Designing innovative therapies for neuropathic pain: pros and cons of target‐based drug discovery

摘要

Among the many challenges facing academic translational research and the pharmaceutical industry is the need for better pre-clinical models to validate drug targets (Plenge et al., 2013). Data have shown that most failures during phase II trials are due to a lack of efficacy (50%) or noted toxicity (25%) (Kola and Landis, 2004; Arrowsmith, 2011). Targeting a specific protein's function is generally the most effective means to approach when developing a pharmacological basis to modify a disease. Establishing a therapeutic hypothesis utilizing a target-based approach ideally requires that the target is clinically validated, mechanistically and structurally elucidated, assayable, pharmacologically tractable, and functionally indistinguishable from the cellular context. The number of these validation criteria required in a program is dependent on what are considered acceptable levels of program risk proportionate to the overall strategy (Table 1). Clearly the establishment of a dose–response relationship between the target and efficacy in humans, also known as “clinical validation,” is a highly desirable piece of evidence to have though often difficult to obtain. Evaluating the relevance of a specific target can be estimated in some cases by examining the relationship or direct involvement in rare inherited diseases. For example, cystic fibrosis (CF) is due to mutations in the CF transmembrane conductance regulator (CFTR) protein. In one form of the disease a missense mutation (G551D) results in a dysfunctional CFTR protein that is expressed on the epithelial cell surface. Ivacaftor, a CFTR potentiator and approved drug developed by Vertex and the CF Foundation, partially restores the chloride transport activity of the impaired protein resulting in clinical benefit. On the other hand, while a target protein may be highly correlated to disease through mutation or deregulated expression, this protein may be refractory toward direct pharmacological intervention. For example, type 1A Charcot-Marie-Tooth peripheral neuropathy occurs in individuals harboring a duplication of the gene for the peripheral myelin protein 22 (Pmp22), and haploinsufficiency of this gene results in HNPP (hereditary neuropathy with liability to pressure palsies). While in this case a target is clearly linked to disease and the nature of the disease is dependent on gene dosage, the unknown structure and function of this membrane-spanning protein precludes it from many of the current approaches used to identify a pharmacological means to directly affect the function of PMP22. Other sources of evidence that may be useful when attempting to establish clinical validity are listed in Table 1 (Dietz, 2010; Fishman, 2013). In many cases, mechanistic and structural knowledge underlying the enzymology or binding interaction mediated by the target are sufficient to enable compound ligands to be rationally designed and optimization through medicinal chemistry (Marmorstein, 2001). Such pharmacological tractability needs to be accompanied by a robust structure–activity relationship (SAR) where ligand structures display a quantifiable relationship to target activity and/or binding affinity. Families of compounds having a high degree of structural similarity are known as a chemotype. For a subset of cases, a chemotype can become the basis for a clinically useful therapeutic agent. For example, development of human immunodeficiency virus (HIV) protease inhibitors based on the peptidomimetic hydroxyethylamine transition state analog chemotype were the first drugs to treat HIV/acquired immunodeficiency syndrome (AIDS) and represents a triumph of rational drug design (Fig. 1). Determining that a target isolated from its pathophysiological context is functionally indistinguishable from that cellular context when incorporated into an assay is an important assessment to make. This can be challenging for targets which are studied as specific domains obtained from modular proteins ofte