Abstract
Translation of scientific discoveries into meaningful human applications, particularly novel therapies of human diseases, requires development of suitable animal models. Experimental approaches to test new drugs in preclinical phases often necessitated animal models that not only replicate human disease in etiopathogenesis and pathobiology but also biomarkers development and toxicity prediction. Whereas the transgenic and knockout techniques have revolutionized manipulation of rodents and other species to get greater insights into human disease pathogenesis, we are far from generating ideal animal models of most human disease states. The challenges in using the currently available animal models for translational research, particularly for developing potentially new drugs for human disease, coupled with the difficulties in toxicity prediction have led some researchers to develop a scoring system for translatability. These aspects and the challenges in selecting an animal model among those that are available to study human disease pathobiology and drug development are the topics covered in this detailed review.
The advancement of medical field is greatly conditioned by the discoveries in pathobiology of human disease and drug development. Translational research is the process of transforming such discoveries into human application. Conventional experimentation in biomedical sciences had hitherto heavily relied on experimentation in vitro models such as cell culture and proteomic studies. Whereas in vitro studies offer distinct advantages for evaluating the effects on a cell system of a given factor in a biological system or a pathophysiological environment, the relevance of such findings to the whole organism is limited for many reasons. Whole animal studies are not only preferred for biomedical experimentation but rather essential for translational research. Developing animal models that simulate any given human disease is a crucial part of translational research and is the primary focus of this review.
IN VITRO EXPERIMENTATION AND LIMITATIONS
Cell culture studies and tissue culture experiments are often performed to test a hypothesis or to examine the response of a cell or tissue preparation to a specific stimulus or inhibitory agent, often a humoral factor or a cytokine. Although such studies answer the question being posed, often with purity and precision, such observations have limited clinical significance. Often the cells used are derived from an established commercial cell line, which involves viral transfection and thus have altered cellular characteristics and the genetic composition. Cells derived from primary cell culture can circumvent this particular disadvantage but are relatively cumbersome to obtain and cannot be maintained past a few cell cycle passages. Often, studies performed in perfused isolated organs will provide more information than cell culture studies, but all the in vitro systems have the same major disadvantage: they lack the complex interactive physiology of the whole animal model. In particular, cell-cell cross talk and the paracrine and endocrine effects of the humoral systems in the whole animal facilitate and/or influence the responses of a specific cell or protein expression and activity. This is a major limitation in the interpretations of cellular responses in cell culture studies or tissue responses in the studies of isolated organ systems. Thus, whereas in vitro studies continue to contribute to understanding the pathophysiology of human disease processes and to drug development, whole animal models are crucial and pivotal for in translational research.
Animal Models for Translational Research
Research models are developed to mimic a disease, and these can be mathematical models such as computer simulations or animal models. An animal model is developed to resemble human physiology or be a surrogate for the human disease process. Such models are used to give deeper insights into the biochemistry and genetics process of human health and disease. It is important to recognize that animal models of human disease do not necessarily have identical features to their human counterparts. However, notwithstanding this fact, animal models have made significant contributions to the understanding of the pathophysiology and molecular pathology of human diseases and are valuable tools for developing innovative therapies for many human disease conditions.
Types of Animal Models
There are several varieties of animal models that have been used historically, and many of them continue to be used while recent technologies now facilitate development of many newer and refined animal models of human disease. Broadly speaking, there are 4 categories of animal models: (1) the inbred strains, which have been used extensively, (2) disease induction models, (3) xenograft animal models, and (4) genetically engineered models (Table 1) . Inbreeding has classically been used to obtain genetically homogeneous animals. Disease induction models are very commonly used to examine pathophysiology and drug development. Streptozotocin-injected rats and mice, which develop diabetes mellitus, is a typical example. Streptozotocin severely injures pancreatic islets and produces an equivalent of human type 1 diabetes. Similarly, 5 of 6 nephrectomy murine models are very commonly used to replicate and study the pathophysiology of human chronic kidney disease. Xenograft animals are used especially in cancer drug development. In general, to create a xenograft model, the tissue or an organ from one species is transplanted into another species. To make the animal more representative, it may be necessary also to inject human marrow cells into the mice to reconstitute the mice’s immune system and “humanize” them so that their responses to a drug are more akin to those in a human tumor microenvironment. Nude nonobese diabetic (NOD)/SCID mice, which lack murine immune responsiveness, are often used to examine the responses of human cancer to novel therapeutic agents.
However, the development of genetically engineered models has made the most valuable contributions to this field. Genetically engineered models are developed by altering the genetic composition of an animal by mutating, deleting, or overexpressing a targeted gene.
