Machine learning has been increasingly used for protein engineering. However, because the general sequence contexts they capture are not specific to the protein being engineered, the accuracy of existing machine learning algorithms is rather limited. Here, we report ECNet (evolutionary context-integrated neural network), a deep-learning algorithm that exploits evolutionary contexts to predict functional fitness for protein engineering. This algorithm integrates local evolutionary context from homologous sequences that explicitly model residue-residue epistasis for the protein of interest with the global evolutionary context that encodes rich semantic and structural features from the enormous protein sequence universe. As such, it enables accurate mapping from sequence to function and provides generalization from low-order mutants to higher-order mutants. We show that ECNet predicts the sequence-function relationship more accurately as compared to existing machine learning algorithms by using ~50 deep mutational scanning and random mutagenesis datasets. Moreover, we used ECNet to guide the engineering of TEM-1 β-lactamase and identified variants with improved ampicillin resistance with high success rates.

Enzyme function annotation is a fundamental challenge, and numerous computational tools have been developed. However, most of these tools cannot accurately predict functional annotations, such as enzyme commission (EC) number, for less-studied proteins or those with previously uncharacterized functions or multiple activities. We present a machine learning algorithm named CLEAN (contrastive learning–enabled enzyme annotation) to assign EC numbers to enzymes with better accuracy, reliability, and sensitivity compared with the state-of-the-art tool BLASTp. The contrastive learning framework empowers CLEAN to confidently (i) annotate understudied enzymes, (ii) correct mislabeled enzymes, and (iii) identify promiscuous enzymes with two or more EC numbers—functions that we demonstrate by systematic in silico and in vitro experiments. We anticipate that this tool will be widely used for predicting the functions of uncharacterized enzymes, thereby advancing many fields, such as genomics, synthetic biology, and biocatalysis.

Tumor repopulation during therapy is an important cause of treatment failure. Strategies to overcome repopulation are arising in parallel with advances in the comprehension of underlying biological mechanisms. Here, we reveal a new mechanism by which high mobility group box 1 (HMGB1) released by dying cells during radiotherapy or chemotherapy could stimulate living tumor cell proliferationInhibition or genetic ablation of HMGB1 suppressed tumor cell proliferation. This effect was due to binding of HMGB1with the member receptor for advanced glycation end-products (RAGE), which activated downstream ERK and p38 signaling pathway and promoted cell proliferation. Furthermore, higher HMGB1 expression in tumor tissue correlated with poor overall survival and higher HMGB1 concentration was detected in serum of patients who accepted radiotherapy. Collectively, the results from this study suggested that interaction between dead cells and surviving cells might influence the fate of tumor. HMGB1 could be a novel tumor promoter with therapeutic and prognostic relevance in cancers.

No abstract