biopharmaceuticals, large molecules of biological origin (e.g. antibodies or enzymes). While this class of drugs shows great promise, rodent model testing has been difficult due to their immunogenic characteristics.
Undesired immune responses can be avoided by generating mouse models which are tolerant to the therapeutic agents, as described in the preclinical enzyme replacement study by Ferla et al.19. In this study, the authors compared the efficacy of protein replacement versus gene therapy in a mouse model of a human lysosomal storage disease lacking the arylsulfatase B (Arsb) gene. Immunotolerance towards the human recombinant enzyme was achieved by ubiquitously expressing an inactive human protein from the Rosa26 mouse locus.
This transgenic mouse was then crossed to a constitutive Arsb KO line to generate animals lacking the endogenous function and expressing inactive human ARSB. Using this model, the authors were able to compare two different therapeutic approaches and open the path for the development of new therapeutic regimens20.
Genetically engineered mouse models play several critical roles in the drug development pipeline, including the study and characterization of disease pathology, target identification, and in vivo evaluation of novel therapeutic agents and treatments21. In the thirty years since their introduction, preclinical and translational scientists have learned a lot about the limitations and challenges involved in using these models to support development of new therapeutic molecules22.
One of the most important lessons so far is that use of suboptimal models might lead to misleading results, wasting valuable resources and time23.
Before initiating any project involving genetically engineered mice, it is prudent to invest time identifying the optimal experimental model. If no suitable transgenic line is available, it might be appropriate to consider generating a new model or modifying an existing one.
Often, however, the time required for the introduction of a specific mutant allele into the mouse genome is not compatible with the pace of the drug development pipeline. This leaves scientists with the difficult choice of slowing down the development of a new therapeutic molecule, or using existing models that have not been designed for the specific experimental goals of the drug development program.
Taconic is introducing new approaches to shorten mouse model generation timelines without compromising the necessary quality controls, with the ultimate goal of allowing researchers to access relevant models within timeframes compatible with their drug development goals.
References:
1. Gibert Y, Trengove M.C., Ward A.C. (2013). Zebrafish as a genetic model in pre-clinical drug testing and screening. Curr Med Chem. 20(19):2458-66.
2. Artal-Sanz M, de Jong L, Tavernarakis N. (2006). Caenorhabditis elegans: a versatile platform for drug discovery. Biotechnol J. Dec;1(12):1405-18.
3. Clark, A. T., Goldowitz, D., Takahashi, J. S., Vitaterna, M. H., Siepka, S. M., Peters, L. L., Frankel W.N., Carlson G.A., Rossant J., Nadeau J.H., Justice M. J. (2004). Implementing large-scale ENU mutagenesis screens in North America. Genetica. 122(1), 51-64.
4. Sean F., L., & Tian, X. (2011). Somatic Genetics Empowers the Mouse for Modeling and Interrogating Developmental and Disease Processes. PLoS Genetics. 7(7), e1002110.
5. Buchovecky, C. M., Turley, S. D., Brown, H. M., Kyle, S. M., McDonald, J. G., Liu, B., Pieper A.A., Huang W., Katz D.M., Russell D.W., Shendure J., Justice M. J. (2013). A suppressor screen in mouse Mecp2 implicates cholesterol metabolism in Rett Syndrome. Nature Genetics. 45(9), 1013-1020.
6. Kheirallah A.K., Miller S., Hall I.P., Sayers I., (2016). Chapter Two - Translating Lung Function Genome-Wide Association Study (GWAS) Findings: New Insights for Lung Biology.Advances in Genetics, Academic Press. Volume 93: Pages 57-145.
7. Sulayman D. Dib-Hajj, Stephen G. Waxman (2014). Translational pain research: Lessons from genetics and genomics. Science Translational: 249sr4.
8. Yazdi, F. T., Clee, S. M., & Meyre, D. (2015). Obesity genetics in mouse and human: back and forth, and back again. PeerJ. 3, e856.
9. Costa C, Hirsch E. More than just kinases: the scaffolding function of PI3K. (2010). Curr Top Microbiol Immunol. 2010;346:171-81.
