The vacuolar H⁺-ATPase – one of the most fundamental ion pumps in nature

H⁺-ATPases function in biological energy conversion in every known living cell. They can generate a proton-motive force (PMF ) by hydrolyzing ATP and may utilize the energy of the PMF for ATP formation (Mitchell, 1968). They also function in homeostasis, maintaining an appropriate cytoplasmic pH and acidifying the interior of several organelles of eukaryotic cells. These activities are among the most fundamental processes in nature.

There are two mechanistically distinct groups of ATP-dependent ion pumps. One group, the P-ATPases, operates with a phospho-enzyme intermediate and its members are usually sensitive to low concentrations of vanadate. Na⁺/K⁺-ATPase and the gastric H⁺-ATPase are notable members of the P-ATPase family. The other group, the families of F-and V-ATPases, functions without a phosphorylated intermediate (Nelson and Taiz, 1989). They are not sensitive to low vanadate concentrations and are sensitive to the specific inhibitor bafilomycin A (Bowman et al. 1988b). The family of F-ATPases functions in eubacteria, chloroplasts and mitochondria, and the family of V-ATPases is present in archaebacteria and the vacuolar system of eukaryotic cells. Eukaryotic F-ATPases are present exclusively in chloroplasts and mitochondria because they require organellar gene products for their functional assembly (Nelson, 1989). In contrast, V-ATPases are composed only of nuclear gene products and are present in organelles of the vacuolar system and in the plasma membrane of specialized cells.

The most fundamental difference between F-and V-ATPases is a consequence of their function and origin. Although the main function of F-ATPases is to synthesize ATP at the expense of a PMF generated by electron transport chains, the main function of V-ATPases in eukaryotic cells is to generate a PMF at the expense of ATP and to cause limited acidification of the internal space of several organelles of the vacuolar system. The origin of F-ATPases is rooted in the eubacteria that evolved into chloroplasts and mitochondria, whereas the origin of V-ATPases is related to archaebacteria. Therefore, the evolution of the two families of proton pumps is highly relevant to their structure, function and mechanism of action.

Both F-and V-ATPases are multisubunit protein complexes built of distinct catalytic and membrane sectors. F-ATPases from Escherichia coli and chloroplasts contain eight and nine subunits, respectively, and we are not certain about the precise number of subunits in the mitochondrial enzyme (Nelson, 1989; Futai et al. 1989). However, the catalytic sector of all of these enzymes is composed of five subunits denoted α, β, γ, δ and ϵ. The membrane sector contains a minimum of three subunits denoted in E. coli as a,b and c (proteolipid), and more than six subunits in mitochondria from various sources. In E. coli, the stoichiometry of the various subunits is 3α, 3β, 1γ, 1δ, lϵ, la, 2b and 6–12c. This stoichiometry is largely maintained for the corresponding subunits in the different F-ATPases from various sources (Penefsky and Cross, 1991).

