Discovery Strategies Conference 2004
Conferences designed to address research needs of the pharmaceutical and biotechnology industry
"Mouse Genetics: Accelerating Drug Discovery and Development"
May 17-19, 2004
Conference Overview:
The mouse is a powerful tool for gene-based drug discovery and
development. As a leading platform for target discovery, the mouse has a genome that can be easily modified. Furthermore, many human diseases can be modeled in the mouse, making it an ideal platform to accelerate the validation process of drugs in the discovery pipeline. This conference explores these and other unique roles this animal model plays during the early stages of drug discovery and development. Topics will include: the identification and validation of druggable targets, optimization of lead compounds, and assessment of risk and toxicity. Round table discussions will cover emerging topics related to drug development.
|
History and Overview of the Mouse in Biomedicine and Pharmaceutical Development | |
|
Dr. Kenneth Paigen |
Genetic Strategies in Drug Discovery: An overview |
|
Dr. Jan Tornell |
Use of Transgenic Mice in the Pharmaceutical industry: From Target Discovery to Humanised Systems and Compound Validation |
|
Target Identification and Validation | |
|
Dr. Susan Ackerman |
Lessons From Loss of Function Mouse Neurodegenerative Mouse Models |
|
Dr. Michael Cawthorne |
Drug discovery for type 2 diabetes and obesityin the pre-genomic era using mouse models |
|
Dr. Luk Van Parijs |
RNAi-based forward genetic screens in the murine immune system |
|
Dr. Gary Peltz |
|
|
Dr. Beverly Paigen |
From Quantitative Trait Loci to Gene: Using Integrative Genomics |
|
Dr. Gary Churchill |
|
|
Dr. Eric Schadt |
|
|
Compound Validation and Optimization | |
|
Dr. Michael Luther |
Lessons From Loss of Function Mouse Neurodegenerative Mouse Models |
|
Dr. Douglas Wallace |
The Mitochondrion: the Next Major Drug Target for Degenerative Diseases and Aging |
|
Dr. Shaoguang Li |
Genetic Identification of Key Signaling Molecules with Therapeutic Potential for Ph+ |
|
Dr. Derry Roopenian |
|
|
Drug Metabolism and Toxicity Sceening | |
|
Dr. Edward Leiter |
New Polygenic Mouse Models of Obesity-Associated Diabetes ("Diabesity") for Pharmacogenetics Research |
|
Daniel Nebert |
|
|
Dr. Frank Gonzalez |
P450 AND PPARa humanized mice for the study of DRUG Metabolism and Safety |
Dr. Frank Gonzalez, NIH/NCI
"P450 AND PPARa humanized mice for the study of DRUG Metabolism and Safety"
Animal models are widely used to examine metabolism, pharmacology, toxicology and carcinogenesis of xenobiotics including therapeutic agents. However, there are marked species differences in the expression and enzymatic functions of P450s. and xenobiotic receptors This is especially notable between mice and rats, commonly used experimental models, and humans. Human P450-expressing transgenic mice were generated using genomic clones, and mouse lines expressing the CYP2D6 and CYP3A4 genes were characterized and found to accurately express the corresponding enzymes and exhibit catalytic activities at levels comparable or higher than those found in human tissues. Highly sensitive LC-MS/MS analytical methods were developed to perform drug metabolism and pharmacokinetic analyses using non-radiolabeled probe drugs. These novel P450 humanized mouse lines overcome the gap of species differences and offer a better system to study drug metabolism and disposition, and drug-drug interactions for the prediction of drug/chemical effects in humans. Moreover, these mice can serve as whole intact animal models for exploring endogenous substrates for human P450s, examining their biotransformations, and elucidating physiological consequences of human P450 expression in an integrated whole animal system. Xenobiotic receptors also exhibit species differences. A large and diverse class of chemicals collectively referred to as peroxisome proliferators that include fibrate hyperlipidemic drugs and pthalates such as diethylhexylpthalate used in the production of plastics, have been widely studied due to their known toxic and carcinogenic effects in rodent model systems. However, humans appear to be resistant to the adverse effects of these chemicals as revealed through epidemiological studies of subjects chronically treated with fibrate drugs. Peroxisome proliferators exert their biological effects through binding to a nuclear receptor called the peroxisome proliferator-activated receptor a (PPARa). There are species differences in the levels of expression and ligand binding specificities of PPARa between mouse and humans. This could account for the species differences in response to peroxisome proliferators. To address this issue, PPARa humanized mice were produced in which a PPARa genomic BAC clone and a doxycycline regulated human PPARa cDNA were introduced onto the PPARa-null mouse background. These mouse lines are responsive to peroxisome proliferators as revealed by target gene expression and triglyceride lowering effects and should be of value in elucidating the mechanisms of species differences in response to peroxisome proliferators.
