Report

Structure, Mechanism, and Regulation of the Neurospora Plasma Membrane H+-ATPase

See allHide authors and affiliations

Science  06 Sep 2002:
Vol. 297, Issue 5587, pp. 1692-1696
DOI: 10.1126/science.1072574

Abstract

Proton pumps in the plasma membrane of plants and yeasts maintain the intracellular pH and membrane potential. To gain insight into the molecular mechanisms of proton pumping, we built an atomic homology model of the proton pump based on the 2.6 angstrom x-ray structure of the related Ca2+ pump from rabbit sarcoplasmic reticulum. The model, when fitted to an 8 angstrom map of theNeurospora proton pump determined by electron microscopy, reveals the likely path of the proton through the membrane and shows that the nucleotide-binding domain rotates by ∼70° to deliver adenosine triphosphate (ATP) to the phosphorylation site. A synthetic peptide corresponding to the carboxyl-terminal regulatory domain stimulates ATPase activity, suggesting a mechanism for proton transport regulation.

P-type ATPases are ion pumps of ∼100 kD that use ATP to transport cations through the cell membrane against a concentration gradient (1). The proton ATPases in the plasma membrane of fungal and plant cells maintain the intracellular pH and membrane potential, providing energy for the uptake of nutrients and exchange of ions by secondary transporters. The pumps cycle between the E1 and E2 states, which have different binding affinities for nucleotides and for the transported ion. A conserved aspartate (Asp378 in the Neurospora proton ATPase) is reversibly phosphorylated (1) after the proton binds to a site in the membrane from the cytoplasmic side. Phosphorylation of the aspartate results in a conformational change. This reduces affinity of the binding site for the proton that is released to the outside. Large conformational changes occurring during the pumping cycle (2, 3) were apparent from a comparison of the structures of the sarcoplasmic Ca2+-ATPase in the E1 (4) and E2 state (5) with one another and with the NeurosporaH+-ATPase (6).

Electron cryomicroscopy enabled the determination of 8 Å maps of the H+-ATPase (6) and the Ca2+-ATPase in the E2 state (5). These maps agreed in the arrangement of 10 transmembrane helices (designated M1 to M10) in the membrane (M) domain but indicated large differences in the cytoplasmic part. The 2.6 Å x-ray structure of the Ca2+-ATPase in the E1 state (4) revealed details of the ion-binding site in the membrane and defined three cytoplasmic domains: the phosphorylation (P) domain, the nucleotide-binding (N) domain, and the A domain of unknown function.

On the basis of a detailed sequence comparison of P-type ATPases to identify conserved, functionally important regions, we built a homology model of the proton ATPase and fitted it to the 8 Å map. The alignment of the H+-ATPase and Ca2+-ATPase sequences is shown in fig. S1. The Ca2+-ATPase has 994 residues, whereas the Neurospora H+-ATPase has only 920, including an NH2-terminal extension of 67 residues and a COOH-terminal extension of 36 residues. Neither extension is present in the Ca2+-ATPase. Overall, 25% of the 817 residues in the four principal domains are identical in both proteins. The homology of individual domains varies from 18% (M domain) to 39% (P domain) (table S1).

An initial comparison of the model to the map (fig. S2, A and B) indicated a very good fit of the M and P domains, whereas the N and A domains were, respectively, largely or partially outside the map density. They were fitted to the map as rigid bodies, and the NH2- and COOH-terminal extensions were added to the remaining, unoccupied map regions. The resulting fitted model of the proton pump is shown in Fig. 1.

Figure 1

Fit of the atomic model of the H+-ATPase to the 8 Å map (6) (blue net). The four principal domains are shown in pink (M), red (P), green (N), and yellow (A). The COOH-terminal regulatory R domain is shown in cyan and the NH2-terminal extension in orange. (A) Side view. Dashed lines indicate the position of cross sections through the M, P/A, and N domains shown in (C), (D), and (E). (B) Front view. (C) Cross section parallel to the membrane plane through the M domain. The map density in the region at the proton-binding site between helices M4, M5, M6, and M8 (red asterisk) is low. (D) Section through the P, A, and R domains. (E) Section through the N domain. The purple net in (C), (D), and (E) is drawn at a higher contour level than the blue net in (A) and (B).

