Second Transmembrane Helix (M2) and Long Range Coupling in Ca²⁺-ATPase

Background: The catalytic A domain of Ca²⁺-ATPase moves substantially and connects to distant Ca²⁺ sites through transmembrane helices M1 and M2.

Results: Systematic mutation along M2 profoundly and differentially affects cytoplasmic and luminal gating and catalysis.

Conclusion: M2 plays region- and catalytic step-specific roles in Ca²⁺ transport.

Significance: M2 is a conduit for coupling A domain movements to transport ion gating.

Abstract

The actuator (A) domain of sarco(endo)plasmic reticulum Ca²⁺-ATPase not only plays a catalytic role but also undergoes large rotational movements that influence the distant transport sites through connections with transmembrane helices M1 and M2. Here we explore the importance of long helix M2 and its junction with the A domain by disrupting the helix structure and elongating with insertions of five glycine residues. Insertions into the membrane region of M2 and the top junctional segment impair Ca²⁺ transport despite reasonable ATPase activity, indicating that they are uncoupled. These mutants fail to occlude Ca²⁺. Those at the top segment also exhibited accelerated phosphoenzyme isomerization E1P → E2P. Insertions into the middle of M2 markedly accelerate E2P hydrolysis and cause strong resistance to inhibition by luminal Ca²⁺. Insertions along almost the entire M2 region inhibit the dephosphorylated enzyme transition E2 → E1. The results pinpoint which parts of M2 control cytoplasm gating and which are critical for luminal gating at each stage in the transport cycle and suggest that proper gate function requires appropriate interactions, tension, and/or rigidity in the M2 region at appropriate times for coupling with A domain movements and catalysis.

Introduction

Sarco(endo)plasmic reticulum Ca²⁺-ATPase (SERCA1a),² a representative member of the P-type ion transporting ATPases, catalyzes Ca²⁺ transport coupled with ATP hydrolysis (Fig. 1A) (for recent reviews, see Refs. 1–3). The enzyme is activated by the binding of two cytoplasmic Ca²⁺ ions at the cytoplasmic facing high affinity transport sites (E2 to E1Ca₂ in Fig. 1) and autophosphorylated at Asp³⁵¹ with MgATP to form an ADP-sensitive phosphoenzyme (E1P), which reacts with ADP to regenerate ATP in the reverse reaction. Upon E1P formation, the two bound Ca²⁺ are occluded in the transport sites (E1PCa₂). The subsequent isomeric transition to the ADP-insensitive E2P form results in rearrangements of the Ca²⁺ binding sites to deocclude Ca²⁺, open the release path, and reduce the affinity, thus releasing Ca²⁺ into the lumen. Finally, the Asp³⁵¹-acylphosphate in E2P is hydrolyzed to form a Ca²⁺-unbound inactive E2 state. During the hydrolysis, the Ca²⁺ release path is closed, thereby preventing possible luminal Ca²⁺ access to the transport sites and Ca²⁺ leakage.

FIGURE 1.

Reaction cycle and structural changes of Ca²⁺-ATPase. A, the structural changes are modeled on the crystal structures E1Ca₂·AlF₄⁻·ADP as the E1∼P·ADP·Ca₂ analog (PDB code 1T5T (10)), E2·BeF₃⁻ as the E2P ground state analog (7) (PDB code 2ZBE (11)), E2·AlF₄⁻(TG) as the transition state analog for E2P hydrolysis (7) (PDB 2ZBG (11)), E2(TG) as the E2 state fixed with thapsigargin (PDB 1IWO (6)), E1Mg²⁺ (PDB 3W5A (14)), and E1Ca₂ (PDB 1SU4 (4)). The structures are aligned with the static M8–M10 helices. The N, P, and A domains and M1–M6 helices are colored as indicated. The approximate position of the membrane is shown by gray lines. The binding sites for two Ca²⁺ (purple spheres) consist of residues on M4, M5, M6, and M8. The yellow, pink, and red arrows indicate the approximate motions of the A and P domains and M2 (depicted in red), respectively, to the next structural state during the EP processing (isomerization and hydrolysis) and the E2 → E1 transition. The broken light blue arrow indicates the unwound M2 part. B, the regions of M2 are named as M2m (transmembrane part), M2c (cytoplasmic part), M2top (top part), and A/M2-junction (junction with the A domain) and indicated with different colors on the α-carbon of residues.

In the transport cycle, the three cytoplasmic domains N, P, and A undergo large movements and change their organizational state, a repositioning that is coupled to rearrangements in transmembrane helices and thereby changes in the transport sites (1–15). Most remarkable is the motion of the A domain, which functions in cytoplasmic and luminal gating to regulate Ca²⁺ binding and release as well as the E2P hydrolysis. The critical importance of the M1/M2 segment, which forms a V-shaped rigid body connected with the A domain, for luminal gating is nicely demonstrated by the crystal structures of catalytic intermediates with bound substrate analogs (9, 11). The long helix M2 connects directly with the A domain at their junction (A/M2-junction) and moves largely together with the A domain and also changes its secondary structure, unwinding/rewinding with consequent length changes, during the Ca²⁺ transport cycle (Fig. 1A). Notably, a long helix structure of M2 and the A domain motions are common features in P-type ion-transporting ATPases (16–20). With the Ca²⁺-ATPase, we previously demonstrated (21, 22) that Tyr¹²² and Leu¹¹⁹ at the A/M2-junction and the top part of M2 (M2top) form a hydrophobic interaction network, which includes a Tyr¹²²-hydrophobic cluster (an interaction with the P domain (Val⁷⁰⁶/Val⁷²⁶), with the loop connecting the A domain and M3 (A/M3-linker, Ile²³²), and with the A domain (Ile¹⁷⁹/Leu¹⁸⁰) in E2P) and that this formation is critical for stabilizing E2P structure with the luminal gate open and the potential for hydrolytic activity at the catalytic site. However, the importance of other M2 regions as well as of the long M2 helix structure itself has not yet been fully explored.

