Unraveling Functional and Structural Interactions between Transmembrane Domains IV and XI of NhaA Na (cid:1) /H (cid:1) Antiporter of Escherichia coli *

A functionally important, interface domain between transmembrane segments (TMSs) IV and XI of the NhaA Na (cid:1) /H (cid:1) antiporter of Escherichia coli has been unraveled. Scanning by single Cys replacements identified new mutations (F136C, G125C, and A137C) that cluster in one face of TMS IV and increase dramatically the K m of the antiporter. Whereas G125C, in addition, causes a drastic alkaline shift to the pH dependence of the antiporter, G338C alleviates the pH control of NhaA. Scanning by double Cys replacements (21 pairs of one replacement per TMS) identified genetically eight pairs of residues that showed very strong negative complemen-tation. Cross-linking of the double mutants identified six double mutants (T132C/G338C, D133C/G338C, F136C/ S342C, T132C/S342C, A137C/S342C, and A137C/G338C) of which pronounced intramolecular cross-linking defined an interface domain between the two TMSs. Remark-ably, cross-linking by a short and rigid reagent ( N , N (cid:1) - o -phenylenedimaleimide) revived the Li (cid:1) /H (cid:1) antiport activity, whereas a shorter reagent (1,2-ethanediyl bismethanethiosulfonate) revived both Na (cid:1)

Similar to many other Na ϩ /H ϩ antiporters, both prokaryotic (1) and eukaryotic (9 -11), one of the most interesting functional characteristics of NhaA is its dramatic dependence on pH. The rate of activity of NhaA changes over 3 orders of magnitude between pH 7 and 8 (4,5). This activation is accompanied by a conformational change that involves the N terminus as probed by a monoclonal antibody (1F6 (12)) and loop VIII-IX ( Fig. 1) as probed by accessibility of NhaA to trypsin (13).
Amino acid residues that affect the transport activity can be identical to or different from those affecting the pH regulation of NhaA. They are clustered in various domains along the protein (14 -17), implying the need of atomic structure for understanding the mechanism of activity and pH regulation of NhaA. Since atomic resolution of NhaA is not yet available, indirect approaches for structure analysis have been applied. Second site suppressor mutations were isolated to mutation G338S, a pH-conditional lethal mutant that is located in TMS XI and grows on high Na ϩ -selective medium at neutral pH but not at alkaline pH (17). The suppressor mutations were found clustered in helix IV (A127T, P129L, and A127V) ( Fig. 1) (17). These results have suggested that residues in TMSs IV and XI are in close proximity and functionally and/or structurally interact (18). We therefore scanned genetically by Cys replacements conserved residues in TMSs IV and XI and identified new residues that participate in the activity and/or pH regulation of the antiporter. We then constructed double Cys replacements, one in TMS IV and the other in TMS XI, and found that many of the pairs of the Cys replacements exhibited a negative interaction, suggesting functional and/or structural interactions between TMSs IV and XI. However, long range effects can explain interaction between pairs of mutants.
Site-directed thiol cross-linking studies have provided considerable insight into the structure of membrane proteins by providing estimates of distances between certain positions in proteins (19 -23). We have used this approach to probe proximities between periplasmic and cytoplasmic loops of NhaA and found very strong cross-linking within the pairs of Cys replacements S146C/L316C and A118C/S352C (18) (Fig. 1), suggesting close proximity between the neighboring TMSs IV and XI. However, since loops can be flexible, the distances obtained are not conclusive. Therefore, in the present work, in addition to the genetic approach, we measured distances between the Cys replacements in TMSs IV and XI by intramolecular site-directed cross-linking and identified several double mutants that strongly cross-linked. Together with the pheno-types of the double mutants, this cross-linking defined an interface between TMSs IV and XI with functional implications. Remarkably, cross-linking of the lethal double mutant T132C/ G338C revived both the Na ϩ and Li ϩ antiporter activities of the mutant and even its regulation by pH, as would be expected if the cross-linking restored an active conformation of the antiporter.
Site-directed Mutagenesis-Site-directed mutagenesis was conducted following a polymerase chain reaction-based protocol (27). All mutations were verified by DNA sequencing of the entire gene, through the ligation junctions with the vector plasmid.
To generate all double mutants, the MunI-MluI fragment (571 bp) of single Cys NhaA mutants in Helix XI, on C-less background, were ligated with MunI-MluI fragment (4.17 kb) of plasmid encoding the single Cys NhaA mutants in Helix IV.
