in 1987, five human genes related to the Na-K-ATPase gene family were isolated using an Na-K-ATPase α-subunit cDNA probe (53, 72, 75). Three of them encode Na-K-ATPase α-subunit isoforms. A fourth one encodes an H-K-ATPase α-subunit that has been isolated from rat colonic, amphibian toad bladder, or human skin cDNA libraries. Further cloning experiments allowed the identification of species variants in guinea pig and rabbit. The functional properties of these nongastric H-K-ATPases have been analyzed in expression systems such as the Xenopus laevis oocyte, the insect Sf9 cells, and the mammalian HEK 293 cells. Excellent reviews have been published recently on the subject (46, 81). The purpose of the present review is to describe our current knowledge of the molecular characteristics of these H-K-ATPase α-subunits. Functional properties will be reported, on the basis of studies performed using exogenous expression in functional expression systems. These data indicate that the nongastric H-K-ATPase may be identified as a subgroup of the Na-K/H-K-ATPase gene family, apart from the Na-K-ATPase and the gastric H-K-ATPase subgroups. Comparison of the functional data with the physiological data reviewed in the accompanying paper by Silver and Soleimani (73a) reveals some discrepancies that will be discussed.

THE NA-K/H-K-ATPASE GENE FAMILY

Classification

P-ATPases are membrane ATPases involved in ion transport (33, 37, 52). This large family includes among others the sarcoplasmic and endoplasmic reticulum Ca-ATPases (SERCA), the membrane Ca-ATPases (PMCA), and the various members of the Na-K/H-K-ATPase group (37). On the basis of the molecular and functional properties of each members, this last group could be subdivided into three different subgroups: the Na-K-ATPase, the gastric H-K-ATPase, and the so-called nongastric H-K-ATPase (Fig.1A). Unique functional properties of the nongastric H-K-ATPases have been demonstrated compared with the Na-K-ATPase and the gastric H-K-ATPase groups, suggesting clearly that the nongastric H-K-ATPases do not functionally belong to the Na-K-ATPase and the gastric H-K-ATPase groups.

Among each subgroup, several subunits have been characterized so far (Fig. 1B). For the Na-K-ATPase, these different α-subunits can be identified as isoforms since1) different genes have been characterized, 2) subunit-specific protein domains have been identified, and3) subunit-specific functional properties have been observed. For the gastric H-K-ATPase subgroup, only one α-subunit has been characterized in various species. It is referred to as HK1.

For the nongastric H-K-ATPases, which are the focus of the present review, several species variants have been cloned. In the currently accepted nomenclature (46, 81), the rat colonic H-K-ATPase is referred to as HK2, the toad bladder H-K-ATPase as HK3, and the human H-K-ATPase (ATP1AL1) as HK4. The recently characterized guinea pig colonic H-K-ATPase (GenBank accession no. D21854) would have to be referred to as HK5 and the one of the rabbit (31) as HK6. By analogy with the Na-K-ATPase α-subunit classification, this nomenclature suggests that the rat, human, toad, and guinea pig α-subunits are subunit isoforms. In the currently accepted nomenclature, the nongastric subunits are also considered to be isoforms of the gastric H-K-ATPase α-subunit (HK1). However, it seems difficult to make such analogies between the H-K-ATPases, including gastric and nongastric H-K-ATPases, and the Na-K-ATPases for which true isoforms have been identified, both at the molecular and functional levels. The molecular and functional data reported in the literature and summarized in the present review indicate that it may be necessary to change the nomenclature in the future and more clearly distinguish the H-K-ATPase expressed in the stomach from those expressed in other organs.

Relationship Between Nongastric H-K-ATPases

Interestingly enough, as already observed by Kone (46), the amino acid homologies between the rat HK2 and the human HK4 are significantly lower (86%) than between the rat HK1 and the human HK1 (98%) or the rat NK1, NK2, or NK3 and the human NK1, NK2, or NK3 (97%, 99%, and 99%, respectively). This also appears clearly on Fig.1B, representing the phylogenetic tree of the α-subunits of the Na-K/H-K gene family. Divergence between rat HK2 and human HK4 is similar to the divergences noted between the Na-K-ATPase α-subunit isoforms, NK1, NK2, and NK3 (Fig.1B). One interpretation may be that the rat HK2 and the human HK4 are true isoforms rather than species variants. By contrast, Sverdlov and collaborators (74) reported high homologies in the 3′-untranslated region of both the human ATP1AL1 (HK4) and the rat colonic H-K-ATPase (HK2), suggesting a possible close relationship of the two genes. Another explanation may be that common posttranscriptional regulatory mechanisms interact with the 3′-untranslated region of these genes.

At the functional level, characterizations of the rat, human, toad, and guinea pig α-subunits do not reveal strong differences. The only functional difference observed so far deals with the ouabain sensitivity of the nongastric H-K-ATPases (see below). It should be noted, however, that the difference between rat HK2 and human HK4 is similar to that observed between rat NK1 and human NK1, two subunits that are not considered as isoforms. Therefore the current molecular and functional data indicate that the relationship between the various members of the nongastric H-K-ATPases remains unclear and needs to be directly addressed. As discussed below, molecular cloning of new members in the same species and/or comparative functional analysis will certainly help to clarify this point.

