Biogenesis and Function of the Yeast Plasma-Membrane H⁺-ATPase

Fungal plasma-membrane H⁺-ATPases were first detected biochemically in the late 1970s, although their existence had been predicted earlier by electrophysiological measurements of a large, ATP-and H⁺-dependent membrane potential in Neurospora crassa (for a review, see Rao and Slayman, 1996). Over the next several years, methods were developed to purify the H⁺-ATPase from three fungal species: Schizosaccharomyces pombe (Dufour and Goffeau, 1978), Saccharomyces cerevisiae (Malpartida and Serrano, 1980) and Neurospora crassa (Bowman et al., 1981; Addison and Scarborough, 1981). In all cases, the ATPase was shown to be composed of a single, 100 kDa polypeptide, firmly embedded in the lipid bilayer and requiring detergents for solubilization. Upon reconstitution into liposomes, the 100 kDa polypeptide carried out ATP-dependent proton pumping (Villalobo et al., 1981; Malpartida and Serrano, 1981; Perlin et al., 1984; Goormaghtigh et al., 1986) with a measured stoichiometry of 1 H⁺ translocated per ATP split (Perlin et al., 1986). Parallel studies with ³²P-labelled ATP pointed to the formation of a covalent β-aspartyl phosphate intermediate as an essential part of the reaction cycle (Amory et al., 1980; Dame and Scarborough, 1980, 1981; Amory and Goffeau, 1982), accompanied by sensitivity to micromolar concentrations of inorganic orthovanadate (Bowman and Slayman, 1979). Taken together, these findings placed the fungal H⁺-ATPases in the group now known as the P-type cation pumps.

The next major step came in 1986, when PMA1 genes encoding the yeast and Neurospora crassa H⁺-ATPases were cloned and the deduced amino acid sequences were compared with those of the sarcoplasmic reticulum Ca²⁺-ATPase and renal plasma-membrane Na⁺,K⁺-ATPase (Serrano et al., 1986; Hager et al., 1986; Addison, 1986). Hydropathy analysis led to a common topological model for all three enzymes (Fig. 1; for a review, see Lutsenko and Kaplan, 1995), with the 100 kDa polypeptide anchored in the membrane by four hydrophobic segments at the N-terminal end of the molecule and six at the C-terminal end. The central region protrudes into the cytoplasm and contains strongly conserved motifs for ATP binding and phosphorylation.

Until recently, the model of Fig. 1 was based only on limited experimental evidence. In 1989, Mandala and Slayman raised antibodies against the N and C termini of the Neurospora crassa H⁺-ATPase and showed that they could bind to inside-out plasma-membrane vesicles but not to intact protoplasts. Likewise, sites close to the N and C termini could be cleaved by trypsin in vesicles but not in protoplasts (Mandala and Slayman, 1988, 1989). These two sets of findings established that the ATPase must have an even number of membrane-spanning segments, with both termini exposed at the cytoplasmic surface of the membrane. Further information came from studies by Lin and Addison (1994), who examined the ability of fusion constructs from the Neurospora crassa H⁺-ATPase to undergo in vitro membrane insertion and glycosylation. As predicted by the model shown in Fig. 1, membrane-spanning segments M5 and M7 behaved as initiate-transfer sequences and M6, M8 and M10 as stop-transfer sequences. However, M9 failed to bring about transfer in vitro even though it became associated with the microsomal membranes, and M10 was able to initiate transfer as well as to stop transfer. Thus, although the data were generally consistent with a model for the H⁺-ATPase containing 10 membrane-spanning segments, it was obvious that another, more direct, approach was needed.

Very recently, an important step was taken by Scarborough, Kuhlbrandt and co-workers, who grew two-dimensional crystals of purified Neurospora crassa H⁺-ATPase, analyzed the crystals using cryoelectron microscopy and successfully computed a three-dimensional map at 8 nm resolution (Auer et al., 1998). The results confirmed the presence of 10 membrane-spanning α-helices, tilted at varying angles from the plane of the bilayer; four of the 10 helices twisted around one another in a distinctive righthanded bundle. The cytoplasmic portion of the ATPase had four distinct domains and was linked to the membrane by a complex stalk-like structure. Similar results were reported for the sarcoplasmic reticulum Ca²⁺-ATPase by Zhang et al. (1998).

Vigorous efforts are now under way to produce higher-resolution structures of P-ATPases. In the meantime, this paper will summarize the progress that has been made towards mapping structure/function relationships within the yeast H⁺-ATPase by means of site-directed mutagenesis.

Immediately after the yeast PMA1 gene had been cloned, work began in several laboratories to construct and characterize site-directed mutants in order to examine the functional role of individual amino acid residues and motifs throughout the protein. It quickly became clear that a special expression strategy was needed because of the central role played by the H⁺-ATPase in cell growth. In particular, the ATPase has been estimated to consume as much as one-quarter of cellular ATP, pumping H⁺ outwards and generating the electrochemical gradient that drives H⁺-dependent nutrient uptake (for a review, see Rao and Slayman, 1996). Thus, knockout of the PMA1 gene is lethal (Serrano et al., 1986), and a variety of studies have established that the ATPase must have at least 30 % of the usual wild-type activity to support cell growth (D. S. Perlin, personal communication; C. W. Slayman’s laboratory, unpublished results).

