New Paths Forward in Translational Medicine: Innovative Biological Models Gabrielle G. Leblanc, PhD, Bioscience Consulting

Introduction Translational research lies at the heart of drug development. “Translational” (or “bench-to-bedside”) research is the process whereby basic science discoveries are harnessed to develop new drugs, devices, and therapeutic approaches for use in human patients.1 A classic success story in translational research was the development of insulin therapy for diabetes, which began in 1869 with the discovery of the pancreatic islets of Langerhans, continued with Nobel Prize-winning experiments on the digestive physiology of dogs in the 1920s, and culminated in the large-scale commercial production of genetically engineered insulin in 1982.2, 3 In a more recent example, the 1995 discovery that spinal muscular atrophy (SMA) is caused by mutations in the Survival Motor Neuron 1 (SMN1) likely contributing to failure of new drug xenograft models: immunodeficient gene led directly to the development of candidates in clinical trials include: mice into which patient-derived tumors the first therapy for SMA, an antisense or tumor cell lines have been trans­ • Suboptimal trial design, poor drug that was approved by the FDA in planted. However, it is has become choices of patient populations (for December 2016.4, 5 increasingly clear that many mouse example, studying patients whose models fail to accurately recapitulate Lately, there has been a sense of crisis disease has already progressed the human disease and/or to predict in the drug discovery field. Despite too far for successful treatment), the efficacy and clinical side effects of huge advances over the last 30 years lack of validated disease and target candidate drugs. in biomedical technology and basic engagement biomarkers, failure science insights into disease mecha- of drugs to engage their intended Fortunately, there is a seismic shift nisms, there has been an increasingly targets at the doses used, and happening in the field of translational high failure rate of new candidate drugs insufficiently sensitive outcome research, as emerging technologies developed during the same period.6 measures. offer new possibilities for creating more Currently, it takes longer than a decade accurate, informative, less expensive, • Lack of rigor in preclinical animal and USD 2.6 billion on average to and higher-throughput biological models studies, including inadequate sample develop a new drug from target discovery for drug discovery. Genome editing sizes, poor study design, inappropriate to market entry, and only one out of ten techniques like TALENS and CRISPR statistical methods and the failure drug candidates entering clinical trials have revolutionized the precision, scale, to seek replication of positive results receives market approval.7 The 90% and speed with which we can generate or report negative results. of drug candidates of that fail in clinical new disease models, in large animals trials have been found to do so because • Poor predictive value of many as well as rodents.18–20 Burgeoning of low efficacy and safety issues.8 currently used preclinical in vitro and databases of human “omics” data are Failure rates are highest for cancer, in vivo model systems. enabling the reverse translation of mental health disorders, cardiovascular clinical findings to inform preclinical Of these, perhaps the most pervasive disease, and neurological disease, studies, generate new animal models, worry on the preclinical side concerns four of the leading causes of morbidity and test the validity of existing models. the inherent predictive validity of and mortality worldwide.9 Furthermore, developments in the field commonly used animal models of of bioengineering are spawning new This situation has prompted extensive disease. Drug discovery relies heavily on in vitro systems for studying disease reappraisals of current approaches genetically engineered animal models biology and the effects of novel drugs to translational research.10–17 Major of disease, especially mice. The cancer on human cells. issues that have been identified as field also makes heavy use of mouse

8 | IN TRANSLATION Building better animal models Failure rates are highest for cancer, mental health disorders, of disease cardiovascular disease, and neurological disease, four of In theory, the ideal animal model the leading causes of morbidity and mortality worldwide.” of disease would have the following features (reviewed in references21, 22): • Lack of genetic diversity in inbred can. Environmental as well as genetic 1. Replicate the human disease mice versus humans. risk factors contribute to most for phenotype (at all levels from the the specific purpose exploring gene/ • Comorbid conditions associated with molecular to the behavioral); environment interactions.33–35 In reverse age-related diseases in humans that translational approaches (see below), 2. Share underlying biological are not reproduced in mouse models. researchers can use human clinical mechanisms with the human disease; • Environmental risk factors that data to inform the creation of animal 3. Have predictive validity with respect contribute to most common human models that more accurately mimic to drug efficacy and safety in humans. diseases but not reproduced in mouse human disease states and compare models of those diseases. disease phenotypes in currently existing In practice, these criteria are rarely, if ever, animal models to those of human met. For example, the Alzforum website Despite these limitations, mouse models patients. Meanwhile, new genome now lists 127 genetically engineered nonetheless have been invaluable editing technologies like CRISPR mouse models of Alzheimer’s disease for establishing causative roles of are enabling the production of more (AD), and not one of them has yet been specific genes and gene variants in accurate genetically engineered animal shown to completely fulfill any of these disease, for understanding biological models.18–20 criteria.23–25 Key pathological features of pathways of disease, and identifying AD in humans include not only amyloid new drug targets. They have also been plaques but also tau tangles and critical in the development of the 10% Reverse