THE NONGASTRIC H-K-ATPASES: MOLECULAR CHARACTERIZATION

Nongastric H-K-ATPase α-Subunits

To date, nongastric H-K-ATPase α-subunits have been characterized in five different species: human (62), rat (24), guinea pig (GenBank accession no. D21854), rabbit (31), and toad (39). Sequence homologies clearly indicate that they all belong to the P-type ATPase family with highly conserved domains such as the ATP-binding domain and the FITC-binding site. Hydropathy plot analysis suggests that these α-subunits have 10 transmembrane domains, with the NH₂ terminus and COOH terminus being located inside the cell.

The extensive description of the amino acid homologies between the subgroups is beyond the scope of the present review. Structure-function relationship analysis of the nongastric H-K-ATPase α-subunit has not been performed so far. However, on the basis of previous studies performed on the Na-K-ATPase and gastric H-K-ATPase subgroups, several domains of the nongastric H-K-ATPase α-subunits could be proposed to be important to support the function of these pumps. As a basis for future structure-function relationship studies, we have included in the present review part of the amino acid comparisons between the members of the Na-K/H-K-ATPase α-subunit gene family (Fig. 2). It is interesting to note that several specific amino acid domains exist. They may support functional specificities between each subgroup members. Some of these domains that appear to be of particular interest will be described.

The intracellular NH₂-terminal domain.

Truncation of the NH₂-terminal domain of the Na-K-ATPase has been reported to induce modifications in the pump cycle, resulting in a lower apparent affinity for K (25, 80). This region may act as a cation-selective gate (37). Kone and Higham (47) recently reported the existence of two splice variants of the rat colonic H-K-ATPase that differ in the NH₂-terminal domain of the Na-K-ATPase. The HKα2b subunit has a truncated NH₂-terminal domain, compared with the HKα2a subunit. It is not known whether the colonic HKα2b functionally differs from the full-length HKα2a, since expression studies in the nonpolarized HEK 293 cells did not examine this point.

Ouabain and Sch-28080 binding sites.

Ouabain and Sch-28080 compounds are thought to be specific inhibitors of the Na-K/H-K-ATPases. The Na-K-ATPase is inhibited by ouabain and other digitalic glycosides but is insensitive to Sch-28080 (or omeprazole, an inhibitor of the gastric H-K-ATPase used for the treatment of peptidic ulcer), whereas the gastric H-K-ATPase is inhibited by both Sch-28080 and omeprazole, but not by ouabain. The nongastric H-K-ATPases have been reported to be sensitive to ouabain and, for some of them, to high concentrations of Sch-28080 (see below). Structure-function analyses have been extensively performed on the Na-K-ATPase and the gastric H-K-ATPase α-subunits. Amino acid sequence homologies/differences between the nongastric H-K-ATPases and the other members of the Na-K/H-K-ATPase gene family may be useful to delineate the ouabain/Sch-28080 binding sites on the nongastric H-K-ATPase α-subunits.

The first putative transmembrane domain has been involved in both ouabain and Sch-28080 sensitivity for the Na-K-ATPase and the gastric H-K-ATPase, respectively (37, 52, 55). Interestingly, the tyrosine residue Tyr¹³⁴, which has been involved in the ouabain sensitivity of the Na-K-ATPase (15, 52), appears to be common to the ouabain-sensitive Na-K-ATPases and H-K-ATPases but not to the ouabain-insensitive gastric H-K-ATPase and the moderately sensitive rat colonic H-K-ATPase.

The VAAA/V domain of the gastric H-K-ATPase in the H1 domain clearly differs from the corresponding one in the other subgroups. This domain has been involved in the Sch-28080 sensitivity of the gastric H-K-ATPase (55). The extracellular domain located between H1 and H2 has been involved in both ouabain and Sch-28080 binding on the Na-K-ATPase and the gastric H-K-ATPase, respectively (52, 68). This domain differs between each subgroup, but any consensus sequences could be identified for each one.

Other amino acids, located in the COOH-terminal half of both the Na-K-ATPase and the gastric H-K-ATPase, appear to participate to the ouabain or Sch-28080 binding pocket. Using a random mutagenesis approach, Burns and Price (14) reported that the mutation of the Thr⁸²⁵ residue to an Asn residue shifts the ouabain sensitivity of the mutated pump by two order of magnitude. This Thr residue is conserved in all the Na-K-ATPase and nongastric H-K-ATPase α-subunits, whereas it is replaced by a Cys residue in the gastric H-K-ATPase α-subunits. Other amino acids located in the H5, H5-H6, or H6 domains appear to be involved in glycoside sensitivity of the Na-K-ATPase. With the exception of Phe⁸¹⁴, which is replaced by a Tyr residue in the gastric and nongastric H-K-ATPases, the others are well conserved. Glu⁸³², located in the H6 domain, has been involved in Sch-28080 binding to the gastric H-K-ATPase (3). This amino acid is replaced by an Asp residue in both the Sch-28080-insensitive Na-K-ATPase and nongastric H-K-ATPase α-subunits.

Most of the mutations that affect ouabain or Sch-28080 binding also affect the pump cycle, lowering the apparent affinity of the K-ATPases for K. This indicates that K binding interferes with ouabain or Sch-28080 binding, resulting in the well-known K dependence of the inhibitory constants of these inhibitors (37).

As discussed below, the ouabain and Sch-28080 sensitivities slightly differ among the various members of the nongastric H-K-ATPase subgroups. Structure-function analyses have not been performed explain these differences.

Putative cation binding sites.