Knockout Models
The knockout models are developed by knocking out one or several genes. These techniques have been so powerful and had such an impact on our understanding of human disease that the scientists who developed this approach won a Nobel prize. A detailed description of the techniques is beyond the scope of this manuscript, but the reader is referred to the cited references.
Transgenic Models
Transgenic animals are generated by inserting a desired gene into the model so the effects of such a gene can be studied on the physiology or pathophysiology of the animal. The genetic DNA can be inserted either into the fertilized egg or by injecting into embryonic stem cells, which are then cultured and introduced into the blastocyst. The transgenic modifications were originally performed predominantly in mice but are now routinely done in rats, primates, and other species.
Characteristics of an Ideal Model to Study Human Disease
Defining the characteristics of an animal model that is ideal for studying human disease is a major challenge. The most important requirement is the similarity of the disease pathogenesis in the animal model to the human disease process. However, verification of this implies detailed understanding of pathobiology in human disease, which creates a vicious circle for investigation.1The situation is compounded by the fact that the molecular pathogenesis of many common human diseases such as hypertension and diabetes remains unclear until today. Dissimilarities in pathogenesis from human disease preclude suitability of an animal model for drug development for the human diseases in question. Furthermore, an ideal animal model should demonstrate the same biomarkers as in the human disease. This feature of biomarkers is important not only for monitoring disease progression but also for predicting drug toxicity.
USE OF MODELS TO DEVELOP BIOMARKERS OF DISEASE
Biomarkers are measurable parameters of biological processes and are very important tools in translational science. Some biomarkers reflect pathophysiology and may represent significant signaling pathways in the disease process. Thus, the presence of similar biomarkers in animal model and human disease indicates that pathogenic pathways are identical in both systems. The effects and effectiveness of a new drug can then be evaluated by the expected changes in the biomarkers further validating the animal models of a given human disease.2Another important advantage in demonstrating useful biomarkers is earlier detection of adverse events in the course of development of a new drug.3As discussed in the following sections, many animal models provide biomarkers for all the applications mentioned here.
USE OF MODELS FOR DRUG DEVELOPMENT AND TOXICITY PREDICTION
In the process of drug development, preclinical evaluation of the drug in animal models is not only very useful but quite essential before embarking on clinical studies (phase 1, phase 2, and phase 3 trials). This phase of the drug development in animals is necessary to demonstrate the anticipated therapeutic responses and to identify any adverse effect. However, a major problem is that there is a wide variability in development of toxic effects in different animal models, and many adverse effects observed in humans may not be seen in an animal model used for drug development.4Thus, novel therapeutic agents, which are deemed safe and effective in animal models, may eventually turn out to be ineffective or even toxic in human studies. For example, aminoguanidine (Pimagedine), which is an inhibitor of advanced glycosylation end products is an illustrious example of this discrepancy. Advanced glycosylation end products are very important in the pathogenesis of diabetic vascular complications and some nondiabetic conditions including aging. During the process of development of aminoguanidine, the preclinical studies in animals as well as clinical phase 1 and phase 2 studies established the safety and efficacy of the drug. However, in phase 3 studies, significant adverse events occurred in the final year of a 4-year study. Thus, unexpected occurrences of serious drug toxicity precluded the completion of phase 3 studies and sealed the fate of aminoguanidine as a potential drug.
In cases of certain other medications, the adverse events were not obvious even in the phase 3 clinical studies. In the postmarket or phase 4 studies, toxic effects may be noticed, which were not detectable in the preclinical studies in the animal models or in clinical phase 1, 2, and 3 studies. One such example is the experience with recombinant human erythropoietin. Epogen (recombinant human erythropoietin) was approved by the Food and Drug Administration rapidly after release of phase 3 data in view of its profound effectiveness to treat the anemia of chronic renal disease. However, occasional cases of Epogen resistance with pure red cell aplasia were reported in the first few years of its use in the dialysis population. In the following years, scores of reports involving hundreds of patients were reported or published. The flurry of reports related to these unanticipated adverse effect necessitated investigation into the chemical nature and manufacturing process of the compound and mechanisms mediating this adverse event.
Thus, an ideal animal model should be capable of demonstrating all the adverse effects that are observed in humans. This often requires the development of suitable biomarkers that can indicate the presence or development of an adverse event. On such biomarker is termed the kidney injury molecule 1 or KIM 1, which predicts the development of acute kidney injury equally well in both humans and animal models. The desirable characteristics of an optimal animal model are listed in Table 2.
Examples of Animal Models of Human Disease
Many animal species such as Drosophila and Caenorhabditis elegans have been used for replicating human diseases,5and they have yielded valuable information in understanding the pathophysiology of the disease and in drug development.6For example, zebrafish is one of the species in which antiangiogenic drugs were developed,7,8tested, and evaluated before human studies.9–11Pigs and nonhuman primate are also used for these purposes and are particularly useful because of their sizes and proximity to human anatomy and physiology.12However, rodent models remain the main species of animal models of human disease.