Isolation of Membrane Vesicles, Assay of Na ϩ /H ϩ Antiporter Activity-EP432 cells transformed with the respective plasmids were grown and everted vesicles were prepared and used to determine the Na ϩ /H ϩ or Li ϩ /H ϩ antiporter activity as described (28 -30). The assay of antiport activity was based upon the measurement of Na ϩ -or Li ϩ -induced changes in the ⌬pH as measured by acridine orange, a fluorescent probe of ⌬pH. The fluorescence assay was performed with 2.5 ml of reaction mixture containing 50 -100 g of membrane protein, 0.5 M acridine orange, 150 mM KCl, 50 mM 1,3-bis[tris(hydroxymethyl)methyl-]amino]propane, 5 mM MgCl 2 , and the pH was titrated with HCl. Where indicated, 10 mM ␤-mercaptoethanol or 20 mM diamide (Sigma) was added to the reaction mixture. After energization with either ATP (2 mM) or D-lactate (2 mM), quenching of the fluorescence was allowed to achieve a steady state, and then either Na ϩ or Li ϩ (10 or 100 mM each) was added. A reversal of the fluorescence level (dequenching) indicates that protons are exiting the vesicles in antiport with either Na ϩ or Li ϩ as indicated (see Figs. 4A and 6A for data of typical experiments). As shown previously, the end level of dequenching is a good estimate of the antiporter activity (31), and the concentration of the ion that gives half-maximal dequenching is a good estimate of the apparent K m of the antiporter (31,32). The concentration range of the cations tested was 0.01-100 mM at the indicated pH values, and the apparent K m values were calculated by linear regression of a Lineweaver-Burk plot.
Membranes treated with cross-linking reagents lost the lactate-dependent respiration. Therefore, when such membranes were used in the antiport assay, the membranes were energized by ATP.
Detection and Quantitation of NhaA and Its Mutated Proteins in the Membrane-NhaA and its mutated derivatives were quantitated by Western analysis using an anti-NhaA monoclonal antibody 1F6 (33) as described previously (34). Total membrane protein was determined according to Ref. 35.
In Situ Site-directed Inter-and Intramolecular Cross-linking-Sitedirected intramolecular cross-linking was conducted in situ on membrane vesicles isolated from TA16 cells expressing the various NhaA double mutants. Membranes (3 mg of protein) were resuspended in a buffer (4.5 ml) containing 100 mM potassium phosphate, 5 mM MgSO 4 (pH 7.4), and one of the freshly prepared homobifunctional cross-linkers: 2 mM BMH (Pierce), 1 mM o-PDM (Sigma), or 2 mM MTS-2-MTS (Toronto Research Chemicals). The stock solutions of the cross-linkers were prepared in DMF at 200 mM, so that the amount of DMF in the reaction mixture does not exceed 1%, a concentration that does not affect the antiporter activity. The suspension was incubated at 26°C with gentle rotation for 60 min. The reaction with BMH or o-PDM was terminated by the addition of 10 mM ␤-mercaptoethanol and that with MTS-2-MTS, by centrifugation (Beckman; TLA 100.4, 265,000 ϫ g, 20 min, 4°C). The protein was Ni 2ϩ -NTA affinity-purified, treated with trypsin at alkaline pH, and separated on SDS-PAGE (nonreducing conditions in the case of treatment with MTS-2-MTS) to identify the proteolytic products according to Ref. 18. Since there is only one trypsin cleavage site at Lys 249 , located between TMSs IV and XI (Fig. 1), trypsinolysis results in two tryptic peptides of mobility faster than the intact protein. On the other hand, intramolecular cross-linking results in one fragment of mobility equal to that of the intact protein (18).
To measure the effect of the cross-linking agents on the activity of the antiporter, everted membrane vesicles were isolated from EP432 expressing the mutations (300 g of protein); treated with one of the homobifunctional cross-linkers as described above; washed in 1.5 ml containing 140 mM choline chloride, 250 mM sucrose, and 10 mM Tris-HCl (pH 7.5); resuspended in 75 l of the same solution; and used to test antiporter activity. The degree of cross-linking in vesicles derived from EP432 and TA16 was indistinguishable.
In situ site-directed intermolecular cross-linking was tested as above but without treatment with trypsin. When intermolecular cross-linking takes place, a band, corresponding in mobility to that of NhaA dimer, appears in SDS-PAGE (15,18).
Accessibility of the Cys Replacements to the Cross-linking Reagents-The procedure was essentially as described in Ref. 36. Following treatment with the cross-linking reagents, the Ni 2ϩ -NTA affinity-purified protein was left bound to the Ni 2ϩ -NTA beads, washed with binding buffer (5 mM imidazol, 500 mM NaCl, 20 mM Tris, and 0.1% n-dodecyl-␤-D-maltoside, pH 7.4) and exposed to 0.2 mM fluorescein 5-maleimide (Molecular Probes, Inc.; dissolved in Me 2 SO) to titrate any Cys residue left free. The incubation was for 30 min at 25°C with gentle tilting. Then the protein was washed and eluted as described in Ref. 36. The degree of fluorescence labeling of the protein resolved in SDS-PAGE was determined under UV light (260 nm) as described (18,36).
