Cowpea weevil (Callosobruchus maculatus) is a major pest that leads to severe damage of the stored leguminous grains. Several management approaches, including physical barriers, biological or chemical methods, are used for controlling bruchid in cowpea. These methods usually target the metabolically active state of weevil. However, it becomes less effective at early stages as egg, larva, or pupa under low temperature and oxygen conditions. Since hypoxia-inducible factor-1 (HIF-1) is known to coordinate multiple gene responses to low oxygen or low temperature signals, we examined the HIF-1α gene expression under low temperature and hypoxic treatments. At −20 °C, it took 4 h to reduce the survival rate for eggs, larvae, and pupae down to 10%, while at 4 °C and 15 °C, the survival rate remained higher than 50% even after 128 h as HIF-1α gene expression peaked at 15 °C. Moreover, HIF-1 protein offers a valuable target for early stage pest control complementary to traditional methods. In particular, HIF-1 inhibitor camptothecin (CPT), one of the five HIF-1 inhibitors examined, achieved a very significant reduction of 96.2% and 95.5% relative to the control in weevil survival rate into adult at 4 °C and 30 °C, respectively. Our study can be used as one model system for drug development in virus infections and human cancer.

Molecular representation learning (MRL) has garnered significant attention due to its pivotal role in downstream applications such as molecular property prediction and drug discovery. In most MRL approaches, molecules are encoded into 2D topological graphs via graph neural network (GNN), which suffers from over-smoothing issues and limited receptive fields. Furthermore, most GNN models fail to utilize the 3D spatial structural information that determines molecular physicochemical properties and biological activity. To this end, here we propose multimodal contrast-enhanced molecular representation learning (MCMRL). This approach utilizes both the 2D topological information and 3D structural information of molecules for contrastive learning to enhance molecular graph representations. Further, it integrates additional molecular fingerprint information and feature fusion techniques to incorporate multimodal knowledge, yielding more reliable and generalizable molecular representations. MCMRL is pre-trained on ~10 million unlabeled molecules from PubChem, followed by various downstream benchmark tasks. Experimental results demonstrate that MCMRL achieves superior performance in 9 out of 13 benchmark tests for molecular property prediction, validating its effectiveness in molecular representation learning. Furthermore, potential molecular drugs binding to biological target protein DRD2 screened by MCMRL representation show promising affinity score, which also demonstrates the efficacy of the proposed method.