Stabilization of the H,K-ATPase M5M6 Membrane Hairpin by K+ Ions

The integral membrane protein, the gastric H,K-ATPase, is an α-β heterodimer, with 10 putative transmembrane segments in the α-subunit and one such segment in the β-subunit. All transmembrane segments remain within the membrane domain following trypsinization of the intact gastric H,K-ATPase in the presence of K+ ions, identified as M1M2, M3M4, M5M6, and M7, M8, M9, and M10. Removal of K+ ions from this digested preparation results in the selective loss of the M5M6 hairpin from the membrane. The release of the M5M6 fragment is directed to the extracellular phase as evidenced by the accumulation of the released M5M6 hairpin inside the sealed inside out vesicles. The stabilization of the M5M6 hairpin in the membrane phase by the transported cation as well as loss to the aqueous phase in the absence of the transported cation has been previously observed for another P2-type ATPase, the Na,K-ATPase (Lutsenko, S., Anderko, R., and Kaplan, J. H. (1995)Proc. Natl. Acad. Sci. U. S. A. 92, 7936–7940). Thus, the effects of the counter-transported cation on retention of the M5M6 segment in the membrane as compared with the other membrane pairs may be a general feature of P2-ATPase ion pumps, reflecting a flexibility of this region that relates to the mechanism of transport.

Recent functional studies of P-type ATPases have focused on two major activities, the transmembrane transport of cations and the hydrolysis of ATP via a phosphoenzyme intermediate, which appear to be carried out by the actions of two separate protein domains, the intramembrane domain that binds cations and the major cytoplasmic loop responsible for ATP binding and hydrolysis (1,2). P-type ATPases can be classified into two subgroups based upon their function and structure (3). A mechanistic aspect of most P 2 -type ATPases is occlusion of the transported cations during the reaction cycle (4). Such occlusion is pictured as an intermediate state where cations are bound within the transmembrane segments of the protein without free access to either aqueous compartment. Occlusion was also recently shown to be retained in a post-tryptic preparation of the Na,K-ATPase composed of the transmembrane segments and lacking most of the cytosolically exposed protein (5).
Chemical modification and site-directed mutagenesis studies of expressed P 2 -ATPases have indicated that the M5M6 transmembrane region may have a special importance in cation occlusion and transport (6 -10). In studies on tryptic membrane preparations of the Na,K-ATPase obtained in the presence of K ϩ , mere removal of K ϩ ions resulted in the selective release of the M5M6 hairpin from the membrane to the aqueous phase (11). This membrane stabilization of the M5M6 region by the counter-transported cation suggested that the organization and folding of the M5M6 hairpin and the interactions between M5M6 and the rest of the protein may be associated with the transport function (11). The uniqueness of this region of the membrane is also shown by the failure of the M5M6 hairpin to behave as an independent unit in the process of synthesis and insertion in in vitro translation studies of the H,K-ATPase (12). Similar conclusions were derived from Cos-1 cell translation of segments of the Na,K-ATPase and from in vitro translation of the endoplasmic reticulum Ca-ATPase (13,14), suggesting a general property of this membrane domain of the P 2 -type ATPases.
The present studies were undertaken to determine whether or not cation-dependent stabilization of the M5M6 hairpin in the membrane was also a feature of the H,K-ATPase, another P 2 -ATPase. Because this protein can be obtained in a sealed vesicular inside out orientation, the directionality of the release of the M5M6 hairpin from the membrane (toward the cytoplasmic or extracellular space) following K ϩ removal could also be determined, unlike right side out oriented vesicles achieved in microsomal preparations of Na,K-ATPase (15). A preliminary report of this work has been presented (16).
Enzyme Preparation and Activity Assay-The H,K-ATPase was isolated from hog gastric mucosa (17). Protein was determined by the method of Lowry et al. (18). Vesicles were aliquoted and stored (3-5 mg/ml protein) in a 34% sucrose solution buffered with 50 mM PIPES (pH 6.8) and frozen at Ϫ80°C until used. H,K-ATPase activity was measured in a standard assay medium containing 1 mM EGTA, 20 mM KCl, 3 mM MgCl 2 , 3 mM Na 2 ATP, 40 mM Tris/HCl (pH 6.9), and 250 mM sucrose in the presence or absence of 1 g/ml nigericin. The suspension was incubated for 15 min at 37°C (10 mg protein/ml), and the liberation of P i measured as described by Brotherus et al. (19). The difference in activity seen upon the addition of nigericin is a measurement of vesicle integrity. Vesicles used were Ͼ70% sealed and oriented cytosolic side out. Maximal enzyme activity was also determined by using 100 mM NH 4 Cl instead of KCl, with NH 4 ϩ acting as a K ϩ congener. Vesicles do not have to be permeablized for NH 4 ϩ to reach the inside because NH 3 can freely diffuse across the bilayer. In experiments to determine the appropriate [SDS] required to permeabilize the vesicles, enzyme activity was measured at various ratios of SDS to protein (see Fig. 2) and compared with maximal activity with nigericin or NH 4 Cl.