Modeling of TMSs IV and XI-Initially, two ideal poly-Ala ␣-helices were built using the program Moleman (37). Then the side chains were altered to those of the respective TMS, according to the putative topology model of NhaA (Fig. 1A) (38) by using the rotamer library in the O software (39). Next, the distances estimated by cross-linking of double Cys replacement mutants located in TMSs IV and XI (see "Results") and on neighboring loops (18) were used as constraints to determine the proximity between the two helices. Alternate orientations of side chains were tested, and those with geometric clashes with neighboring residues were ruled out. Energy minimization of the putative model was carried out using the program CNS (40) to eliminate further clashes between neighboring residues.

Construction of Single and Pairs of Cys Replacements in
TMSs IV and/or XI of NhaA-In this study, site-directed Cys replacements and site-directed thiol cross-linking were used to identify functional interactions and determine proximities between TMSs IV and XI of NhaA. For this purpose, we used the single Cys replacements mutants (D133C, T132C, P129C, and A127C (14)) that had already been constructed in TMS IV and constructed new single Cys replacement mutants: G125C, F136C, and A137C in TMS IV and L334C, G338C, and S342C in TMS XI (Tables I and II and Fig. 1). All old and new mutations replace conserved amino acid residues (Fig. 1). We then constructed double mutants, each containing one replace-ment in TMS IV and one in TMS XI (Table II and Fig. 1). All mutants, both single and double, were constructed in plasmidencoded CL-NhaA that is very similar to the wild type protein in transport activity and pH regulation (25) (Table II). To determine the level of expression, growth phenotype, and antiporter activity, the plasmids encoding the mutants were transformed into EP432, an E. coli strain devoid of both Na ϩspecific antiporters NhaA and NhaB (3). This strain can grow in the presence of high Na ϩ or Li ϩ only when transformed with a plasmid expressing active NhaA (3) and Table II (compare EP432/pCL-HAH4 with EP432/pBR322). The results summarized in Table II show that the single and double mutants were expressed to a level that ranges between 10 and 100% of the control level (EP432/pCL-HAH4). It should be stressed that because all mutants are expressed from multicopy plasmids, even the lowest level of expression observed here is significant, CGGTTTTACGATGT*GTATCTTTATTGCC TCT 3 TGT End primers for mutations G125C, F136C, A137C, and G338C TTTAACGATGATTCGTGGCGG (sense primer) None GCTCATTTCTCTCCCTGATAAC (antisense primer) End primers for mutations L334C and S342C GTGTGGTTGTCGACGCACGGGCG (sense primer) None GTGGAGTTAAATAAAGCGCC (antisense primer)

TABLE II Single and double Cys replacements in TMSs IV and XI of nhaA
For characterization of the mutations, EP432 cells transformed with the respective plasmids were used. The expression level was expressed as a percentage of the control cells (EP432/pCL-HAH4). Growth experiments were conducted on agar plates with high Na ϩ (0.6 M) or high Li ϩ (0.2 M) at the pH values indicated in parentheses. ϩϩϩ, the number and size of the colonies after 24 h of incubation at 37°C was identical to that of the wild type; ϩϩ and ϩ, the number of colonies was similar but size was slightly or much smaller than that of the wild type, respectively; Ϫ, no growth. Na ϩ /H ϩ and Li ϩ /H ϩ antiporter activity at pH 8.5 was determined with 10 mM NaCl or LiCl.

Mutation Expression
Growth  readily detected by Western analysis, and way above the level expressed from a single chromosomal gene that is hardly detected by Western analysis (17,18).
The Effect of the New Single Mutations in TMSs IV or XI on the Growth Phenotype and the Na ϩ /H ϩ and Li ϩ /H ϩ Antiporter Activity-As summarized in Table II, all new single mutants retained certain growth capacity on the selective media, implying that none of the replaced residues are essential for the Na ϩ /H ϩ antiport activity of NhaA. However, the results also show that residues Gly 125 , Phe 136 , and Gly 338 are essential for certain properties of NhaA that support growth in the presence of high Na ϩ , at alkaline pH, whereas residues Phe 136 and Gly 338 are, in addition, important for growth on high Li ϩ at neutral pH (Table II).
The Na ϩ /H ϩ antiporter activity, at various pH values, was monitored in everted membrane vesicles of EP432 expressing the various single mutants, as recovery (dequenching) from the respiration-dependent fluorescence quenching of acridine or- ange upon the addition of the ion (Table II and Fig. 2). EP432 transformed with CL-NhaA-expressing plasmid (Table II and Figs. 2 and 4A) (3,25) or the vector plasmid served as a positive and negative control, respectively (Table II) The apparent K m values for Na ϩ and Li ϩ were determined for each mutant at pH 8.5 (Table II), whereas the pH dependence was monitored at low (10 mM) and high (100 mM) concentrations of NaCl. The latter procedure allowed us to differen-tiate between three types of mutations in NhaA (14) as follows. (a) Mutations that affect solely the apparent K m of the antiporter have a pH profile of activity that is similar to that of the wild type when measured at saturating Na ϩ concentrations. (b) Mutations that shift the pH dependence with hardly any effect on the apparent K m exhibit the pH shift, whether tested at saturating Na ϩ concentrations or not. (c) Mutations that affect both the apparent K m and the pH dependence of NhaA.