10. Vaillancourt F.H., Brault M., Pilote L., Uyttersprot N., Gaillard E.T., Stoltz J.H., Knight B.K., Pantages L., McFarland M., Breitfelder S., Chiu T.T., Mahrouche L., Faucher AM., Cartier M., Cordingley M.G., Bethell R.C., Jiang H., White P.W., and Kukolj G. (2012). Evaluation of Phosphatidylinositol-4-Kinase IIIα as a Hepatitis C Virus Drug Target. J. Virol. 86:21 11595-11607.
11. He, G., Luo, W., Li, P., Remmers, C., Netzer, W., Hendrick, J., Bettayeb, K., Flajolet, M., Gorelick, F., Wennogle, L.P., Greengard, P. (2010). Gamma-secretase activating protein, a therapeutic target for Alzheimer's disease. Nature. 467(7311), 95-98.
12. RNAi.
13. Scheer N., Wilson I.D. (2016). A comparison between genetically humanized and chimeric liver humanized mouse models for studies in drug metabolism and toxicity. Drug Discovery Today. Volume 21, Issue 2, Pages 250-263.
14. Scheer N., Kapelyukh Y., Rode A., Oswald S., Busch D., McLaughlin L.A., Lin D., Henderson C.J., and Wolf R.C. (2015). Drug Metabolism Pathways in a Multiple Humanized Mouse Model. Drug Metabolism and Disposition. 43(11):1679-1690.
15. Zhang J, Heimbach T, Scheer N, Barve A, Li W, Lin W, He H. (2016). Clinical Exposure Boost Predictions by Integrating Cytochrome P450 3A4-Humanized Mouse Studies With PBPK Modeling. J Pharm Sci. 105(4):1398-404.
16. Saitoh A, Kimura M, Takahasi R, Yokoyama M, Nomura T, Izawa M, Sckiya T, Nishimura S, Katsuki M. (1990). Most tumors in transgenic mice with human c-Ha-ras gene contained somatically activated transgenes. Oncogene. 5(8):1195-1200.
17. Urano K, Tamaoki N, Nomura T. (2012). Establishing a laboratory animal model from a transgenic animal: rasH2™ mice as a model for carcinogenicity studies in regulatory science. Vet Pathol. 49(1):16-23.
18. Svensson KA, Heinz BA, Schaus JM, Beck JP, Hao J, Krushinski JH, Reinhard MR, Cohen MP, Hellman SL, Getman BG, Wang X, Menezes MM, Maren DL, Falcone JF, Anderson WH, Wright RA, Morin SM, Knopp KL, Adams BL, Rogovoy B, Okun I, Suter TM, Statnick MA, Gehlert DR, Nelson DL, Lucaites VL, Emkey R, DeLapp NW, Wiernicki TR, Cramer JW, Yang CR, Bruns RF. (2017). An allosteric potentiator of the dopamine D1 receptor increases locomotor activity in human D1 knock-in mice without causing stereotypy or tachyphylaxis. J Pharmacol Exp Ther. 360(1):117-128.
19. Ferla R, Claudiani P, Cotugno G, Saccone P, De Leonibus E, Auricchio A. (2014). Similar therapeutic efficacy between a single administration of gene therapy and multiple administrations of recombinant enzyme in a mouse model of lysosomal storage disease. Hum Gene Ther. Jul;25(7):609-18.
20. Alliegro M, Ferla R, Nusco E, De Leonibus C, Settembre C, Auricchio A. (2016). Low-dose Gene Therapy Reduces the Frequency of Enzyme Replacement Therapy in a Mouse Model of Lysosomal Storage Disease. Molecular Therapy. 24(12):2054-2063.
21. Sharpless, N. E., & DePinho, R. A. (2006). The mighty mouse: genetically engineered mouse models in cancer drug development. Nature reviews Drug discovery. 5(9):741-754.
22. McGonigle P, Ruggeri B. (2014). Animal models of human disease: challenges in enabling translation. Biochem Pharmacol. 87(1):162-71.
23. Taconic Insight: Why Your Transgenic Project Failed.