It is proposed that cation specificity resides in (or is highly dependent on) the structure of the transmembrane domains, which could form some kind of channel or pocket for the cations to be exchanged between the extracellular and the intracellular media (52, 79). Amino acid comparison indicates that there are more homologies between the nongastric H-K-ATPases and the Na-K-ATPases than homologies between the nongastric H-K-ATPases and the gastric H-K-ATPases. Specific amino acid domains of the gastric H-K-ATPase subgroup could be identified (Fig.2). For example, the LQC domain of the gastric H-K-ATPase that is located in H1 is gastric H-K-ATPase specific, as well as Met¹²³ and Ala¹²⁶. Ala¹²⁶ (Glu in the Na-K-ATPases and nongastric H-K-ATPases) has been involved in the K affinity of the gastric H-K-ATPase (3). The tryptophan residue (Trp¹³¹) appears to be specific for the nongastric H-K group (Phe for Na-K-ATPase, Leu for gastric H-K-ATPase).

The H4 putative transmembrane domain has been involved in the cation specificity of the gastric H-K-ATPase compared with the Na-K-ATPase. The Asn³⁵¹ residue in the H4 putative transmembrane domain plays an important role in determining the Na affinity of the Na-K pump (78). The gastric H-K-ATPase counterpart is a tyrosine residue, whereas the asparagine residue is conserved in the nongastric H-K-ATPase subgroup. The overall sequence homologies between the Na-K and nongastric H-K-ATPase subgroups may account for the functional similarities between these two pumps (see below). Two exceptions are Ile³⁴¹replaced by valine in the Na-K-ATPase and methionine in the gastric H-K-ATPase and the Thr³⁶³ replaced by a cysteine in both the Na-K and H-K-ATPases. These amino acids are therefore specific for the nongastric H-K-ATPase subgroup.

Amino acid residues located in the H4 putative transmembrane domain that differ between the Na-K-ATPase α-subunit and the gastric H-K-ATPase-specific amino acids have been recently mutated by Mense and collaborators (60). Specific amino acids for the gastric H-K subgroup (Phe³⁴⁴, Tyr³⁵¹, Ser³⁶⁵) have been introduced instead of the corresponding residues present in the Na-K-ATPases and generally conserved in the nongastric H-K-ATPases. Data obtained with this elegant mutagenesis approach clearly indicate that these amino acids participate in the cation transport selectivity since the mutated Na-K-ATPase has an apparent K affinity different from the wild-type form. Moreover, the results obtained strongly suggest that these changes allow protons to substitute for sodium ions in the mutated pump (60). That the amino acids mutated in this study are almost fully conserved between the Na-K-ATPase and the nongastric H-K-ATPase may explain functional homologies between members of this two subgroups (see below).

The H5 and H6 transmembrane domains also appear interesting (Fig. 2). For example, Ser⁸⁰³, which has been proposed to be involved in K binding during K translocation (2), is specific to the Na-K-ATPase group, but is replaced by a charged lysine residue in both the gastric H-K-ATPase and the nongastric H-K-ATPases. Glu⁸³² of the gastric H-K-ATPase is involved in both K affinity and Sch-28080 binding (3). A recent study indicates that this negatively charged amino acid inhibits the dephosphorylation process. Interaction of K with Glu⁸³² (neutralizing the negative charge of Glu⁸³²) allows the cycle to proceed (76). This amino acid is replaced by an aspartic acid residue in both the Na-K and the nongastric H-K-ATPase α-subunits. Indeed, Asp⁸³² and Asp⁸³⁶ in the Na-K-ATPase have been involved as cation coordinating residues (49), emphasizing the importance of these residues in the K-ATPase pump cycle.

The H8 and H9 putative transmembrane domains appear to be the most divergent between each subgroup. Several domains appear to be quite specific for each subgroup and may reflect functional specificities. Mutagenesis studies suggest that some amino acid residues within these putative transmembrane domains of the Na-K-ATPase represent candidates to form part of the cation binding pocket (44). Vilsen and collaborators (79) have recently proposed a working model for the cation binding pocket. They proposed that Glu³⁵⁴ is closely associated with the gating mechanism at the cytoplasmic entrance of the cation binding pocket in the E1P form, whereas Glu⁸⁰⁷ is important in Na binding in E1 form, as well as Thr⁸³⁵. Tyr⁷⁹⁹ affects Na affinity, probably due to its cytoplasmic position in E1 form. Asp⁸³² is important for K affinity, probably due its extracellular position in the E2P conformation.

Sorting signals.

The H4 putative transmembrane domain of the gastric H-K-ATPase has been recently reported to act as a dominant sorting signal. When the gastric H-K-ATPase H4 domain is transferred into an Na-K-ATPase chimera, it allows targeting of the mutant pump to the apical membrane rather than to the basolateral membrane of LLC-PK₁ transfected cells (28). The nongastric H-K-ATPases are believed to be targeted to the apical membrane (see below). However, the H4 transmembrane domain of the nongastric H-K-ATPases appears to be homologous to the Na-K-ATPase rather than to the gastric H-K-ATPase. Functional expression experiments in a polarized cell line are required to address this question. Indeed, recent data obtained in our laboratory suggest that, when it is expressed in a rat cortical collecting duct cell line, the rat colonic HKα2a subunit is targeted to the basolateral domain (Jaisser and Beggah, unpublished results).

The putative α/β interaction domain.