Back to Top
Dr. Shaoguang Li, The Jackson Laboratory
"Genetic Identification of Key Signaling Molecules with Therapeutic Potential for Ph+ B-ALL"
Human Philadelphia chromosome-positive (Ph+) leukemias include chronic myeloid leukemia (CML) and B-cell acute lymphoblastic leukemia (B-ALL) and are induced by the BCR-ABL oncogene. The recent success of the BCR-ABL tyrosine kinase inhibitor STI571 (Gleevec, imatinib mesylate) in treatment of chronic phase CML has generated great excitement in cancer therapy. However, emerging clinical resistance to Gleevec and its limited effectiveness in treating both CML blast crisis and B-ALL suggest that use of STI 571 as a single agent may not prevent eventual disease progression to terminal blast crisis or cure CML and more synergic therapeutic strategies need to be developed for advanced Ph+ leukemia. We took genetic approaches to identify in vivo the key signaling pathways required for BCR-ABL-induced leukemias, and to test therapies directed against them. Using our mouse models of human BCR-ABL-induced CML and B-ALL, we show that Bcr-Abl activates the Src kinases Lyn, Hck and Fgr in B-lymphoid cells. Upon retroviral transduction of BCR-ABL, marrow from mice lacking all three Src kinases efficiently induces CML in recipient mice but is defective for induction of B-ALL. The Src kinase inhibitor CGP76030 inhibits these Src kinases but not BCR-ABL, impairs the proliferation of Bcr-Abl-expressing B-lymphoid cells in vitro, and prolongs the survival of mice with B-ALL but not CML. The combination of CGP76030 and imatinib is superior to imatinib alone in treating B-ALL mice. These results implicate Src family kinases as therapeutic targets in Ph+ B-ALL and suggest that simultaneous inhibition of Src and BCR-ABL kinases may be of benefit in patients with Ph+ acute leukaemia. Our genetic study will provide the foundation for multiple-target therapy of advanced Ph+ leukemia.
Back to Top
Dr. Luk Van Parijs, Massachusetts Institute of Technology
"RNAi-based forward genetic screens in the murine immune system"
RNAi provides an efficient and flexible approach to study gene function in mammalian cells, tissues, and animals. We have developed lentivirus-based transgenesis systems that have allowed us to perform RNAi-based screens for genes that modify the development of immune cell cancer and autoimmune disease. The results of these screens will be presented.
Back to Top
Dr. Gary Peltz, Roche Palo Alto
From Genetics to Therapeutics
Our understanding of genetic factors effecting complex disease and biologic processes in humans is advanced by analysis of disease-related murine genetic models. We have used an integrated genetic and genomic approach to identify the genetic factors effecting disease-susceptibility in murine models of human disease. As shown by our anlaysis of a murine genetic model of osteoporisis, the information obtained from mouse genetic analysis can direct biological experimentation and genetic analysis in human cohorts that enables identification of novel therapeutic targets.
We have developed computational tools to markedly accelerate the rate at which murine genetic models can be analyzed. A haplotypic map of the genome for 13 commonly used laboratory mouse strains was produced, which enabled development of a method for rapid computational mapping of complex phenotypes onto haplotypic blocks. This method correctly predicted the genetic basis for strain-specific differences in multiple biologically important phenotypic traits. It accurately discriminated among neighboring, co-linear genes in the MHC region; identified the individual polymorphism responsible for strain-specific differences in aromatic hydrocarbon response; and identified a factor effecting the metabolism of a drug. We have recently utilized this computational genetic method to imrove our understanding of how drugs are metabolized, and to investigate the mechanism of action of commonly used pharmaceuticals.