In this study, we focus on each region of the long helix M2: transmembrane M2m, cytoplasmic M2c, M2top, and A/M2-junction (Fig. 1B). We explored their roles and the functional significance of the changes in secondary structure and length for the coupling of the different conformational steps that are required to efficiently convert the chemical energy of ATP hydrolysis into the changes in accessibility, orientation, and affinity of the Ca²⁺ binding transport sites.

In extensive preliminary experiments, we first introduced a series of mutations at various positions throughout: insertions of 1–5 glycine residues, deletions of 1–4 successive residues, glycine substitutions of 2 or 3 successive residues, and some specific substitutions. In detailed kinetic analyses, we found that results were most clear with the insertions of five glycine residues (5Gis), which disrupt the helix structure and elongates, showing profound region-specific and catalytic step-specific effects. Therefore, we present here the results obtained for the 5Gi insertions. Our results demonstrate that different parts of the A/M2 link play a critical role in synchronizing gating of the two Ca²⁺ at the transport sites on both sides of the membrane, and each is therefore part of the mechanism for coupling catalytic and transport site structural events. M2 and its connection with the A domain have a role distinct from that of the A/M1′-linker loop (23–25), although both are needed to coordinate coupling.

EXPERIMENTAL PROCEDURES

Mutagenesis and Expression

The pMT2 expression vector (26) carrying rabbit SERCA1a cDNA with a desired mutation was constructed as described previously (24). Transfection of pMT2 DNA into COS-1 cells and preparation of microsomes from the cells were performed as described (27). The amount of expressed SERCA1a was quantified by a sandwich enzyme-linked immunosorbent assay (28).

Ca²⁺-ATPase Activity and Ca²⁺ Transport Activity

Activities of expressed SERCA1a were obtained essentially as described previously (29). The rate of ATP hydrolysis was determined at 25 °C in a mixture containing 1–5 μg of microsomal protein, 0.1 mm [γ-³²P]ATP, 0.1 m KCl, 7 mm MgCl₂, 0.1 mm CaCl₂, 5 mm potassium oxalate, and 50 mm MOPS/Tris (pH 7.0). The Ca²⁺-ATPase activity of expressed SERCA1a was obtained by subtracting the ATPase activity determined in the presence of 1 μm thapsigargin (TG), a specific inhibitor of SERCA, with conditions otherwise as above. The rate of Ca²⁺ transport was determined with ⁴⁵Ca²⁺ and nonradioactive ATP, otherwise as above. The Ca²⁺ transport activity of expressed SERCA1a was obtained by subtracting the activity determined in the presence of 1 μm TG, with conditions otherwise as above.

Formation and Hydrolysis of EP

Phosphorylation of SERCA1a in microsomes with [γ-³²P]ATP or ³²P_i and dephosphorylation of ³²P-labeled SERCA1a were performed under conditions described in the legends to Figs. 4–9. The reaction was quenched with ice-cold trichloroacetic acid containing P_i. Precipitated proteins were separated by 5% SDS-PAGE at pH 6.0 according to Weber and Osborn (30). The radioactivity associated with the separated Ca²⁺-ATPase was quantified by digital autoradiography as described (31). The amount of EP in expressed SERCA1a was obtained by subtracting the background radioactivity determined in the presence of 1 μm TG, with conditions otherwise as above.

FIGURE 4.

Total amount of EP (EP_total) at steady state and E2P fraction. A, microsomes expressing wild type or the 5Gi mutant were phosphorylated with [γ-³²P]ATP at 0 °C for 1 min in 50 μl of a mixture containing 1–5 μg of microsomal protein, 10 μm [γ-³²P]ATP, 1 μm A23187, 0.1 m KCl, 7 mm MgCl₂, 0.1 mm CaCl₂, and 50 mm MOPS/Tris (pH 7.0). The EP_total formed (gray bars) was determined by acid quenching. For determination of E2P (closed bars), an equal volume of a mixture containing 2 mm ADP, 1 μm A23187, 0.1 m KCl, 7 mm MgCl₂, 10 mm EGTA, and 50 mm MOPS/Tris (pH 7.0) was added to the above phosphorylation mixture, and the reaction was quenched at 1 s after the ADP addition. ADP-sensitive EP (E1P) disappeared entirely within 1 s after the ADP addition. B, the E2P fraction in EP_total (E1P plus E2P) is shown. Statistical significance compared with the wild type is shown for EP_total (A) and for E2P/EP_total (B): *, p < 0.05; ‡, p < 0.01. C, the mutation effects of residues in B are visualized with α-carbon coloring: green, E2P/EP_total less than 30% (almost no or slight E2P increase); light blue, 30–50% (moderate increase); purple, higher than 50% (marked increase). Error bars, S.D.