The extracellular segment between amino acid Asn⁹¹⁷ and Ala⁹⁴² has been reported to be involved in the interaction between the α- and β-subunits of the Na-K-ATPase (51). For the gastric H-K-ATPase, similar function was assigned to the extracellular segment between Arg⁹⁰⁸ and Arg⁹³² (59). Therefore the minimal α/β interaction domain may reside between amino acid residues 917 and 932. Recently, using the two-hybrid system and the alanine scanning mutagenesis approach, the results obtained in the laboratory of D. Fambrough indicate that the highly conserved SYGQ motif was required for α/β interactions (20). Whether the specific amino acids for each of the subgroups (like Glu⁹²⁰, Tyr⁹²⁹, Glu⁹³⁰) are important for specific α/β interactions remains to be analyzed. As discussed below, when expressed in heterologous expression systems, the nongastric H-K-ATPase α-subunits are able to functionally associate with either the Na-K-ATPase β1-subunit, the Bufo marinus bladder β-subunit, or the gastric H-K-ATPase β-subunit. By contrast, the gastric H-K-ATPase α-subunit appears to require specific interaction with its physiological counterparts for functional expression (34, 58), whereas the Na-K-ATPase is more permissive (41, 43). Functional expression of mutated or chimeric subunits would be interesting to determine whether in vivo interactions of the α-subunit of each subgroup with a specific β-subunit depend on the amino acid composition of this extracellular domain.

Putative phosphorylation sites.

Protein kinase C (PKC) phosphorylation sites have been identified in the NH₂-terminal domain of the Na-K-ATPase α1-subunit (6, 8, 32). Phosphorylation may also occur (but at lower level) in the rat axolemma α3-subunit (32). Similarly, putative PKA and PKC phosphorylation motifs are found in the rat colonic H-K-ATPase α-subunit (47).

PKA phosphorylation has been consistently reported for the Na-K-ATPase α1-subunit on residue Ser⁹⁶⁴ (6,8, 9). The consensus PKA phosphorylation site (RR/KXS) is conserved between the Na-K and nongastric H-K-ATPase α-subunits. The residue X is Asn in both the Na-K and nongastric H-K-ATPase α-subunits, whereas it is replaced by Leu in the gastric H-K-ATPase group. It is not known whether the nongastric H-K-ATPases are regulated by PKA phosphorylation.

H-K-ATPase β-Subunits

The P-ATPases of the Na-K/H-K-ATPase group are heterodimers formed by α- and β-subunits. It is generally accepted that the natural heterodimer is an α2β2 complex, composed of two α- and two β-subunits (37). Several biochemical and functional evidences (see below) indicate that the nongastric H-K-ATPases are also heterodimers that consist of α- and β-subunits. Which β-subunit is associated in vivo with the cloned nongastric H-K-ATPase α-subunits remains under debate (see below).

Various P-ATPase β-subunits have been characterized (37): the β-subunit of the gastric H-K-ATPase; the β-subunits of the Na-K-ATPase that include β1 and β2 (also called AMOG, for “adhesion molecule on glia”); a β3-subunit, first characterized in amphibian and recently isolated in human (56) and in the rat (69); and a β-subunit isolated from the toad urinary bladder that may be the amphibian homolog of the mammalian Na-K-ATPase β2-subunit (40).

Immunoprecipitation using β-subunit-specific antibodies (73) indicates that the Na-K-ATPase β1- and β2-subunits interact in vivo with the α1- and α3-subunits of the Na-K-ATPase, respectively, as does the gastric H-K-ATPase β-subunit with the α-subunit of the gastric H-K-ATPase (17). Using a rat colonic H-K-ATPase α-subunit-specific antibody, it has been recently reported that the colonic H-K-ATPase α-subunit physically associates with a so-called β1-subunit of the Na-K-ATPase in a renal medullary membrane preparation (18, 48) and in membrane preparation from the distal colon (18). This indicates that the β1-subunit of the Na-K-ATPase or a β1-related isoform may associate in vivo with the colonic H-K-ATPase α-subunit in the kidney and in the distal colon.

The colonic H-K-ATPase α-subunit has been reported to be expressed in the apical membrane of the distal colon and the principal cells of the collecting duct (50). This may be relevant to the finding of Marxer et al. (57) that described a β1-related protein in the apical membrane of the rat distal colon. Whether this β1-related protein is the known Na-K-ATPase β1-subunit or a β-subunit isoform sharing a common epitope with Na-K-ATPase β1 remains to be established. Indeed, a so-called β3-subunit has been recently isolated in the rat distal colon (69). It is only 35–50% homologous to the previously characterized rat P-ATPase β-subunits (69). Its colonic expression is upregulated by K depletion. This β3-subunit physically interacts with the colonic H-K-ATPase α-subunit (70). Therefore, the rat colonic β-subunit recently cloned by the group of H. Binder may be the proper β-subunit that functionally associates in vivo in the distal colon with the colonic H-K-ATPase α-subunit in case of K depletion. The respective roles and properties of the HK2/β3 vs. HK2/β1 heterodimers remain to be determined.

THE NONGASTRIC H-K-ATPASES: FUNCTIONAL CHARACTERIZATION

To analyze the functional properties of the nongastric H-K-ATPases that have been characterized so far and, subsequently, to understand their physiological roles in ion homeostasis, functional expression of the α-subunits of the nongastric H-K-ATPases has been done in various expression systems, including the X. laevis oocyte, the insect Sf9 cell line, and the nonpolarized human cell line HEK 293.