The subunit structure of V-ATPases from various sources has been studied in several laboratories during the last few years (Xie and Stone, 1986; Cidon and Nelson, 1986; Arai et al. 1988; Moriyama and Nelson, 1989a,b,c; Bowman et al. 1989; Lai et al. 1988; Kane et al. 1992; Umemoto et al. 1991). The study underwent the usual evolution of biological observations, from conflicting data to general agreement into reasonable doubts. Recent studies utilizing chaotropic agents and cold-inactivation in the presence of MgATP have revealed that the catalytic sector of V-ATPases may be composed of five different subunits denoted as subunits A to E in order of decreasing relative molecular mass from 70×10³ to 26×10³, respectively (Nelson, 1989). The genes encoding four of these subunits (A, B, C and E) in mammalian, plant and fungal cells have been cloned and sequenced (Hirsch et al. 1988; Zimniak et al. 1988; Bowman étal. 1988a, c; Nelson et al. 1989, 1990; Südhof et al. 1989; Beltrán et al. 1992; Hirata et al. 1990; Puopolo et al. 1991, 1992a). Disruption of these genes in yeast cells gave similar phenotypes, leaving little doubt that they are indeed subunits of the V-ATPase (Nelson and Nelson, 1990; Hirata et al. 1990; Noumi et al. 1991). Subunit D has been identified by immunological cross reactivity in preparations of mammalian V-ATPases and in the supernatants of cold-inactivated enzymes (Moriyama and Nelson 1989a,b,c). Cloning of the gene encoding this polypeptide in yeast may verify its integrity as a subunit of the enzyme. A stoichiometry of 3A, 3B, 1C and 1E has been reported for the catalytic sector of V-ATPases from clathrin-coated vesicles and chromaffin granules (Arai et al. 1988; Nelson, 1989). Two recent studies provided some doubts concerning the number of subunits in V₁. It was demonstrated that the lack of a 40×10³M_r polypeptide (subunit C?) did not prevent reconstitution of the solubilized catalytic sector onto depleted membranes (Puopolo et al. 1992b). Recently, a yeast gene encoding a protein of about 54×10³M_r was cloned and shown to be necessary for the assembly of the enzyme (Anraku et al. 1992). Moreover, the observation that this protein was present in the assembled enzyme suggests additional subunits for the V₁ sector from yeast vacuoles. Additional polypeptides of about 50×10³M_r have been reported in preparations of mammalian V-ATPases (Moriyama and Nelson, 1989b,c). Further studies on the structure of V-ATPases and the crystallization of this enzyme may help to settle the arguments regarding its structure.

Owing to its hydrophobic nature, the study of the structure of the membrane sector of V-ATPases is lagging behind. A pioneering work of over ten years ago revealed the presence of a 16×10³M_r proteolipid in membranes containing V-ATPases (Sutton and Apps, 1981). This observation has been verified in every enzyme of eukaryotic cells studied so far (Nelson, 1989). The gene encoding this proteolipid has been cloned from a large variety of sources and its disruption in yeast cells has given a phenotype similar to that of mutants in which other genes encoding subunits have been interrupted (Nelson and Nelson, 1990). Therefore, there is no doubt that the proteolipid is an integral part of the membrane sector of V-ATPases. A stoichiometry of six proteolipids per enzyme has been reported for V-ATPases from clathrin-coated vesicles and plant vacuoles (Arai et al. 1988; Lai et al. 1991). An analogy to the membrane sector of F-ATPases suggests that additional subunits should function in the membrane sector of V-ATPases. Recently it was proposed that the 115×10³M_r subunit of V-ATPases functions as the counterpart of the proteolipid in proton conduction across the membrane (Perin et al. 1991). This subunit is present in most if not all V-ATPases of eukaryotic cells (Kane et al. 1992). Its presence in all V-ATPases is a prerequisite for assigning it as an integral functional subunit in the membrane sector. An equivalent subunit could not be detected in archaebacteria, and a highly purified and active enzyme from plant vacuoles has been shown to lack the 115×10³M_r polypeptide (Ward and Sze, 1992). Therefore, the admission of the 115×10³M_r polypeptide to the club of the integral V-ATPase membrane sector is contingent upon its presence in all enzymes, including those of plant sources. However, the detection of this subunit in the enzyme from yeast cells suggests an important role for the 115×10³M_r polypeptide in the correct assembly and proper functioning of the enzyme. Another polypeptide of 39×10³M_r has been detected in the membrane sector of V-ATPase from several mammalian sources (Cidon and Nelson, 1986; Arai et al. 1988). Cloning of the gene encoding this polypeptide revealed no potential transmembrane helices and, therefore, it may be associated with the membrane by binding onto another membrane protein (Wang et al. 1989). A highly hydrophobic protein of about 20×10³M_r has been detected in a few preparations of mammalian V-ATPases (Moriyama and Nelson, 1989b,c). Cloning of the gene encoding this polypeptide should clarify whether this protein is analogous to subunit a of F-ATPases. When the dust has settled, we anticipate a complicated membrane sector for V-ATPases with more subunit and accessory polypeptides.

