Some have referred to modern molecular biology as "the new physics," in the sense that it is now what physics was at the beginning of the 20th century—an exciting but immature field that had immense potential but had yet to produce many practical applications. The excitement surrounding molecular biology was recently heightened after the publication of numerous genome sequences, particularly those of humans. Much of the future and promise of the field, in fact, will center on the genome-based approach that has just begun.
According to a recent study, there are only 500 molecular targets in the current medical armamentarium. However, there may be as many as 5000 to 10,000 potential targets that affect or directly initiate disease pathways for the 150 most pressing multifactorial diseases alone.1 The current model of drug development describes the abnormal function in the "diseased" cell, identifies target molecules affecting this state, and develops molecules that can alter these targets. However, bottlenecks exist in the methods used to identify molecules that cause or modify a disease state. Genome-based mRNA, DNA, and protein approaches have the potential to alleviate such barriers.
One of the more immediately available genomic strategies is to use microarrays to monitor genome-wide patterns and changes in gene expression. In this technique, cDNA from different pools of RNA are labeled and hybridized to spotted arrays of fixed-identity DNA clones. Because labeled cDNAs are less abundant than their immobilized DNA targets, the resultant differences in signal intensities of each spot reflect different levels of these transcripts in the original RNA pool.2 The underlying rationale is that on a large scale, gene expression patterns will emerge that correlate with clinical variables like disease, predisposition, and drug effectiveness. Such patterns may also point to a family of functionally related genes that are coregulated as a response, thus elucidating pathophysiologic mechanisms. These data can be useful even in the absence of an understanding of the molecular and cellular implications of the patterns.
One strategy to identify drug targets is to use microarrays to find receptors or enzymes that are upregulated or downregulated in diseased tissue. The advantage of microarrays is their capacity to describe the genome-wide changes that represent the process of a disease. These changes, however, are not necessarily important mediators of the disease. Thus, this approach for drug target selection lacks clear focus. The power of microarrays to recognize patterns may be more useful to catalogue normal patterns of gene expression in different cells in response to stimulations of various cellular processes. These normative databases of gene expression patterns could be compared to those of diseased tissue, and the differences would suggest which physiological and cellular responses correspond to a given insult. Using microarrays in this way could potentially yield drug targets capable of regulating these processes.
Genomic DNA-based approaches have also recently attempted to correlate genetic polymorphisms with their functional significance. The most successful example to date is the analysis of polymorphisms in human mitochondrial DNA, for which single-base resolution can be attained in minutes.3 However, it is not yet clear whether this technology is applicable to the much larger human genome. As polymorphism resolution improves, it will improve the ability to link loci with diseases or predispositions. This will shorten the interval between clinical characterization of a disease and identification of underlying causes.
Genomic approaches can also be applied to study pharmacogenomics, which examines the relationship between genetic identity and the metabolism and efficacy of an existing drug therapy. A drug must first enter the body, then be distributed to the proper compartments, and finally be metabolized to active or inactive components before it can exert its biological action. Genes responsible for many elements of drug pharmacokinetics may exhibit polymorphisms that alter their function. Many genes that affect drug metabolism, including drug targets and transporter molecules, contain known polymorphisms. But these likely represent only a small fraction of functionally relevant polymorphisms. The combination of these genetic changes may influence the effective dosage and drug effect more so than the commonly considered clinical criteria like renal and hepatic function, patient age, nutritional status, and concomitant illnesses.4 If so, the future of drug development and prescription will move away from dosing by weight, a crude approximation for many reasons, and be guided more by a patient's genetic constitution.
A recent article5 proved the potential of the genomic approach to drug design by identifying vaccine candidate genes for serogroup B meningococcus in the scope of a single study. The pathogen's sequence was screened for genes encoding surface proteins based on conserved domains; these genes were then expressed in E coli and used to immunize mice. This approach allowed the pathogen's genome to be restricted to a group of 7 proteins conserved between 22 strains of group A, B, and C pathogens. All of these proteins evoked effective antibacterial antibodies.
