The list of known drugs provide a means to a reverse definition of a druggable target, but in light of the genome project, and a now extended list of gene products with variable chances of being targeted with a small molecule, how do we determine a 'forward' definition of a druggable target, and triage this to a workable list? In other words, can we come up with a reasonable set of criteria which is predictive of druggability and relevance to disease? By way of an example, an analogy in the chemical world was provided by Chris Lipinski who established a set of criteria based on physicochemical and pharmacokinetic characteristics of known drugs to define the 'rule-of-five' for oral bioavailability.9 This study showed that poor absorption or permeability of a drug is more likely when there are more than five hydrogen-bond donors, the lipophilicity is high (ClogP >5), the molecular weight is greater than 500 Da, and the sum of all nitrogen and oxygen atoms is greater than 10. Now, drug discovery groups use the 'rule-of-five' as a guideline for prioritizing drug leads to go into animal efficacy studies or even into humans. By analogy, then, in the biological world, are there guidelines to be gleaned from the list of known druggable targets, based on gene family, structure, expression, tissue distribution, function, biological pathway connectivity, etc. to help narrow the list of genes in the genome for drug targeting exercises?
An attempt to assess the number of putative targets that represent an opportunity for therapeutic intervention was made by Hopkins and Groom. Applying the constraint that a druggable target must be physically capable of binding compounds, preferably to those with druglike features as set forth by the Lipinski 'rule-of-five,' this group determined that only 3051 of the predicted ~ 30 000 or so genes code for a protein with some precedent (or likelihood, by analogy to family members), of binding a druglike small molecule.2 Almost half of these druggable genes fall into only six of 130 protein families, including the usual suspects G protein-coupled receptors (GPCRs), tyrosine and serine/threonine kinases, zinc metallopeptidases, serine proteases, nuclear hormones, and phosphodiesterases. Of course, these are purely estimates based on a small number of proteins which are documented to bind small molecules, whereas the true number of druggable entities may be much larger. Furthermore, this analysis does not take into account those targets for which chemical modulation has not been profoundly demonstrated nor those which are composed of different material (DNA, RNA, lipid, etc.).
Another issue tangentially related to druggability, but germane to the discussion of what constitutes a 'good' drug target, is the potential selectivity of a small molecule inhibitor against a chosen target. Although historically GPCRs and enzymes such as kinases, proteases, phosphodiesterases, etc. have been the preferred targets of the drug industry, many of the compounds developed for such targets have considerable cross-reactivity with other family members, often leading to untoward and significant side effects. Conventional wisdom suggests that the increased specificity of a compound against its target increases the likelihood of efficacy, a position supported by the considerable cross-reactivity of some of the kinase inhibitors which have efficacy but dose-limiting toxicity ascribed to their lack of selectivity.10 However, some marketed drugs may have positive attributes based on their lack of selectivity. Even Gleevec, the first molecularly-targeted anticancer drug designed to inhibit BCR-abl in chronic myelogenous leukemia, has cross-reactivity with platelet-derived growth factor receptor (PDGFR) and v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog (c-kit) which may contribute to possible future success in treating other types of cancer, and potentially asthma11 and diabetes.12'13 Of course, this type of serendipity is not unprecedented as many drugs are efficacious for off-label indications, likely based on cross-reactivity and off-target effects. The extent to which this strategy will be leveraged in target prioritization is yet to be determined. Drug discovery companies now routinely profile compounds in vitro against panels of related targets, such as kinases, GPCRs, and proteases to find either potential sources of toxicity or new opportunities for a compound of interest. What is clear is that the genome project has significantly expanded and ostensibly defined the number of possible cross-reactive targets within a family or enzyme class, and thus enabled a more complete assessment of a compounds selectivity-based advantages and disadvantages.
Was this article helpful?