Functional Expression in X. laevis Oocytes

The first H-K-ATPase α-subunit to be expressed inXenopus oocytes was the α-subunit isolated from the amphibian toad urinary bladder, a functional equivalent of the mammalian renal cortical collecting tubule. K transport capacity was evaluated by measuring influx of the radioisotope ⁸⁶Rb (as a tracer of K ion transport) in Xenopus oocytes expressing various combinations of P-ATPase α- and β-subunits (39). Proton transport was evaluated by measuring either steady-state intracellular pH in oocytes or external pH in the surrounding medium. The results indicated that the toad bladder P-ATPase works as an H-K-ATPase when expressed in Xenopusoocytes (39). K_mfor external K was ∼300 μM. Functional expression resulted in a strong intracellular alkalinization of the oocytes. Because of technical constraints, it was not possible to evaluate the stoichiometry of K vs. H transport. This H-K-ATPase had a novel pharmacological profile, since it was both moderately ouabain sensitive (K_i = 50 μM in the presence of 200 μM external K) and poorly Sch-28080 sensitive (K_i = 500 μM in the presence of 200 μM external K). The bladder H-K-ATPase was therefore the first ouabain-sensitive H-K-ATPase to be described. This pharmacological profile clearly correlates the amino acid homologies of the nongastric H-K-ATPases with both the Na-K-ATPase and the gastric H-K-ATPase, opening the way to the interesting dissection of the structure and the function relationship for ion or inhibitors selectivities.

The second nongastric H-K-ATPase that was functionally expressed inXenopus oocyte was the human skin P-ATPase. This P-ATPase α-subunit is encoded by the human ATP1AL1 gene (62). Its functional properties are roughly identical to the amphibian one, i.e., it encodes a moderately ouabain-sensitive, poorly Sch-28080-sensitive H-K-ATPase. This emphasizes the high conservation of the properties of the nongastric H-K-ATPases during evolution, a feature usually considered to reflect an important and specialized role in organism homeostasis.

Functional expression of the rat colonic P-ATPase inXenopus oocytes has been reported by two independent groups (Fig. 3). This pump is a moderately ouabain-sensitive, Sch-28080-insensitive K-ATPase (19, 23). Proton transport capacity was evaluated by Cougnon et al. (23). The authors showed that this pump could be considered as an H-K-ATPase, allowing K-dependent, ouabain-sensitive internal pH alkalinization.

The two groups also evaluated the influence of various P-type ATPase β-subunits on the functional properties of the rat colonic H-K-ATPase. Codina et al. (19) showed that this α-subunit could indifferently associate with the rat Na-K-ATPase β1- and the rat gastric H-K β-subunit. Their functional properties are not affected by the Na-K-ATPase β1- or the gastric H-K-ATPase β-subunits (19) nor by the amphibian Na-K-ATPase β3- or bladder P-ATPase β-subunits (Cougnon et al., unpublished results). These results differ with what has been previously reported for the Na-K-ATPase. The Na-K-ATPase β-subunit has been shown to be involved in the pump cycle through its interaction with the α-subunit (12, 29, 30, 41). However, one should keep in mind that the oocyte system may be inappropriate to detect subtle differences related to pharmacological properties (see below). These data also indicate that in such an expression system, heterologous α/β association are possible, a feature that may reflect a permissive α/β interaction. It would be interesting to compare the functional properties of the colonic HK2/rat β1-subunits and the colonic HK2/rat β3-subunits. Putative distinct properties may explain the functional differences that have been reported about K-ATPase activities in the colon and in the kidney (see Table 1).

More recently, Cougnon and coworkers (21) reported evidences in support of a yet unrecognized functional property that appears to be common to all nongastric H-K-ATPases. They measured the intracellular Na (Na_i) activity in steady-state conditions in Xenopus oocytes expressing various subunits of the Na-K/H-K-ATPase gene family (21). As anticipated, steady-state Na_i was markedly decreased in oocytes expressing theB. marinus Na-K-ATPase compared with oocytes expressing the gastric H-K-ATPase or a β-subunit alone (Fig.3). Interestingly enough, a similar decrease in Na_i was also observed when the rat colonic H-K-ATPase was expressed, regardless of the coexpressed β-subunit (21). This effect on Na_i depends on the presence of extracellular K. Due to electrochemical concerns, it was shown that this effect relies to an active transport process. Ouabain, 2 mM, increased Na_i value inXenopus oocytes expressing the rat colonic H-K-ATPase or the B. marinusNa-K-ATPase, indicating that Na transport was ouabain sensitive (21). These data support the hypothesis that the nongastric H-K-ATPase may have a direct role in Na⁺extrusion and strongly suggest that the rat colonic H-K-ATPase is indeed a (Na/H)-K-ATPase, when expressed inXenopus oocytes. It should be emphasized that this “Na transport” capacity occurs in “physiological” experimental conditions and that this finding is specific for the nongastric H-K-ATPase and not for the canonical gastric H-K-ATPase functionally tested in the same experimental conditions. This clearly differs with the “Na transport” capacity of the gastric H-K-ATPase or the “proton transport” capacity of the Na-K-ATPase reported to occur in very special experimental conditions (65, 66).

Interestingly, this property appears to be common to the nongastric H-K-ATPase subfamily, i.e., the rat colonic H-K-ATPase (21), theB. marinus bladder H-K-ATPase (21), and the human ATP1AL1 H-K-ATPase (35, 36). Whether Na may be transported in vivo in exchange for K by the corresponding K-ATPases remains to be examined.