Several animal models currently exist for many human diseases such as hypertension, heart failure,13atherosclerosis, stroke,14,15and diabetes. Most of these models have at least some, often severe, limitations and deviations from the human disease16,17and hence, there is an ongoing major effort to refine and improve them into ideal animal models. Invertebrates such as Drosophila have evolved as excellent models for studying pathogenic mechanisms and for drug testing in neurodegenerative diseases.18,19Similarly, nonhuman primates have been used also as models of developmental psychopathology.20Transgenic mice have been developed with selective knockouts of mitochondrial transcription factors to examine the mechanisms underlying neurodegenerative disorders.21–24Furthermore, apolipoprotein E–deficient and low density lipoprotein receptor–deficient mice have been used successfully as models of atherosclerosis and drug development of the same.25
Infections
Understandably, animal models are quite important in examining the pathogenic mechanisms of infectious disease and testing the safety and efficacy of antimicrobial agents. Accordingly, several models have been established in the recent decades to study various infections. For example, development of transgenic mice expressing hepatitis C viral proteins helped evaluate virus-host interactions and test new antiviral agents.26
Diseases caused by Streptococcus pneumoniae are multiple and continue to be challenging. A recent review summarizes the techniques used in developing models of such diseases and how choosing appropriate animal models results in better understanding of these disorders.27Another area where animal models such as mice, rabbits, and guinea pigs were used to study immunopathogenesis and efficacy of therapy is human tuberculosis.28
The major advances in pathogenesis of sepsis and demonstration of sepsis-induced tissue damage by novel biomarkers was described recently in the animal models of sepsis.29Host responses to infection and factors that determine such responses have often been elucidated by optimal animal models as exemplified by certain mouse models of pneumonia.30
Cancers
In the field of carcinogenesis and chemotherapy for cancer, the role of animal models need not be overemphasized. Human tumor xenografts have been useful in identifying tumor responsive histotypes and murine models have been extensively used in drug development and toxicology studies.31Furthermore, the chemical basis of carcinogenesis was exemplified with N-methyl-N-nitrosourea (MNU)-induced gastric cancer in mice, which also provided more insights into pathogenesis.32Because of greater similarity to humans in genetic aspects, immune system, and pathophysiology, nonhuman primates have been used in cancer research to the study the chemical and biological carcinogens. Such models are proving to be of great benefit, and the guidelines to optimally use them in cancer research continue to evolve.33Additionally, animal models have been extensively used for determination of drug dosing using pharmacokinetics/pharmacodynamics data in cancer and noncancerous conditions in humans.34
Pulmonary Diseases
The understanding of several human pulmonary diseases and advances in therapy of such conditions has been significantly affected by development of suitable animal models. Such diseases include asthma35and other airway allergic disorders,36idiopathic pulmonary fibrosis,37pulmonary arterial hypertension,38and chronic obstructive pulmonary disease.39The characteristics of these animal models illustrate the importance of strategies to optimize such animal models to simulate human pulmonary diseases.
Liver and Gastrointestinal Diseases
In general, there has been a paucity of models of human hepatic and gastrointestinal disorders. However, in the last few years, spontaneous and induced models of primary biliary cirrhosis40as well as surgical, viral, and toxic models of acute hepatic failure41have been developed, which have aided in understanding not only the etiopathogenesis of these conditions but also transplant and other liver assist devices in the management of those conditions. Similarly, chemically induced models42and genetically engineered murine models43of inflammatory bowel disease have been used to study the pathogenesis and the efficacy of new therapeutic agents. Some nutritional issues, such as use of ketogenic diet in seizures,44and biological mechanisms leading to complications of obesity have been evaluated in relevant animal models.45
Psychiatric Diseases
Optimizing mouse models to evaluate the genetic basis and therapeutics of neurodegenerative diseases has been a challenging field46–48and has been well discussed in recent literature. Using animal models to evaluate predominantly subjective49and psychosomatic conditions16is also equally complex and continuing to evolve.