Trypsin Digestion-H,K-ATPase (1 mg/ml) was suspended in medium containing 1 mM EDTA, 200 mM KCl, 20 mM Tris/HCl (pH 6.8) and incubated at 4°C for 16 -24 h. Then TPCK-treated trypsin (1:10 w/w with respect to H,K-ATPase) was added, and incubated at 37°C for 1.5 h. The reaction was terminated with the addition of the serine protease inhibitor AEBSF (2 mM); AEBSF was present at 1 mM in all subsequent steps. Soluble peptide fragments were separated from membrane-bound fragments by centrifugation (436,000 ϫ g, 30 min, 4°C). The membrane fraction was resuspended with buffer and pelleted once more. The pellet (ϳ500 g) was resuspended in buffer containing 1 mM EDTA, 20 mM Tris (pH 6.8), and 250 mM sucrose. Samples were split in half and diluted to 1 mg/ml with buffer containing either 200 mM KCl or 5 mM MgP i ; the [sucrose] was adjusted accordingly to maintain isosmotic conditions. Samples were incubated (37°C, 15 min) and centrifuged (436,000 ϫ g, 30 min, 4°C). Supernatants containing peptides from the extravesicular space were removed and saved. Pellets were resuspended (1 mg/ml) in the same buffer (including KCl and MgP i , respectively) with the addition of 0.05% SDS (w/w) and incubated (37°C, 15 min) and then centrifuged (436,000 ϫ g, 30 min, 4°C). Supernatants containing peptides from the intravesicular space were removed and saved. The pellet from the sample-containing KCl tube was resuspended in buffer (1 mg/ml), but MgP i was added instead of KCl. The suspension was incubated (37°C, 15 min) and then centrifuged (436,000 ϫ g, 30 min, 4°C). The supernatant was removed and saved. All supernatants were treated with 1 mM CPM for better visualization of low molecular weight peptides (15). Peptides were precipitated with acetone (9:1, v/v) at Ϫ20°C for 16 h and then were run on a 16.5% Tricine gel (20). After electrophoresis, protein fragments were transferred onto PVDF membranes by electroblotting in 10 mM CAPS, 10% MeOH (pH 11.0) (21). Fluorescent CPM-labeled bands were cut from the PVDF and subjected to N-terminal amino acid sequencing.

RESULTS
The two specific questions addressed by our studies were (i) in the H,K-ATPase post-tryptic preparation does the removal of K ϩ ions lead to the destabilization and selective release of the M5M6 hairpin from the membrane and (ii) if such release occurs, is the hairpin released to the cytoplasmic or extracellular compartment?
SDS Treatment of the H,K-ATPase Vesicle Preparation-As shown in Fig. 1, SDS treatment of the vesicle preparation is used to gain access to the intravesicular (extracellular) space. If release of M5M6 occurs to the extracellular space the vesicles will need to be disrupted in order for the M5M6 hairpin to be seen in the supernatant; if release is to the cytoplasmic space, disruption of the vesicles by detergent will not be required, and centrifugation alone will separate the released hairpin from the vesicles. It was necessary first to determine an appropriate SDS concentration that would permeabilize the vesicles but not denature or disrupt the H,K-ATPase. To do this, we made use of the fact that for maximal H,K-ATPase activity K ϩ ions must gain access to the intravesicular space. Fig. 2 demonstrates that SDS ratios Յ0.1% (w/w) were able to disrupt vesicle integrity without denaturing the H,K-ATPase.
