Evolution and isoforms of V-ATPase subunits

The V-ATPases that energize the endomembranes of eukaryotes have evolved from the archaebacterial coupling factor ATPases (Gogarten et al. 1989a; Kibak et al. 1992). These ATPases show a high degree of sequence similarity; in particular the A and B subunits are surprisingly conserved between the eukaryotic and the archaebacterial enzyme (about 50% identical amino acid residues), whereas both ATPase types show a much lower degree of similarity to the homologous F-ATPase subunits (α and β subunits, only about 25% identity). Based on this high degree of similarity one might be tempted to classify the archaebacterial coupling factor ATPase as a V-type ATPase. However, with regard to function, tertiary and quaternary structure, the archaebacterial ATPases appear to be more similar to the eubacterial F-ATPases. Therefore, in the following discussion the term V-ATPase will be used to denote the eukaryotic V-ATPases only; the archaebacterial ATPases will be abbreviated as A-ATPases. The whole group of homologous ATPases will be denoted as VFA-ATPases.

The catalytic subunits of VFA-ATPases are homologous (orthologous) to each other. In addition, the catalytic and non-catalytic subunits of these ATPases were also derived from a common ancestral gene, i.e. these subunits are also homologous to each other; however, their relatedness does not reflect the evolution of the species, but the ancient gene duplication event. This type of homology is termed paralogy (see Fig. 1).

An ATPase-like peptide has been found to be encoded as part of the Fla operon (open reading frame no. 4) of Bacillus subtilis (Galizzi et al. 1991). A thermosensitive mutant of the homologous flil gene in Salmonella typhimurium exhibits normal flagellar usage at the restrictive temperature; only the assembly of the flagella is interrupted in the mutant (Vogler et al. 1991 ). The two gene products from B. subtilis and S. typhimurium are 47% identical to each other and about 23% identical each to the catalytic and non-catalytic subunits of the VFA-ATPases; the ATPase-like flagellar proteins have about the same degree of similarity to the catalytic and non-catalytic subunits as the paralogous VFA-ATPase subunits have among themselves (see Fig. 3). On the basis of this similarity, it is appealing to assume that the ATPase-like flagellar protein is involved in the generation of local proton gradients. However, the phenotype of the thermosensitive mutants (see the above description) is more compatible with the role of this ATPase in the export of flagellum-specific proteins. A phylogenetic tree based on pairwise comparisons of the protein sequences is depicted in Fig. 2. The flagellar ATPase-like peptide joins the VFA-ATPase subunits at the innermost branch before the lines leading to the ATPases of the extant species separate from each other.

The ancient gene duplication that gave rise to the paralogous catalytic and non-catalytic subunits had already occurred in the last common ancestor of eubacteria, archaebacteria and eukaryotes (Gogarten et al. 1989a). Rooting the tree of life by means of this ancient gene duplication (see Fig. 1) shows the archaebacteria to be more closely related to the eukaryotes than to the eubacteria. The root is placed between eubacteria on one side and eukaryotes and archaebacteria on the other. Sequences of other archaebacteria and eukaryotic ATPase subunits have since been obtained (see Fig. 2). When these sequences were subjected to phylogenetic analyses, the results confirmed the above conclusion (Gogarten et al. 1989b; Ihara and Mukohata, 1991; Linkkila and Gogarten, 1991). Analyses of other duplicated genes (t-RNAmet, dehydrogenases, elongation factors; Iwabe et al. 1989; Cammarano et al. 1992) agree with the conclusion obtained for the ATPases: namely, all archaebacteria branch off from the line leading from the root to the eukaryotes (see Fig. 2).

The analysis of the ATPase sequences results in phylogenetic trees of the same topology as the ones that are obtained using other markers for the nuclear/cytoplasmic component (e.g. Iwabe et al. 1989; Cammarano et al. 1992). This confirms the usefulness of the V-ATPase subunits as a marker for the nuclear cytoplasmic component of the eukaryotic cell. Furthermore, it suggests that the ATPases energizing the eukaryotic endomembranes and the endomembrane system itself evolved within the cell that functioned as the host for the symbioses that evolved into the present-day eukaryotes, i.e. the endomembrane system itself is not due to another symbiotic event similar to the ones that gave rise to plastids and mitochondria. The evolution of V-ATPases from the coupling factor ATPase of archaebacterial plasma membrane furthermore suggests that the endomembrane system evolved from invaginations of the plasma membrane of the cell that contributed the bulk of the nuclear genome.