Rheumatologic and Cutaneous Disorders
Mouse models of epidermolysis bullosa50and canine models of atopic dermatitis51have not only established the pathogenic theories but also have helped in rapid screening of new treatment options. A recent review summarized the usefulness of animal models of rheumatoid arthritis in identification of susceptible genes and examining the pathogenic pathways.52Rat models of osteoarthritis have been used to evaluate new intra-articular drugs.53The role of nitric oxide generated by endothelium through endothelial nitric oxide synthase in leading to endothelial dysfunction and contributing to vasculitis and atherosclerotic complications was established by transgenic mouse models.54A vast array of findings has accumulated from animal studies to understand the basis of febrile seizures in infective and other neurological condition.55
Kidney Diseases
Many animal models have been used to investigate the genetic aspects of chronic kidney disease,56pathophysiology, and drug development for various renal diseases including chronic kidney disease,57podocytopathies,58human immunodeficiency virus–associated nephropathy,59nephrogenic diabetes insipidus,60polycystic kidney disease.61,62Although these animal models may have contributed to understanding these disorders, there are several problems with these models that limit their usefulness. Notwithstanding such drawbacks, in many renal disorders, animal models have provided great insights. One example is diabetic nephropathy.
Animal Models of Diabetic Nephropathy
The following section focuses on animal models that are used to study the pathogenesis of diabetic nephropathy as an example of a common human disease and also because that has been the focus of our laboratory. These models may simulate nephropathy in type 1 and type 2 diabetes,63and these may be derived from rodents, (mice and rats) as well as pigs and nonhuman primates. As in most conditions, rodents are the most extensively used species to examine this disease, especially after the advent of the transgenic and knockout technology. Type 1 diabetes is often induced by streptozotocin injection into rats64and mice, whereas type 2 is often due to genetic manipulations.65Driven by the need to develop models to simulate diabetic complications, National Institute of Diabetes and Digestive and Kidney Diseases/National Institutes of Health funded a major initiative termed Animal Models of Diabetes Complication Consortium (AMDCC). Several studies funded by the AMDCC initiative were focused on transgenic and knockout mouse models and help us to further our understanding of the pathophysiology and pathogenesis of diabetic nephropathy. A detailed review of these models is beyond the scope of this manuscript. Models like db/db mouse and eNOS knockout helped us in the standing the significance of several mediators of pathogenesis, but none of the models could replicate the human diabetic nephropathy entirely.
ZSF Rats
After examining several popular models of diabetic nephropathy In our laboratory, we characterized the renal disease in ZSF rat and a newer rodent model of obesity and diabetes. ZSF rats were developed in Indiana University by crossbreeding rat strains with 2 separate leptin receptor mutation, Zucker diabetic fatty rat and lean male spontaneously hypertensive heart failure–prone rat, derived from the obese spontaneously hypertensive rat. Phenotypically, they are normal until the eighth week when they start developing hypertension and spontaneous hyperglycemia, obesity, and hyperlipidemia, particularly hypertriglyceridemia. The renal manifestations develop pari passu with the other features; however, they become full blown after the 12th week.66Initial hyperfiltration with subsequent decreasing renal function, increasing proteinuria, and progressive systemic hypertension accompanied by histological pictures of thickened glomerular basement membrane, mesangial expansion, arteriolar sclerosis, and diffuse glomerulosclerosis completes the renal manifestations in this model. Furthermore, many therapies shown to be effective in humans in slowing the progression of diabetic nephropathy such as glycemic and blood pressure control and inhibitors of renin angiotensin system are also very effective therapies in ZSF rats. Thus, ZSF rats represent an excellent model to study the pathogenesis and potential new therapies in diabetic nephropathy.
Translatability Score
In view of the myriads of problems encountered during preclinical and clinical phases of drug development, a scoring system was recently developed to minimize the challenges due to attrition of projects in tertiary stages and to enhance the translatability of potential drugs.67A scoring system takes into account several factors including the preclinical68and clinical studies, biomarker validation, genetics,69and pharmacogenomics70and scored to 1 to 5 times a weighted factor (Table 3). Such scoring system may be used at the stage of animal model selection.
SUMMARY
In conclusion, animal models of human disease are crucial for translational research studies and particularly those involving the development of novel therapeutic agents. Included in such animal studies are biomarker development and toxicity prediction. Many drugs often fail in phase 2 and phase 3 human studies owing to lack of accurate toxicity prediction, which in turn is often due to the lack of an ideal animal model for the disease state for which the drug is being studied. To avoid enormous costs involved in drug development for various human diseases, some investigators have proposed a scoring model for translatability. However, the use of these scores still remains to be validated. The salient criteria that need to be considered while selecting the suitable and right animal model for translational studies were discussed in this review. Although the existing models represent powerful tools for translational research, several problems with them continue to challenge us. Thus, even with recent technological advances in animal model design, this area remains a fertile field for ongoing research and development.
ACKNOWLEDGMENTS
The author thanks Ms SangYi Park for helping with the manuscript and Dr McMahon for some of the work done in our laboratory with the ZSF rats cited in this manuscript. The author also sincerely thanks Larry and Jane Woirhaye for the generous research endowment fund which was used for the conduct of experiments with ZSF rats.