FIGURE 5.

EP isomerization rate. A, for the wild type and mutants that accumulate mostly E1P at steady state (E2P less than 30% of EP_total; cf. Fig. 4), the E1P to E2P isomerization rate was determined by E1P decay kinetics, which represent the rate-limiting E1P to E2P isomerization (mutant 5Gi-114/115 (right inset) is a typical example). Here, the Ca²⁺-ATPase was first phosphorylated with [γ-³²P]ATP at 0 °C for 1 min as in Fig. 4; phosphorylation was terminated by Ca²⁺ removal and the addition of an equal volume of a buffer containing 10 mm EGTA, 0.1 m KCl, 7 mm MgCl₂, and 50 mm MOPS/Tris (pH 7.0) at 0 °C; and the amounts of EP_total and E2P were determined at the indicated times. The amount of E1P was calculated by subtracting E2P from EP_total (EP_total − E2P). Solid lines show the least squares fit to a single exponential, and the E1P decay rates thus obtained are shown in the main panel (light gray bar). For the mutants that accumulate E2P more than 30% of EP_total (cf. Fig. 4), the EP isomerization rate was determined as the apparent rate of E2P formation from E1P to reach steady state as follows (mutant 5Gi-121/122 (left inset) is a typical example). Here, the Ca²⁺-ATPase was phosphorylated for the indicated periods after ATP addition, and the amounts of EP_total and E2P were determined; otherwise, conditions were as described above. The formation of EP (EP_total, representing E1PCa₂ formation from E1Ca₂) was very fast (reaching a steady state within ∼1 s) and was followed by E2P formation from E1PCa₂. Solid lines show the least squares fit to a single exponential, and the apparent rates of E2P formation are shown in the main panel (dark gray bar). Statistical significance compared with the wild type is shown: *, p < 0.05; ‡, p < 0.01. B, the mutation effects of residues in A are visualized with α-carbon coloring: green, transition rate less than 0.15 s⁻¹ (almost no effect or only a slight effect); light blue, 0.15–0.3 s⁻¹ (moderate acceleration); purple, higher than 0.3 s⁻¹ (marked acceleration). Error bars, S.D.

FIGURE 6.

Ca²⁺ occlusion in EP. A, phosphorylation was performed with ATP in 10 μm ⁴⁵Ca²⁺; otherwise conditions were as in Fig. 4. The amount of Ca²⁺ occluded was determined as described under “Experimental Procedures” and is shown as the value relative to EP_total (mean ± S.D. (n = 8–18)). Statistical significance compared with the wild type is shown: *, p < 0.05; ‡, p < 0.01. B, the mutation effects of residues in A are visualized with α-carbon coloring; green, Ca²⁺ occluded/EP_total above 1.0; red, below 0.15. Error bars, S.D.

FIGURE 7.

E2P hydrolysis and luminal Ca²⁺-induced inhibition. A, microsomes expressing wild type or mutant were phosphorylated with ³²P_i at 25 °C for 10 min in 5 μl of a mixture containing 1–5 μg of microsomal protein, 0.1 mm ³²P_i, 1 μm A23187, 0.2 mm EGTA, 7 mm MgCl₂, 50 mm MOPS/Tris (pH 7.0), and 35% (v/v) Me₂SO. The mixture was then cooled and diluted at 0 °C by the addition of 95 μl of a mixture containing 2.1 mm non-radioactive P_i, 105 mm KCl, 7 mm MgCl₂, 50 mm MOPS/Tris (pH 7.0), and 3 mm EGTA (open bar) or 3.2 mm (gray bar) and 21.1 mm (closed bar) CaCl₂ in place of EGTA, and E2P hydrolysis was followed (see typical examples with wild type and 5Gi-106/107 in the inset). The amounts of E2P formed with ³²P_i at zero time are normalized to 100%. Solid lines show the least squares fit to a single exponential, and the rates thus obtained are shown in the main panel. In the absence of Ca²⁺, statistical significance of the mutant rate compared with the wild-type rate is shown above an open bar: *, p < 0.05; ‡, p < 0.01. In each of the mutants and the wild type, statistical significance of the rate in 3 and 20 mm Ca²⁺ compared with that in the absence of Ca²⁺ is shown above a gray bar and closed bar: V, p < 0.05; ↓, p < 0.01. B, the marked retardation and acceleration in the absence of Ca²⁺ (open bars in A) are visualized with α-carbon coloring, red and purple, respectively. Error bars, S.D.

FIGURE 8.

EP formation from E2 and E1Ca₂ states. A, microsomes expressing wild type or mutant were preincubated for 20 min at 25 °C in 50 μl of a mixture containing 1–5 μg of microsomal protein, 1 μm A23187, 0.1 m KCl, 7 mm MgCl₂, 50 mm MOPS/Tris (pH 7.0), and 1 mm EGTA with and without 1.2 mm CaCl₂ (to form the Ca²⁺-bound (open bar) and unbound (closed bar) states, respectively). After cooling, an equal volume of a phosphorylation mixture containing 10 μm [γ-³²P]ATP and 1 mm EGTA with 1.2 or 2.4 mm CaCl₂ (to give 0.2 mm Ca²⁺ for both Ca²⁺-bound and Ca²⁺-unbound states) (otherwise as above) was added at 0 °C, and the EP formation time course was followed. In the inset, typical examples with the wild type and mutant 5Gi-119/120 are shown. Solid lines show the least squares fit to a single exponential, and the rates thus determined are shown in the main panel. The Ca²⁺-unbound state was denoted as E2 for simplicity, and the ratio of the two rates is shown in B. In A and B, statistical significance compared with the respective wild-type value is shown: *, p < 0.05; ‡, p < 0.01. C, the observed effects on the rate-limiting process E2 → E1 in B are visualized with α-carbon coloring; green, no effect or acceleration (ratio higher than 35%); yellow, retardation (ratio 30–15%); red, marked retardation (ratio lower than 15%). Error bars, S.D.