translation neuronal cell death. Most AD model of drug candidates that do make it In “reverse translational” (or “bedside- mice do develop amyloid deposits, but through clinical trials. To give just a few to-bench”) research, data from human most don’t develop tau tangles or show examples: subjects is used to develop new neuronal cell death. In addition, it’s still • Despite the fact that mouse hypotheses for testing in the laboratory unclear how well any of these mouse models of rheumatoid arthritis (RA) and to develop new animal models models reproduce the underlying mech­ don’t perfectly model human RA, and therapeutics. Although reverse anisms of AD, especially for sporadic these models were pivotal in the translation is a recently coined term, this AD. And, to date, none of the many drug development of anti-tumor necrosis kind of research has been done for candidates developed in mouse models factor (anti-TNF; a translational centuries. For example, Edward Jenner’s of AD has proven effective in halting success that helped launch the 1796 discovery of the first smallpox or even slowing disease progression in biopharmaceutical industry);31 vaccine was based on the observation human clinical trials. that milkmaids who had previously • The recent discovery of the first Issues that likely contribute to the caught cowpox developed resistance therapy for SMA would likely have imperfect accuracy of mouse models to smallpox, and the vaccine’s success been impossible without SMN2 of disease include: helped lay the foundation for modern transgenic mice;4, 5 immunology.36 • Major differences in the basic biology • The development of most successful of rodents and humans. Over the past few years, molecular anti-cancer drugs has relied on mouse profiling of human patients has yielded • Lack of homology of molecular targets: xenograft models.32 vast quantities of “omics” data that can e.g., mice and humans express It is also possible that many cases be harnessed for reverse translational different isoforms of beta-amyloid.26 of “failure to translate” are due to poor research. For example, genome-wide • Lack of homology in molecular design and interpretation of preclinical association studies (GWAS) and next- pathways: transcription factors bind studies rather than to inadequacies generation sequencing (NGS) have to overlapping but different sets of in the animal models studied.13 identified hundreds of novel genes and genes in mouse vs. humans, and in Nonetheless, given the expense and gene variants associated with risk of some cases transcriptome changes ethical issues of working with vertebrate or protection against human diseases, in mouse models of a particular animals, we should not only improve including common sporadic disorders. disease barely resemble those seen the design of preclinical studies, but These data can be used to create new in humans.25, 27–30 also develop the best models we genetically engineered iPSC (induced

NEW PATHS FORWARD IN TRANSLATIONAL MEDICINE | 9 Although reverse translation is a recently coined term, this kind of research has been done for centuries.”

pluripotent stem cell) models as well can also suggest new targets for drug The microbiome is the array of micro­ as animal models of disease, which development.42–45 Molecular profiling organisms (, fungi, and other then can be used both to explore the can also be used to screen and optimize single celled organisms) that populate functions of the newly identified genes, cell-based therapeutics. In one recent the gut, skin, respiratory tract, and and to discover new disease pathways study, molecular profiling of over 100 other parts of the body that are directly and candidate drug targets.37–39 This different preparations of dendritic cell exposed to the environment. The gut kind of reverse translational approach (DC) vaccines targeting prostate cancer microbiome (the most thoroughly to generating animal models offers identified a signature of DC gene and studied) has now been shown to play a special boon for the study of rare protein expression that correlated essential roles in nutrient digestion, inherited diseases, for many of which with the induction of strong anti-tumor drug metabolism, and the development no animal models have previously responses in patients.46 In a modern and function of the immune and been available.40, 41 spin on the development of the smallpox nervous systems.51, 52 vaccine, immune profiling of humans Molecular profiling of patient tissue The contains who show resistance to certain diseases samples can be used to identify around 1,000 species of bacteria, (Alzheimer’s disease, progressive patterns of RNA and protein expression whose exact numbers and proportions multifocal leukoencephalopathy) has that correlate with disease resistance vary from person to person.53 The been used to develop antibody therapies and/or responsiveness to therapeutics. composition of an individual’s for these diseases.47, 48 In the cancer field, for example, gene microbiome can also change over time and protein expression profiling of Another important application of the in response to environmental factors, tumors has begun to define molecular reverse translational approach is in the including diet and sex hormones.54 signatures associated with better analysis of results of failed clinical Alterations in the gut microbiome have responses to immunotherapy and higher trials.49, 50 In one example, the anti-IL-12B been linked to a growing list of diseases, patient survival rates; these signatures p40 antibody, which showed promise as including obesity, diabetes, irritable a therapeutic for multiple sclerosis (MS) bowel syndrome, cardiovascular based on results in mice and marmoset disease, cancer, and autism.55, 56, 57 The models of experimental autoimmune gut microbiome also helps determine encephalomyelitis (EAE), failed in human drug efficacy and side effects.58, 59 trials. Subsequent analysis of disease Transfer of gut from one progression in the mouse and marmoset animal to another allows direct testing EAE models vs. human MS showed of suspected roles of the microbiome that (1) the initiation and progression in disease, and may also enable phases of the disease are driven by the creation of new