Cryo-EM structure of the ATP11C Q79E mutant reveals the structural basis for altered Phospholipid recognition

Abstract

Closely related P4-ATPases, ATP11A and ATP11C, act as major phospholipid flippases in the plasma membrane of mammalian cells, with strict substrate specificity for phosphatidylserine (PS) and phosphatidylethanolamine (PE), but not for phosphatidylcholine (PC), thereby contributing to the asymmetric distribution of PS and PE across bilayers. A previously reported disease-associated Q84E mutation in ATP11A confers the ability to flip PC, implicating the involvement of this conserved residue in substrate specificity. We performed cryo-EM analysis for the equivalent mutant Q79E of ATP11C to address the structural basis for its unusual substrate specificity. Measurement of ATPase activity revealed that the ATP11C Q79E mutant retained PS-dependent activity, whilst gaining robust PC-dependent activity, indicative of expanded substrate specificity, consistent with reported properties in ATP11A Q84E. The cryo-EM structure of ATP11C Q79E mutant in the PC-occluded E2-P_i state revealed a PC molecule in a reshaped binding pocket. Due to the Q79E mutation and associated conformational changes in its surrounding residues, including Ser91and Asn352, the binding pocket has additional space to accommodate the bulky choline headgroup. Our results provide structural and functional insights into how a single point mutation can alter substrate specificity in a P4-ATPase.

Keywords: P-type ATPase; P4-ATPase; cryo-electron microscopy; flippase; membrane protein; phospholipid; transporter.

Figure 1

Reaction scheme and phospholipid (PL)-dependent ATPase activity of ATP11C.A, the cycle consists of four principal conformational states: E1 (not show in the figure), E1P, E2P, and E2, where “P” denotes a covalently bound phosphate group. E2P reflects the phosphorylated state prior to substrate binding, while E2-P_i captures the enzyme conformation after substrate binding but just before phosphate release. The models in surface representation show the outward-open E2P state (brown box, left) and PL-occluded (PL)E2-P_i state (blue box, right). Phospholipids are shown as green spheres. The red arrowhead indicates the outward gate exposed to the exoplasmic leaflet in E2P state, while it is closed in the PL-occluded E2-P_i state (black arrowhead). B, sSpecific ATPase activity of ATP11C WT and Q79E mutant measured in the absence of phospholipid (W/O, grey) and presence of 100 μM DOPS (PS, red) or DOPC (PC, blue) added to the sample with 10 or 500 column volume (CV) wash. The value obtained from the sample with 1 mM BeF₃^- serves as a blank. Data plotted are mean ± S.D. from three independent experiments. Asterisks indicate the p values between the indicated data sets are ≦ 0.05.

Figure 2

Cryo-EM structure of ATP11C Q79E in E2P state.A, the atomic model (left) and cryo-EM map (middle) of ATP11C Q79E E2P PC-bound state. Bound PC in the TM region is shown as green sticks, and its close-up view with a density map (only showing within 4 Å from the PC model in mesh representation) is shown on the right. A-, P-, N-domains and DGES motif are indicated in the figure. BeF₃^- are shown as cyan spheres. B–D, close-up view of phospholipid binding site of E2P state for WT (B) and Q79E mutant (C) and their comparison (D), viewed parallel to the membrane plane, from above the exoplasmic side. Cryo-EM density maps (transparent surface) and fitted atomic models of PS-bound ATP11C WT (B, PDB: 7BSU) and PC-bound ATP11C Q79E (C) are shown. Yellow dotted lines in B and C indicate probable hydrogen bond or salt bridge (with distance less than 4 Å) involved in phospholipid coordination. The arrow in D indicates the different rotamer conformation between WT and Q79E. E and F, cross-sectional surface representations of the lipid-binding pocket for WT (E) and Q79E (F). Phospholipids (sticks) are shown with their van der Waals surfaces displayed, for PS (orange) and PC (green), from the viewpoint where TM2 is located. Red arrowheads in both E and F show the unoccupied part of the binding pocket, where the pocket doesn’t tightly constrain the phospholipid headgroup. A small inset in E shows the viewing angle used for the main snapshot, as context for the orientation of the binding pocket shown in panels E and F.

Figure 3

Cryo-EM structure of ATP11C Q79E in E2-P_i state.A and B, the atomic model (left) and cryo-EM map (middle) of ATP11C Q79E E2-P_i state with occluded PC (A) or PS (B). PC and PS in the TM region are shown as green and orange sticks, respectively, and their close-up views are as shown in Figure 2A. AlF₄^- is shown as cyan spheres. C–H, c-up view of phospholipid binding site of E2-P_i state for WT (C, PDB: 7BSV), Q79E mutant with PC (D) and Q79E with PS (F) and their comparison (E, G, and H) as shown in Figure 2. Residues in transparent sphere representations are located within 3.5 Å from the occluded phospholipids, likely involved in van der Waals contact. Comparison of TM helices between ATP11C WT with PS (gray), ATP11C Q79E with PC (blue) and PS (pink), is viewed from the exoplasmic side for (H), while close-up views from parallel to the membrane plane are shown for PS-WT and PC-Q79E (E), and PS-WT and PS-Q79E (G). Arrows indicate different rotamer conformation between WT and Q79E.

Figure 4

Phospholipid-occlusion in the binding pocket of ATP11C. Cross-sectional surface representations of the lipid-binding pocket in ATP11C WT with PS (A) and Q79E with PS (B) or PC (C) in E2-P_i state. For comparison, only the PC molecule in Q79E mutant is superimposed on the WT structure in D. Phospholipids are shown as sticks with their van der Waals radii in mesh representation. Black arrowheads in C indicate residues with altered rotamer conformation due to the Q79E mutation. Red arrowheads indicate expected steric clash between PC and the WT binding pocket.

Figure 5

Cryo-EM structure of ATP11C Q79A in E2-P_i state.A, specific ATPase activity of ATP11C Q79A mutant sample purified with 500 CV as in Figure 1B. Data plotted are mean ± S.D. from three independent experiments. B, the atomic model (left) and cryo-EM map (right) of ATP11C Q79A PS-occluded E2-P_i state as in Figure 2. C, cross-sectional surface representations of the lipid-binding pocket in ATP11C Q79A as in Figure 4. D, the PC molecule observed in the structure of Q79E in PC-occluded E2-P_i state was superimposed onto the Q79A lipid-binding pocket as in Figure 4C.

Figure 6

Structural comparison with PL-occluded ATP8B1.A–C, comparison of the phospholipid binding models in PS-occluded ATP8B1(8OXA, yellow), PC-occluded ATP8B1 (8OXB, salmon) and PS-occluded ATP11C WT (7BSV, gray) as indicated in figures. Only amino acids close to the phospholipid binding site are shown as sticks. Amino acids and their numbering are indicated according to ATP8B1, and their corresponding residues in ATP11C WT are displayed in parentheses. PS from ATP8B1 is colored pink, while PS from ATP11C WT is colored orange. Dashed circles in C indicate residues with different rotamer conformations between PS- and PC-occluded forms of ATP8B1.

Figure 7

A model of PC recognition in ATP11C Q79E. In the ATP11C WT (top), canonical substrate PS is recognized by hydrogen bonds in its occluded form, and the binding pocket fits well to the head group of PS, which is too narrow to accommodate PC. In contrast, given the different rotamer conformation of Q79E (red arrow), the binding pocket in Q79E (bottom) is widened due to subsequently altered rotamer conformations of Ser91 and Asn352 (black arrows), thus allowing PC occlusion.