FIGURE 9.

Ca²⁺ affinity determined with EP formation. The microsomes expressing wild type or mutant were preincubated for 20 min at 25 °C in 45 μl of a mixture containing 1–5 μg of microsomal protein, 2 mm EGTA, 1 μm A23187, 0.1 m KCl, 7 mm MgCl₂, 50 mm MOPS/Tris (pH 7.0), and various concentrations of CaCl₂ to give the desired Ca²⁺ concentrations. After cooling, the Ca²⁺-ATPase was phosphorylated with 10 μm [γ-³²P]ATP for 15 s at 0 °C, and the amount of EP formed was determined. In the inset, the typical Ca²⁺ dependence of EP formation was shown for the wild type and the two mutants including 5Gi-110/111, which gave the highest K_d value. The K_d value and Hill coefficient were estimated by least squares fit to the Hill equation (solid line). The Hill coefficient obtained was ∼2 (actually 1.6–2) in the wild type and all of the mutants. Statistical significance compared with the wild type is shown: *, p < 0.05; ‡, p < 0.01. Error bars, S.D.

Ca²⁺ Occlusion in EP

Microsomes were phosphorylated for 1 min at 0 °C in a mixture containing 1–5 μg of microsomal protein, 5 μm ATP, 0.1 m KCl, 7 mm MgCl₂, 10 μm ⁴⁵CaCl₂, 1 μm A23187, and 50 mm MOPS/Tris (pH 7.0) and immediately filtered through a 0.45-μm nitrocellulose membrane filter (Millipore). The filter was washed extensively with a washing solution (1 mm EGTA, 0.1 m KCl, 7 mm MgCl₂, and 20 mm MOPS/Tris (pH 7.0)), and ⁴⁵Ca²⁺ remaining on the filter was quantified as described (24). The amount of Ca²⁺ occluded at the transport sites of EP in the expressed SERCA1a was obtained by subtracting the amount of nonspecific Ca²⁺-binding determined in the presence of 1 μm TG, with conditions otherwise as above. The amount of EP formed was determined with nonradioactive Ca²⁺ and [γ-³²P]ATP under otherwise the same conditions as above by membrane filtration, and the radioactivity remaining on the filter was quantified.

Miscellaneous

Protein concentration was determined by the method of Lowry et al. (32) with bovine serum albumin as a standard. Data were analyzed by nonlinear regression using the program Origin (MicroCal Software, Inc., Northampton, MA). Three-dimensional models of the enzyme were reproduced by the program VMD (33). The data represent the mean ± S.D. for 2–6 independent experiments (or 8–18 experiments in Fig. 6). Statistical analysis was performed by one-way analysis of variance with Dunnett's post hoc test using SPSS software version 22.

RESULTS

Protein Expression Level

For understanding possible structural roles of each of the M2 regions and the functional significance of its helix structure and changes during the transport cycle (unwinding/elongation) (Fig. 1), we introduced 5Gis throughout M2 and performed functional analyses. We first determined the expression levels of the wild type and mutants in the microsomes prepared from COS-1 cells (Fig. 2). The expression level of wild-type SERCA1a in the microsomes is ∼2% of total microsomal protein. Those of the 5Gi mutants are comparable, although a few show a level that is somewhat reduced but still high enough to perform all functional analyses.

FIGURE 2.

Expression levels. The expression levels of wild-type and 5Gi mutant SERCA1a in the microsomes prepared from COS-1 cells were determined and shown as values relative to the total amount of protein in the microsomes. Statistical significance compared with the wild type is shown as follows: *, p < 0.05; ‡, p < 0.01. Error bars, S.D.

ATP Hydrolysis, Ca²⁺ Transport, and Their Coupling

First, we explored possible effects of 5Gi mutations on the overall transport cycle. Fig. 3A shows the specific Ca²⁺-ATPase and oxalate-dependent Ca²⁺ transport activities of SERCA1a mutants and wild type at 10 μm Ca²⁺ in the presence of 5 mm oxalate. Here, the activities were determined during the linear part of P_i liberation and Ca²⁺ transport (inset) and defined as the fraction that is sensitive to inhibition by 1 μm TG, a highly specific and subnanomolar affinity SERCA inhibitor (34). We first confirmed that the mutants all retain TG sensitivity by observing that TG (1 μm) completely inhibits EP formation from [γ-³²P]ATP in 10 μm Ca²⁺, conditions essentially the same as the activity measurements. We found that TG reduces the EP value in all cases to a background radioactivity level as in the wild type (i.e. less than 1% of the maximum EP level). The background radioactivity level is actually the same as that obtained in the absence of Ca²⁺ without TG.

FIGURE 3.