animal models different mechanisms in primates, (2) of disease.60, 61 the mouse replicates only the initiation The metabolome is the full set of small mechanism, and (3) the drug blocks molecule chemicals (sugars, amino only the initiation mechanism.50 acids, lipids, etc.) found in a given bio­ logical sample, including metabolites Targeting environment-gene generated by the microbiome as well as interactions: the microbiome those produced by an individual’s own and metabolome cells. The metabolome lies at the direct interface of the environment with the Most common human diseases result genome and microbiome, and provides from interactions between environmental a dynamic readout of the current state factors such as diet and exercise, with of an individual’s health.62, 63 genetic risk factors. However, outside the areas of cancer and metabolic High-throughput profiling of the diseases, environmental risk factors are microbiome and metabolome, made typically not built into animal models. possible by recent developments in This situation is changing now, in part genome sequencing and chemical as a result of increased awareness analytic technologies (e.g., automated, of the roles of the microbiome and the quantitative NMR and liquid or gas metabolome in health and disease. chromatography coupled with mass spectrometry) can identify new disease

10 | IN TRANSLATION signatures and translatable bio­­ 20,000 gene targets, the human Whereas the human markers, and generate hypotheses for microbiome offers several million, and genome provided reverse translational research.57, 62, 64 the metabolome offers not only targets These approaches now have been but also natural product drug leads.56 20,000 gene targets, the used to discover the first pre-clinically human microbiome offers successful microbiome-targeting “Disease-in-a-dish” models several million.” drugs in the areas of cancer and cardiovascular disease.55 In the latter Cell culture systems are advantageous case, an untargeted for preclinical studies because they normally function and communicate screen in human patients suggested offer simplified biological models in with one another, and (2) employ non- that a microbe-derived metabolite which environmental factors can be human cells and/or immortalized cell trimethylamine N-oxide (TMAO) was tightly controlled. Compared to in vivo lines that have been selected based associated with greater disease risk.65 studies, in vitro studies are generally on their ability to grow under non- Subsequent animal studies confirmed much faster, can be done at much physiological conditions. In addition, a causal link between elevated TMAO higher throughput and lower cost, and most 2D culture systems used in drug and atherosclerosis, and identified a are largely free of the ethical issues. screening include only one cell type. small molecule inhibitor of the microbial To date, 2D cell culture models have However, we now know that many TMAO pathway that attenuates disease been the norm for early-stage drug human diseases involve dysfunctional progression in mouse models.55 screening. However, some traditional interactions between two or more cell The microbiome and the metabolome 2D cell culture models are inherently types. For example, defects in glial, each offer huge, virtually untapped non-physiological because they (1) lack immune, and/or vascular cell function sources of potential drug targets. the 3D architecture under which cells initiate or contribute to neuronal Whereas the human genome provided degen­eration in many neurological

NEW PATHS FORWARD IN TRANSLATIONAL MEDICINE | 11 diseases, including Huntington’s disease, contain many classes of neurons (as gene-editing technologies like CRISPR ALS, and Alzheimer’s disease. Similarly, well as astrocytes), develop functional or TALENS, it is possible to introduce the growth of tumor cells is strongly synapses and circuits, and show precise mutations or combinations of affected by their interactions with region-specific patterning (e.g., cortical mutations into iPSCs for the purpose stromal, immune, and vascular cells. layers).69, 71 A key feature of iPSC of analyzing genotype-phenotype technology is that it can be used to relationships—a boon for the study of Over the past decade, advances in generate cultures of specific cell types both single-gene and complex genetic stem cell biology, tissue engineering, carrying a patient’s own, individual disorders.70, 72 and microfluidics have spawned a genetic makeup. In many cases, patient- plethora of exciting new in vitro model Bioprinting: In bioprinting, 3D tissue- or derived iPSCs recapitulate cellular systems that mimic in vivo 3D cellular organ-like structures are constructed phenotypes similar to those seen in architecture and can include multiple layer-by-layer by 3D printing machines human patients in vivo.67, 70 For example, cell types. These include the following: that deposit precisely patterned sheets iPSC-derived neurons generated from of living cells, extracellular matrix, Induced pluripotent stem cell (iPSC) fibroblasts of patients with genetic and other bioreagents.73, 74 Bioprinted cultures: iPSCs have caused a revolution forms of Parkinson’s disease show organs can incorporate multiple cell in translational research because many disease-appropriate abnormalities, types and can be created from either they provide unlimited supplies of including reduced neurite outgrowth, primary cells (including tumor cells) human cells for in vitro studies.66 increased sensitivity to oxidative stress, or iPSCs. One challenge has been to iPSCs can be cultured in 2D, or in 3D and elevated α-synuclein levels.70 Thus, provide these artificial organs with a to create “organoids” with complex iPSCs offer powerful tools for precision blood supply, as tissue development architectures.67–70 For example, human medicine, including ex vivo testing of and function in vivo requires functional, iPSC-derived neural precursor cells can patient-specific disease mechanisms hierarchically organized vasculature. self-organize into “mini-brains” that and drug responses. Moreover, using Recently, methods have been

12 | IN TRANSLATION Up to five different organ types have been functionally coupled on a single hardware platform.”