As expected from the high sensitivity of mutant F136C to Na ϩ and Li ϩ (Table II), its apparent K m for Na ϩ increased dramatically (510-fold), its rate decreased, and it lost the Li ϩ /H ϩ antiporter activity (Table II and Fig. 2B). Due to its very low activity, it was impossible to determine accurately the pH profile of this mutant. Although to a lesser extent, the other two new mutations in TMS IV, G125C and A137C, also strongly increase the apparent K m of the antiporter for both Na ϩ (60-and 36-fold, respectively) and Li ϩ (28-and 40-fold, respectively) (Table II). However, whereas the pH dependence of A137C was very similar to that of CL-NhaA at saturating concentration of Na ϩ (Fig. 2A), G125C caused a drastic alkaline shift (of one pH unit) to the pH dependence of NhaA, which was independent of the concentration of Na ϩ (Fig. 2B). Hence, whereas A137C affects the apparent K m of NhaA to the ions, G125C affects both the apparent K m and the pH dependence of NhaA.
Mutants L334C and S342C of TMS XI show a maximal end level of fluorescence dequenching (Fig. 2C) and kinetic parameters that are similar to those of the wild type (Table II). However, whereas the pH dependence of L334C is identical to that of CL-NhaA, the pH profile of S342C is drastically shifted to the acidic range by one pH unit (Fig. 2C) in a Na ϩ concentration-independent fashion. Similar to the previously described mutant G338S in TMS XI (17), G338C alleviates the pH control of NhaA and has a small effect on the kinetic parameters of the antiporter (Table II and Fig. 2D).
The Effect of the Double Mutations (One in TMS IV and the Other in TMS XI) on the Growth Phenotype and the Na ϩ /H ϩ and Li ϩ /H ϩ Antiporter Activity-We used the single mutants to construct double Cys replacements, one replacement in TMS IV and the other in TMS XI. By comparing the growth and transport phenotypes of the single mutants with that of the 21 new double mutants, two patterns of phenotypes were identified as follows. (a) Some growth and transport phenotypes were similar to that of both respective single mutants (Table II and Fig.  3). Thus, the growth and transport of the double mutants A127C/L334C, A127C/S342C, P129C/L334C, P129C/S342C, D133C/L334C, and D133C/S342C were very similar to the respective single mutants and the control. (b) Surprisingly, in many of the double mutants, a drastic negative interaction between the double mutations was found ( Fig. 3 and Table II), and the mutants G125C/L334C, G125C/G338C, G125C/S342C, T132C/G338C, D133C/G338C, F136C/L334C, F136C/G338C, and F136C/S342C completely lost antiporter activity and growth capacity (Fig. 3B and Table II). The negative interaction between residues in TMSs IV and XI can be due to a direct physical interaction with functional consequences. Alternatively, it is also possible that the two TMSs are wide apart in the NhaA molecule and that long range effects, mediated through the protein, exert the phenotypes of the double mutants. Since the atomic structure of NhaA has not yet been solved, we applied intramolecular cross-linking to estimate proximities between TMSs IV and XI.
Effect of Reducing Conditions on the Na ϩ /H ϩ and Li ϩ /H ϩ Antiporter Activity of the Double Cys Replacement Mutants-Since eight of the double Cys replacement mutants were inac-FIG. 2. Na ؉ /H ؉ antiporter activity in everted membrane vesicles of the single NhaA mutants in TMSs IV or XI. Everted membrane vesicles were prepared from EP432 cells, expressing the indicated NhaA mutants and grown in LBK (pH 7). The Na ϩ /H ϩ antiporter activity was determined at the indicated pH values, using acridine orange fluorescence to monitor ⌬pH in the presence of 10 mM (continuous line) or 100 mM (broken line) NaCl as described under "Experimental Procedures." The results are expressed as end level of dequenching (%). All experiments were repeated at least three times with practically identical results. tive ( Fig. 3B and Table II), we considered the possibility that spontaneous cross-linking by oxidation occurs between the two vicinal SH groups, forming S-S bonds. Therefore, the antiporter activity of the inactive double mutants was tested in the presence of ␤-mercaptoethanol (reducing conditions). The double mutant F136C/S342C was the only mutant that was activated significantly by reduction; following the addition of ␤-mercaptoethanol, the rate of fluorescence dequenching re-flects the restored antiporter activity, which reaches a fluorescence dequenching end level of 45% (Fig. 4, compare B and C). Furthermore, the subsequent addition of diamide (20 mM) that impose oxidative conditions restored the inactivation (Fig. 4C). Following the addition of diamide, the drastic drop in fluorescence reflects the fast rate of activity of the respiratory proton pumps that restore the ⌬pH. No activity was observed with LiCl, or at different pH values. Either in the presence or absence of ␤-mercaptoethanol, diamide had no effect on Cys-less NhaA (data not shown). The formation of an S-S bond between F136C and S342C, in the presence of diamide, was verified by SDS-PAGE (under nonreducing conditions) as described below (data not shown). Hence, an S-S bond exists between F136C and S342C and inhibits the activity of the double mutant. These results imply the close proximity between positions F136C and S342C of TMSs IV and XI (Fig. 1, B and C). Since in no other case did reducing conditions affect the antiporter activity, we chose to assess proximities between TMSs IV and XI by using thiol-specific homobifunctional cross-linking reagents.