Due to the rapidly developing abilities to examine genome-wide alterations in gene expression or genetic polymorphisms, many alterations will be found in sequenced but functionally uncharacterized genes. An equally strong effort, therefore, needs to be focused on bioinformatics. The challenge to bioinformatics will be to infer a role for a protein, its structure, and possible interactions from the primary genomic sequence alone. The method most widely used now is the comparison of amino acid sequences between the gene in question and all other known genes. This comparison reveals the degree of similarity between protein domains, thereby implying functional similarity. While linear sequence is sometimes a good surrogate level at which to compare function, 3-dimensional structure is more relevant, necessitating development of novel bioinformatic approaches to help obtain 3-dimensional structure from sequence without the crystallization of each protein. The improved ability to model 3-dimensional structure stems from the considerable work done over the years in the field of structural biology.
Better understanding of protein structure will facilitate the production of custom-made drugs designed to interact only with specific sites on their target molecules, a process that is currently only carried out empirically during drug development. Structure-guided small molecule drug design was recently used to develop the neuraminidase inhibitor oseltamivir.6 This effort demonstrates the utility of 3D structural modeling in drug design, but represents only a fraction of the sophistication possible in the future.
In contrast to sequence or expression analyses, a genome-wide approach to protein expression is still in its infancy and faces great technical hurdles. This field, often referred to as proteomics, seeks to profile "the complete set of proteins that is expressed, and modified following expression, by the entire genome in the lifetime of a cell."7 A recent study found that the correlation between mRNA and protein expression levels is only 0.48, suggesting that information from microarray-based expression analyses will give only a partial picture of the levels of the proteins produced, which are the direct effectors of the gene's function.8 Arrays also give no information on the effect of posttranslational modifications on the activity of these proteins. Current proteomic techniques have great limitations especially in detecting low-abundance proteins. Ultimately the technical considerations will be overcome, just as similar ones were for the human genome project. The long-term goal is the simulation of cellular processes and prediction of outcome. This would be analogous to the way computer simulations have replaced car crash experiments, optimizing materials and designs for maximum safety.9
With such rapid advances already within sight, it is important also to consider what role the government should play in protecting the public interest by regulating the conversion of emerging knowledge into useful technology. Of particular relevance to this issue is the nature of the academic-industry relationship. The flow of intellectual property from academia to industry for the eventual creation of royalty streams has accelerated since the Bayh-Dole Act of 1980, which encouraged patenting by universities and subsequent licensing to industry of inventions stemming from federally funded research, with little government interference. According to a General Accounting Office report, this law likely contributed to the ongoing expansion in biotechnology and high technology.10 A concern is that, due to the exclusivity of many licensing agreements, drug prices can escalate far beyond the cost of their development. An issue to be considered, therefore, is whether there is a governmental responsibility to subsidize or regulate drug prices, so as to return some of the benefit of federally funded research to the people underwriting it. The danger is that such governmental decisions, even based on actuarial analysis, may prove arbitrary and might limit the incentive of industry to develop future drugs with the prospect of lower profitability margins.
Financial conflicts of interest may also arise from increased collaboration between academia and industry. Financial incentives may encourage inadequate oversight of clinical trials with the potential to harm study subjects, discourage publication of negative results, or bias study design. Universities may not have adequate measures in place to defend scientific integrity and proper patient care against such influences. These points remain largely unresolved and have the potential to seriously undermine public faith in university research if they are not properly addressed by prospective policy decisions.11
The challenge at hand is to use knowledge about genome sequences and gene expression patterns in the development of new therapies and for the optimization of current ones. The resulting field of functional genomics is still at its earliest stage of technological maturation, and many ethical dilemmas and legal questions have yet to be addressed. Considerable insight, innovation, and proactive policy development will be required to make genomic technology integral to the practice of medicine, just as decades of work in theoretical physics preceded the development of tools like the transistor.
Etkin A. Drugs and Therapeutics in the Age of the Genome. JAMA. 2000;284(21):2786-2787. doi:10.1001/jama.284.21.2786-JMS1206-4-1