The sequences of cDNAs and genes encoding subunits A and B of V-ATPases have shed light on the relationship between F-and V-ATPases and the evolution of these two families of proton pumps (Bowman et al. 1988a,c; Zimniak et al. 1988; Manolson et al. 1988). The A subunit of V-ATPases has been shown to be homologous to the /3 subunit of F-ATPases. The alignment of the two amino acid sequences created a gap of about 150 amino acids in the middle of the sequence of the β subunit. The origin and function of the non-homologous sequence in the middle of the A subunit are not known; it might have been removed during the evolution of the β subunit. The B subunit of V-ATPases has been shown to be homologous to the a subunit of F-ATPases. Several stretches of amino acids in the a and β subunits implicated in ATP and ligand binding were conserved in the corresponding B and A subunits of V-ATPases (Zimniak et al. 1988; Bowman et al. 1988a,c). It is assumed that the A subunit has a similar function to the β subunits and the B subunit has a similar function to the a subunit in the catalytic and regulatory activities, respectively, of these proton pumps. The amino acid sequence homologies suggest that the a, β, A and B subunits evolved from a common ancestral gene. The property that separates the F-and V-ATPases into two distinct families of proton pumps is the lack of any sequence homology among the C, E, γ, δ and ϵ subunits (Nelson et al. 1991). The possible presence of extra subunits in the catalytic sector of V-ATPases may further separate two families (Anraku et al. 1992 ).

The most likely time for the early evolution of the catalytic sector of F-ATPase from that of the V-ATPase was when oxygen rapidly accumulated in the atmosphere as a result of the proliferation of organisms capable of water photolysis (Nelson and Nelson, 1989). The amino acid compositions of V-ATPases suggest higher sensitivity to oxygen than their related F-ATPases. There are several tryptophan and cysteine residues positioned in catalytically sensitive places in V-ATPases. Most notable is a cysteine residue right inside the P-loop structure of the A subunit implicated in the binding of ATP (Zimniak et al. 1988). This cysteine is sensitive to oxidation and its modification by ligands such as N- ethylmaleimide (NEM) inactivates the enzyme. This cysteine is missing in the V₁ sector of archaebacteria, indicating the necessity of its removal for operating in an oxygenic environment (Denda et al. 1988). In eukaryotic cells, V₁ is present on the cytoplasmic side of the membranes that maintain a rather reduced environment.

The membrane sector of F-and V-ATPases followed a separate evolutionary pathway. So far we have sequence information only for the proteolipids, which bind dicyclohexylcarbodiimide (DCCD), of the two families of proton pumps (Mandel et al. 1988; Nelson and Nelson, 1989; Umemoto et al. 1990). It is likely that the ancestral proteolipid was of the short version (8×10³M_r) present in all F-ATPases and archaebacterial V-ATPases (Nelson, 1989). The proteolipid of eukaryotic V-ATPases evolved by gene duplication and fusion into a double-size proteolipid. The archaebacterial proteolipid may have diverged from the F-ATPase proteolipid prior to the gene duplication event that led to the present shape of the eukaryotic V-ATPases. The mechanistic implication of this event of gene duplication was far reaching; several specific properties of the eukaryotic V-ATPases may be correlated with the presence of a 16×10³M_r proteolipid in these enzymes.