In Situ Intramolecular Cross-linking between TMSs IV and XI of NhaA-Using the cross-linking reagents described below, we did not find any intermolecular cross-linking between monomers of the single or double mutants (data not shown).
We have previously shown that site-specific intramolecular cross-linking by thiol-specific homobifunctional cross-linking reagents, is a very powerful tool to measure proximities between double Cys replacements introduced into the CL-NhaA molecule (18). In this protocol, the cross-linkers are applied, in situ, on the membrane and the unique trypsin cleavable site of NhaA at Lys 249 (Fig. 1A) (13, 18) is exploited to detect changes in the mobility of cross-linked products of double Cys replace-  (2), and the fluorescent quenching (Q) was recorded until a steady state level of ⌬pH (100% quenching) was reached. NaCl or LiCl, at the indicated concentrations, was then added (1), and the new steady state of fluorescence obtained (dequenching) after each addition was monitored. Where indicated, 10 mM ␤-mercaptoethanol or 20 mM diamide were added to the reaction mixture. The experiments were repeated at least three times with practically identical results. ments introduced, each on an opposite side of the trypsin cleavage site. In the untreated control, the resulting two tryptic products can easily be identified by SDS-PAGE as 24-and 17-kDa fragments (Fig. 5A, lane b, HF and LF fragments, respectively), whereas the intact CL-NhaA banded at 32.5 kDa (Fig. 5A, lane a) (13,18). Treatment of membrane vesicles expressing CL-NhaA by the cross-linking reagents BMH, o-PDM, or MTS-2-MTS had no effect on the proteolytic products of the protein (Fig. 5A, lanes c-e, respectively). Hence, the cross-linking reagents are specific to the Cys replacements in the protein.
The results presented in Fig. 5 show that the double Cys replacement mutations, T132C/S342C (Fig. 5B, lanes c-e) and T132C/G338C (Fig. 5C, lanes c-e) were cross-linked by the three cross-linking reagents BMH, o-PDM, and MTS-2-MTS. Following cross-linking, instead of the two tryptic products (Fig. 5, B and C, lanes b), one prominent band of mobility similar to that of the intact untreated protein appeared on SDS-PAGE (Fig. 5, B and C, compare lanes c-e with lane a). The double mutants A137C/S342C (Fig. 5D) and A137C/G338C (Fig. 5E) were cross-linked only by MTS-2-MTS. The double mutant D133C/G338C was cross-linked significantly by o-PDM (Fig. 5F, lane d) and partially by BMH (Fig. 5F, lane c) but not by MTS-2-MTS (2 mM).
It was previously shown (41) that cross-linking might be overlooked when the concentration of the reagent is too high. This could cause separate labeling of both cysteine residues rather than cross-linking. In line with these results in one mutant, D133C/G338C, treated with MTS-2-MTS, cross-linking appeared by reducing the concentration of the reagent from 2 mM to 500 M (Fig. 5F, lane f) and even to 50 M (Fig. 5F, lane  g). We therefore applied various concentrations of the reagents (2 to 0.05 mM) in all cross-linking attempts, but no further improvement was obtained.
Accessibility of the Double Cys Replacement Mutants to the Cross-linking Reagents-Apart from the five double Cys replacement mutations described above none of the rest (16 mutants) showed cross-linking with any of the three reagents. These negative results can be explained at least in two ways: (a) the distances between the double Cys replacements are longer than the distance between the bifunctional reactive groups of the reagent, so that cross-linking cannot take place; (b) one (or both) Cys replacement in the pair is not accessible to the cross-linking reagents.
Therefore, it was important to determine the accessibility of the three cross-linking reagents to the double Cys replacement mutations that either did not cross-link at all or cross-linked partially. For this purpose, the membranes of the double mutants were treated in situ with the cross-linking reagents as described above. Then the affinity-purified protein, bound to the Ni 2ϩ -NTA beads, was exposed to fluorescein maleimide to detect by fluorescence the Cys residue left free in the protein.
