Decoding efficacy and resistance space at a drug binding site

Interactions between drugs and their targets impact efficacy and, when altered by mutation, can result in resistance 1-3. Assessing and understanding the impacts of all possible mutations at a drug binding-site remain challenging, however. Here we use Multiplex Oligo-Targeting (MOT) for mutational profiling, and computational modelling, to decode efficacy and resistance space at the otherwise native binding-site for a low nanomolar potency, anti-trypanosomal, proteasome inhibitor. We saturation-edited twenty codons in the Trypanosoma brucei proteasome b5 subunit. MOT libraries were then subjected to stepwise drug selection, amplicon-sequencing, and codon variant scoring, yielding dose-response profiles for >100 resistance-conferring mutants, among 1,280 possible codon variants. Codon variant scores were predictive of relative resistance observed using a bespoke set of mutants, while fitness profiling revealed otherwise extensive constraints on mutational fitness and resistance space. The resistance profile that emerged allowed us to readily predict routes to spontaneous drug-resistance observed within accessible, single nucleotide mutational space. In silico analysis of b5 subunit mutations predicted impacts on ligand-affinity via steric effects, hydrogen-bonding and lipophilicity, which when combined with predictions of proteasome structure - function perturbing mutations, were closely aligned with observed impacts on drug resistance. We conclude that MOT-library profiling facilitates assessment of all possible mutations at a drug binding-site. Further decoding of drug-target structure-activity relationships and drug resistance space will facilitate the design of more effective and durable drugs.

药物与其靶点之间的相互作用会影响药效，而当该相互作用因突变发生改变时，可引发耐药性[1-3]。然而，评估并阐明药物结合位点上所有潜在突变的影响仍存在不小挑战。本研究借助多重寡核苷酸靶向（Multiplex Oligo-Targeting, MOT）技术开展突变谱分析，并结合计算建模，对一款低纳摩尔活性的抗锥虫蛋白酶体抑制剂的天然结合位点处的药效与耐药性空间进行了解码。我们对布氏锥虫（Trypanosoma brucei）蛋白酶体β5亚基的20个密码子实施了饱和编辑。随后将构建的MOT文库进行梯度药物筛选、扩增子测序及密码子变异打分，在1280种可能的密码子变异中，获取了超过100种耐药相关突变体的剂量反应谱。密码子变异打分可有效预测使用定制突变体集观测到的相对耐药性水平，而适应性谱分析则揭示了突变适应性与耐药性空间中广泛存在的约束机制。所得耐药性谱可使我们快速预测在可及的单核苷酸突变空间内观测到的自发性耐药途径。对β5亚基突变的计算机模拟分析预测，突变可通过空间位阻效应、氢键作用和亲脂性改变影响配体亲和力；若将此类预测与蛋白酶体结构-功能扰动突变的预测相结合，则与观测到的耐药性影响高度契合。本研究最终证实，MOT文库谱分析能够实现对药物结合位点上所有潜在突变的系统性评估。进一步解析药物-靶点结构-活性关系与耐药性空间，将助力开发更高效、更持久的治疗药物。