The COVID-19 pandemic caused by SARS-CoV-2 remains a global public health threat and has prompted the development of antiviral therapies. Artificial intelligence may be one of the strategies to facilitate drug development for emerging and re-emerging diseases. The main protease (M^pro) of SARS-CoV-2 is an attractive drug target due to its essential role in the virus life cycle and high conservation among SARS-CoVs. In this study, we used a data augmentation method to boost transfer learning model performance in screening for potential inhibitors of SARS-CoV-2 M^pro. This method appeared to outperform graph convolution neural network, random forest and Chemprop on an external test set. The fine-tuned model was used to screen for a natural compound library and a de novo generated compound library. By combination with other in silico analysis methods, a total of 27 compounds were selected for experimental validation of anti-M^pro activities. Among all the selected hits, two compounds (gyssypol acetic acid and hyperoside) displayed inhibitory effects against M^pro with IC50 values of 67.6 μM and 235.8 μM, respectively. The results obtained in this study may suggest an effective strategy of discovering potential therapeutic leads for SARS-CoV-2 and other coronaviruses.

Human immunodeficiency virus-1 (HIV-1) recognizes one of its principal coreceptors, CXC chemokine receptor 4 (CXCR4), on the host cell via the third variable loop (V3 loop) of HIV-1 envelope glycoprotein gp120 during the viral entry process. Here, the mechanism of the molecular recognition of HIV-1 gp120 V3 loop by coreceptor CXCR4 was probed by synthetic peptides containing the full-length V3 loop. The two ends of the V3 loop were covalently linked by a disulfide bond to form a cyclic peptide with better conformational integrity. In addition, to probe the effect of the changed side-chain conformations of the peptide on CXCR4 recognition, an all-D-amino acid analog of the L-V3 loop peptide was generated. Both of these cyclic L- and D-V3 loop peptides displayed comparable binding recognition to the CXCR4 receptor, but not to another chemokine receptor, CCR5, suggesting their selective interactions with CXCR4. Molecular modeling studies revealed the important roles played by many negative-charged Asp and Glu residues on CXCR4 that probably engaged in favorable electrostatic interactions with the positive-charged Arg residues present in these peptides. These results support the notion that the HIV-1 gp120 V3 loop-CXCR4 interface is flexible for ligands of different chiralities, which might be relevant in terms of the ability of the virus to retain coreceptor recognition despite the mutations at the V3 loop.