Previous results (18,36) and the results presented below show that under these incubation conditions, free Cys replacements in the protein are accessible to fluorescein maleimide. However, it should be emphasized that total or drastic abolition of the fluorescence could occur only when both Cys replacements are accessible to the cross-linker (Fig. 5B). When only one of the double Cys replacements is accessible or when both are inaccessible, the free Cys replacements can be labeled by the fluorescent probe. However, in the former case, the fluorescence labeling would be less. We did not differentiate between these two possibilities and scored both-inaccessible to the crosslinkers. Nevertheless, very good agreement was found between the cross-linking data and the accessibility results, implying the credibility of the fluorescein maleimide-based accessibility test. Thus, the results of two accessibility tests conducted with the double mutants P129C/L334C that did not show any crosslinking (data not shown) and T132C/S342C that showed crosslinking with all three reagents (Fig. 5B) are presented in Fig. 5, G and H, respectively. It is apparent that mutant P129C/ L334C is not accessible to all three cross-linkers, since the level of fluorescence labeling (Fig. 5G, lower panel, lanes b-d) was identical to the untreated control (Fig. 5G, lower panel, lane a). In marked contrast, as expected, the mutant T132C/S342C is accessible to all three cross-linking reagents, since pretreatment with these cross-linkers inhibited the subsequent labeling by the fluorescent probe (Fig. 5H, lower panel, lanes b-d). Accordingly, mutants that cross-linked only with MTS-2-MTS (A137C/S342C and A137/G338C) (Fig. 5, D and E) or partially with BMH (D133C/G338C) (Fig. 5F) were found accessible only to the respective cross-linkers (data not shown).
The mutants A127C/L334C and P129C/L334C were not accessible to all three cross-linking reagents and in agreement did not perform any cross-linking (data not shown). This inaccessibility can be due to a hydrophobic environment surrounding the Cys replacements in the TMS in which Cys residues are less reactive. Similar to the metal-tetracycline/H ϩ antiporter (42), many Cys replacements in NhaA (14,25) are not accessible to SH reagents (reagents that react with Cys residues in proteins). Most interestingly, four double mutants G125C/ S342C, G125C/L334C, G125C/G338C, and A127C/G338C were accessible to all cross-linkers and yet were not cross-linked by them (data not shown). Most probably, these double mutants are separated by more than 15.6 Å, the span of BMH, the longest cross-linker used. The double mutants A127C/S342C, P129C/S342C, P129C/G338C, T132C/L334C, D133C/L334C, F136C/G338C, F136C/S342C, and A137C/L334C, were all accessible to MTS-2-MTS but were not cross-linked by this reagent. The double mutant D133C/S342C was accessible to BMH and slightly to MTS-2-MTS but was not cross-linked by either. Hence, the lack of cross-linking of these double Cys replacements cannot be ascribed to inaccessibility to the cross-linking reagents. Most probably, these double Cys replacements are not close enough to be cross-linked by these short reagents (23). It should be noted that we optimized the cross-linking conditions only with respect to concentrations of the three reagents (see above) but not with respect to other conditions such as pH, temperature, and prolonged incubation time. Therefore, the number of the residues that cross-linked may be underestimated.
Effect of the Bifunctional Cross-linking on the Na ϩ /H ϩ and Li ϩ /H ϩ Antiporter Activity-Cross-linking of the Cys replacement mutant A137C/G338C by MTS-2-MTS (Fig. 5E) inhibited totally (data not shown) the very low Na ϩ /H ϩ antiporter activity (Fig. 3C) of the mutant. Cross-linking of the double Cys replacement mutant T132C/S342C (Fig. 5B) that had a significant Na ϩ /H ϩ antiporter activity (Fig. 3B) was 80% inhibited by treatment with the three cross-linking reagents (data not shown). Although the double mutants A137C/S342C and D133C/G338C were cross-linked by certain of the cross-linkers (Fig. 5, D and F, respectively), their activity (Fig. 3B) was not affected by the cross-linking (data not shown).
Most surprising was the behavior of the double mutant T132C/G338C that cross-linked by all three cross-linking reagents (Fig. 5C). This mutant has neither Na ϩ /H ϩ nor Li ϩ /H ϩ antiporter activity (Figs. 3B and 6A and Table II), yet crosslinking of the dead mutant with o-PDM restored Li ϩ /H ϩ activity only at pH 8.5 and at high Li ϩ concentrations (50 -100 mM; Fig. 6B). Most importantly, BMH (that also cross-linked T132C/G338C (Fig. 5C)) did not restore activity (Fig. 6B). Hence, it is mere cross-linking, rather than the type of crosslinker, that determines whether the cross-linked sample is active or not.