Both families of F-and V-ATPases share several structural and mechanistic properties. They are composed of several subunits assembled into distinct catalytic and membrane sectors. The underlying mechanism of ATPase and proton pumping activities is quite similar for the members of the two families (Nelson, 1989). F-ATPases operate close to thermodynamic equilibrium and, depending on the level of the PMF, may operate in ATP synthesis or ATP-dependent proton pumping. In contrast, V-ATPases of eukaryotic cells operate exclusively as ATP-dependent proton pumps. Apparently these enzymes operate below thermodynamic equilibrium and therefore can no longer synthesize ATP at the expense of a PMF (Moriyama and Nelson, 1988). It was proposed that a proton slip (a voltage-dependent change in ATP/H⁺ stoichiometry) in eukaryotic V-ATPases rendered them incapable of ATP synthesis while gaining better control over the extent of the PMF in the various organelles. Can we find structural features that may be responsible for these properties of the different proton pumps? It has been proposed that the size of the proteolipid may be a major factor in transforming eukaryotic V-ATPases into exclusive proton pumps (Nelson, 1989). It has been demonstrated that all F-and V-ATPases capable of ATP synthesis contain a short proteolipid of about 8×10³M_r (Futai et al. 1989; Denda et al. 1989). We proposed that concomitantly with the introduction of V-ATPases to eukaryotic cells the gene encoding the proteolipid was duplicated and fused to give a protein of about 16×10³M_r that spans the membrane four times. This event completed the shaping of the eukaryotic V-ATPase and may have generated the proton slip. Recently, we observed that removal of the catalytic sector of eukaryotic V-ATPases from the membrane failed to generate increased proton conduction across the depleted membranes (Beltrán and Nelson, 1992). This observation contrasts with F₁-depleted membranes and archaebacterial V₁.depleted membranes, which exhibited a marked increase in proton conduction (Racker, 1976; Nelson, 1980). Fig. 1 illustrates the results of removing F₁ and V₁ from membranes of chloroplasts and chromaffin granules, respectively. Upon removal of V₁ from the chromaffin granule membrane, no increase in proton conduction could be observed for proton uptake in the presence of valinomycin in K⁺-loaded granules or for a proton leak in NH₄Cl-loaded vesicles (Beltrán and Nelson, 1992). In contrast, chloroplasts depleted of Fi by NaBr treatment exhibited proton conduction of over 10⁶ protons per second for each F_o in the depleted membranes (Nelson, 1980). Therefore, the capacity of an H⁺-ATPase to operate as an ATP synthase is positively correlated with the presence of a short proteolipid and the generation of a proton leak by the removal of the catalytic sector from the membrane. Moreover, both the catalytic and membrane sectors of eukaryotic V-ATPases are interdependent in their partial activity; the separated V₁ sector cannot catalyze the ATPase reaction (Moriyama and Nelson, 1989a). In contrast, the ATPase activities of F₁ and archaebacterial Vj are enhanced when the sectors are removed from the membranes (Racket, 1976; Lübben et al. 1987). I have proposed that all of these differential properties evolved concomitantly with the duplication of the proteolipid in eukaryotic V-ATPases (Nelson, 1989, 1992).

Studies of the yeast V-ATPase provided additional insight into the specific properties of the eukaryotic enzyme. Mutant strains with chromosomal disruptions of the genes encoding subunits A, B, C and E of V₁ and the proteolipid of V₀ exhibit a similar phenotype that cannot grow at a pH above 7 and is sensitive to low and high Ca²⁺ concentrations in the medium (Nelson and Nelson, 1990;Foury, 1990; Noumi et al. 1991; Ohya et al. 1991). It was proposed that these mutants survive the inactivation of V-ATPase because the acidification of the vacuolar system proceeds via fluid-phase endocytosis (Nelson and Nelson, 1990). Inactivation of the gene encoding the proteolipid prevented the assembly of the catalytic sector (Noumi et al. 1991). Inactivation of genes encoding subunits of the catalytic sector prevented the assembly of the other subunit of this sector, but the assembly of the membrane sector was not disrupted (Noumi et al. 1991; Beltrán et al. 1992; Kane et al. 1992). This observation supports the hypothesis that the removal of the catalytic sector does not cause a proton leak through the membrane sector. If the vacuolar membranes were to become permeable to protons, the cytoplasmic pH would drop to the external pH, killing the mutants. Therefore, the chromosomal disruptant yeast mutants add credence to the biochemical observations that the V₁ and V₀ sectors are interdependent in their partial activities.

The vacuolar H⁺-ATPase (V-ATPase) is one of the most fundamental enzymes in nature. Very little attention was given to this enzyme by the most prestigious scientific journals. Now that the enzyme has been implicated in human diseases this attitude will change drastically.