developed to incorporate preformed vascular beds into bioprinted tissues, which can then form functional connections with the vascular system of a living host animal.75, 76 Organ-on-a-chip, patient-on-a-chip: Bioprinted tissues can be combined with microfluidic, “lab-on-a-chip” platforms that provide tissue perfusion, delivery of compounds, and continuous measurements of tissue responses. This “organ-on-a-chip” approach has now been used to create in vitro models of several diseases, including cancers, lung diseases and nonalcoholic steatohepatitis (NASH).47, 48, 77, 78 In the latest generation of this technology, called “body-on-a-chip” or “patient-on- a-chip,” up to five different organ types have been functionally coupled on a single hardware platform, allowing the study of disease and drug effects on and are readily amenable to genetic • High-throughput phenotype-based complex organ system interactions.48, 78 manipulations that enable the study screening (see section above). In addition to the advantages of basic mechanisms of biology and • Reverse genetic screens to identify previously cited, these new “disease- disease.81 Compared to in vitro systems, new molecular partners of disease in-a-dish” models can all be used for intact animals can be used to assess genes and new drug targets. high-throughput drug screens and cell- a much broader range of phenotypes, and organ-specific toxicology screens. including behavior, gut motility, and • Enhancer/suppressor screens for They can be also used for high- cardiovascular function and zebrafish drug discovery. throughput phenotype-based screening. even possess a blood brain barrier.82 • Combined screening and counter- Phenotype-based screening used to C. elegans and zebrafish have the screening (for therapeutic and adverse be basis of all drug development, but additional advantage of being transpar­ effects of drugs) in the same assay. in recent decades had been largely ent, so their cells can be fluorescently replaced by target-based approaches. labeled and visualized over time in Automated phenotypic assays have been Phenotype-based screens are now living animals. Many genetic and developed for several of these organisms, having a renaissance, because they biological pathways are conserved as have species-specific mechanism of may be better for discovering “first in from lower organisms to humans. For action discovery tools.79, 81, 86, 87 class” drugs, and can identify drugs example, over 80% of human disease High-throughput drug screens that exert beneficial effects by acting genes are conserved in zebrafish, and have now been conducted in worm, on multiple biochemical pathways 60% in the fruit fly, and mutation of fruit fly, and/or zebrafish for drugs simultaneously.79, 80 human disease genes in lower organisms to combat infectious diseases, often produces cellular phenotypes cancer, neurodegeneration, aging, comparable to those seen in humans.83, 84 Non-rodent models cardiomyopathies, and many other In addition, invertebrates and lower diseases.79, 81, 87–89 A number of approved vertebrates often have unique regen­ Non-traditional small animal models cancer drugs, including crizotinib, erative abilities (zebrafish, for example, gefitinib, and vandetanib, were developed Non-traditional small animal models, can regenerate heart tissue), and so or validated in the fruit fly88 and a drug such as the nematode worm C. elegans, are of particular interest in regenerative discovered in zebrafish is now in the fruit fly Drosophila( melanogaster) medicine.81, 85 clinical trials for treatment of hemato­ and zebrafish, can provide a bridge Additional advantages of these logical malignancies.79 between in vitro and rodent models. non-traditional animal models are that They are cheap, fast, and easy to breed, they can be used for:

NEW PATHS FORWARD IN TRANSLATIONAL MEDICINE | 13 Large animal models applied in pigs (over 25 lines of gene- 4. Singh NN, Howell MD, Androphy EJ, Singh edited mini-pigs have now been RN. How the discovery of iss-n1 led to the first The anatomy and physiology of rodents medical therapy for spinal muscular atrophy. generated) and non-human primates.94, 95 differs from that of humans in many Gene therapy. 2017 crucial ways, including the following 5. Awano T, Kim JK, Monani UR. Spinal muscular (reviewed in reference 90): Conclusion atrophy: Journeying from bench to bedside. Neurotherapeutics : the journal of the American • Immune mechanisms. For example, During the past few years, researchers Society for Experimental NeuroTherapeutics. lymphocytes are the predominant involved in translational science and 2014;11:786–795 population of white blood cells in drug discovery have been harnessing 6. Pammolli F, Magazzini L, Riccaboni M. The productivity crisis in pharmaceutical r&d. Nature mouse blood, but neutrophils are the exciting new technologies to enable reviews. Drug discovery. 