Strikingly, MTS-2-MTS, a shorter (5.2-Å) cross-linking reagent compared with o-PDM (7.7-10.5 Å (23, 43)), was much more potent in activating T132C/G338C. Cross-linking by MTS-2-MTS restored the Li ϩ /H ϩ antiporter activity at all pH values with a pH dependence between pH 6.5 and 8.5 very similar to the wild type control (Fig. 6C). Furthermore, crosslinking by MTS-2-MTS also restored the Na ϩ /H ϩ antiporter activity (about 50% dequenching) and its activation by pH between 7 and 8 (Fig. 6C). We therefore suggest that crosslinking between TMSs IV and XI, at positions Thr 132 and Gly 338 by the short cross-linking reagents, restores an active conformation of the antiporter that is crucial for antiporter activity and pH regulation.

DISCUSSION
Using Cys-scanning mutagenesis of either single conserved residues in TMS IV or XI or double residues, one in each TMS, of the NhaA antiporter, we identified new mutations that affect the transport activity and/or pH regulation of NhaA. Furthermore, we revealed positions that, when replaced simultaneously in the two TMSs, showed negative interaction in trans, implying functional and/or structural interactions between the two TMSs. By scanning the double Cys replacement mutants with cross-linking reagents of various length and flexibility, we have proven that many of these positions in TMSs IV and XI are indeed within very close proximity. Strikingly, for the first time, cross-linking revived activity and even the pH dependence of a completely inactive double mutant (T132C/G338C) of NhaA. These results highly suggest that the cross-linking restores an active conformation of NhaA in which both TMSs IV and XI participate.
Although only conserved residues were replaced, none were found essential for growth in high Na ϩ -selective media, at least at neutral pH (Table II). Nevertheless, this mutagenesis identifies new residues that are important for the antiporter activity. In TMS IV, F136C, G125C, and A137C increase the apparent K m for Na ϩ and Li ϩ of the antiporter activity (Table II); in addition to its effect on the apparent K m , the mutation G125C drastically shifts the pH dependence of the antiporter (by one pH unit) to the alkaline range (Fig. 2B). We have previously isolated mutations in TMS IV (P129L, T132C, and D133C) that affect drastically the apparent K m of the antiporter for Na ϩ and Li ϩ (14). It is remarkable that these mutations together with the new mutations cluster in one face of TMS IV (Fig. 1, B and  C), implying its importance in determining the apparent K m for the ions of the antiporter. Notably, TMS IV is one of the most conserved TMS in NhaA (14).
The mutations L334C, G338C, and S342C of TMS XI had a minor effect on the apparent K m of the antiporter for both ions (Table II). However, whereas L334C had a pH profile very similar to that of the wild type, the pH profile of S342C was shifted to the acidic range by one pH unit (Fig. 2C), and G338C lost entirely its control by pH (Fig. 2D). We have previously found that G338S lacks the pH control of NhaA (17). Taken together, these results imply the importance of TMS XI in the pH response of NhaA and corroborate our previous results, showing that residues affecting the pH dependence of NhaA can be separated from those affecting the apparent K m of the antiporter (14).
Comparison of the growth and transport phenotypes of the double mutants with those of the respective single mutants shows a very pronounced difference in many double mutants that cannot be explained by the phenotype of the single mutants or by differences in expression levels between the single and the double mutants (Table II and Figs. 2 and 3). Among the 21 new double mutants, only six (A127C/L334C, A127C/S342C, P129C/L334C, P129C/S342C, D133C/L334C, and D133C/ S342C) were very similar to the wild type in apparent K m for Na ϩ and pH dependence. In seven double mutants, mostly negative effects were introduced by the second mutation (A127C/G338C, P129C/G338C, T132C/L334C, T132C/S342C, A137C/L334C, A137C/G338C, and A137C/S342C), as compared with the respective single mutations. Nevertheless, these mutations still retained a significant antiporter activity. Most strikingly, eight of the double mutants (G125C/L334C, G125C/ G338C, G125C/S342C, T132C/G338C, D133C/G338C, F136C/ L334C, F136C/G338C, and F136C/S342C) lost entirely their antiport activity, although certain activities were found in the respective single mutants (Table II; compare Fig. 2 with Fig. 3). Hence, in addition to being important for activity, certain residues in either TMSs IV or XI have to be compatible, in trans, with residues in the other TMS. This need of compatibility, in trans, suggests that there is a strong functional and/or structural interaction between many residues in these two TMSs that are needed for ion translocation and/or pH regulation.