Directionality of Release of the M5M6 Hairpin-Just as was originally reported for the Na,K-ATPase (5), extensive tryptic digestion of the H,K-ATPase in the presence of K ϩ ions produces a membrane residue that contains the ␤-subunit (largely undigested) and four sets of transmembrane segments, M1M2, M3M4, M5M6, and M7-M10 (22). The post-tryptic preparations of both the Na,K-and H,K-ATPases retained the ability to occlude K ϩ ions (5,23). In the Na,K-ATPase, removal of K ϩ ions from this preparation resulted in a loss of the ability to occlude K ϩ associated with the release of the M5M6 hairpin from the membrane to the aqueous phase (11). Fig. 3 shows a single fluorescently labeled peptide (ϳ10 kDa) that was released from the post-tryptic H,K-ATPase preparation upon the removal of K ϩ ions. N-terminal amino acid analysis gave a single sequence (NAADMIL . . . ) that corresponds to the residues in the M5M6 hairpin (Table I). It was shown earlier that both the N and C termini of this fragment are cytoplasmic (24), whereas either (or both) cysteine 813 and 822 within this stretch are located at the extracytoplasmic side (25). These observations demonstrated that the M5M6 segment indeed spans the membrane twice. Furthermore, this peptide was not released to the supernatant until after SDS treatment of the proteolyzed vesicles, thus it is released toward the intravesicular (extracellular) space (Fig. 3A). In some experiments, a faint band was apparent (ϳ10 kDa) in the supernatant prior to SDS treatment (data not shown). This peptide was most likely the M5M6 hairpin, because it was only observed when K ϩ ions were not present in the trypsinized preparation (i.e. treatments outlined in 3a of Fig. 1). Additionally, because the intensity of this band appeared to vary inversely with the fraction of tight vesicles (as determined by K ϩ ionophore stimulated ATPase activity), we conclude that the M5M6 hairpin is released to the intravesicular space and then exits from leaky vesicles. Consistent with this conclusion is that SDS permeablization of these preparations subsequently revealed a vivid band at 10 kDa that subsequent amino acid sequencing confirmed as M5M6 (Table I).
These findings cannot directly rule out an SDS effect on the post-tryptic H,K-ATPase preparation. That is, does SDS itself promote the release of the M5M6 fragment? To answer this question, we also treated post-tryptic H,K-ATPase vesicles with SDS in the presence of 200 mM K ϩ . Under these conditions, we observed no peptide release (Fig. 3B, lane 1). However, when K ϩ was subsequently removed from these permeablized vesicles the M5M6 hairpin appeared in the supernatant (Fig. 3B, lane 2).
In the experiments discussed above, the post-tryptic membrane preparation was obtained in the presence of 200 mM KCl. In two initial experiments, when trypsin digestion was performed in the presence of Յ20 mM KCl, we obtained two separate peptide sequences from the released fragment (ϳ10 kDa; Table I). The first sequence corresponded to M5M6, whereas the second sequence corresponded to the M7M8 transmembrane pair (Table I). It has been previously shown that in the absence of K ϩ ions an additional trypsin cleavage takes place in the H,K-ATPase between M8 and M9 (22). It seems that this cleavage also takes place at low K ϩ (i.e. Յ 20 mM), thus separating the M7M8 hairpin from the ϳ21-kDa C terminus. DISCUSSION The membrane domain of the P 2 -type ATPases contains the ion transport pathway, and the results of a large number of studies have been interpreted as showing involvement of M4, M5, M6, and perhaps M8 in ion translocation and ion occlusion (2,7,26). Various methods including sequencing of membraneembedded tryptic fragments (5,22), labeling with sided thiophilic reagents (23), or extracytoplasmic photoactivatable re-agents (27) and in vitro translation provided strong evidence for 10 membrane segments in the ␣-subunit of the P 2 -type ATPases.
Hydropathy algorithms of P 2 -type ATPases all predict four membrane sequences in the N-terminal sector but predict only a single sequence in the M5M6 region and vary in terms of prediction of the last four segments (3). In accord with the ambiguity of these predictions, in vitro translation of individual segments or truncated constructs of the H,K-ATPase were not able to demonstrate transmembrane insertion of M5, M6, and M7 (13). Expression of segments or truncated constructs of the Na,K-ATPase also showed that M5 and M6 were not membrane inserted (14,28). These results suggested that peptidepeptide interactions were more significant for these segments as compared with the N-terminal segments. Indeed, it was suggested earlier that the M5M6 hairpin of the Na,K-ATPase was involved in protein-protein interactions with other transmembrane segments within the lipid portion (11). Furthermore, our results (Ref. 11 and this work) suggest that cations are also required for stabilization in the membrane of the M5M6 hairpin of P 2 -type ATPases.
