STRUCTURALMECHANISMOF TRANSPORT

Given the abundant structural and biochemical evidence for conformational changes, we are now faced with describing how these changes couple the local effects of calcium on the transmembrane sites to phosphorylation of D351 some 40˚ A away. For some time, the M4/S4 connection between E309 and D351 has been discussed as most likely to mediate cross-talk between these sites (45). The X-ray structures show us that the links between the P domain and both M4 and M5 are indeed well structured and potentially capable of conveying long-range conformational changes. Nevertheless, the actual sequence of events between binding of calcium and domain movement is still a matter for educated guesswork, guided by the conformational criteria in Table 1. In the starting conformation, calcium ligands are protonated (E2 ¢H3) and a distinct step of deprotonation is required prior to binding cytoplasmic calcium. This step must involve opening the partially occluded proton sites seen in the E2 ¢ TG structure to the cytoplasm, e.g., by reorientation of E309, and has been followed by kinetic studies of NBD fluorescence (97). This probe labels the P1 helix wedged between the top of M4 and the P domain, and the fluorescence changes indicate that there must be some structural change in this region. However, the T2 tryptic cleavage site in the A domain is still protected at pH 7, even in the presence of nucleotide, indicating that the A domain is still docked with the P domain in this deprotonated state (20, 43). Thus, a high pH, calcium-free conformation, which has typically been called E1, appears to be intermediate between E2 ¢H3 and E1 ¢ Ca2 with cytoplasmically exposed calcium sites but with the A domain still in the E2 conformation. Its properties may prove important in defining an access pathway for calcium, which has been problematic from the two existing X-ray structures. In contrast to NBD fluorescence, increases in tryptophan fluorescence (12) and small changes in circular dichroism (31) occur only after binding of calcium to E1 and are probably correlated with the cooperative changes required to create the second calciumsite,whose occupation activates phosphorylation. Froma structural point of view, the X-ray structures reveal an unbending ofM5 and displacement of M4,which seemto represent levers for rotating the P domain, but themechanismis not obvious. Locally, M6 undergoes the largest structural rearrangement with the backbone winding up upon calcium binding (Figure 3). Perhaps this winding puts strain on the L67 loop, which then induces the bending ofM5. Binding of calcium by M4 might also induce its vertical movement as the main chain carbonyls in its unwound portion move upward toward D800 to provide ligands for cooperative binding of the second calcium ion. In subsequent steps of the reaction cycle, the P domain is clearly the center of operations and is seen to adopt two distinct orientations in response to calcium binding. In the presence of calcium, the P domain appears to be available for phosphate transfer fromATP boundwithin theNdomain. In the absence of calcium, the P domain is seen to interact with the TGES184 loop of the A domain. Thus, it is possible that movement of the P domain represents a kind of switch, under control of calcium binding that selects between the N and the A domains. This simple pivot might represent the mechanism for calcium-induced activation of D351 for phosphorylation, which cannot otherwise be explained by rearrangement of residues within the catalytic site (Figure 4a). Although we do not yet have a structure of a phosphorylated intermediate,we postulate that consequent changes in the P domain would facilitate its interaction with the A domain and that formation of this A-P domain interface would represent the E1 »PtoE2-P transition that lowers calcium affinity and induces calcium release to the lumen. In contrast to the P domain, the N and A domains are covalently tethered by flexible, unstructured loops and their noncovalent interactions with other cytoplasmic domains appear to be transient. Although there is convincing evidence for mobility ofNandAdomains, there is little indication ofwhat causes their dramatic movements. A likely answer is thermal energy, or Brownian motion, which has been hypothesized as a driving force in the mechanisms of a wide variety of other macromolecular motors such as F0F1 (70), myosin (40), kinesin (6), as well as protein translocation into mitochondria (69) and the endoplasmic reticulum (58). Thermal energy would ensure that these weakly bound and flexibly tethered domains would be moving extensively, thus sampling a large range of orientations and potential binding interfaces. In the absence of tethers, these domains could be viewed as separate subunits as seen, for example, in the family of response reg- ulators, although the increased efficiency of a covalently attached domain should be advantageous for the continual turnover of ion pumps.The specific changes that accompany phosphorylation must be precipitated by local events near D351 and, by analogy with G protein