The in situ intramolecular site-directed chemical cross-linking between TMSs IV and XI has proven that six positions are within very close proximity (Figs. 4 and 5). The determination of distances by cross-linking is based on the premise that Cys cross-links are a measure of proximity. However, it should be emphasized that cross-link formation detects dynamic collisions and chemical reaction between residues, not simply their proximities. We therefore conducted the cross-linking reaction in situ on the membranes. The in situ approach increases the specificity of the cross-linking reaction in several ways (44): (a) the membrane is the environment in which a membrane protein adopts its native conformations; (b) in the lipid bilayer, the protein motions are restricted, and collisions can only occur within the two-dimensional space of the membrane; (c) a strong correlation was found between collision rates in the membrane and proximity (45,46); and (d) the recently solved atomic structure of the Lac permease (47) showed in many positions, good correlation between cross-linking data and the actual distances within the protein. In addition, two essential controls were carried out to validate the specificity of the crosslinking reactions to the double Cys replacements in NhaA. (a) As described above (Fig. 5A), there was no cross-linking in CL-NhaA. (b) None of the single mutants cross-linked intermolecularly.
All of the double Cys replacement mutants of NhaA were tested with three bifunctional cross-linking reagents that differ in flexibility and the distance spanning the reactive groups. BMH is flexible and can span 3.5-15.6 Å (43), o-PDM is rigid and spans 7.7-10.5 Å (43), and MTS-2-MTS is the shortest (5.2 Å (23)). The double mutants T132C/G338C, T132C/S342C, and D133C/G338C cross-linked with all three cross-linking reagents. These results are consistent with the suggestion that the cross-linking reagents are not so rigid, and the distances spanned by them overlap to a certain extent (43). Another possibility that cannot be excluded is that the respective TMSs at these positions possess certain mobility, allowing the crosslinking to take place with all three reagents. The double mutants A137C/S342C and A137C/G338C are inaccessible to the longer cross-linkers but cross-link efficiently with MTS-2-MTS. We assume that the distance spanning all five double Cys replacements that are cross-linked by MTS-2-MTS is 5.2 Å. The distance between F136C and S342C that perform the S-S bond is considered to be around 2 Å (48).
The cross-linking data were used to construct a distance map between the respective Cys replacements in TMSs IV and XI. We assume that the estimated distances between the double Cys replacements are similar to the respective positions in native NhaA. Our cross-linking data are most consistent with the packing models of TMSs IV and XI shown in Fig. 1, B and C. Most importantly, these putative models are most consistent with the cross-linking, accessibility, and physiological results obtained. Thus, approximately in the middle of TMS IV and TMS XI, all residues that cross-link cluster in one face of each helix (Fig. 1, B and C). We suggest that this is an interface domain where the helices are in very close proximity. The double Cys replacements that are accessible to all three crosslinkers but do not cross-link (data not shown) are located wide apart (pairs of G125C) or on the opposite face of the helix (A127C/G338C). Most interestingly, the double Cys replacements that are active (Table II) are located far apart from the interface domain (all pairs of A127C and P129C). In marked contrast, in many double mutants that show strong negative complementation, at least one member of the double Cys replacements is located in the interface domain (pairs of F136C, D133C/G338C, T132C/G338C, G125C/G338C, and G125C/ S342C) (Table II and Fig. 1, A and B), implying that the interface is very sensitive to replacements of its native residues.
In this paper, we also show that testing the effect of crosslinking on the activity of the respective mutants is a very powerful tool to discover positions in two different TMSs where their proximity in trans has most important functional implications. Cross-linking inhibited drastically the mutant T132C/ S342C but not the mutant A137C/S342C. This result may suggest that whereas, at the former position, flexibility between the two helices is important, it is not at the latter position. However, we cannot exclude the possibility that the cross-linking affected the activity of T132C/S342C and A137C/ G338C by chemical modification rather than by cross-linking. In marked contrast, cross-linking of the dead mutant T132C/ G338C by o-PDM revived its Li ϩ /H ϩ antiport activity, whereas cross-linking by BMH had no effect. Hence, it was the correct proximity between T132C and G338C, rather than the chemical modification, that restored activity. Cross-linking by the shorter reagent, MTS-2-MTS was even more efficient in restoring activity; it revived both Na ϩ /H ϩ and Li ϩ /H ϩ antiporter activities and even the pH response of the T132C/G338C mutant.
The revival of the antiporter activity by cross-linking is remarkable. Since NhaA is a dimer with functional implications (15), it is possible that the double Cys replacements T132C/ G338C destabilizes NhaA dimer, and the cross-linking brings the two subunits together. However, we believe that this intriguing interpretation is unlikely for the following reasons: (a) we did not find any inter-molecular cross-linking between the single or double mutants; (b) the trypsin digest of the crosslinked double mutants of NhaA (Fig. 5C) shows only the products expected from the cross-linked monomer. We therefore suggest that the cross-linking at the interface between the helices, at position T132C/G338C, restores a conformation of NhaA that is essential for its transport activity and pH regulation.
In summary, we unraveled closed proximity between TMSs IV and XI of NhaA and identified a functionally important interface domain between the two TMSs. In addition, we show that scanning activity of single and double Cys replacements combined with cross-linking provides a most powerful tool for determination of structure and function relationship between pairs of TMSs in polytopic membrane proteins.