2011;10:428–438 main type in humans. the reappraisal and fine-tuning of 7. DiMasi JA, Grabowski HG, Hansen RW. The traditional models, and to create new • Composition of the microbiome. cost of drug development. The New England models. As detailed above, there is journal of medicine. 2015;372:1972 • Cardio- and cerebrovascular increasing emphasis on overcoming 8. Waring MJ, Arrowsmith J, Leach AR, Leeson architecture. limitations of currently available in PD, Mandrell S, Owen RM, et al. An analysis of vitro and in vivo models by improving the attrition of drug candidates from four major • Complexity of the CNS and pharmaceutical companies. Nature reviews. ease and cost of use, and predictive behavioral repertoires. Drug discovery. 2015;14:475–486 validity. These innovative model 9. BIO. Clinical development success rates • Proportions of glial cells. In the systems include new animal models 2006–2015. 2016 cerebral cortex, the glial: neuron that more closely recapitulate human 10. Benatar M. Lost in translation: Treatment ratio is 10:1 in humans, compared disease syndromes, and new in vitro trials in the sod1 mouse and in human als. to 1:1 in mice,91 a difference that models that support easier, deeper Neurobiology of disease. 2007;26:1–13 may be critical when modeling study of human disease pathways and 11. Shineman DW, Basi GS, Bizon JL, Colton CA, neurodegenerative diseases, where “personalized” drug testing. Greenberg BD, Hollister BA, et al. Accelerating glia are now believed to play major drug discovery for alzheimer's disease: Best The mission of the inaugural Charles practices for preclinical animal studies. roles in most diseases. River World Congress on Animal Models Alzheimer's research & therapy. 2011;3:28 Large mammals often reproduce key in Drug Discovery & Development is to 12. van der Worp HB, Howells DW, Sena ES, features of human biology and disease describe and discuss some of these Porritt MJ, Rewell S, O'Collins V, et al. Can animal models of disease reliably inform human more closely than rodents do and are cutting-edge innovations and break­ studies? PLoS medicine. 2010;7:e1000245 closer to humans genetically: rodents throughs in translational tools and 13. Jucker M. The benefits and limitations of have only 48-66% genetic homology methods in drug discovery research animal models for translational research in with humans, whereas swine and New and development. The conference has neurodegenerative diseases. Nature medicine. World monkeys have 80% and Old World been designed to create a platform 2010;16:1210–1214 monkeys (e.g. baboons) up to 99%.92 that fosters the sharing of knowledge, 14. Karran E, Hardy J. A critique of the drug Animals that are closer to humans in views and collaborations through a discovery and phase 3 clinical programs targeting the amyloid hypothesis for alzheimer size are necessary for testing surgical mix of presentations and interactive disease. Annals of neurology. 2014;76:185–205 devices and procedures, and also are networking sessions. We anticipate 15. Herter-Sprie GS, Kung AL, Wong KK. New more suitable for dosage, delivery and that conference attendees will be cast for a new era: Preclinical cancer drug safety studies. Large animals that inspired to be a part of breakthroughs in development revisited. The Journal of clinical have proven useful in studying human translational tools and methods in drug investigation. 2013;123:3639–3645 diseases include the pig (cardiovascular discovery research and development. 16. Scannell JW, Blanckley A, Boldon H, disease, obesity, cystic fibrosis)90,­ 93–95 Warrington B. Diagnosing the decline in pharmaceutical r&d efficiency. Nature reviews. and non-human primates (neurological Drug discovery. 2012;11:191–200 and psychiatric disease).92, 96 References 17. Mak IW, Evaniew N, Ghert M. Lost in translation: Animal models and clinical trials Apart from the issues of cost and ethics, 1. Emmert-Buck MR. Translational research: in cancer treatment. American journal of From biological discovery to public benefit the major limitation to using large translational research. 2014;6:114–118 animal models in translational research (or not). Advances in Biology. 2014;2014 18. Moore JD. The impact of -cas9 on 2. Bliss M. The discovery of insulin: Twenty-fifth has been the difficulty of creating target identification and validation. Drug anniversary edition. University Of Chicago Press; genetically engineered animals. This discovery today. 2015;20:450–457 2007. situation is changing now with the 19. Dow LE. Modeling disease in vivo with 3. Cooper T, Ainsberg A. Breakthrough: Elizabeth advent of gene editing technologies, crispr/cas9. Trends in molecular medicine. hughes, the discovery of insulin, and the making 2015;21:609–621 which have now been successfully of a medical miracle. Publisher: St. Martin's Griffin; 2011.