Recent data collected by Cougnon and collaborators (22) indicate that understanding the physiological properties of the rat colonic H-K-ATPase (and probably those of the other nongastric H-K-ATPases) is more difficult than anticipated. Using the well-characterized oocytes as an expression system, these authors report that ammonium ions can also be transported by the rat colonic H-K-ATPase in “standard” experimental conditions (22). Ammonium competes with potassium for transport by the colonic H-K-ATPase. Similar to K⁺, ${NH}_4^{+}$appears to be exchanged for proton or Na⁺, since, in the absence of K⁺ but the presence of ammonium in the extracellular medium, proton or Na⁺ could be extruded from the oocyte intracellular compartment. Ammonium transport is also sensitive to ouabain, similar to that of K⁺, proton, or Na⁺transports.

Ammonium transport by the colonic H-K-ATPase may be of physiological relevance in the kidney medulla in certain physiopathological conditions. Interestingly, these results contrast with the data reported recently indicating that, in K⁺ depletion, the colonic H-K-ATPase can function in${NH}_4^{+}$/K⁺exchange mode (1). Indeed, Silver and Soleimani (73a) propose in the accompanying review that the colonic H-K-ATPase could mediate the transport of intracellular ${NH}_4^{+}$in exchange for luminal K⁺ in the case of potassium depletion, mediating secretion of${NH}_4^{+}$ in the lumen. One possible explanation for this apparent discrepancy may be that, in the inner medullary collecting duct, the rat colonic H-K-ATPase is indeed targeted to the basolateral membrane, instead of the apical membrane. If ammonium is transported in place of K⁺, then this would result in a net secretion of ammonium from the interstitium to the lumen.

Functional Expression in Sf9 Insect Cells

The rat colonic HKα2a subunit has been expressed in the insect Sf9 cell line using the baculovirus expression system. Lee and collaborators (50) reported that this α-subunit encodes for an Sch-28080- and ouabain-resistant K-ATPase. It should be noted that the sole α-subunit was expressed without coexpression of an exogenous β-subunit. An association with a presently unrecognized endogenous β-subunit cannot be excluded. On the other hand, expression of α/α homodimers of the Na-K-ATPase α-subunit has been previously reported in Sf9 cells (10, 11). This leads to a functional ATPase with peculiar properties different from those of the natural αβ heterodimer (10, 11).

Functional Expression in HEK 293 Cells

Functional expression of the human ATP1AL1 H-K-ATPase, the rat colonic H-K-ATPase, and the guinea pig H-K-ATPase has been reported in HEK 293 cells, a nonpolarized cell line derived from the embryonic kidney (4,35, 47). The human ATP1AL1 H-K-ATPase requires coexpression of α- and β-subunits for optimal activity. It encodes a ouabain-sensitive, poorly Sch-28080-sensitive H-K-ATPase (35). Kone and Higham (47) reported that the rat colonic H-K-ATPase allows ouabain-sensitive, Sch-28080-insensitive K transport when the α2b variant is expressed in HEK 293 cells. Functional expression of the guinea pig nongastric H-K-ATPase α-subunit indicates that this pump also allows ouabain-sensitive, Sch-28080-insensitive K-ATPase activities, when coexpressed in HEK 293 cells with the gastric H-K-ATPase β-subunit (4). Interestingly, these authors confirm that these pumps could associate indiscriminately with the Na-K-ATPase β1-subunit or the gastric H-K-ATPase β-subunit (4).

Grishin and coworkers (35) showed that the proton fluxes mediated by the human ATP1AL1 H-K-ATPase was 10-fold lower than those of K⁺, suggesting that the stoichiometry of H/K transport differs from 1/1 or that this observation results from the exchange of another cation against K⁺. Interestingly, the full inhibition of the endogenous ouabain-sensitive Na-K-ATPase that leads to cell death in wild-type HEK 293 cells was compensated by the expression of the skin ATP1AL1 K-ATPase (35, 36) or the rat colonic H-K-ATPase (47). This result supports the fact that the nongastric H-K-ATPases may also transport Na⁺, resulting in a (Na/H)-K-ATPase, as proposed by Cougnon et al. (21). Grishin and Caplan (36) recently reported that the human ATP1AL1 H-K-ATPase mediates ouabain-sensitive Na⁺ efflux, when the ATP1AL1 α-subunit is coexpressed with the gastric H-K-ATPase β-subunit in HEK 293. Correlation with⁸⁶Rb influx (as K⁺ surrogate) indicates that the pump formed by the human ATP1AL1 α-subunit and the gastric H-K-ATPase β-subunit mediates primarily Na/K rather than H/K exchange, when expressed in HEK 293 cells. Analysis of Na⁺ and proton transports by the various nongastric H-K-ATPases remains to be done by direct measurement of Na⁺ dependence and proton dependence of the K-ATPase activities mediated by the nongastric H-K-ATPases, a difficult task in expression systems such as theXenopus oocyte or transfected cell lines.

DISCREPANCIES BETWEEN THE FUNCTIONAL PROPERTIES OF THE CLONED NONGASTRIC H-K-ATPASES AND THOSE OF THE NATIVE COLONIC AND RENAL K-ATPASES

The molecular and the functional characterizations of the various nongastric H-K-ATPases should help to achieve a difficult goal: to identify the molecular entities encoding the physiologically relevant K-ATPases. So far, several major discrepancies exist between the physiological studies analyzing the cell-specific expression and the functional properties of the endogenous colonic and renal K-ATPases on one hand and the functional characteristics of the cloned nongastric H-K-ATPases α-subunits on the other hand.