The strikingly good fit of the M domain (Fig. 1, A to C) indicates that its structure is highly conserved, even though this would not be expected from the low sequence homology. The match was almost perfect for M3, M4, and M5, whereas the remaining transmembrane helices required minor repositioning. The root mean square (RMS) deviation of α carbons in M1 through M10 from the Ca2+-ATPase structure was 2.1 Å.

The ion-binding site in the membrane is defined by Asp800 in M6 and Glu908 in M8 of the Ca2+-ATPase. Both residues are highly conserved in the P-type ATPases, highlighting their importance for ion transport. In the H+-ATPase, Asp730 and Glu805 are found in the corresponding proton-binding site. Mutagenesis data on Glu805 have not been reported, but Asp730 is critical for proton pumping (7). The map density in this region was conspicuously low (Fig. 1C), presumably due to the presence of water molecules near the proton-binding site. Nearby polar groups include the main-chain carbonyls of Ile331, Ile332, and Val334 in the unwound part of M4 and the side chains of Tyr694 and Ser699 in M5, which are both essential for proton pumping (7).

All other ionic and polar side chains in the Ca2+-binding site (Glu58, Glu309, Glu771, Asn796, and Thr799) are replaced by valines or alanines in the H+-ATPase. By contrast, the proton-binding site of the latter contains the basic residues Arg695 and His701 (Fig. 2A). Both are completely conserved in the yeast plasma membrane proton ATPases and are essential for ATPase activity and proton pumping (7). In the plant proton pumps, His701 is replaced by an arginine and Arg695 is replaced by an alanine. An arginine and two acidic side chains thus appear to be the key residues for proton transport, as in bacteriorhodopsin (8–10), another proton pump.

Figure 2

Stereo diagrams of two functionally important regions of the proton ATPase. (A) The proton-binding site in the center of the M domain. Basic side chains (Arg695 and His701 on M5) are blue, acidic side chains (Asp730 on M6, Glu805 on M8) are red, and polar side chains (Tyr694 and Ser699 on M5, Thr733 on M6) are yellow. (B) The nucleotide-binding site at the interface of the P (red), N (green), and R (cyan) domains. The side chains of residues defining the binding-site Lys456 (in front), Lys474, and Phe451 are shown, with Lys in blue and Phe in pink. The model suggests that the regulatory R domain locks the N domain in position, thus rendering the ATPase inactive.

The conserved structure of the M domain must reflect the conserved mechanism of ion transport and a correspondingly conserved ion path. An analysis of conserved polar and ionic residues in the structures of two family members should therefore reveal those that are involved in ion conduction, especially if the proteins are not closely related, as in this case. Access to the ion-binding site from above is blocked by large, hydrophobic side chains. The E2-state model of the Ca2+-ATPase (11) does not indicate a major reorientation of the helices in this region. The most likely access to the ion-binding site is therefore through a pocket of ionic and polar side chains between the cytoplasmic ends of M1 and M2. The conserved side chains of Glu108 on the cytoplasmic side of M1 and Glu162 on M2 are in strategic positions for proton conduction, as are the polar side chains of Gln161 and Asn154 in M2. Gln and Asn residues facilitate rapid proton conduction in cytochrome f (12) and in the bacterial reaction center (13). The low map density in this region is consistent with the presence of water molecules, as might be expected along the proton path (8–10).

The upper half of M2 is polar and is unlikely to be embedded in the lipid bilayer. In the H+-ATPase model, the cytoplasmic end of this helix is ∼4 Å closer to M1 than in the E1 conformation of the Ca2+-ATPase (4). However, in the E2 state (4), M2 is tilted toward M1 by 14° so that the two helices are close together and more or less parallel. This conformational change would block the ion path back to the cytoplasm, thus imparting a direction to the transport process. The reorientation of M2 may facilitate the 90° rotation of the A domain in the E1-E2 transition (11) because this domain is tethered to the cytoplasmic end of the helix.

The exit channel on the exocytoplasmic side is likely to be the shortest route from the proton-binding site to Asn718 on M6, which is only 15 Å below. Asn718 is at the deepest point of a shallow cavity in the exocytoplasmic protein surface. This cavity is wider in the H+-ATPase because its exocytoplasmic loops are shorter than those of the Ca2+-ATPase. In the H+-ATPase map, a region of low density extends from Asn718 to the ion-binding site, which suggests that protons may be able to leave along this path (14).