14 | IN TRANSLATION 20. Fellmann C, Gowen BG, Lin PC, Doudna JA, 34. Asgharpour A, Cazanave SC, Pacana T, 47. Feaver RE, Cole BK, Lawson MJ, Hoang SA, Corn JE. Cornerstones of crispr-cas in drug Seneshaw M, Vincent R, Banini BA, et al. A Marukian S, Blackman BR, et al. Development of discovery and therapy. Nature reviews. Drug diet-induced animal model of non-alcoholic fatty an in vitro human liver system for interrogating discovery. 2017;16:89–100 liver disease and hepatocellular cancer. Journal nonalcoholic steatohepatitis. JCI insight. of hepatology. 2016;65:579–588 2016;1:e90954 21. Nestler EJ, Hyman SE. Animal models of neuropsychiatric disorders. Nature 35. Morris EM, McCoin CS, Allen JA, Gastecki 48. Vernetti LA, Senutovitch N, Boltz R, DeBiasio neuroscience. 2010;13:1161–1169 ML, Koch LG, Britton SL, et al. Aerobic capacity R, Shun TY, Gough A, et al. A human liver mediates susceptibility for the transition from microphysiology platform for investigating 22. McGonigle P, Ruggeri B. Animal models steatosis to steatohepatitis. The Journal of physiology, drug safety, and disease of human disease: Challenges in enabling physiology. 2017;595:4909–4926 models. Experimental biology and medicine. translation. Biochemical pharmacology. 2016;241:101–114 2014;87:162–171 36. Morgan AJ, Parker S. Translational mini-review series on vaccines: The edward 49. Stephenson D, Hu MT, Romero K, Breen K, 23. Laurijssens B, Aujard F, Rahman A. jenner museum and the history of vaccination. Burn D, Ben-Shlomo Y, et al. Precompetitive data Animal models of alzheimer's disease and Clinical and experimental immunology. sharing as a catalyst to address unmet needs drug development. Drug discovery today. 2007;147:389–394 in parkinson's disease. Journal of Parkinson's Technologies. 2013;10:e319–327 disease. 2015;5:581–594 37. Manolio TA, Fowler DM, Starita LM, Haendel 24. Onos KD, Sukoff Rizzo SJ, Howell GR, Sasner MA, MacArthur DG, Biesecker LG, et al. Bedside 50. t Hart BA, Jagessar SA, Kap YS, Haanstra KG, M. Toward more predictive genetic mouse back to bench: Building bridges between Philippens IH, Serguera C, et al. Improvement models of alzheimer's disease. Brain research basic and clinical genomic research. Cell. of preclinical animal models for autoimmune- bulletin. 2016;122:1–11 2017;169:6–12 mediated disorders via reverse translation 25. Burns TC, Li MD, Mehta S, Awad AJ, of failed therapies. Drug discovery today. 38. Mitchell KJ, Huang ZJ, Moghaddam B, Sawa Morgan AA. Mouse models rarely mimic the 2014;19:1394–1401 A. Following the genes: A framework for animal transcriptome of human neurodegenerative modeling of psychiatric disorders. BMC biology. 51. Sharon G, Sampson TR, Geschwind DH, diseases: A systematic bioinformatics-based 2011;9:76 Mazmanian SK. The central nervous system and critique of preclinical models. European journal the gut microbiome. Cell. 2016;167:915–932 of pharmacology. 2015;759:101–117 39. da Silva Xavier G, Bellomo EA, McGinty JA, French PM, Rutter GA. Animal models of 52. Shreiner AB, Kao JY, Young VB. The gut 26. Cavanaugh SE, Pippin JJ, Barnard gwas-identified type 2 diabetes genes. Journal microbiome in health and in disease. Current ND. Animal models of alzheimer disease: of diabetes research. 2013;2013:906590 opinion in gastroenterology. 2015;31:69–75 Historical pitfalls and a path forward. Altex. 2014;31:279–302 40. Shen T, Lee A, Shen C, Lin CJ. The long 53. Human Microbiome Project C. Structure, tail and rare disease research: The impact of function and diversity of the healthy human 27. Seok J, Warren HS, Cuenca AG, Mindrinos next-generation sequencing for rare mendelian microbiome. Nature. 2012;486:207–214 MN, Baker HV, Xu W, et al. Genomic responses disorders. Genetics research. 2015;97:e15 in mouse models poorly mimic human 54. Org E, Mehrabian M, Parks BW, Shipkova P, Liu inflammatory diseases. Proceedings of the 41. Miller JN, Kovacs AD, Pearce DA. The novel X, Drake TA, et al. Sex differences and hormonal National Academy of Sciences of the United cln1(r151x) mouse model of infantile neuronal effects on composition in mice. States of America. 2013;110:3507–3512 ceroid lipofuscinosis (incl) for testing nonsense Gut microbes. 2016;7:313–322 suppression therapy. Human molecular genetics. 