Back to Top
Dr. Eric Schadt, Rosetta Inpharmatics, Merck
"Combining genotypic, transcription, and clinical trait data in segregating mouse populations to infer causal relationships between gene expression and complex disease traits"
The reconstruction of genetic networks in mammalian systems is one of the primary goals in biological research, especially as such reconstructions relate to elucidating not only common, polygenic human diseases, but living systems more generally. Here I present a statistical procedure for inferring causal relationships between gene expression traits and more classic clinical traits, including complex disease traits. The statistical procedure tests whether variations in relative transcript abundances statistically support a causative or reactive function in the disease state. More generally, I demonstrate how this statistical procedure can be incorporated into a novel gene network reconstruction algorithm, derived from classic Bayesian network methods, that utilizes naturally occurring genetic variations as a source of perturbations to elucidate disease-specific gene networks. This method incorporates relative transcript abundance and genotypic data from segregating populations and results in a generalized scoring function of maximum likelihood commonly used in Bayesian network reconstruction problems. The utility of this novel algorithm is demonstrated via application to gene expression data from multiple segregating mouse populations. I demonstrate that the network derived from these data using the novel network reconstruction algorithm is able to capture causal associations between genes that result in increased predictive power, compared to more classically reconstructed networks derived from the same data. This procedure allows for the objective identification of key drivers of common human diseases.
Back to Top
Dr. Jan Tornell, AstraZeneca Pharmaceutical
"Use of Transgenic Mice in the Pharmaceutical industry: From Target Discovery to Humanised Systems and Compound Validation"
Pharmaceutical companies are faced with the challenge that only approximately 10 % of the compounds tested in clinical trials eventually become a registered drug. The reason for failure is both related the to target (lack of efficacy, mechanism related toxicology) and to the compound (pharmacokinetics, chemistry related toxicology). Without having an optimal compound acting on a well-validated target there will be no successful new drug.
Transgenic technology represents an attractive approach to reduce the attrition rate of compounds entering clinical trials by increasing the quality of both the target and the compound. The technology can impact decision making at many points in the discovery process including target discovery, provision of better models for human diseases and generation of models designed to understand specificity and safety of compounds. Considerable efforts are being made to establish new, rapid and robust tools with general utility for in vivo validation but so far only transgenic animals work reliably on a wide range of targets.
The phenotypes discovered in transgenic animals are often predictive to understand the effects of drugs acting on the specific target. Moreover, the phenotyping of mice has advanced during the recent years making it possible to measure the same parameters in mice as in human patients. This, in combination with the recently published striking similarities of human- and mouse DNA sequences and chromosomal arrangements, suggest that genetically modified mice will be very valuable in understanding the function of human genes and their relationship to disease.
Back to Top
M.A. Cawthorne, Clore Laboratory,University of Buckingham, UK
"Drug discovery for type 2 diabetes and obesity in the pre-genomic era using mouse models"
The spontaneous mutation leading to obesity, insulin resistance and hyerinsulinaemia noted by Ingalls, Dickie and Snell at the Jackson Laboratory gave rise to the availability of the C57Bl/6 ob/ob mouse as an animal model for the investigation of the metabolic defects associated with the phenotype and a tool for evaluating potential therapeutics. Although it was known from the 1950’s that a single gene mutation was responsible for the phenotype, the identity of the gene was not realised until 1994. Nevertheless, the C57Bl/6 ob/ob mouse was widely used by pharmaceutical companies in drug discovery efforts from the late sixties to identify potential new therapies for both obesity and type 2 diabetes. In the diabetes arena, the C57Bl/6 ob/ob mouse was complemented by the C57Bl/Ks db/db mouse. The elegant cross-circulation studies of Coleman at the Jackson Laboratory demonstrated that the db gene product was probably the receptor for the ob gene product and hence one would expect identical phenotypes. This is the case if the mutations are on the same genetic background. However, the db mutation is commonly available on the C57Bl/Ks background and is significantly more diabetic than the C57Bl/6 ob/ob mouse. This is a clear demonstration that the phenotype is a product of the mutation and genetic background (and the environment including diet).
In the author’s laboratory, these models were used as primary and secondary screens in the discovery and progression into clinical development of a range of potential anti-obesity and anti-diabetic agents including b3-adrenoceptor agonists and insulin sensitisers. One of these, rosiglitazone, is marketed worldwide.
Back to Top
"Genetic Strategies in Drug Discovery: An overview"
Recent advances in genetic technologies have dramatically increased our ability to identify key molecular components and their interactions in physiological pathways. The implications for drug development are several, including identification of new drug targets, predicting efficacy and suggesting possible side reactions.
The key elements are the parallel genetic and physiological systems of human and mouse; the ability to genetically engineer mice; the extreme phenotypic diversity of inbred mouse strains, especially with respect to disease susceptibility; the ability to identify the key genes underlying this diversity, and the fact that humans and mice share the same key, susceptibility genes.
We will review the conceptual background for using mouse genetics in such studies using several case histories focusing on cardiovascular parameters, hypertension and obesity.