The A domain was moved sideways into the map by ∼10 Å as a rigid body to fit the density. Because the cytoplasmic end of M2 had moved by ∼5 Å, the linking peptide was not strained by this repositioning. The NH2-terminal extension of the H+-ATPase is continuous with the A domain. It was modeled on the structures of homologous sequences from two other proteins. Three short α helices, linked by extended chains (orange in Fig. 1), were fitted to the map density around three sides of the A domain, in keeping with the predominant negative charge of this extension, which requires a solvent-exposed position. The location of a major part of this acidic extension close to the membrane surface, directly above the entrance to the proton path, suggests that it acts as a pH sensor or a local proton reservoir. The local membrane potential would alter its protonation state, affecting the mobility of the A domain and of the attached M2 helix. This would be consistent with the presumed role of the NH2-terminal extensions of the heavy metal–pumping P-type ATPases that contain several metal-binding motifs (15), which may have an analogous function in sensing toxic heavy metal ions and in sequestering them to the channel entrance.

Most helices of the P domain Rossmann fold fitted the map density very well (Fig. 1, A, B, and D) and required only minor readjustment. Striking exceptions were the short peripheral helices 4 and 4′ (4), which were clearly outside the map density. Helix 4′ required an upward tilt of ∼35° around Ala603 to fit the map. As a result, its NH2-terminal end came into close contact with the N domain. Excepting helices 4 and 4′, the α-carbon RMS deviation from the Ca2+-ATPase P domain was 1.48 Å.

The N domain is connected to the P domain by two adjacent peptide strands (4). Rigid-body rotation by 73° around this hinge at Asn386 and Asp534 resulted in an optimal fit (Fig. 1, A, B, and E). As a result, Gln466 at the tip of the N domain moved by ∼50 Å. Delivery of ATP to the phosphorylation site by tethered Brownian motion of the N domain has been proposed as a part of the Ca2+-ATPase mechanism (11), even though, surprisingly, the N and P domains are in nearly the same relative positions in the E1 structure and in the E2 model (11). The H+-ATPase model shows that the postulated movement of the N domain does indeed occur.

In the E1 state of the Ca2+-ATPase, the distance between the nucleotide-binding site in the N domain and the phosphorylation site in the P domain is at least 25 Å (4). In the H+-ATPase model, this distance is shortened to ∼15 Å, which still seems too far for phosphate transfer. The N domain cannot move closer to the phosphorylation site without a steric clash with the P domain (Fig. 2B), partly as a result of the reoriented helix 4′, which appears to act as a doorstop for the N-domain hinge movement. Direct transfer of the phosphate to Asp378 would require this helix to revert to its E1 position and would also require some reorientation of the P domain.

The distance between the phosphorylation site in the P domain and the ion-binding site in the membrane is ∼50 Å (4). The two sites are linked by M4 and M5, which contain most of the key residues in the ion-binding site. The P domain fits around the top of M5 like a hand around a pole. Any movement of this domain is therefore transmitted to the ion-binding site, changing the local chemical environment of proton-binding side chains, and vice versa. In bacteriorhodopsin, small movements of proton-binding side chains cause shifts in their pK a (whereK a is the acid dissociation constant), resulting in proton pumping (8–10). We propose that the proton-binding residues in the H+-ATPase—particularly Arg695 and His701 on M5, as well as Asp730 and Glu805—undergo similar pK a changes due to a change in their local environment, resulting in the release of bound protons to the outside.

The autoinhibitory COOH-terminal extension of the H+-ATPase contains a regulatory site at Ser913and Thr914 (16–18), phosphorylated by a specific kinase (19). A yeast double mutant of these residues locks the enzyme in the inactive state, whereas deletion of the autoinhibitory extension results in constitutive activation (16–18). The fit of the four principal domains and the NH2-terminal extension left one small but prominent part of the H+-ATPase map unoccupied to accommodate the COOH-terminal extension. This part of the map, located above the cytoplasmic end of M10 (fig. S2D), had roughly the shape and size of a membrane-spanning helix. We therefore modeled residues 884 to 920 as a predominantly α-helical structure (cyan in Fig. 1) with kinks at Pro893, and at Ser913 and Thr914 to fit the map, and linked it to M10 by a stretch of extended chain. The well-defined shape and its position in the H+-ATPase hexamer (fig. S3) indicate that it must be regarded as a separate domain. In accordance with its function, it is termed the regulatory (R) domain. The R domain fit puts Ser913/Thr914next to the N and P domains. This is consistent with numerous second-site revertants (7), which require a physical interaction of these residues with the main body of the enzyme.