28. Takao K, Miyakawa T. Genomic responses 55. Brown JM, Hazen SL. Targeting of microbe- 2015;24:185–196 in mouse models greatly mimic human derived metabolites to improve human health: inflammatory diseases. Proceedings of the 42. Kwa M, Makris A, Esteva FJ. Clinical utility of The next frontier for drug discovery. The Journal National Academy of Sciences of the United gene-expression signatures in early stage breast of biological chemistry. 2017;292:8560–8568 States of America. 2015;112:1167–1172 cancer. Nature reviews. Clinical oncology. 2017 56. Garber K. Drugging the gut microbiome. 29. Weidner C, Steinfath M, Opitz E, 43. Lee JS. Exploring cancer genomic data from Nature biotechnology. 2015;33:228–231 Oelgeschlager M, Schonfelder G. Defining the the cancer genome atlas project. BMB reports. 57. Org E, Blum Y, Kasela S, Mehrabian M, optimal animal model for translational research 2016;49:607–611 Kuusisto J, Kangas AJ, et al. Relationships using gene set enrichment analysis. EMBO 44. Hendrickx W, Simeone I, Anjum S, Mokrab between gut microbiota, plasma metabolites, and molecular medicine. 2016;8:831–838 Y, Bertucci F, Finetti P, et al. Identification of metabolic syndrome traits in the metsim cohort. 30. Galatro TF, Vainchtein ID, Brouwer N, genetic determinants of breast cancer immune Genome biology. 2017;18:70 Boddeke EW, Eggen BJ. of microglia phenotypes by integrative genome-scale 58. Dubin K, Callahan MK, Ren B, Khanin R, Viale and immune infiltrates from mouse and primate analysis. Oncoimmunology. 2017;6:e1253654 A, Ling L, et al. Intestinal microbiome analyses central nervous system. Methods in molecular 45. Shen T, Pajaro-Van de Stadt SH, Yeat NC, identify melanoma patients at risk for checkpoint- biology. 2017;1559:333–342 Lin JC. Clinical applications of next generation blockade-induced colitis. Nature communications. 31. Feldmann M. Development of anti-tnf sequencing in cancer: From panels, to exomes, 2016;7:10391 therapy for rheumatoid arthritis. Nature reviews. to genomes. Frontiers in genetics. 2015;6:215 59. Enright EF, Gahan CG, Joyce SA, Griffin BT. Immunology. 2002;2:364–371 46. Castiello L, Sabatino M, Ren J, Terabe M, The impact of the gut microbiota on drug 32. Klinghammer K, Walther W, Hoffmann J. Khuu H, Wood LV, et al. Expression of cd14, metabolism and clinical outcome. The Yale journal Choosing wisely—preclinical test models in the il10, and tolerogenic signature in dendritic of biology and medicine. 2016;89:375–382 era of precision medicine. Cancer treatment cells inversely correlate with clinical and 60. Ellekilde M, Selfjord E, Larsen CS, Jakesevic reviews. 2017;55:36–45 immunologic response to tarp vaccination M, Rune I, Tranberg B, et al. Transfer of gut in prostate cancer patients. Clinical cancer 33. Koch LG, Pollott GE, Britton SL. Selectively microbiota from lean and obese mice to research : an official journal of the American bred rat model system for low and high antibiotic-treated mice. Scientific reports. Association for Cancer Research. 2017 response to exercise training. Physiological 2014;4:5922 . 2013;45:606–614

NEW PATHS FORWARD IN TRANSLATIONAL MEDICINE | 15 61. Zhu W, Gregory JC, Org E, Buffa JA, Gupta 75. Richards D, Jia J, Yost M, Markwald R, 90. Tsang HG, Rashdan NA, Whitelaw CB, N, Wang Z, et al. Gut microbial metabolite Mei Y. 3d bioprinting for vascularized tissue Corcoran BM, Summers KM, MacRae VE. Large tmao enhances platelet hyperreactivity and fabrication. Annals of biomedical engineering. animal models of cardiovascular disease. Cell thrombosis risk. Cell. 2016;165:111–124 2017;45:132–147 biochemistry and function. 2016;34:113–132 62. Wishart DS. Emerging applications of 76. Zhu W, Qu X, Zhu J, Ma X, Patel S, Liu J, et al. 91. Herculano-Houzel S. The glia/neuron metabolomics in drug discovery and precision Direct 3d bioprinting of prevascularized tissue ratio: How it varies uniformly across brain medicine. Nature reviews. Drug discovery. constructs with complex microarchitecture. structures and species and what that means 2016;15:473–484 Biomaterials. 2017;124:106–115 for brain physiology and evolution. Glia. 2014;62:1377–1391 63. Beger RD, Dunn W, Schmidt MA, Gross SS, 77. Zhang YS, Zhang YN, Zhang W. Cancer-on- Kirwan JA, Cascante M, et al. Metabolomics a-chip systems at the frontier of nanomedicine. 92. Eaton SL, Wishart TM. Bridging the gap: enables precision medicine: "A white paper, Drug discovery today. 2017 Large animal models in neurodegenerative perspective". Metabolomics : Official research. Mammalian genome : official journal 78. Skardal A, Shupe T, Atala A. Organoid-on- journal of the Metabolomic Society. 