背景:PICK1PDZ结构域已被确定为神经系统疾病的潜在药物靶标。经过多年的努力,一些抑制剂,例如TAT-C5和mPD5,已经通过实验发现以相对高的结合亲和力与PDZ结构域结合。随着计算研究的迅速发展,迫切需要更有效的计算方法来设计靶向蛋白质的可行配体。
方法:最近,一个新开发的程序,名为AfDesign(ColabDesign的一部分),位于https://github.com/sokrypton/ColabDesign),一个基于AlphaFold的开源软件,已经表明能够产生与目标蛋白质结合的配体,从而潜在地促进配体发展过程。为了评估这个程序的性能,我们探索了它靶向PICK1PDZ域的能力,鉴于我们目前对它的理解。我们发现,配体的指定长度和再循环次数在生成具有最佳性能的配体中起着至关重要的作用。
结果:利用序列长度为5的配体的AfDesign产生了与先前鉴定的配体最高的可比较的配体。此外,与手动创建的序列相比,这些设计的配体显示出明显更低的结合能。
结论:这项工作表明,AfDesign可能是促进探索配体空间以靶向PDZ结构域的强大工具。
BACKGROUND: The PICK1 PDZ domain has been identified as a potential drug target for neurological disorders. After many years of effort, a few inhibitors, such as TAT-C5 and mPD5, have been discovered experimentally to bind to the PDZ domain with a relatively high binding affinity. With the rapid growth of computational research, there is an urgent need for more efficient computational methods to design viable ligands that target proteins.
METHODS: Recently, a newly developed program called AfDesign (part of ColabDesign) at https:// github.com/sokrypton/ColabDesign), an open-source software built on
AlphaFold, has been suggested to be capable of generating ligands that bind to targeted proteins, thus potentially facilitating the ligand development process. To evaluate the performance of this program, we explored its ability to target the PICK1 PDZ domain, given our current understanding of it. We found that the designated length of the ligand and the number of recycles play vital roles in generating ligands with optimal properties.
RESULTS: Utilizing AfDesign with a sequence length of 5 for the ligand produced the highest comparable ligands to that of prior identified ligands. Moreover, these designed ligands displayed significantly lower binding energy compared to manually created sequences.
CONCLUSIONS: This work demonstrated that AfDesign can potentially be a powerful tool to facilitate the exploration of the ligand space for the purpose of targeting PDZ domains.