Structural basis for ligand binding modes of CTP synthase

Abstract

Cytidine triphosphate synthase (CTPS), which comprises an ammonia ligase domain and a glutamine amidotransferase domain, catalyzes the final step of de novo CTP biosynthesis. The activity of CTPS is regulated by the binding of four nucleotides and glutamine. While glutamine serves as an ammonia donor for the ATP-dependent conversion of UTP to CTP, the fourth nucleotide GTP acts as an allosteric activator. Models have been proposed to explain the mechanisms of action at the active site of the ammonia ligase domain and the conformational changes derived by GTP binding. However, actual GTP/ATP/UTP binding modes and relevant conformational changes have not been revealed fully. Here, we report the discovery of binding modes of four nucleotides and a glutamine analog 6-diazo-5-oxo-L-norleucine in Drosophila CTPS by cryo-electron microscopy with near-atomic resolution. Interactions between GTP and surrounding residues indicate that GTP acts to coordinate reactions at both domains by directly blocking ammonia leakage and stabilizing the ammonia tunnel. Additionally, we observe the ATP-dependent UTP phosphorylation intermediate and determine interacting residues at the ammonia ligase. A noncanonical CTP binding at the ATP binding site suggests another layer of feedback inhibition. Our findings not only delineate the structure of CTPS in the presence of all substrates but also complete our understanding of the underlying mechanisms of the allosteric regulation and CTP synthesis.

Keywords: CTP synthase; allosteric regulation; cryo–electron microscopy; cytoophidium.

Fig. 1.

Overall structures of substrate-bound and product-bound Drosophila CTPS tetramer. (A and B) Cryo-EM reconstructions of substrate-bound (A, 2.48-Å resolution) and product-bound (B, 2.65-Å resolution) dmCTPS tetramers. (C and D) The models and maps of the tetramer–tetramer interaction interface of substrate-bound (C) and product-bound (D) dmCTPS polymers. (E and F) Models of substrate-bound (E) and product-bound (F) dmCTPS tetramers. (G) Models of dmCTPS monomers display binding site of each ligand in substrate-bound and product-bound states. Residues^556-627 are disordered in the models.

Fig. 2.

Conformational changes of Phe373 reveals its role in regulating the binding of glutamine and GTP. (A) The structural comparison of the glutamine binding sites in glutamine-bound E. coli. CTPS (5TKV, pink), Thermus thermophilus HB8 CTPS (1VCO, gray), and dmCTPS^+Sub (green) model. (B) DON is located at the glutamine binding site at the GAT domain. The electron density map demonstrates the covalent bond between DON and Cys399 (arrow). (C) The structural comparison of the Phe373 in dmCTPS^+Sub (green) and dmCTPS^+Pro (gray) models. The Phe373 is ordered and more conductive to stacking with guanine in the dmCTPS^+Sub state. (D) B factors are shown on the putative glutamine entrance in dmCTPS^+Sub and dmCTPS^+Pro models. (E) The Phe373 (yellow) is located at the putative entrance (arrow) to the glutamine binding site, which is adjacent to the GTP binding site. In A, C, and D, we aligned the GAT domain (297 to 556) for comparison.

Fig. 3.

Determination of GTP binding site and relevant conformational changes reveal the mechanism of allosteric regulation. (A) The GTP binding site at GAT domain. Residues interacting with GTP are indicated. (B) The structure comparison of the GTP binding sites in dmCTPS^+Sub (green) and dmCTPS^+Pro (gray) models shows the general conformational change of the GAT domain and the switch of the wing region. Phe373 is shown in yellow and orange in dmCTPS^+Sub and dmCTPS^+Pro models, respectively. The wing structure is indicated by arrows. The AL domain (1 to 280) was used for the alignment. (C) Analysis for generation of CTP by wild-type dmCTPS (dmCTPS^WT) in conditions with 0, 0.2, and 2 mM GTP. The absorption of 291 nm (Abs_291nm) represents the concentration of CTP in samples. For the analysis, 2 μM dmCTPS was mixed with 2 mM UTP, 2mM ATP, 10 mM MgCl₂, 25 mM Tris HCl (pH 7.5), and GTP at different concentrations before the supplement of 10 mM glutamine for initiating the reaction. (Error bars, SD.) (D) B factors are shown on GTP binding sites in dmCTPS^+Sub and dmCTPS^+Pro models. Arrows indicate the wing structure. (E) Analysis for generation of CTP by mutant dmCTPS (F50A and L444A) in condition with 0.2 mM GTP.