This is outlined by Silver and Soleimani in the accompanying review (73a) and clearly shown in the data summarized in Table 1. All the data reported in Table 1 deal with the rat colonic H-K-ATPases, excluding possible species differences.

It is tempting to deduce from physiological studies that the type III K-ATPase from the kidney is encoded by the rat colonic H-K-ATPase. However, the pharmacological properties of the rat colonic H-K-ATPase when exogenously expressed in oocytes, Sf9 cells, or HEK 293 cells differ drastically from the pharmacological properties of the native K-ATPases (Table 1).

A major task for the next few years would be to explain these discrepancies. The following reasons may be involved:1) the functional expression systems are not appropriate to analyze the functional properties of the cloned nongastric H-K-ATPase α-subunits;2) posttranslational modifications or splicing variants exist in vivo in the “normal” cell context. This may explain why the full-length or the unprocessed subunits display different functional properties than do the endogenous αβ complexes; 3) other subunits may exist and need to be characterized. We would like to discuss these points, and we hope that the bulk of results that will be accumulated in the future will help to discriminate between all these nonmutually exclusive possibilities.

Functional Expression Systems May Not Be Appropriate To Analyze the Functional Properties of the Cloned Nongastric H-K-ATPase α-Subunits

Although various expression systems have proven to be useful in the past for the functional characterization of the P-ATPase α- and β-subunits, these systems may be inappropriate to recapitulate all the functional properties of the natural nongastric H-K-ATPases. Indeed, several points have not been taken into account in the previous studies that have been summarized above.

Putative γ-subunit.

Recent experiments indicate that the γ-subunit of the Na-K-ATPase is required for proper function of the Na-K-ATPase at early stages of embryonic development (45). The γ-subunit also affects the functional properties of the αβ heterodimer when expressed in theXenopus oocyte (7). The γ-subunit modifies the K affinity of the αβ complexes, without effects on the other functions tested (7). Minor et al. (61) recently reported that the human Na-K-ATPase γ-subunit induces cation channel activity when expressed in Xenopus oocytes. Moreover, these authors observed an increase in the Na and K uptake mediated by the exogenously expressed Na-K-ATPase, when the human Na-K-ATPase γ-subunit was coexpressed with the rat α1- and β1-subunits in Sf9 cells (61).

It cannot be excluded that, in vivo, the Na-K-ATPase γ-subunit or another specific γ-subunit is associated with the nongastric H-K-ATPases. This may affect the functional properties of the nongastric H-K-ATPases. One candidate may be the channel-inducing factor (CHIF), an Na-K-ATPase γ-subunit homolog (5). Indeed Capurro and coworkers (16) reported a good correlation between the expression of the rat colonic HKα2-subunit and CHIF in the distal colon and along the nephron.

Oligomerization of the α-subunits.

The P-ATPases have been reported to act as a heterodimer consisting of at least two α- and two β-subunits. The dimerization of the gastric α-subunit is required for proper function (63, 71). If this finding is extended to the other K-ATPases, then one could imagine that the natural nongastric H-K-ATPase α-subunit dimer could be present in the target cells as an α/α homodimer between two nongastric HKα2a or HKα2b subunits, or as an α/α oligodimer between the two α2a and α2b splice variants, or even between two different α-subunits such as the Na-K-ATPase α-subunit/nongastric H-K-ATPase α-subunit. Indeed, Kone and Higham (47) reported that the NH₂-terminal truncated colonic α2b-subunit is expressed at a higher levels (up to 5-fold) than the α2a-subunit in both the colon and the kidney medulla. The ratio between the two HKα2a and HKα2b splice variants may influence the functional properties of the native H-K-ATPase. It would be easy to test this hypothesis using functional expression systems.

Functional importance of the β-subunit.

It has been reported that the nature of the β-subunit involved in the Na-K-ATPase αβ heterodimer influences the functional properties of the αβ complexes. For example, the K affinity, as well as Na affinity, depends of the β-subunit present in the Na-K-ATPase αβ heterodimer (12, 29, 30, 41). Na-K-ATPase α1β3 complexes have a lower K affinity than Na-K-ATPase α1β1 complexes. The β-subunit also influences Na interactions in the Na-K-ATPase pump cycle (30). However, neither ouabain nor Sch-28080 affinities of the nongastric H-K-ATPase appear to be influenced by the β-subunit (19,23). One cannot exclude that an unrecognized β-subunit could behave differently.

Indeed, the β-subunit associated in vivo with the various nongastric H-K-ATPases remains to be clearly identified. The β1-subunit of the Na-K-ATPase that has been recognized as a partner of the colonic H-K-ATPase in both the distal colon and the renal medulla (18) could be a β-subunit related to, but different from, the known Na-K-ATPase β1-subunit, sharing a common epitope allowing immunoprecipitation with an anti-Na-K-ATPase β1 antibody. This β1/β1-related subunit displays a different glycosylation pattern in the distal colon and in the kidney medulla (18). Whether different glycosylation patterns could be of importance for a specific function of the nongastric H-K pumps has not been examined ex vivo. A complex situation may exist in the distal colon, since the colonic H-K-ATPase α-subunit is also associated with the recently described rat colonic β3-subunit (70). A different ratio of HKα2/β1 and HKα2/β3 complexes in the colon and the kidney may be important and physiologically regulated.