Ca²⁺-ATPase and oxalate-dependent Ca²⁺ transport activities. A, the specific activities of the expressed SERCA1a 5Gi mutants were determined and shown as values relative to the respective wild-type activities (ATP hydrolysis, 0.594 ± 0.028 μmol of P_i/min/mg of SERCA1a protein (n = 5); oxalate-dependent Ca²⁺ transport, 0.116 ± 0.006 μmol Ca²⁺/min/mg SERCA1a protein (n = 5); very similar to the values obtained by our group and other groups under optimum conditions with the microsomes prepared from the COS cells (e.g. see Refs. 29, 47–49)). Typical time courses of P_i liberation and Ca²⁺ accumulation in the wild type and mutant 5Gi-91/92 are shown in the inset. B, the coupling ratio (i.e. Ca²⁺ transport activity per Ca²⁺-ATPase activity (Ca²⁺/ATP)), is shown as a percentage of the wild-type ratio. In A and B, statistical significance compared with the respective wild-type value is shown; *, p < 0.05; ‡, p < 0.01. C, the mutational effects of residues in B are visualized with α-carbon coloring on M2; green, Ca²⁺/ATP higher than 80% of the wild type (coupled transport); yellow, 80 to 60% (slightly uncoupled); red, less than 60% (severely uncoupled). Error bars, S.D.

Most insertions reduce both activities (Fig. 3A). Importantly, some mutations at specific regions decrease Ca²⁺ transport/ATPase activity ratios (Ca²⁺/ATP, Fig. 3B). The insertions in M2m and those in the A/M2-junction cause severe uncoupling (i.e. almost no Ca²⁺ transport despite fair ATP hydrolysis). This contrasts with mutations at M2c, which result in normal coupling. At M2top (aa 116–119), uncoupling increases with proximity of the 5Gi insertions to the A/M2-junction.

EP Formation from ATP and E1P → E2P Isomerization

We then analyzed each of the steps and intermediates in the Ca²⁺ transport cycle. In Figs. 4–7, we assessed possible effects of 5Gi mutations on the properties of the phosphorylated intermediates E1P (occludes Ca²⁺ at the transport sites) and E2P (releases Ca²⁺ into the lumen). We first determined in Fig. 4 the total amount of EP (EP_total, sum of E1P and E2P) and the E2P fraction at steady state. The analysis was made in the presence of K⁺, which strongly accelerates hydrolysis of E2P and therefore suppresses its accumulation in the wild type (35). The 5Gi insertions at aa 103–115 in M2m to M2c markedly decrease EP_total (Fig. 4A). The decrease seems to correlate well with Ca²⁺ insensitivity of the EP hydrolysis rate in these mutants (see Fig. 7). It is also possible that a very rapid E2P hydrolysis and a very slow E2 → E1 transition in these mutants (cf. Figs. 7 and 8) may contribute to the decrease in EP_total.

Otherwise, 5Gi mutants in M2c are largely wild type-like; E2P accumulation, the E1P to E2P isomerization rate (Figs. 4B and 5), and coupled Ca²⁺ transport are unremarkable. At M2m, E2P accumulation is considerably higher in some 5Gi mutants (5Gi-91/92, 5Gi-92/93, 5Gi-97/98, and 5Gi-101/102 to 5Gi-104/105) (Fig. 4B); EP isomerization is slightly faster than in the wild type (Fig. 5); and coupling in Ca²⁺ transport is substantially reduced as noted above.

At the M2top and A/M2-junction (aa 116–119 and 120–126), almost all EP_total is E2P in the 5Gi mutants despite the presence of K⁺ (Fig. 4). In these mutants, EP isomerization is markedly accelerated (Fig. 5), and E2P hydrolysis is strongly retarded (as shown in Fig. 7), consistent with the dominant E2P accumulation. Ca²⁺ transport is substantially uncoupled from ATPase activity as noted above.

Ca²⁺ Occlusion in EP

In Fig. 6, the amount of ⁴⁵Ca²⁺ occluded in EP was determined after EP formation under otherwise the same conditions as in Fig. 4. In the wild type, approximately two Ca²⁺ ions were occluded in EP_total (comprising mostly E1P; Fig. 4B), in agreement with established mechanisms. 5Gi mutants at the luminal end of M2m (5Gi-87/88 and 5Gi-89/90) and those at the M2c region (aa 106–115) occluded approximately two Ca²⁺ in E1P (cf. Fig. 4B), which fits with their coupled Ca²⁺ transport although turnover is severely inhibited.

In contrast, in 5Gi mutants at M2m aa 91–105, all of which show severe uncoupling (cf. Fig. 3B), no Ca²⁺ is occluded despite reasonable E1P accumulation. Bound Ca²⁺ must escape from the E1P state, probably to the cytoplasmic side, to account for the uncoupling. For 5Gi mutants at M2top to the A/M2-junction region (aa 116–126), we also did not observe Ca²⁺ occlusion, but in this case, it is rather due to the dominant E2P accumulation and very rapid EP isomerization (Figs. 4 and 5). Nevertheless, in the uncoupled mutants, especially at the A/M2-junction, the Ca²⁺ must escape from E1P or during EP isomerization to the cytoplasmic side, showing the critical importance of this junctional region to keep the gate closed. The increased severity of uncoupling as the 5Gi insertions approach the A/M2-junction emphasizes its importance in coupling. In summary, 5Gi insertions in the central section of M2 have little effect on cytoplasmic gating, whereas those on either side block gate closure, although mutations at M2top (i.e. at the border region between M2c and A/M2-junction) exhibit mixed properties.

