Fig 1: Structural annotation of phosphosites on metabolic enzymes identifies overrepresentation of pY in functional and dimerization domains.(A) Joint plot representing assessment of published structural data for metabolic enzymes based on resolution and refinement cross-validation. High quality structures are indicated by red dotted lines at resolution ≤ 2.6Å and Rfree – Rwork ≤ 0.05 to filter for high quality x-ray crystal structures. (B) Stacked histogram plot of phosphosite distance from domains. Protein functional domains curated from UniprotKB were annotated on corresponding structures using PyMol. For each phosphosite on a given enzyme, distance was measured from the hydroxyl group of the apo residue to the center of mass (COM) of any defined functional domain, distances were averaged across all available PDB structures of each enzyme for phosphosite-domain pairs. Stacked histogram plot represents frequency of distances for each pair of phosphosite and functional domain, where 50% of the data is indicated by red dotted line at 23.29Å. (C) Functional domains for each enzyme were classified into domain types based on UniprotKB annotation, and stacked bar plot represents the proportion of phosphosite residues within 23Å of each domain type. (D) Bar plot representation of hypergeometric-distribution test with multiple hypothesis correction of the cumulative distribution function done via Benjamini-Hochberg test. (E) Stacked histogram plot of complex formation score for dimer interfaces that contain phosphosites. High quality protein structures of metabolic enzymes were used to identify dimerization interfaces and the residues that form contacts important for complex formation using PISA analysis tool. The data was filtered for phosphosites only, and stacked bar plot represents frequency of complex formation score (CSS) obtained from PISA. CSS≥0.3 indicates that the interface plays a strong role in dimerization. (F) Heatmap for the frequency of interaction between phosphosite hydroxyl group (OH) or backbone (Ca) with residues and ligands found in the interface region. (G) 4 examples of phosphosites on metabolic enzymes illustrate the versatile role phosphosites serve in functional domains. Structure of AKR1C1 showing hydrophobic interactions (blue dotted line) of Y55 with cofactors (NADH or NADPH), Y24 with steroid substrates (progesterone shown here), and pi-stacking interaction (green dotted line) of Y216 with the nicotinamide ring of NADP+. G6PD protein structure showing pi-stacking interactions between Y401, Y503, and NADP within the interface of two monomers. ACAT1 structure showing interaction between the hydroxyl group on Y219 with cations (e.g., K+, Mg2+) as well as a water bridge and hydrogen bond with the adenosine ring of Coenzyme A (CoA). Structure of UMPS showing hydrogen bonds and pi-stacking interaction between Y37 and OMP. Illustrations were generated using SwissModel103,104,138 and PLIP105 tool and the following PDBs: AKR1C1 (1mrq)139, ACAT1 (2iby)140, G6PD (7sni)55, UMPS (2wns)102.
Fig 2: Integrative omics analysis yields pY sites that predict HFD-induced changes in metabolites.(A) Selected integrative map depicting pY sites and metabolites in pathways for purine degradation, oxidative phosphorylation (OXPHOS), redox homeostasis, fatty acid (FA) metabolism, and anaplerosis/cataplerosis. (B) Scatter plot of metabolite PLSR predictive score (Q2) vs fold change in Log2(HFD/NCD) where metabolites with valid model are color-coded by pathway. Each metabolite (Y-matrix) was regressed against the top 50% of pY sites with greatest change in magnitude (X-matrix) with k-fold cross-validation (see Methods). A model prediction score (Q2) ≥ 0.4 was used as an indicator of metabolites predicted by pY sites without overfitting. (C) Stacked histogram plot illustrating the frequency of predictive pY sites and their respective distances from functional domains and interface region. (D) Scatter plot of representative PLSR model coefficients vs Log2(HFD/NCD) of pY sites for uridine monophosphate (UMP), each datapoint is a pY site. PLSR coefficient indicates a positive (red) or negative (blue) relationship between UMP and the pY site. Highlighted sites are on enzymes in the de novo pyrimidine synthesis pathway: Umps Y37 and Cps1 Y590, Y1450, and Y852.
Fig 3: Interrogation of pY functional role by coupling isotope labeled tracing and CRISPRi-rescue validates regulatory role of IDH1 Y391 and UMPS Y37 on enzyme activity.(A) U-13C5-Glutamine stable isotope tracing scheme to evaluate the role of IDH1 Y391 which is positioned in the NADP binding domain (PDB:1t0l)92. (B) Fraction labeling of citrate and aspartate via oxidative (blue) and reductive (red) TCA cycle over 0, 60, or 120min in A549 sgIDH1 CRISPRi-rescue cell lines (N=4). Knockdown-rescue was achieved by adding 500ng/mL of Dox, replenished daily, to culture media for a total of 96 hours. Total ion counts were normalized to Norvaline internal standard and total protein content, and fraction labeling was calculated by taking the ratio of each labeled species divided by the total pool size. (C) Pool sizes of citrate and aspartate at 120min (N=4). Pool sizes were calculated by summing all species of each metabolite per condition. Data available in Table S8(A – C). (D) Model showing pY391 activating IDH1 to induce reductive carboxylation of aKG to citrate. (E) U-13C5-Glutamine stable isotope tracing scheme for 0, 12, or 24 hours to measure changes in de novo pyrimidine synthesis pathway in A549 sgUMPS CRISPRi-rescue cell lines treated with dox for 96 hours total. (F) Total ion counts of M+4 labeled and total de novo pyrimidine synthesis intermediates carbamoyl aspartate, dihydroorotate, and orotate (N=4). (G) Extracellular orotate and dihydroorotate were measured from media collected at the 24 hours timepoint. Data available in Table S8(D – G). (H) model for pY37-mediated inhibition of UMPS enzyme activity.
Supplier Page from DNASU for UMPS (Homo sapiens) in pDONR221 (Gateway donor/master vector)