Influence of the functional expression system.

Finally, the nature of the functional expression system may also directly affect the functional properties of the exogenously expressed K-ATPases. This could explain the discrepancies reported in studies using either the insect Sf9 cell or the X. laevis oocyte for the functional expression of the rat colonic HKα2a subunit. In this case, the ouabain sensitivity was opposite in the two functional expression systems (Table 1).

The endogenous properties of the expression system could affect the function of the exogenous pumps. For example, the experimental temperature (30°C for the Xenopusoocyte) may affect proper α/α dimerization or α/β oligomerization; the absence of glycosylation in the baculovirus expression system, the functional coupling with an unrecognized protein in the insect Sf9 cells, the protein/lipid composition of the plasma membrane in the cell lines used for functional expression, the absence of membrane polarization in the functional expression systems used so far, and subsequently the absence of the physiological cytoskeleton interactions all could affect the function of the exogenously expressed pumps.

To minimize these potential problems, it appears critical to develop other functional expression systems that share a more physiological cellular environment, such as an inducible expression system in a polarized cell line derived from the rat collecting tubule or the rat distal colon. Genetically modified animals allowing conditional expression of the nongastric H-K-ATPase α-subunit would certainly be of great interest.

Posttranslational Modifications of the Nongastric H-K-ATPases

The effects of posttranslational modifications on the properties of the nongastric H-K-ATPases have not been analyzed so far. This contrasts with the extensive analysis of the Na-K-ATPase α-subunit phosphorylation and its physiological relevance. It has been reported that phosphorylation/dephosphorylation events affect the functional activity of the Na-K-ATPase, such as K and Na transport capacity (9). Such posttranslational modifications could also affect the functional properties of the nongastric H-K-ATPases, even if it seems yet unlikely that it could drastically influence its pharmacological properties.

The recent cloning of two splice variants of the rat colonic H-K-ATPases α-subunit may be the first step in recognizing such a phenomenon. The two splice variants have a different NH₂ terminus, with the α2b having a shorter NH₂-terminal domain than the α2a and missing two putative PKA and PKC phosphorylation sites (47). In the Na-K-ATPase α1-subunit, one of these sites has been shown to be phosphorylated in vivo by the PKC (6, 32, 54). Such a phosphorylation event has not been analyzed either ex vivo or in vivo for the nongastric H-K-ATPases.

Other Nongastric H-K-ATPase Subunits Have To Be Characterized

Finally, one explanation for the discrepancy between the functional expression ex vivo and the functional characteristics of the colonic and renal K-ATPase activities in vivo may be that other yet unrecognized isoforms of the nongastric H-K-ATPases have to be identified.

Against this is the fact that only one gene has been identified in human (called ATP1AL1, encoding the skin H-K-ATPase α-subunit) (75) and in the mouse (P. Meneton, personal communication). To clone such unrecognized isoforms of the nongastric H-K-ATPase subgroup, we have previously used degenerated oligonucleotides directed against highly conserved domains of the α-subunits of the Na-K/H-K-ATPase gene family (38). This approach was successful for the cloning of the colonic H-K-ATPase α-subunit cDNA both in the distal colon (38) and in the rat collecting tubule (Jaisser, unpublished data). We have not been able to isolate another H-K-ATPase isoform from these tissues. A negative result is of course a weak argument.

In favor of the “missing α-subunit” hypothesis, it should be noted that two different K-ATPase activities with different pharmacological profiles have been consistently reported to be present in the distal colon. Recently, Rajendran et al. (67) described a highly ouabain-sensitive K-ATPase in the crypt cells of the distal colon. The colonic H-K-ATPase α2-subunit is not expressed in the crypt cells. Its expression is strictly restricted to the surface cells (38, 42,50). As shown in Table 1, the ouabain- and Sch-28080-sensitive type III K-ATPase described by Doucet and collaborators (13, 82) may be encoded by a new isoform, which could be identical to the colonic “crypt” isoform. Another explanation proposed by some authors is that the cloned colonic H-K-ATPase acquires Sch-28080 sensitivity (and an increased ouabain sensitivity) in vivo during K depletion (13). On the basis of the functional studies reported in the literature for the Na-K-ATPase, the gastric H-K-ATPase, and the nongastric ATPases, the “missing subunit” hypothesis is more attractive than a drastic change of the pharmacological profile dependent on the physiopathological status. Further cloning efforts are required to argue in favor or against the “missing H-K-ATPase α-subunit isoform” hypothesis.

CONCLUSION

The purpose of the present review was to describe the molecular and functional properties of the cloned nongastric H-K-ATPases. The molecular data indicate that these nongastric H-K-ATPases could be identified as a subgroup of the Na-K/H-K-ATPase gene family, apart from the canonical Na-K-ATPases and gastric H-K-ATPases. Functional data obtained from ex vivo expression in various functional expression systems support this classification. The unique cation transport capacity of the nongastric H-K-ATPase as well as its unique pharmacological profile indicate that these H-K-ATPases may have a specialized physiological role. The physiological relevance of the cell-specific expression of these H-K-ATPases and their regulations are discussed in the accompanying review (73a). It should be noted that several questions remain to be answered, particularly the existence of yet unrecognized isoforms or the existence of functional variants due to gene variants or to posttranslational modifications.

FOOTNOTES

• This work was supported by Institut National de la Santé et de la Recherche Médicale. A. T. Beggah is supported by a fellowship of the Swiss National Fund for Scientific Research.