In the case of the P 2 -ATPases, results from chemical modification (8) and mutagenesis (6 -10) have suggested that amino acid residues in M5M6 are intimately associated with cation occlusion and transport, and this hairpin is now believed to play a central role in cation binding and complexation. The loss of the M5M6 hairpin from the membrane and its prevention by K ϩ (Ref. 11 and this work) suggest that the occluded K ϩ ions play a role in causing a rearrangement of these membrane segments. One plausible mechanism for the action may be the neutralization of repulsive negative charges in Asp residues of M6 and the Glu in M5. Cation complexation by the electrondonating residues of M5M6 substitute for hydration and stabilize the presence of the charged cation in the membrane. At the same time the positive cation apparently serves to stabilize the anionic and hydrophilic transmembrane segments of M5 and M6 within the membrane.
The finding that the M5M6 hairpin of about 10 -11 kDa (or about 90 amino acid residues) is lost to the extracellular space shows that this redistribution involves the mobilization across the membrane of a considerable number of charged and hydrophilic amino acid residues. We believe that M5M6 is probably surrounded by other protein helices rather than the membrane lipid (3,11). In this way protein-protein interactions would be greatly modified as cations (e.g. K ϩ ) are bound to and released ϩ (NH4Cl) for K ϩ . The difference between no treatment (None) and nigericin (or NH 4 Cl) represents the population of tightly sealed vesicles (ϳ66 -75%). Activity was then measured after vesicle treatment with the indicated amounts of SDS. by the pump. Another line of evidence suggested mobility in this region of the enzyme. Pantoprazole (5-difluoromethosy-2-[3,4-methosy-2-pyridyl)methylsulfinyl]-1H-benzimidazole) converts to a cationic thiophilic sulfenamide and binds to Cys 813 and Cys 822 in the loop joining M5 and M6 forming disulfide bonds. Prior to this covalent reaction tryptic cleavage occurs largely at Arg 775 . After derivatization, cleavage is found only at Lys 791 (29), showing that formation of disulfides within the connecting loop of M5 and M6 results in a change in accessibility of the N-terminal region of M5 to trypsin. Furthermore, the exposure of Cys 983 in M10 of the Na,K-ATPase to sulfhydryl-reactive probes (but not of Cys residues in M1, M2, or M4) following the loss of the M5M6 hairpin also provides evidence for interactions between C-terminal segments and M5M6 (15). Similar conclusions have been reached from in vitro translation studies of both the Na,K-ATPase and the Ca-ATPase (13,14).
In addition to the release of the M5M6 hairpin, the M7M8 transmembrane pair was also released from the membrane following digestion by trypsin in low K ϩ conditions (Table I) where the 21-kDa fragment is cleaved between M7M8 and M9M10, also suggesting a weak interaction between the bilayer and this pair of transmembrane segments. Prior to cleavage between M8 and M9, the M5M6 tryptic fragment remains associated with the 21-kDa C terminus; conversely, in the absence of K ϩ , when cleavage occurs between M8 and M9 the association vanishes, suggesting that M5 and M6 are in the vicinity of M7-M10 (22). Further, although M8 acts as a stop transfer sequence in in vitro translation, M7 is unable to membrane insert either as a stop transfer or a signal anchor sequence (its putative role). However, in in vivo expression in frog oocytes, M7 is able to act as a membrane inserted sequence provided the ␤-subunit is co-translated, consistent with the strong interaction observed between the region preceding M8 and the ␤-subunit of the H,K-ATPase (12,30). The hydropathy profile (31,32) of the M7 segment in the H,K-ATPase is significantly lower than that of corresponding segment of the Na,K-ATPase consistent with its easier release from the membrane after cleavage of the connecting cytoplasmic linkage.
The striking mobility of the M5M6 hairpin with respect to the membrane that is modified by the presence of the occluded and transported cation probably plays a significant role in the transport cycle. Indeed, the direct link between the M5M6 hairpin and the large intracellular loop, shown to contain the ATP binding site (33)(34)(35), is consistent with a role in coupling ATP hydrolysis with cation transport. We have speculated earlier (11) that movements of M5M6 in the Na,K-ATPase protein in a direction that is perpendicular to the plane of the mem-brane may play a role in active transport; on the basis of the present results such a mechanism may apply to other P 2 -ATPases.