Back to Top
"New Polygenic Mouse Models of Obesity-Associated Diabetes ("Diabesity") for Pharmacogenetics Research"
The more common forms of obesity and Type 2 diabetes are under multigenic control in humans. Hence, monogenic mutations in mice producing morbid obesity, impaired glucose tolerance, and diabesity in mice do not accurately reflect the etiopathogenesis of the human diseases. We have developed new polygenic mouse models that are more relevant to the human conditions. New Zealand Obese (NZO) is a mouse strain selected for polygenic obesity. NON is a mouse strain selected for impaired glucose tolerance. Obesity-driven type II diabetes (diabesity) develops in 90-100% of (NZO x NON) F1 males by 24 wk. Genomic analysis of this "negative heterosis" has identified a number of quantitative trait loci (QTL) controlling various obesity and diabetes subphenotypes contributed by both NZO and NON parental backgrounds. We developed 10 interval-directed Recombinant Congenic Strains (RCS) by transferring selected NZO diabesity QTL to the recipient NON/Lt background and then inbreeding at second backcross. The presentation will contrast the phenotypic differences of 5 of these new RCS, and will demonstrate their utility for pharmacogenetic research.
Back to Top
"From Quantitative Trait Loci to Gene: Using Integrative Genomics"
Hundreds of quantitative trait loci (QTL) have been identified; the challenge now is to identify the genes responsible. We present several ways to move from QTL to gene using bioinformatics resources including combining crosses statistically, using mouse/human/rat homologies, haplotype analysis, and sequence comparisons of candidate genes. We illustrate the progress that can be made for HDL, blood pressure, and atherosclerosis QTL. The combination of these methods has accelerated the pace of gene finding.
If the same QTL is found in multiple crosses, the data can be combined thus effectively increasing the number of mice. This technique can (1) narrow the QTL region substantially as demonstrated by combining four crosses for a QTL for HDL cholesterol on Chr 4, (2) demonstrate that a broad QTL splits into two separate peaks as demonstrated by a QTL for gallstones on Chr 6, and (3) cause suggestive QTLs to become significant. QTL can also be narrowed by haplotype analysis. We illustrated this by comparing SNP data for an atherosclerosis susceptibility QTL, Ath17, that differs between strains C57BL/6 and 129. Furthermore, haplotype analysis combined with phenotype data from the Mouse Phenome Database can be used to prove a candidate gene as illustrated by the identification of the gene for an HDL cholesterol QTL on Chr 1. Lastly, the QTL for several traits are found in homologous regions in mouse, rat, and human genomes. For such traits, comparative genomics can be used to narrow the QTL region. Finally, after finding a candidate gene in the mouse, one can immediately test it in human populations characterized for the trait.
Back to Top
"Knockout Mouse Lines for Cytochromes P450 Enzymes: their Utility in Pharmacogenetics and Pharmacokinetics"
There are 57 human cytochrome P450 (CYP) genes and 102 mouse Cyp genes in their respective genomes. Whereas human CYP3A, CYP2C and CYP2D enzymes metabolize >60%, ~40%, and ~30% of all commonly prescribed drugs, respectively, dioxin-inducible CYP1 enzymes metabolize a few drugs but many environmental carcinogens. In vitro and cell culture studies have shown that polycyclic hydrocarbons (e.g. benzo[a]pyrene, BaP) are oxidized by CYP1A1 and CYP1B1 to reactive intermediates capable of binding to nucleic acids and proteins (adducts) and causing mutations and cancer; in fact, pharmaceutical companies routinely test candidate drugs for CYP1A1 inducibility and, if a candidate drug shows this property, further testing of it is generally discontinued--for fear of possibly causing cancer. We therefore had expected the Cyp1a1(-/-) knockout mouse to be protected against BaP toxicity. However, oral BaP causes toxic chemical depression of the bone marrow, immunosuppression, and elevated levels of BaP-DNA adducts in the marrow, spleen, liver, and GI tract of Cyp1a1(-/-) mice, much more than that in Cyp1a1(+/+) wild-type mice. By pharmacokinetic studies of BaP (area-under-the-curve, AUC), we found that detoxication by CYP1A1 is far more important than metabolic activation by CYP1A1 in the intact animal.