E2P Hydrolysis

Then we assessed possible effects of 5Gi mutations on the E2P hydrolysis rate and luminal Ca²⁺-induced back-inhibition of hydrolysis that reflect luminal gate closure during hydrolysis. The closure prevents luminal Ca²⁺ access to the transport sites and leakage. Here we determined the E2P hydrolysis rate directly by first phosphorylating the enzyme with ³²P_i in the absence of Ca²⁺ and K⁺ and the presence of 35% (v/v) Me₂SO, which strongly favors E2P formation in the reverse reaction (36), and then diluting the phosphorylated protein with a large volume of nonradioactive P_i and K⁺ without Ca²⁺ (open bars in Fig. 7). In regions M2m (aa 87–102), M2top (aa 116–119), and the A/M2-junction (aa 120–126), the 5Gi insertions strongly retard E2P hydrolysis. In contrast, at the top of M2m (aa 103–105) and the entire M2c region (aa 106–115), the insertions stimulate hydrolysis 3–10-fold.

Importantly, those mutants showing marked acceleration of E2P hydrolysis are insensitive to luminal Ca²⁺-induced back-inhibition. Here, the E2P hydrolysis rate was determined in the presence of 3 (gray bars) or 20 mm (black bars) Ca²⁺ and ionophore A23187. In the wild type, E2P hydrolysis is completely blocked by 3 mm Ca²⁺ due to Ca²⁺ binding to the open luminally oriented transport sites (open luminal gate) in E2P, in agreement with previous findings. Thus, accessibility to the Ca²⁺ sites from the luminal side in E2P, reflecting open gate status, can be assessed by the back-inhibition (22, 24, 37–42). All of the mutants with accelerated E2P hydrolysis are resistant against inhibition by Ca²⁺ even at 20 mm. The luminal gate is tightly closed, preventing access of Ca²⁺ to its binding site, or Ca²⁺ binding is possibly no longer coupled to a large change in the rate of hydrolysis. The results, either way, indicate the importance of the M2c region for coupling catalysis and luminal gate closure during E2P hydrolysis. Notable also, 5Gi mutants at aa 95–101 with retarded hydrolysis exhibit partial resistance to inhibition by luminal Ca²⁺ although less than those at aa 103–115.

E2 → E1 Transition

After E2P hydrolysis, the enzyme in the inactive E2 state is reactivated for the next transport cycle by two Ca²⁺ binding from the cytoplasmic side to the high affinity transport sites forming E1Ca₂. In this activation, the E2 state is isomerized first to a transient E1 state (ready to accept Ca²⁺) with an open cytoplasmic gate and a high Ca²⁺ affinity (E2 → E1 transition; Fig. 1). Here we measured possible effects of 5Gi mutations on the transition itself (Fig. 8) as well as the affinity of the Ca²⁺ binding sites (Fig. 9).

The rate of the transition can be quantified by measuring the rate of phosphoenzyme formation following the addition of Ca²⁺ plus ATP to Ca²⁺-deprived Ca²⁺-ATPase. The assay takes advantage of the fact that subsequent steps (Ca²⁺ binding, ATP binding, and phosphorylation) are relatively fast. The transition is pH-sensitive, and we have found that pH 7.0 is ideal for studying the effect of mutations because, even in wild-type, at lower pH, the rate is rather slow, and at higher values, it approaches that of the subsequent steps, notably the rate of phosphorylation of Ca²⁺-bound Ca²⁺-ATPase (data at pH 7.0 shown in Fig. 8; other data not shown).

First, the rate of phosphorylation of Ca²⁺-bound Ca²⁺-ATPase E1Ca₂ is only marginally affected by the mutations, although the change 5Gi-89/90 is not without consequences. The transition itself is markedly slowed by 5Gi insertions in most of M2 aa 94–119 (from the upper half of M2m to M2top).

5Gi insertions higher up at the A/M2-junction (aa 120–126) are less inhibitory, and effects diminish closer to the A domain. At the luminal end of M2, mutations are without effect, except for 5Gi-89/90 and 5Gi-87/88, which have unexpected and contrary effects. In general then, most mutations in M2 affect the transition.

Possible changes in Ca²⁺-binding properties were assessed by measuring the Ca²⁺ dependence of phosphorylation. Apparent K_d values increase somewhat (up to 5-fold) (Fig. 9) yet still remain in the high affinity (1–1.5 μm) range. The Hill coefficients hover around the usual 2. Evidently, M2 has little to do with cytoplasmic Ca²⁺ binding itself, consistent with the fact that the Ca²⁺ binding sites consist of M4, M5, M6, and M8 (Fig. 1).

DISCUSSION

We find that almost the entire M2 section plays a crucial role in coupling movements of the A domain and catalysis with gating at the transport sites. Although 5Gi insertions in border regions of the M2 sections mostly exhibit mixed kinetic properties, in general, two main regions can be distinguished: namely that in the middle (M2c), where the long M2 helix breaks during E2P hydrolysis, and that on either side (M2m, M2top, and A/M2-junction). Our results show that the former controls luminal gating, whereas the latter controls cytoplasm gating. In addition, insertions in part of M2m (aa 94–103) have mixed kinetic effects, probably reflecting its structural role in the M1/M2 V-shaped body and involvement in gating on both sides of the membrane with long range effects on the catalytic site (9, 11). The results are best interpreted with reference to Fig. 10.

