LITMIR.BIZ

Скачать книгу

of methods has been most extensively developed for protein‐ligand binding affinity analysis [77–82]. Likewise, important work is ongoing to extend this class of methods to more comprehensively support ADMET property optimization, including solubility and binding selectivity [91, 92].

      With a better understanding of these essential techniques in‐hand, we will now turn our attention to how these techniques have been used effectively by multiple parties to accelerate the discovery of novel and more efficacious drug therapies. We will illustrate these explanations through a variety of published case studies, which we hope will highlight the wide variety of ways such computational analyses can positively contribute to discovery project execution.

2.3 Illustrative Applications

      2.3.1 Modeling Support of Target Validation, Feasibility Assessment, and Hit Discovery for Acetyl‐CoA Carboxylase

De novo fatty acid synthesis is rate limited by the enzyme ACC [93, 94]. A wealth of animal study data has suggested that inhibition of ACC might be palliative for NASH, which is believed to be an increasing cause of liver failure [95–102]. The development of a drug therapy to inhibit this target has been a highly nontrivial challenge, and conventional drug discovery efforts targeting the carboxyltransferase (CT) domain of ACC were found to yield drug candidates with poor physical chemical properties [93,103–110]. Nimbus therapeutics decided to take a different approach to drugging the target, and utilized a computational solvent analysis method, WaterMap, to investigate if it might be possible to develop improved drug‐like inhibitors that bind to the biotin carboxylase (BC) domain of the enzyme [111, 112]. These WaterMap calculations indicated the presence of multiple high energy hydration sites adjacent to an earlier known binding site of a natural product inhibitor of the BC domain, which might be exploited to design tight‐binding and ligand‐efficient chemical matter [112] (Figure 2.1).

This information was viewed as crucial by the project team to support initiation of the discovery project, since the presence of these high energy hydration sites was expected to greatly increase the likelihood the project team would be able to find satisfactory chemical matter [111]. The displacement of the water bound to these hydration sites was used as a constraint in virtual screening efforts, where only compounds expected to displace the water molecules were purchased and assayed. In agreement with the expectations of the computational modeling, the virtual screening campaign identified compound ND‐022 (Figure 2.2), which displaced the solvating water occupying these high‐energy hydration sites and was observed to bind to the target with 3.9 μM affinity [112]. Although in this case study, we have focused on the contributions that computational modeling made to the target validation, feasibility assessment, and hit discovery phases of this discovery project, ND‐022 was advanced into lead optimization, which resulted in the discovery of an exceptional development candidate molecule, ND‐630, which has been advanced into phase II clinical trials [4, 112, 113].

Figure 2.1 (a) An overlay of ND‐022 with the natural product soraphen and (b) the high energy hydration sites that were crucial to the decision to initiate the ACC drug discovery project and the discovery of ND‐022.

      Source: Reproduced with permission. Copyright© 2017, Elsevier [112].

Figure 2.2 Structures of ND‐022 and ND‐630.

      2.3.2 Optimizing Selectivity in Lead Optimization for Tyrosine Kinase 2

      The Janus kinase (JAK) family, comprised of Janus kinase 1 (JAK1), Janus kinase 2 (JAK2), Janus kinase 3 (JAK3), and tyrosine kinase 2 (TYK2), controls a variety of inflammation pathways, which may be relevant to the treatment of autoimmune disorders such as psoriasis, inflammatory bowel disease, and rheumatoid arthritis [114]. A variety of drug therapies inhibit these kinases, including Ruxolitinib, Tofacitinib, Baricitinib, Fedratinib, and Upadacitinib [114–118]. However, significant side effects are sometimes observed with these therapies. For example, inhibition of JAK2 can lead to anemia, and inhibition of JAK1 and JAK3 can lead to increased risk of infection [112]. Interestingly, genome‐wide and phenome‐wide association studies suggest the safety profile of a TYK2 selective drug therapy might have an improved safety profile while remaining efficacious for the treatment of autoimmune disorders [119–121].

The preceding opportunity notwithstanding, designing a highly selective TYK2 inhibitor is a substantial challenge. The four Janus family kinases have a high degree of structural and sequence similarity, especially so in the active site regions of the receptors [112, 122] (Figure 2.3). This has led to very few selective TYK2 drug therapies being advanced into clinical trials [123]. Nimbus therapeutics' approach to solving this difficult challenge in late‐stage lead optimization was to employ rigorous physics‐based methods at an unprecedented scale, including free energy calculation‐based scoring of 4000 design ideas over a six month period to identify molecules predicted by the free energy calculations to be potent and selective inhibitors of TYK2 [112]. Of these 4000 evaluated compounds, 46 were advanced to synthesis and assay with 9 of the compounds meeting the targeted property profile. This computational modeling hit rate of 20% identifying satisfactory chemical matter should be viewed as outstanding given the difficulty of the challenge. Preclinical studies of several of these compounds further demonstrated outstanding efficacy in mouse model of psoriasis, and the discovery project has succeeded to advance a TYK2 inhibitor into clinical studies [112, 124].

Figure 2.3 Superimposed crystal structures of TYK2, JAK1, JAK2, and JAK3 cocrystalized with tofacitinib.

      Source: Reproduced with permission. Copyright© 2017, Elsevier [112].

      2.3.3 Discovery of Novel Allosteric Covalent Inhibitors of KRAS^G12C

KRAS mutations are widely observed in a variety of cancers [125]. The G12C mutation has received a great deal of recent attention, since it creates an allosteric pocket that can be exploited by covalent ligands to inactivate the protein [125]. Seeking to build on this growing body of work, researchers at Bayer Pharmaceuticals employed a modern computational approach to identify novel allosteric inhibitors, as depicted in Figure 2.4 [125]. In this workflow, seven million putative design ideas were first computationally enumerated and then advanced to a computational screening funnel where more approximate methods, i.e. pharmacophore analysis and MM‐GB/SA scoring, were used to reject unpromising design ideas, and then more accurate and computationally intensive methods, i.e. free energy calculations, were used to finalize those molecules which were advanced to synthesis. The results of this effort led to only four of the original seven million design ideas being designated for synthesis and assay. This effort led directly to the development of a novel congeneric series with KRAS^G12C activity, and may constitute an important step along the path of the discovery of KRAS‐targeted drug therapies.

Скачать книгу