Fig. 4.

Conformational changes of ammonia tunnel reveals its open states in both substrate-bound and product-bound dmCTPS. (A) Protein surface of the GTP binding cleft (selected by red dashed line) between the GAT domain and AL domain. An opening (yellow arrow) connecting the active site of the GAT domain and the ammonia tunnel would be fully covered by GTP. (B) Overview of the ammonia tunnel in substrate- and product-bound dmCTPS. The estimated tunnel is shown in magenta. (C and D) Models showing the ammonia tunnel and surrounding residues in (C) substrate- and (D) product-bound states. The tunnel is displayed by serial white spheres. The size of each sphere represents the estimated space in the tunnel at each point. The distance (dashed lines) is shown between the His55 side chain and uracil O4 and cytosine NH₂. (E) Graph of the diameter of the ammonia tunnel along its length in substrate- and product-bound dmCTPS. (F) The structural comparison of the ammonia tunnel structures in dmCTPS^+Sub and dmCTPS^+Pro models. The estimated routes of the tunnel in dmCTPS^+Sub and dmCTPS^+Pro models are shown in cyan and magenta, respectively. The region of ammonia tunnel (50 to 58) was used for superposition. (G) Electron density map shows that the putative “gate” His55 does not enclose the ammonia tunnel in the dmCTPS^+Pro state. Estimated interior space of ammonia tunnel is displayed in magenta. (H) B factors are shown on ammonia tunnels (arrows) in dmCTPS^+Sub and dmCTPS^+Pro models. (I) The interactions of Phe50 with Tyr42 and His55 may contribute to the stabilization of the ammonia tunnel. The distance between aromatic rings is shown.

Fig. 5.

Ligands binding at AL domain reveal mechanisms of UTP phosphorylation and feedback inhibition. (A) Electron density map showing the mechanism of UTP phosphorylation. (B) Analysis for generation of CTP by K16A and K38A mutant dmCTPS in condition with 0.2 mM GTP. The absorption of 291 nm (Abs_291nm) represents the concentration of CTP in samples. For the analysis, 2 μM dmCTPS was mixed with 2 mM UTP, 2 mM ATP, 10 mM MgCl₂, 25 mM Tris HCl (pH 7.5), and 0.2 mM GTP. Glutamine (10 mM) is supplied for initiating the reaction. (Error bars, SD.) (C and D) The ATP (C) and UTP (D) binding site at AL domain. ATP and UTP have become ADP and phosphorylated UTP in our dmCTPS^+Sub model. Residues interacting with ADP/UTP are indicated. (E and F) The competitive binding of CTP with ATP (E) and UTP (F). Water molecules are shown as red spheres in all panels.

Fig. 6.

Model of conformational changes of dmCTPS mediating glutamine-dependent CTP synthesis. (A) The binding of ATP/UTP results in an open-to-close state transition. The cleft between GAT domain and AL domain contracts to form the GTP binding site. (B) The γ-phosphate of ATP is transferred to the 4-oxygen atom of the uracil base. (C) Glutamine enters the binding site and stabilizes Phe373. (D) GTP enters the binding site, and the binding is further stabilized by Phe373 and also Leu444 on the wing structure. (E) GTP covers the gap between the active site of the GAT domain and stabilizes the ammonia tunnel. (F) Nascent NH₃ is transported to the AL domain through the tunnel and interacts with the iminophosphate intermediate to form CTP. (G) Ligands detach from binding sites, and CTPS returns to the open state. (H) CTP binds CTPS to competitively inhibit the binding of ATP and UTP.