杜克大学Pranam Chatterjee小组的一项最新研究提出了利用掩码语言建模的肽结合物靶序列条件设计。该研究于2025年8月13日发表于国际一流学术期刊《自然—生物技术》杂志上。

在这里，该研究团队介绍PepMLM，一个新的线性肽结合物的目标序列条件设计器。通过将同源肽序列定位于靶蛋白序列的C端，PepMLM对ESM-2蛋白语言模型进行微调，以完全重建结合区，实现低复杂度匹配或改进已验证的肽-蛋白序列对。在基于AlphaFold对接的硅基准测试成功后，小组通过结合和降解实验验证了PepMLM的有效性。PepMLM衍生的肽显示出与癌症和生殖靶点（包括NCAM1和AMHR2）的序列特异性结合，并能够靶向降解各种疾病背景下的蛋白质，从亨廷顿病到活病毒感染。总之，PepMLM可以设计任何靶蛋白的候选结合物，而不需要结构输入，促进了治疗开发的广泛应用。

研究人员表示，基于蛋白质的结合剂的计算设计为接近“不可药物”靶标提供了独特的机会，但有效的结合剂设计通常依赖于稳定的三维结构或结构影响的潜在空间。

附：英文原文

Title: Target sequence-conditioned design of peptide binders using masked language modeling

Author: Chen, Leo Tianlai, Quinn, Zachary, Dumas, Madeleine, Peng, Christina, Hong, Lauren, Lopez-Gonzalez, Moises, Mestre, Alexander, Watson, Rio, Vincoff, Sophia, Zhao, Lin, Wu, Jianli, Stavrand, Audrey, Schaepers-Cheu, Mayumi, Wang, Tian Zi, Srijay, Divya, Monticello, Connor, Vure, Pranay, Pulugurta, Rishab, Pertsemlidis, Sarah, Kholina, Kseniia, Goel, Shrey, DeLisa, Matthew P., Chi, Jen-Tsan Ashley, Truant, Ray, Aguilar, Hector C., Chatterjee, Pranam

Issue&Volume: 2025-08-13

Abstract: The computational design of protein-based binders presents unique opportunities to access ‘undruggable’ targets, but effective binder design often relies on stable three-dimensional structures or structure-influenced latent spaces. Here we introduce PepMLM, a target sequence-conditioned designer of de novo linear peptide binders. Using a masking strategy that positions cognate peptide sequences at the C terminus of target protein sequences, PepMLM finetunes the ESM-2 protein language model to fully reconstruct the binder region, achieving low perplexities matching or improving upon validated peptide–protein sequence pairs. After successful in silico benchmarking with AlphaFold-based docking, we experimentally validate the efficacy of PepMLM through both binding and degradation assays. PepMLM-derived peptides demonstrate sequence-specific binding to cancer and reproductive targets, including NCAM1 and AMHR2, and enable targeted degradation of proteins across diverse disease contexts, from Huntington’s disease to live viral infections. Altogether, PepMLM enables the design of candidate binders to any target protein, without requiring structural input, facilitating broad applications in therapeutic development.

DOI: 10.1038/s41587-025-02761-2

Source: https://www.nature.com/articles/s41587-025-02761-2