To make it possible to study a full range of mutants, including those with low activity, our laboratory undertook to develop a transient expression system for the PMA1 ATPase and other homologous and heterologous plasma membrane proteins (Nakamoto et al., 1991). As illustrated in Fig. 2, this approach relies upon a temperature-sensitive mutation in the SEC6 gene to arrest newly synthesized mutant ATPase in intracellular secretory vesicles, just prior to fusion with the plasma membrane. The vesicles are abundant and readily isolated by gel filtration (Walworth and Novick, 1987) or sucrose gradient centrifugation (Ambesi et al., 1997). Equally important, they are uniformly oriented inside-out relative to the plasma membrane, allowing ATP-dependent H⁺ pumping to be assayed by Acridine Orange fluorescence quenching (Nakamoto et al., 1991; Ambesi et al., 1997). In the work to be described, PMA1 mutants were constructed, expressed in secretory vesicles and examined in detail for their ability to split ATP and pump protons.

As the list of known fungal H⁺-ATPase genes has grown, it has become possible to obtain a provisional view of functionally important regions by aligning the deduced amino acid sequences and mapping the residues that have remained unchanged over 1.1 billion years of fungal evolution. Fig. 1 shows that only 40 % of residues are conserved throughout a group of 10 fungal PMA1 and PMA2 ATPases. The identical residues are scattered along the small cytoplasmic loop between M2 and M3 and the large loop between M4 and M5, sometimes in regions with recognizable function (e.g. the ATP-binding and phosphorylation sites) but often with no direct evidence as to functional role. Most of the membrane-spanning segments have also been relatively well conserved, while three (M3, M7 and M10) have diverged considerably.

On the basis of a variety of considerations, it seems likely that the 60-amino-acid stretch from the start of M4 to the end of the phosphorylation region and the 60-amino-acid stretch from stalk segment 5 (S5) to M6 are central to the reaction mechanism of the H⁺-ATPase. Aspartyl and glutamyl residues in M4, M5 and M6 of the mammalian Na⁺,K⁺-, H⁺,K⁺-and Ca²⁺-ATPases are now known to be required for high-affinity cation binding (Clarke et al., 1989, 1990; Nielsen et al., 1998), making it reasonable to think that these helices line the transport pathway of the H⁺-ATPase as well. Furthermore, the stalk segments linking M4 and M5 to the central catalytic domain of the ATPase are strong candidates for transmitting the E1→E2 conformational change into the membrane. We have performed alanine-scanning (or, in the case of S5, cysteine-scanning) mutagenesis along the entire length of both regions to maximize the chance of identifying functionally important residues. Further details about many of the mutants can be found in recent papers by Ambesi et al. (1996), Nakamoto et al. (1998), Dutra et al. (1998), DeWitt et al. (1998) and Sen Gupta et al. (1998); results on the S4 and S5 mutants are now being prepared for publication.

Because some mutations could prevent proper folding or insertion of newly synthesized ATPase into the endoplasmic reticulum membrane, the first step was to determine the relative amount of mutant protein that reached the secretory vesicles. This was achieved by quantitative immunoblotting with polyclonal ATPase antibody. Indeed, mutations at 12 of the 120 positions studied completely prevented normal biogenesis. In most cases, the mutant polypeptides have been epitope-tagged and shown by indirect immunofluorescence to accumulate in intracellular structures derived from the endoplasmic reticulum (e.g. Harris et al., 1994; Portillo, 1997; DeWitt et al., 1998; Sen Gupta et al., 1998). The same mutants are exceedingly sensitive to low concentrations of trypsin, consistent with the idea that they are poorly folded and arrested by the endoplasmic reticulum quality control machinery. Many of them behave genetically in a dominant lethal fashion, causing co-expressed wild-type ATPase to be trapped in the endoplasmic reticulum (Harris et al., 1994; Portillo, 1997; DeWitt et al., 1998); studies are under way to determine whether the wild-type and mutant forms interact directly by forming oligomers or whether they compete for some component of the endoplasmic reticulum biosynthetic machinery such as a translocation channel or chaperone.

Perhaps the most unexpected finding has been the nearly uninterrupted stretch of biogenesis-defective mutants in the phosphorylation region of the H⁺-ATPase, from Asp378 (the phosphorylation site) to Thr384. When the data described above were superimposed upon secondary structural predictions for this region, there was a clear correspondence between the cluster of biogenesis-defective mutants and a likely β-turn in the vicinity of Asp378. According to the algorithm of Garnier et al. (1978), substitution by Ala at each position should introduce a more helical nature to the β-turn and could therefore perturb the structure of the cytoplasmic loop, preventing the insertion of M5–M10 into the lipid bilayer. It will be interesting to see whether these predictions are borne out when a high-resolution structure becomes available for the H⁺-ATPase.