It was reported recently that the proton-pumping ATPase from Thermus aquaticus (=Thermus thermophilus strain HB8; Yokoyama et al. 1990) and the Na⁺-pumping ATPase from Enterococcus hirae (Kakinuma et al. 1991) are vacuolar-type ATPases. Based on rRNAs and other biochemical characters (cell wall, lipids, antibiotic resistance), the genus Thermus is classified as eubacterial, with closest similarities to the Deinococci (Woese, 1987; Hensel et al. 1986). However, note that the malate dehydrogenase from a Thermus species groups together with its eukaryotic, rather than its eubacterial, counterparts (Iwabe et al. 1989).

The recently published sequence of the A subunit of this enzyme (Tsutsumi et al. 1991) has 50 –55% identical amino acid residues when aligned to the A subunits of the eukaryotic V-ATPase and 53 –60% identical residues when compared to the various archaebacterial subunits. If this sequence is included in a phylogenetic reconstruction (Fig. 2), the Thermus sequence groups within the archaebacterial domain together with the subunits of Halobacterium and Methanosarcina. The same grouping is found for the B subunit (Fig. 2). The Thermus ATPase shares a long history of evolution with the archaebacterial ATPases, and it does not represent an ‘earlier’ V-ATPase branch. This grouping of the Thermus sequence within the archaebacterial domain contradicts scenarios that assume that both F- and V-type ATPases were already present in the last common ancestor and that one or the other was lost in the lines leading to the extant species. The grouping of the Thermus ATPase within the archaebacteria indicates a lateral gene transfer across domain boundaries between an archaebacterial-like organism and a eubacterium.

With respect to function and with respect to quaternary structure the archaebacterial coupling factor ATPase appears to be more similar to F-than to V-ATPases (Schafer et al. 1990). The switch in function from ATP synthase/ATPase (F-type and archaebacterial-type) to an exclusively proton-pumping ATPase (V-type) appears to be correlated with a change in subunit stoichiometry and an increased proteolipid size (Cross and Taiz, 1990; Gogarten and Taiz, 1992). The V-ATPase proteolipid is thought to have evolved by a gene duplication and fusion event from a smaller F-ATPase-like proteolipid (Mandel et al. 1988). The first two and the last two membrane-spanning helices are each similar to the F-ATPase proteolipid (Mandel et al. 1988). Using these front and back halves of the V-ATPase proteolipid as separate entries for phylogenetic reconstruction, the gene duplication that gave rise to these two parts is mapped to the early evolution of the eukaryotes (Fig. 3). The duplication occurred after the bifurcation of the Sulfolobus sequence and before the radiation of the eukaryotes (Fig. 3). In accordance with the finding that the carboxy terminal helix of the V-ATPase proteolipid is more intimately involved with proton translocation (the carboxylic acid residue thought to participate in the proton passage through the membrane, Glul39, is part of the fourth membrane-spanning helix; Mandel et al. 1988), the two amino terminal helices have changed much faster since the duplication event occurred. This high rate of change might reflect the changed, and probably relaxed, selection pressure acting on this portion of the proteolipid.

As discussed elsewhere in this volume, V-type ATPases were found to energize a variety of endomembrane-enclosed organelles in eukaryotes (Nelson, 1992), and in some cases these enzymes are also found to be responsible for proton secretion through the plasma membrane (Gluck, 1992; Chatterjee et al. 1992). Gene disruption experiments in yeast that led to a complete loss of V-ATPase activity gave no indications for multiple isoforms in Saccharomyces cerevisiae (e.g. Umemoto et al. 1990; Nuomi et al. 1991; Foury, 1990); also, in other fungi only one gene per subunit and per genome has been identified.

In the case of humans and in higher plants, different genes encoding the same subunit type have been reported. Sudhoff et al. (1989) reported a cDNA from human kidney encoding the V-ATPase B subunit. A different, partial cDNA encoding the same subunit type has been obtained from hippocampal mRNA (Bernasconi et al. 1990). The part of the encoded proteins for which both sequences have been determined is 91% identical on the amino acid level. The two isoforms are much more similar to each other than either is to the homologous subunit from plants or fungi. 75 –80% identical amino acid residues are found for the comparisons between each of the two human isoforms and the sequences from Arabidopsis, Neurospora, Candida and Saccharomyces. This suggests that the two isoforms observed in humans evolved during the evolution of the metazoans but that these, probably tissue-specific, isoforms are not characteristic of multicellular eukaryotes in general. In contrast to these reports demonstrating different isoforms for the non-catalytic subunit, Puopolo et al. (1991) reported that only a single gene encodes the catalytic A subunit in cattle. From these findings it appears that, if different V-ATPase isoforms exist in higher animals, then some subunits can be shared between the different isoforms.