FIGURE 10.

Detailed inspection of structure and interactions of M2 in crystal structures. The structural changes are modeled as in Fig. 1A. The N domain and M7–M10 helices are not depicted for simplicity. The red (A and B), ice blue (A), yellow (A and B), and pink (B) arrows indicate the approximate motions of M2, M1′, A domain, and P domain, respectively. The broken light blue arrow (A and B) indicates the unwound part of M2. The TGES¹⁸⁴ loop is depicted with orange van der Waals spheres. Interactions involving M2 regions are shown in transparent circles in enlarged views. See “Discussion” for the details and the roles of M2 regions.

Cytoplasmic Gating in E2 → E1

Opening of the cytoplasm gate takes place during the E2 → E1 transition. Our results show that 5Gi insertions along most of M2 (aa 94–119) markedly retard this transition, indicating the critical importance of the helix structure for a rapid transition. Consistently, the relevant crystal structures show that the unwound Asn¹¹¹–Asn¹¹⁵ part of M2c in E2 is rewound in E1Mg²⁺, a change associated with the ∼110° A domain rotation (14), and M2 becomes a contiguous helix in E1Mg²⁺, as in E1Ca₂. In the critical region aa 94–119, the lower half (aa 94–111) and the upper half (aa 111–119) change from broken and disengaged entities in E2 to making strong helix-helix contacts with M6 and M4C in E1, respectively (gray circle and yellow circle in panel b in Fig. 10A). These helix-helix interactions probably must stabilize the E1 structure, facilitating the E2 → E1 transition and opening of the cytoplasmic gate.

Cytoplasmic Gating in E1P and Its Isomerization

Once the two Ca²⁺ bind to the transport sites from the cytoplasmic side, closing of the gate to occlude the Ca²⁺ is effected by phosphorylation to E1P (9, 43, 44). Two sets of mutations, those in M2m and those in the M2top∼A/M2-junction, but not in the middle at M2c, uncouple the pump and fail to occlude Ca²⁺, indicating the importance of these regions in stabilizing the closed cytoplasmic gate.

The immediate gating residue is Glu³⁰⁹, and Leu⁶⁵ in M1 fastens it down as demonstrated previously (9, 43, 44). As seen (gray circle in panel d in Fig. 10B), the region around Leu⁹⁸ in M2m is in close contact with Leu⁶⁵ by hydrophobic interactions producing the M1/M2 V-shaped body and also directly with Glu³⁰⁹. The disruption of the hydrophobic Leu⁹⁸ region probably allowed greater freedom of movement of Glu³⁰⁹, thereby permitting the Ca²⁺ to escape. It is also possible that luminal gate opening is also impaired because EP isomerization and opening depends on the motion of the M1/M2 V-shaped body, which is formed by the interactions of the hydrophobic Leu⁹⁸ region with M1 (11).

In mutants at the A/M2-junction and M2top, we did not observe Ca²⁺ occlusion because of the low level of E1P (Figs. 5 and 6), but it is evident from the lack of transport that Ca²⁺ escaped from the transport sites either in E1P or during the isomerization to E2P. A properly stabilized M2 and A domain must be critical for maintaining a closed cytoplasmic gate. Interestingly, the severe uncoupling is accompanied by acceleration of EP isomerization. It is as if freedom from coupling rechannels the energy into a faster transition. The extensive slack caused by the 5Gi insertions evidently uncoupled and accelerated E1P → E2P. In the change, E1Ca₂·AlF₄⁻·ADP → E2·BeF₃⁻, the A/M2-junction∼M2top loop region does indeed seem to be strained, because, for example, the distance between Glu¹¹⁷-αC and Glu¹²⁵-αC increases from 14.3 Å in E1Ca₂·AlF₄⁻·ADP (PDB code 1T5T) to 15.7 or 17.3 Å in E2·BeF₃⁻ (PDB code 2ZBE or 3B9B), thus by 1.4 or 3.0 Å.

The closer the M2top mutations get to the A/M2-junction the more severe the uncoupling, suggesting that M2top stabilizes the junction and a closed Glu³⁰⁹ gate. The M2top mutations also strongly accelerate EP isomerization; therefore, in the wild type, there must be some structural restriction here. Indeed, in the crystal structure of E1P (E1Ca₂·AlF₄⁻·ADP), M2top interacts with the cytoplasmic half of M4 (M4C, yellow circle in panel c in Fig. 10B), and in E2P (E2·BeF₃⁻), it interacts with the bottom of the P domain and α-helix 7 (yellow circle in panel e in Fig. 10B).

A mechanistic scenario for the EP isomerization and gating with respect to the A/M-junction∼M2top may be as follows. During the E1P → E2P isomerization, the A domain rotates and pulls on A/M2-junction∼M2top, straining it to detach from M4C. The P domain/M4/M5 entity is now able to incline toward the underside of the A domain to produce new interactions with M2top (Fig. 10B). In these coupled motions, the M1/M2 V-shaped body and M4/M5 probably move coordinately, keeping the Glu³⁰⁹ gate closed. Kinetically, coupling and proper Ca²⁺ handling have an energy cost: the rate-limiting slow EP isomerization. Ca²⁺ deocclusion and release from E2PCa₂ involves further inclination of the A and P domains due to strain in the A/M1′-linker, with the A domain lodging above the P domain in E2PCa₂, and the luminal gate opens to release Ca²⁺ (24, 25). Thus, the A domain's M2 link and A/M1′-linker have distinct functions; the former keeps the cytoplasmic gate closed, and the latter is needed for opening the luminal gate, both regions coupling A domain motions with gating.

