Inflammatory cytokines are key signaling molecules that can promote an immune response, thus their RNA turnover must be tightly controlled during infection. Most studies investigate the RNA decay pathways in the cytosol or nucleoplasm but never focused on the nucleolus. Although this organelle has well-studied roles in ribosome biogenesis and cellular stress sensing, the mechanism of RNA decay within the nucleolus is not completely understood. Here, we report that the nucleolus is an essential site of inflammatory pre-mRNA instability during infection. RNA-sequencing analysis reveals that not only do inflammatory genes have higher intronic read densities compared with non-inflammatory genes, but their pre-mRNAs are highly enriched in nucleoli during infection. Notably, nucleolin (NCL) acts as a guide factor for recruiting cytosine or uracil (C/U)-rich sequence-containing inflammatory pre-mRNAs and the Rrp6-exosome complex to the nucleolus through a physical interaction, thereby enabling targeted RNA delivery to Rrp6-exosomes and subsequent degradation. Consequently, Ncl depletion causes aberrant hyperinflammation, resulting in a severe lethality in response to LPS. Importantly, the dynamics of NCL post-translational modifications determine its functional activity in phases of LPS. This process represents a nucleolus-dependent pathway for maintaining inflammatory gene expression integrity and immunological homeostasis during infection.

Efficient identification of chemical probes for the manipulation and understanding of biological systems demands specificity for target proteins. Computational means to optimize candidate compound selection for experimental selectivity evaluation are being sought. The active learning virtual screening method has demonstrated the ability to efficiently converge on predictive models with reduced datasets, though its applicability domain to probe identification has yet to be determined. In this article, we challenge active learning’s ability to predict inhibitory bioactivity profiles of selective compounds when learning from chemogenomic features found in non-selective ligand-target pairs. Comparison of controls versus multiple molecule representations de-convolutes factors contributing to predictive capability. Experiments using the matrix metalloproteinase family demonstrate maximum probe bioactivity prediction achieved from only approximately 20% of non-probe bioactivity; this data volume is consistent with prior chemogenomic active learning studies despite the increased difficulty from chemical biology experimental settings used here. Feature weight analyses are combined with a custom visualization to unambiguously detail how active learning arrives at classification decisions, yielding clarified expectations for chemogenomic modeling. The results influence tactical decisions for computational probe design and discovery.