In carrots at least some of the different endomembranes appear to be energized by different V-ATPase isoforms. Carrot cells were transformed via Agrobacterium- mediated gene transfer with genes that expressed an antisense mRNA to the A subunit of the vacuolar ATPase (Gogarten et al. 1992). Constructs utilizing either the coding or the 5 ′ noncoding region gave the same surprising result: in contrast to the various yeast mutants lacking V-ATPase activity (Nuomi et al. 1991), the transformed plant cells grew without noticeable impairment as callus cultures; plants expressing the antisense constructs were readily regenerated from these cultures.

These transgenic carrot cells still contain vacuolar-type ATPase activity (measured as bafilomycin-sensitive ATPase activity and immuno-detectable A subunit), but the level of V-ATPase at the tonoplast (the membrane surrounding the central vacuole) is reduced by more than 70%. The reason that the cells and plants regenerated from these cells grow without noticeable impairment is probably because of the presence of a proton-pumping pyrophosphatase that can provide at least some energization of the tonoplast. However, hexose transport into the vacuoles of these transgenic plant cells, a process that apparently depends on an acidified vacuole (Rausch et al. 1987), is greatly impaired (Gogarten et al. 1992; J. P. Gogarten, unpublished results). Table 1 depicts results from 3-O-methyl [U-^l4C]glucose (3-OMG) efflux experiments. In the control transformants the bulk of 3-OMG is located inside the vacuole after 20 h of incubation, whereas in the V-ATPase-deficient line most of the 3-OMG remains in the cytoplasm. The higher unidirectional flux from the cytoplasm to the outside in the latter case is due to transstimulation of this flux by the higher external 3-OMG concentration (Gogarten and Bentrup, 1989). Transport across the tonoplast is greatly reduced, whereas the cytoplasmic concentrations remain nearly unchanged. Note that in both cases the main accumulation occurs across the plasmalemma; the amount of 3-OMG in the vacuole of the controls is larger than the amount in the cytoplasm, but the vacuole has about 10 times the volume of the cytoplasm.

Surprisingly, the V-ATPase activity in Golgi-enriched microsomal fractions was not inhibited but stimulated in both types of antisense-expressing cells. In both cases, the bafilomycin-inhibitable ATPase activity increased by about 30% in the Golgi-enriched fraction.

Western blots of two-dimensional gels probed with a monospecific polyclonal antiserum directed against the A subunit revealed two cross-reacting A subunit types with different isoelectric points. The concentration of the more acidic isoform did not change in microsomes isolated from the antisense-expressing cell lines, whereas the concentration of the more alkaline form was greatly reduced. This more alkaline isoform was the only isoform that was detectable in vacuole preparations from control plants.

In higher plants, two genes encoding the A subunit can be characterized by the size of an intervening sequence (Starke et al. 1991; T. Starke and J. P. Gogarten, unpublished results). This intron is surprisingly conserved and its unequal size in the two isoform-encoding genes allows their easy detection using the polymerase chain reaction (PCR) across the exon-intron boundaries. This intron and 90 bp of the surrounding exon region were used to investigate the origin and evolution of land plants. The analysis of the PCR product revealed that the A subunit is encoded by at least two genes in all vascular plants except in Arabidopsis thalliana. In addition, all fragments amplified from 13 different land plants (with the exception of the horn moss Anthoceros) show the presence of the intervening sequence. This intron is surprisingly conserved throughout at least 400 million years of evolution, suggesting the presence of a functional constraint on the intron sequence. This intervening sequence is also found in Coleochaete scutata, a green alga that is considered to be closely related to the higher plants, whereas other green algae (e.g. Zygnema), various fungi and Euglena have no intron in the corresponding position. These findings suggest that this intron was already present in the ancestor of the land plants. Phylogenies inferred from these intron/exon sequences (see Fig. 4) are in general agreement with phylogenies based on other characters. The duplication that gave rise to the two isoforms, distinguished by the differently sized introns, occurred early in the evolution of land plants. Analyses using parsimony, maximum likelihood and evolutionary parsimony methods point to two duplication events that occurred independently in the lineages leading to the fems and to the seed plants. However, it is likely that the greater similarity of the two isoforms within these two lineages is due to a gene conversion event that occurred in one of the lineages shortly after a unique duplication event that gave rise to the two isoforms.

These two isoforms are not a characteristic of higher eukaryotes in general; their presence appears to be restricted to higher land plants.

Research in J.P.G.’s laboratory was funded by NSF grant BSR-9020868.