The importance of an intact A/M2-junction∼M2top interaction is further underpinned by the fact that mutations in the M2c region do not have the same deleterious effects (uncoupling with accelerated EP isomerization) because the interactions of M2top with the M4C/P-domain and of M2m (M1/M2 V-shaped body) to fix the Glu³⁰⁹ gate are normal in these mutants. Also, M2c does not interact strongly with other parts both in E1P and E2P and does not change structure (is not strained) during the EP isomerization. The mutations here are silent. Also, as noted above, the A domain motion driving Ca²⁺ release into the lumen after the EP isomerization is controlled to a significant extent by the A/M1′-linker, another cytoplasmic link (24, 25), and evidently not by M2c.

Luminal Gating

Opening of the luminal gate and Ca²⁺ release is effected during E1P → E2P, and the latter Ca²⁺-free state becomes susceptible to Ca²⁺ binding from the lumen at high Ca²⁺ concentrations (back-inhibition). The gate closes during E2P hydrolysis E2P → E2∼P^‡ with most of Ca²⁺-binding residues protonated, according to both biochemical evidence (7) and relevant crystal structures (9, 11, 12).

We find that E2P hydrolysis is markedly accelerated by 5Gi insertions at aa 103–115 in the M2c region, in sharp contrast to the marked retardation by those at adjacent regions aa 87–102 on M2m and aa 116–125 on M2top∼A/M2-junction (Fig. 7). Significantly, acceleration is accompanied by the almost complete resistance against inhibition by luminal Ca²⁺ even up to 20 mm. Most of these mutants (aa 106–115) are well coupled in terms of Ca²⁺ transport/ATP hydrolysis ratios. Therefore, Ca²⁺ must be released to the lumen, but the binding sites are unavailable for Ca²⁺ rebinding from the lumen due to the rapid E2P hydrolysis and closure of the luminal gate. Alternatively, luminal Ca²⁺ binding is no longer coupled to a large change in the rate of hydrolysis.

The results indicate that a proper unwinding/elongation on M2c with properly controlled E2P hydrolysis may be critical for coupling the change in E2P → E2∼P^‡ at the catalytic site with the luminal gate closure. The 5Gi insertions exaggerated both coupled structural events. Physiologically, the setting of the luminal Ca²⁺ concentration is governed by the Ca²⁺-induced back-inhibition of E2P hydrolysis (22, 24, 37–42, 45). In the wild type, the regionally limited partial unwinding/elongation in M2c appears critical for the set point, but it comes with an energy cost: slow but properly controlled gate closing coupled to E2P hydrolysis.

Then, in the structural change mimicking E2P → E2∼P^‡ (Fig. 10B), a part of helical M2c, Asn¹¹¹–Ala¹¹⁵, unwinds and extends 7 Å, which is due to strain exerted through a 25° rotation of the A domain upon water attack at the active site to effect hydrolysis (11). Thereby, the lower part of M2 moves downward 6 Å, inclines, and presses the M1/M2 V-shaped rigid body on M3 and M4 to close the luminal gate (11). In E2∼P^‡, the interactions between M2c and the rest of the protein are rather weak, whereas those between the regions on either side and protein are helix to helix, extensive, and therefore fixed (yellow and green circles in panel g in Fig. 10B). Thus, it appears that M2 is literally pulled apart at M2c. Evidently, the 5Gi mutations in this weak region facilitate and exaggerate (possibly causing elongation of up to ∼17 Å) these processes, resulting in an activation of hydrolysis and fast and tight closure of the transport sites (or uncoupling). Consistently, mutations that disrupt the strong helix to helix interactions at the bottom of M2c (aa 103–110) with the M1′ helix alongside resulted in accelerated and luminal Ca²⁺-resistant E2P hydrolysis.

In contrast, mutations higher up, in the M2top∼A/M2-junction, impede hydrolysis and closure, probably by affecting interactions with the A and P domains and the Tyr¹²²-hydrophobic cluster, which was previously demonstrated to be critical for proper catalytic site formation in E2P with hydrolytic ability (yellow circle in panel g in Fig. 10B) (21, 22, 46). Mutations lower down, in aa 87–102 of M2m, disrupt arrangement of the transmembrane helices and probably cause a long range effect on the A and P domain interaction and catalytic site, thus retarding E2P hydrolysis. Note that mutants at the region aa 95–101 showed some resistance against luminal Ca²⁺-induced back-inhibition. This region forms the M1/M2 V-shaped body, and the downward movement of M2 with inclination of the body is critical for the luminal gate closure (9, 11). The 5Gi elongation in this region probably exaggerated such motions and tightened gate closure.

The long M2 helix and probably the A domain's motion are common structural features in P-type ion-pumping ATPases (16–20). The strategy employed here of systematically introducing 5Gi insertions along M2 to disrupt the helix and relieve strain may be helpful for exploring the role of M2 in coupling A domain movements with gating and catalysis in other P-type ATPases.