A Systems Chemoproteomic Analysis of Acyl-CoA/Protein Interaction Networks

Abstract

Acyl-coenzyme A (CoA)/protein interactions are essential for life. Despite this importance, their global scope and selectivity remains undefined. Here, we describe CATNIP (CoA/AcetylTraNsferase Interaction Profiling), a chemoproteomic platform for the high-throughput analysis of acyl-CoA/protein interactions in endogenous proteomes. First, we apply CATNIP to identify acetyl-CoA-binding proteins through unbiased clustering of competitive dose-response data. Next, we use this method to profile the selectivity of acyl-CoA/protein interactions, leading to the identification of specific acyl-CoA engagement signatures. Finally, we apply systems-level analyses to assess the features of protein networks that may interact with acyl-CoAs, and use a strategy for high-confidence proteomic annotation of acetyl-CoA-binding proteins to identify a site of non-enzymatic acylation in the NAT10 acetyltransferase domain that is likely driven by acyl-CoA binding. Overall, our studies illustrate how chemoproteomics and systems biology can be integrated to understand the roles of acyl-CoA metabolism in biology and disease.

Keywords: acetyl-CoA; acetylation; acetyltransferase; activity-based protein profiling; chemical proteomics; epigenetics; malonylation; metabolism; pharmacology; systems biology.

Figure 1.

Diverse consequences of acyl-CoA interactions on protein activity and signaling. Metabolic acyl-CoAs can interact with proteins as cofactors, inhibitors, allosteric modulators, or covalent modifiers.

Figure 2.

Profiling acetyl-CoA/protein interactions using CATNIP. (a) Schematic for chemoproteomic analyses of acyl-CoA/protein interactions. (b) Normalized dNSAF values across 4 acetyl-CoA concentrations (0, 3, 30, 300μM) were t-SNE transformed and plotted in two dimensions for all proteins competed in CATNIP experiments (n=3 LC-MS/MS experiments). (c) Dose-response profiles of acetyl-CoA CATNIP clusters. Colored lines indicate the capture profiles of individual proteins at each concentration of acetyl-CoA competitor. Black lines indicate the mean capture profile for all proteins in a given cluster. (d) Clusters 1–3 are enriched in Uniprot-annotated CoA-binding proteins (“CoA binders”) as well as members of acetyltransferase complexes (”AT interactors”). (e) Analysis of annotated CoA binders exhibiting 2-fold competition at each concentration. Fold-change in d and e were calculated by QPROT. (f) Gene ontology analysis of CoA binders and AT interactors lying in CATNIP clusters 1 and 2. Fold enrichment of functional terms are plotted versus statistical significance (−log10[FDR]). Circle size reflects the number of proteins matching a given term. Functional enrichment was performed with the tool DAVID (https://david.ncifcrf.gov).

Figure 3.

Applying CATNIP to profile the comparative pharmacology of CoA metabolites. (a) CoA metabolites (1-6) analyzed in this study. (b) Venn diagram depicting overlap between proteins whose capture was competed more than four-fold by acetyl-CoA (1) or all other CoAs (2-6). (c) Venn diagram depicting overlap between proteins whose capture was two-fold competed by acyl-CoAs 3-6. (d) Comparison of acyl-CoA (x-axis) and acetyl-CoA (y-axis and acyl-CoA competition. Uniprot-annotated CoA binders and AT interactors are highlighted in dark blue and light blue, respectively. Log2FC values for b-d were calculated using QPROT. (e) Comparative CATNIP analysis highlights distinct signatures of metabolite interaction amongst families of CoA binders. Relative capture was calculated by comparing average dNSAF values. White = more competition by CoA metabolite, blue = less competition by CoA metabolite. All graphs are from n=3 control/competitor datasets.

Figure 4.

Applying CATNIP to profile AT feedback inhibition. (a) Exemplary dose-response profiles of proteins that interact strongly with CoA. (b) Exemplary dose-response profiles of proteins more weakly with CoA. (c) Scheme for isotopic tracing of metabolic source of the acetate group in ac4C. Heavy (U-¹³C) glucose or acetate were applied in separate metabolic labeling experiments. Incorporation into ac4C was assessed by mass isotopomer analysis of digested total RNA. (d) The major source of ac4C’s N4-acetyl group is glucose. (e) Disruption of ACLY reduces levels of ac4C in poly(A)-enriched, but not total RNA. Values represent ≥ 3 replicates, analyzed by two-tailed student’s t-test (ns = not significant, *P<0.05, **P<0.01, ***P<0.001).

Figure 5.

Applying CATNIP to assess the annotation of the CoA-binding proteome. (a) Schematic for differentiating CATNIP-enriched proteins from background based on significant competition and absence from common contaminant databases. (b) Proteins from diverse AT families display statistically significant CoA/acyl-CoA competition. (c) Topological network analysis of proteins exhibiting significant interaction with ≥ 3 CoA metabolites. Protein nodes are colored based on the metric PCA2. Color bar: red = high values; blue = low values. Node size is proportional to the number of proteins in the node. (d) Combining CATNIP competition (QPROT Log2FC ≤ −2 and FDRdown<0.05) and acetylation stoichiometry filters greatly enriches CoA-binders and AT-interactors relative to either measure alone. (e) Comparing annotated sites of lysine malonylation (top) and competition by malonyl-CoA mimic 5 (bottom) for CoA-binders detected by de novo CATNIP analysis. (f) A conserved site of lysine malonylation lies in close proximity to the acetyl-CoA binding site of bacterial NAT10. (g) Overexpressed FLAG-NAT10 is malonylated in the presence of malonyl-CoA, consistent with a non-enzymatic acylation mechanism.