switching (34), may well involve changes in the liganding of magnesium. Initially the magnesium is bound by two oxygens from the ¯ and ° phosphates of ATP and by four other ligands provided by the protein (including bound waters). Analogy with the X-ray structures of phosphoserine phosphatase (99) suggests that ligating oxygens of Ca2C-ATPase would come from the phosphate, the D351 carboxyl, the D703 carboxyl, the T353 main chain carbonyl, and two from bound water molecules (Figure 4).Homologous ligands are also found for severalmembers of theCheY response regulator family (53) and appear to be consistent in an unpublished structure for the Mg2F4 complex of the Ca2C-ATPase (C. Toyoshima & H. Nomura, unpublished data) andwithmutagenesis of theDGVND707 loop ofNaC/KC-ATPase (72). Initially, magnesium bound by Ca2C-ATPase is freely exchangeable (78), but after transfer of phosphate to D351 and loss of ADP, the magnesium becomes tightly bound (koff < 0.5 s¡1) and is released only after hydrolysis (98); similar results have been obtained with NaC/KC-ATPase (29). This represents a substantial difference from the phosphatases and response regulators, which bind magnesium loosely throughout their reaction cycles. We have previously suggested (90) that this occlusion of magnesium by Ca2C-ATPase could imply a change in its ligation, which could initiate further conformational changes required for calcium occlusion and the E1 »PtoE2-P transition. We can now specify that formation of the A-P domain interface may stabilize the magnesium ligand cage and that interactions at this interface might induce further conformational changes within the transmembrane domain.In the CheY family, a modest conformational change of the ®-4, ¯-4 loop(homologous to TGD627 of Ca2C-ATPase) follows formation of a new hydrogen bond to the covalently linked phosphate and induces interactions with responsive subunits (e.g., FliMandCheZ).Asimilar effectmay be occurring in theDGVND707 loop in Ca2C-ATPase, which is significantly farther from D351 than the analogous loop from response regulators and phosphatases (Figure 4). We hypothesize that the presence ofMg ¢ PO4 will pull this loop closer to provide the requisite ligands,thus producing a binding site for the TGES184 loop from the A domain, which would in turn confer tighter magnesium binding. In terms of the reaction cycle, phosphoenzyme formation and movement of the DGVND707 loop would produce the ADP-sensitive E1 »P intermediate and the docking of the A domain would initiate the transition to E2-P. The elements of this A-P domain interface include the conserved TGES184 And DGVND707 loops as well as the P6 helix and the MAATEQ244 loop connectingthe A domain to M3. The P6 helix directly follows DGVND707and contains K712,whichmakes hydrogen bondswithM239,T242, andQ244 in theE2 ¢ TGstructure. Significantly, thisMAATEQ244 region contains proteolytic cleavage sites that are only accessible in the E1 conformations (42, 50), consistent with its burial at a domain interface in E2. Both mutagenesis of TGES184 (2) and excision of the MAATE243 sequence with proteinase K (65) yield enzymes defective in the E1 »PtoE2-P transition. Results of Fe-catalyzed cleavage of NaC/KC-ATPase place TGES184 near the magnesium sites in E2-P (71). Taken together, these results indicate that rotation of the A domain and formation of the A-P domain interface are essential for producing E2-P. It is tempting to speculate that TGES184 contributes ligands to the aspartyl-PO4 ¢Mg complex, similar to Q147 of CheZ, which ligates magnesium at the active site of CheY and ultimately directs a water molecule to hydrolyze the aspartylphosphate (108). Although the original analyses of TGES184 mutations showed no change in phosphorylation levels from either ATP or Pi (4), more thorough studies of magnesium binding might now be possible using larger-scale expression systems. Both the formation of E2-P and its subsequent hydrolysis are linked to events at the calcium transport sites. The corresponding 90± 110±rotation of the A domain is likely to place stress on M1, M2, and M3 thereby accounting for their large movements in E2 ¢VO4 and E2 ¢ TG structures. These movements may represent an indirect mechanism for altering the calcium binding properties, by opening the lumenal gate in the M3/M4 loop and by perturbing the ion binding sites between M4, M5, M6, and M8. The dramatic changes in E58 probably reflect this perturbation, as this residue is hydrogen bonded to E309 in E1 ¢ Ca2 but is pulled completely out of the site in E2 ¢ TG owing to the kinking of M1 (Figures 2c, 3), perhaps destabilizing E309 as a first step in lowering calcium affinity. Release of the calcium ions and/or protonation of the transport sites stimulate hydrolysis of E2-P, and in analogy with the postulated effects of calcium binding, events at the ion binding sites are likely to be propagated through the M4/M5 helices and the L67 loop, perhaps inducing changes in the A-P domain interface similar to those seen in the structure of E2 ¢ TG relative to E2 ¢VO4 in order to stimulate hydrolysis.

Date: 2016-01-03; view: 688