Developing a novel pipeline to determine monoclonal antibody solution structures and stabilities in bioprocessing conditions to optimise manufacture

背景：人IgG1和IgG4单抗已成为生物技术行业中重要的治疗药物。抗体主要是为对抗特定靶点而设计的，在下游的生物过程环境中非常容易聚集，特别是当抗体浓度增加到最终产品浓度50-200 mg/ml时。这对生物技术行业来说是一个重大挑战，因为聚集可能严重危及安全性。理解如何管理抗体聚集越来越重要，这是由BBSRC组织的一年两次的金砖四国会议上讨论的一个主要主题，提倡者参加了会议。直接阐明启动蛋白质聚集的物理参数将允许设计出聚集较少的更强大的候选铅。这些因素包括缓冲液pH、温度(T)、缓冲液离子强度(I)和抗体浓度(C)。系统地研究这四种因素对抗体溶液结构的影响以及它们形成大分子寡聚体的可能性，将揭示它们聚集的关键阶段，并提出更合理的蛋白质工程方法来阻断它。我们开发了使用散射和超离心法来确定免疫球蛋白抗体原子化溶液结构的新方法(Rayner等人，2013、2014、2015)。今年，我们的方法得到了进一步的改进，在CCP-SAS项目中融入了先进的结构建模方法。例如，在实验上，我们的测试显示，IgG4抗体在pH 3，然后pH 7分两步改变其构象，形成一个紧凑的结构，然后沉淀，我们可以模拟所有这些构象变化(Perkins，未发表)。因此，我们非常适合在BBSRC.AIMS的职权范围内开展这一及时的项目：(1)通过研究两种人源单抗IgG1和IgG4的溶液构象随pH、T、i和c的变化，我们将构建抗体构象和聚集倾向的相图。(2)使用数据的精确原子模拟，我们将确定这些不同条件下的抗体结构，以将主要构象变化与实验条件相关联。(3)纳入另外四种免疫球蛋白单抗将使我们能够开发一条快速评估新抗体产品的管道。实施计划：富士胶片将提供克量的单抗IgG1和IgG4用于实验，并就免疫球蛋白单抗的制造工艺进行现场培训。使用法国格勒诺布尔ESRF同步加速器的高通量X射线光束线BM29和96孔微板，将迅速和系统地研究pH、T、I和C的影响，以得出相图。如果需要，法国格勒诺布尔ILL的D22仪器上的高通量中子数据收集将补充X射线数据集。UCL的分析性超速离心法将显示在选定的pH、T、i和c值下的构象或寡聚变化。我们新的CCP-SAS建模软件(Huu等人2015,2016)通过高性能计算将生成多达70万个随机但物理准确的构象。这个结构库将与每个数据集进行比较，以识别每个缓冲区中的最佳匹配，从而识别构象。最终，我们将设计一个操作流水线，以快速表征新开发的抗体结构的可制造性。

BACKGROUND: Monoclonal human IgG1 and IgG4 antibodies have become important therapeutics in the biotechnology industry. Primarily engineered for efficacy against specific targets, antibodies are highly susceptible to aggregation in downstream bioprocess environments, particularly as antibody concentrations increase up to final product concentrations of 50-200 mg/ml. This represents a major challenge for the biotechnology industry where aggregates can seriously compromise safety. It is increasingly important to understand how to manage antibody aggregation, and this is a major theme addressed in the twice-yearly BRIC meetings organised by the BBSRC, attended by the proposers. A direct elucidation of the physical parameters that initiate protein aggregation would allow the design of more robust lead candidates that aggregate less. These factors include buffer pH, temperature (T), buffer ionic strength (I), and antibody concentration (c). A systematic study of the effects of these four factors on antibody solution structures and their prospensity to form large oligomers will reveal the key stages towards their aggregation and suggest avenues for more rational protein engineering to block it. We have developed new methods using scattering and ultracentifugation to determine atomistic solution structures for IgG antibodies (Rayner et al, 2013, 2014, 2015). This year, our methods have been improved further by the incorporation of advanced structural modelling methods in the CCP-SAS project. For example, experimentally, our tests showed that an IgG4 antibody changed its conformation in two steps at pH 3, then pH 7, to form a compact structure that precipitates, and we could model all these conformational changes (Perkins, unpublished). We are thus ideally poised to undertake this timely project within the remit of the BBSRC.AIMS: (1) By studying the solution conformations of two human monoclonal IgG1 and IgG4 antibodies as a function of pH, T, I and c, we will construct phase diagrams of antibody conformation and aggregation propensity. (2) Using accurate atomistic modelling of the data, we will determine antibody structures in these different conditions to correlate major conformational changes with experimental conditions. (3) The inclusion of four other IgG monoclonals will enable us to develop a pipeline for the rapid assessment of new antibody products.EXPERIMENTAL PLAN: Fujifilm will provide gram amounts of monoclonal IgG1 and IgG4 for experiments, and on-site training on the manufacturing processes for IgG monoclonals. Using the high-throughput X-ray beamline BM29 at the ESRF synchrotron, Grenoble, France, with 96-well microplates, the effects of pH, T, I and c will be explored rapidly and systematically to yield phase diagrams. If required, high-throughput neutron data collection on instrument D22 at the ILL, Grenoble, France, will supplement the X-ray data sets. Analytical ultracentrifugation at UCL will reveal conformational or oligomerisation changes at selected pH, T, I and c values. Our new CCP-SAS modelling software (Hui et al 2015, 2016) via high performance computing will generate as many as 700,000 randomised but physically-accurate conformations. This structural library will be compared with each data set to identify the best fit in each buffer, hence identifying the conformation. Ultimately, we will design a pipeline of operations to characterise newly-developed antibody structures rapidly for their manufacturability.