In vitro and cell culture studies have shown that arylamines (e.g. the human bladder carcinogen, 4-aminobiphenyl, ABP) are oxidized by CYP1A2 to reactive intermediates to form adducts and cause mutations and cancer. We thus had expected the Cyp1a2(-/-) knockout mouse to be protected against ABP toxicity. However, topical ABP causes elevated liver and urinary bladder ABP-DNA adducts in Cyp1a2(-/-) mice, much more than in Cyp1a2(+/+) wild-type mice. These two examples suggest that, in the intact animal: [a] the cell type- or tissue-specific coupling and amount of these enzymes becomes important, depending on the route of administration and target organ; [b] other non-CYP enzymes "take over" in the absence of CYP1A1 (for BaP) or CYP1A2 (for ABP); and/or [c] the genome of knockout mice is able to "compensate" for the loss of a gene.
Theophylline, widely used in the treatment of asthma and other obstructive pulmonary diseases, is metabolized to 1-methyl- and 1,3-dimethyluric acid and 3-methylxanthine by CYP1A2. Studying theophylline urinary metabolites and clearance in Cyp1a2(-/-) mice, we found that a second dioxin-inducible enzyme is responsible for a portion of the 1-methyluric acid formation.
Polymorphisms exist in the CYP1A2 gene, accounting for >60-fold differences in mRNA, protein and enzyme activity levels in human populations. Polymorphisms exist in all human drug-metabolizing genes. Studies of human variant alleles, in place of the mouse ortholog in "humanized" mice, should provide greater understanding of drug therapy, drug toxicity, pharmacogenetics, and pharmacokinetics in the intact animal.
Back to Top
"The role of the IgG receptor FcRn in autoimmune disease and the persistence of therapeutic antibodies"
Improving the pharmokinetics of therapeutic drugs remains a pressing need in medicine. Increasingly, therapies based on monoclonal antibodies, in particular, IgG, offer one of the most promising strategies for treating a variety of diseases, including cancer and autoimmunity. In comparison to other protein-based therapeutics, IgG offers the advantage of a substantially persistence in body fluids, thus prolonging the effectiveness of the drug. Similarly, fusion proteins that include the Fc region of IgG show a similar persistence and improved therapeutic efficacy. Therapeutic proteins engineered to bind albumin also show an extended serum. We will review studies which convincingly show that FcRn, originally described as the neonatal Fc receptor, is the molecule responsible for all of these effects. However, while the therapeutic benefits of engaging FcRn is clearly beneficial for drug pharmacology, FcRn can also be deleterious in patients with autoimmune diseases caused by autoantibodies. This is because FcRn facilitates the accumulation of autoantibodies, thus exacerbating autoimmune diseases, such as systemic lupus erythematosus and rheumatoid arthritis. In such diseases, the blockade of FcRn would be a promising therapeutic goal. We will describe recent insights into the the biology of FcRn, the fundamental role it plays in IgG homeostasis, novel mice to monitor humanized antibody pharmokinetics, and the therapeutic opportunities that FcRn provides.
Back to Top
"Lessons From Loss of Function Mouse Neurodegenerative Mouse Models"
Although neurodegenerative disorders are prevalent in the aging human population, the molecular mechanisms underlying these diseases are not well understood. To identify molecules that are necessary for the survival of terminally differentiated neurons in the aging nervous system, we are studying spontaneous, loss-of-function mutations that cause progressive ataxia and neurodegeneration in the adult mouse. We have identified the molecular defect, and the associated pathways that are disrupted by these mutations, in several of these unique models. These mutations and their consequential role in neurodegeneration will be discussed.
Back to Top
"Inference of Genetic Networks"
Experimental perturbations of a complex system can provide observations that shed light on the system's structure. However when one is faced with a large multivariate data structure, it is often difficult to establish the direction of causal relationships from the observed correlation. In biological systems analysis, experimental crosses provide unique features that admit causal inference. Naturally occurring genetic variations represent a large set of perturbations that can be distributed in a combinatorial fashion among a set of cross progeny. The transmission of genetic variants through the process of meiosis serves as a natural randomization mechanism. Thus genetic crosses share properties of statistically designed experiments that allow us to infer causation. This is consistent with the intuitive notion that causation flows from genes to phenotypes.
We have applied methods of network inference to genetically randomized populations of laboratory mice to investigate a variety of phenotypes. A typical study involves the measurement of a number of correlated phenotypes on the same set of mice. The problems to be address are two. First, how to identify the genetic loci involved in the system. Second, how to infer the network of causal relationships among genes and phenotypes. We illustrate this concept with application to bone density, blood pressure and body mass. We are able to identify genetic loci that are causal for variation in multiple phenotypes. Moreover we are able infer directed causal connections among the phenotypes that provide insight into the pleiotropic nature of the genetic effects.
Back to Top