For two other residues that are essential for normal folding and biogenesis, further studies have uncovered a likely mechanism. When either Arg695 in M5 or Asp730 in M6 is replaced by a neutral amino acid, the ATPase becomes trypsin-sensitive and is retained in the endoplasmic reticulum. But when Arg695 and Asp730 are replaced simultaneously by Ala, or when they are swapped to produce an R695D, D730R double mutant, the ATPase undergoes proper maturation and displays normal rates of ATP hydrolysis and ATP-dependent H⁺ pumping (Sen Gupta et al., 1998). These results indicate that neither Arg695 nor Asp730 is an essential residue; instead, the two take part in a charge–charge interaction, presumably lending stability to the M5–M6 hairpin.

Also encountered in the present study have been mutants with a characteristic pattern of kinetic changes, including an elevated IC₅₀ for orthovanadate, a decreased K_m for MgATP and, in some cases, a higher-than-normal pH optimum for ATP hydrolysis (Ambesi et al., 1996; Dutra et al., 1998; A. Ambesi, M. Miranda and C. W. Slayman, in preparation). The most likely interpretation is that these mutants have undergone a shift in equilibrium from the E₂ conformation, which is expected to bind vanadate with high affinity and ATP with low affinity, towards the E₁ conformation, which should have a high affinity for ATP and a low affinity for vanadate. Whether the alkaline pH optimum seen in some but not all of the mutants reflects a greater affinity of the E₁ form for the transported proton, or whether it is simply a spurious pKa change at some unrelated site, remains to be established.

An interesting aspect of the E₁→E₂ mutants is their distribution within the ATPase polypeptide (Fig. 3). Recent work (A. Ambesi, M. Miranda and C. W. Slayman, in preparation) has identified a stretch of eight consecutive residues in the stalk 4 region along which mutations elevate the IC₅₀ for vanadate from 1 μmol l⁻¹ to 10–200 μmol l⁻¹ and decrease the K_m for MgATP from 1.5 mmol l⁻¹ to 0.1–0.5 mmol l⁻¹. This stretch contains Ser368, where a mutation to Phe had previously been isolated on the basis of hygromycin B resistance and found to display similar IC₅₀ and K_m changes (Perlin et al., 1989). The properties of the kinetically altered mutants suggest that the middle of stalk segment 4 may play a critical role in the E₁→E₂ transition and thus in communication between the cytoplasmic and membrane portions of the H⁺-ATPase.

A third type of mutant, found during a systematic survey of charged residues within the transmembrane segments of the H⁺-ATPase (V. V. Petrov, K. P. Padmanabha, R. K. Nakamoto and C. W. Slayman, in preparation), is altered in H⁺:ATP coupling. When secretory vesicles expressing the wild-type ATPase are incubated with increasing concentrations of MgATP, a linear relationship is seen between the initial rate of ATP-dependent Acridine Orange fluorescence quenching (as a measure of H⁺ transport) and the initial rate of ATP hydrolysis, consistent with the idea that the pump operates at a constant stoichiometry over a considerable range of velocities (Ambesi et al., 1996). For most mutants studied in the same way, the data points have been found to fall along the wild-type line, suggesting little or no change in H⁺:ATP coupling. For E703Q, E703L and E803N, however, the slope of the line was significantly lowered, pointing to a partial uncoupling of the pump; and for E803Q, the slope was elevated, as if the pump had somehow increased its functional stoichiometry (V. V. Petrov, K. P. Padmanabha, R. K. Nakamoto and C. W. Slayman, in preparation).

Glu703 and Glu803 are of special interest because the corresponding residues have been implicated in high-affinity cation binding by mammalian Ca²⁺-and Na⁺,K⁺-ATPases (Clarke et al., 1989, 1990; Nielsen et al., 1998). Studies are under way to express the E703Q, E703L, E803N and E803Q H⁺-ATPases in the yeast plasma membrane, where it should be possible to analyze the charge-transfer step of the transport cycle by patch-clamping and therefore to test the effect of the mutations directly.

The mutants described in this brief review, together with mutants that have been constructed to study M1 and M2 (Seto-Young et al., 1996), the small cytoplasmic loop (Wang et al., 1996), the stalk segments extending into the cytoplasm from M2 and M3 (Soteropoulos and Perlin, 1998) and several other parts of the polypeptide (Maldonado and Portillo, 1995; Portillo, 1997), have produced a broad picture of structure/function relationships throughout the H⁺-ATPase. Important questions remain concerning the architecture of the ATP-binding and phosphorylation sites, the way in which ATP hydrolysis and phosphorylation lead to the E₁→E₂ conformational change, the coupling between that change and cation transport, and the detailed pathway along which cations move through the membrane. Progress towards answering these questions will speed up dramatically once high-resolution structures are available for the E₁ and E₂ conformations, which will provide a framework for refining existing data and for pursuing new biochemical, biophysical and genetic tests of the pump mechanism.

Work in our laboratory has been supported by NIH grant GM15761 and by postdoctoral fellowships to A.A. (NIH), M.M. (Fogarty) and V.P. (Yale School of Medicine).