2016;12:149 of the International Mammalian Genome Society. a-chip and body-on-a-chip systems for drug 2017 64. Kulkarni H, Mamtani M, Wong G, Weir JM, screening and disease modeling. Drug discovery Barlow CK, Dyer TD, et al. Genetic correlation today. 2016;21:1399–1411 93. Keiser NW, Engelhardt JF. New animal of the plasma lipidome with type 2 diabetes, models of cystic fibrosis: What are they 79. MacRae CA, Peterson RT. Zebrafish as prediabetes and insulin resistance in mexican teaching us? Current opinion in pulmonary tools for drug discovery. Nature reviews. Drug american families. BMC genetics. 2017;18:48 medicine. 2011;17:478–483 discovery. 2015;14:721–731 65. Wang Z, Klipfell E, Bennett BJ, Koeth R, 94. Whitelaw CB, Sheets TP, Lillico SG, Telugu 80. Moffat JG, Vincent F, Lee JA, Eder J, Levison BS, Dugar B, et al. Gut flora metabolism BP. Engineering large animal models of Prunotto M. Opportunities and challenges of phosphatidylcholine promotes cardiovascular human disease. The Journal of pathology. in phenotypic drug discovery: An industry disease. Nature. 2011;472:57–63 2016;238:247–256 perspective. Nature reviews. Drug discovery. 66. Scudellari M. How ips cells changed the 2017 95. Fan N, Lai L. Genetically modified pig models world. Nature. 2016;534:310–312 for human diseases. Journal of genetics and 81. Strange K. Drug discovery in fish, flies, and genomics = Yi chuan xue bao. 2013;40:67–73 67. Avior Y, Sagi I, Benvenisty N. Pluripotent worms. ILAR journal. 2016;57:133–143 stem cells in disease modelling and drug 96. Kaiser T, Feng G. Modeling psychiatric 82. Neuwelt EA, Bauer B, Fahlke C, Fricker discovery. Nature reviews. Molecular cell biology. disorders for developing effective treatments. G, Iadecola C, Janigro D, et al. Engaging 2016;17:170–182 Nature medicine. 2015;21:979–988 neuroscience to advance translational research 68. Liu F, Huang J, Ning B, Liu Z, Chen S, Zhao in brain barrier biology. Nature reviews. W. Drug discovery via human-derived stem Neuroscience. 2011;12:169–182 cell organoids. Frontiers in pharmacology. 83. Fernandez-Hernandez I, Scheenaard E, 2016;7:334 Pollarolo G, Gonzalez C. The translational 69. Quadrato G, Brown J, Arlotta P. The promises relevance of drosophila in drug discovery. EMBO and challenges of human brain organoids as reports. 2016;17:471–472 models of neuropsychiatric disease. Nature 84. Shim H, Kim JH, Kim CY, Hwang S, Kim medicine. 2016;22:1220–1228 H, Yang S, et al. Function-driven discovery of 70. Shi Y, Inoue H, Wu JC, Yamanaka S. Induced disease genes in zebrafish using an integrated pluripotent stem cell technology: A decade genomics big data resource. Nucleic acids of progress. Nature reviews. Drug discovery. research. 2016;44:9611–9623 2017;16:115–130 85. Yin VP, Lepilina A, Smith A, Poss KD. 71. Quadrato G, Nguyen T, Macosko EZ, Regulation of zebrafish heart regeneration Sherwood JL, Min Yang S, Berger DR, et by mir-133. Developmental biology. al. Cell diversity and network dynamics in 2012;365:319–327 photosensitive human brain organoids. Nature. 86. Chao HT, Liu L, Bellen HJ. Building dialogues 2017;545:48–53 between clinical and biomedical research 72. Karakikes I, Termglinchan V, Cepeda DA, Lee through cross-species collaborations. Seminars J, Diecke S, Hendel A, et al. A comprehensive in cell & developmental biology. 2017 talen-based knockout library for generating 87. O'Reilly LP, Luke CJ, Perlmutter DH, human-induced pluripotent stem cell-based Silverman GA, Pak SC. C. Elegans in high- models for cardiovascular diseases. Circulation throughput drug discovery. Advanced drug research. 2017;120:1561–1571 delivery reviews. 2014;69–70:247–253 73. Nguyen DG, Pentoney SL. Bioprinted three 88. Yadav AK, Srikrishna S, Gupta SC. Cancer dimensional human tissues for toxicology drug development using drosophila as an in vivo and disease modeling. Drug Discovery Today: tool: From bedside to bench and back. Trends in Technologies. 2017;23:37–44 pharmacological sciences. 2016;37:789–806 74. Peng W, Unutmaz D, Ozbolat IT. Bioprinting 89. Huiting LN, Laroche F, Feng H. The zebrafish towards physiologically relevant tissue models as a tool to cancer drug discovery. Austin for pharmaceutics. Trends in biotechnology. journal of pharmacology and therapeutics. 2016;34:722–732 2015;3:1069

16 | IN TRANSLATION