We investigated the effect of the R domain on the activity of the Neurospora plasma membrane H+-ATPase with a synthetic peptide of the 38 COOH-terminal residues. Addition of this peptide stimulated the ATPase activity by as much as a factor of 10, depending on pH (Fig. 3), whereas other peptides of similar size had no effect. This suggests that the R domain exerts its regulatory function by attaching to the N domain, restricting its mobility by tethering it to the membrane. We hypothesize that the R domain is released upon phosphorylation, leaving the N domain free to move and able to deliver ATP to the P domain. An excess of R domain peptide would have the same effect, replacing the enzyme's own R domain in the binding site and thus enabling the hinge movement of the N domain. The resulting proposed mechanism of proton pumping and enzyme regulation is shown in Fig. 4.

Figure 3

Stimulation of ATPase activity by the regulatory R domain. A peptide of the 38 COOH-terminal residues of theNeurospora proton ATPase was added to a standard ATPase assay (24). At peptide concentrations of 200 μM and above, ATPase activity (circles) increased by more than a factor of 10 relative to the activity in the absence of added peptide. Maximum stimulation was observed at pH 6.8. Control peptides of similar size (diamonds, thiocalcitonin; triangles, insulin chain B) had no effect. The peptide itself (squares) had no ATPase activity.

Figure 4

Proposed mechanism of proton transport and regulation. The proton pump is activated by reversible phosphorylation of the regulatory R domain. In the open E1 state of the H+-ATPase, protons have access to the proton-binding site in the M domain. Proton binding causes a conformational change that is transmitted via M4 and M5 to the P and A domains, causing them to reorient. The A domain movement pulls M2 into a position that blocks the proton path to the binding site in the membrane. Phosphorylation of Asp378 in the P domain reduces the affinity of the binding site in the membrane for the proton that is released to the outside. The enzyme then returns to the E1 state via the E2 state, and another ion pumping cycle starts. E1-P and E2-P refer to the intermediate, phosphorylated states of the enzyme. As nutrients are depleted and the cell metabolism shuts down, the autoinhibitory R domain is dephosphorylated and attaches to the N domain, rendering it unable to deliver ATP to the phosphorylation site. Domains are color-coded as in Figs. 1and 2.

The H+-ATPase model indicates that the R domain interacts with the next-door monomer at Gln624 and Arg625 in helix 5 of the P domain. The arginine is completely conserved in the hexamer-forming fungal H+-ATPases, which suggests that the R domain links adjacent monomers and thus has a critical role in hexamer formation. Characteristic crystalline patches of rosette-shaped particles are common in freeze-fracture replicas of starving yeast (20) and Neurospora cells (21). The arrays have the same unit cell parameters and morphology as single-layer two-dimensional crystals of the Neurospora H+- ATPase (22). We conclude that the H+-ATPase hexamers are a storage form of the inactive enzyme. The minor domain movements observed in low-resolution maps of isolated ATPase hexamers in the presence and absence of ADP (23) are unlikely to reflect the well-documented large conformational changes in fully active P-type ATPases.

The striking structural similarity between the H+-ATPase and the distantly related Ca2+-ATPase implies that all other P-type ATPases—including the Na+,K+-ATPase, the H+,K+-ATPase, and the heavy metal pumps—have essentially similar structures and can be modeled on the Ca2+-ATPase. The reason for this remarkable conservation of structural detail must be strong evolutionary pressure to maintain the functional sites of each domain in their exact spatial relationship for efficient ion pumping.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1072574/DC1

Methods

Figs. S1 to S3

Table S1

References

  • * Present address: Laboratory of Molecular Biophysics, University of Oxford, South Parks Road, Oxford OX1 3QU, UK.

REFERENCES AND NOTES

View Abstract

Navigate This Article