CMAdb: a structurally annotated database of chaperone-mediated autophagy substrates

Abstract

Chaperone-Mediated Autophagy (CMA) is a highly selective pathway of protein degradation implicated in multiple diseases. The precise role of the pathway under pathological conditions is speculated to be largely governed by the nature of the substrate undergoing degradation. Although numerous proteins containing KFERQ-like motifs have been identified as CMA substrates, the information remains scattered across the literature. Importantly, the structural determinants governing CMA substrate recognition are poorly understood, particularly, in proteins with multiple potential targeting motifs, whose functionality may depend on motif exposure, solvent accessibility, and secondary structure preferences. In this work, we developed the Chaperone-Mediated Autophagy substrate database (CMAdb), the first dedicated CMA substrate database integrating 103 curated UniProt entries corresponding to experimentally validated CMA substrates, 463 annotated CMA-targeting motifs, structural annotations of 158 canonical motif instances, full-length protein structural information, and ortholog-based sequence and structural comparisons. These assessments incorporate both experimentally resolved and AlphaFold (AF)-predicted structures. Where experimentally resolved structures of the same protein are available, they are used to assess agreement with the corresponding AF model. For homologous proteins, sequence- and structure-based comparisons are provided to facilitate structural interpretation. In total, the database provides approximately 8000 superimposed structures together with annotations of structural features and CMA-targeting motifs. Analysis of the curated canonical CMA substrate dataset showed that approximately 72% of the identified CMA-targeting motifs are located within α-helical regions. ~81% of the motifs are observed to be solvent-exposed overall, while the central residue exhibits the highest proportion of buried conformations. Additionally, we provide an interactive web server that allows users to query any protein and seamlessly retrieve all integrated information, including CMA-targeting motif predictions, solvent accessibility analysis, sequence and structural alignments, and structural comparisons of AlphaFold-predicted models with experimentally resolved and homologous protein structures. The CMAdb developed in this study is freely accessible via the webserver at https://cma-substrates.co.in . In this study, we leveraged the tool to identify context-dependent motifs in CMA substrates through the integration of sequence and structural annotations, literature-curated evidence of context-dependent conformational transitions, and with template-based structural comparisons.