Negative dominance studies demonstrate the oligomeric structure of EmrE, a multidrug antiporter from Escherichia coli.

EmrE, the smallest known ion-coupled transporter, is an Escherichia coli 12-kDa protein 80% helical and soluble in organic solvents. EmrE is a polyspecific antiporter that exchanges hydrogen ions with aromatic toxic cations such as methyl viologen. Since it is many times smaller than the classical consensus 12 transmembrane segments transporters, it was particularly interesting to determine its oligomeric state. For this purpose, a series of nonfunctional mutants has been generated and characterized to test their effect on the activity of the wild-type protein upon mixing. As opposed to the wild type, these mutants do not confer resistance to methyl viologen, ethidium bromide, or a series of other toxicants. Co-expression of each of the nonfunctional mutants with the wild-type protein results in a reduction in the ability of the functional transporter to confer resistance to several toxicants. To perform mixing experiments in vitro, all the mutants have been purified by extraction with organic solvents, reconstituted in proteoliposomes, and found to be inactive. When co-reconstituted with wild-type protein, they inhibit the activity of the latter in a dose-dependent form up to full inhibition. We assume that this inhibition is due to the formation of mixed oligomers in which the presence of one nonfunctional subunit causes full inactivation. A binomial analysis of the results based on the latter assumptions do not provide statistically significant answers but suggests that the oligomer is composed of three subunits. The results described provide the first in vitro demonstration of the functional oligomeric structure of an ion-coupled transporter.

A great diversity of multidrug transporters are known to us today. They actively remove a wide variety of toxicants in an energy-dependent process and thereby decrease the concentration of the offending compounds near their target. We can group these transporters into several different families based on structure similarities (1)(2)(3)(4)(5). A unique family (Smr or Mini-TEXANs) is represented by very small proteins, about 100 amino acids long, that render bacteria resistant to a variety of toxic cations (6,7). Two MiniTEXANs, Smr (8,9) and EmrE (10) have been characterized, purified, and reconstituted in a functional form. Both proteins catalyze H ϩ /cation antiport in proteoliposomes reconstituted with purified transporter, and they are capable of recognizing a wide range of inhibitors and substrates. In addition, EmrE has been shown to display unique properties of solubility in organic solvents such as a mixture of chloroform and methanol (10). After solubilization in the above solvent, the protein retains its ability to transport, as judged from the fact that it can be reconstituted in a functional form. Using transmission FTIR and oriented attenuated total reflection-FTIR spectra, the protein was found to be a 4-membered transmembrane antiparallel helical bundle. The helices in EmrE are oriented perpendicular to the lipid bilayer with a tilt angle of 27°with respect to the bilayer normal (11). Residues in the substrate translocation pathway have been identified with thiol reactive substrates (12).
Although it is evident that EmrE is an ion-coupled transporter, it is three to eight times smaller than the classical 12 transmembrane segments (TMS) 1 consensus. Thus, the question whether this small protein can function as a monomer was raised and is addressed in this work. A successful approach, used with channel proteins such as Shaker (13), MinK (14), and others, is based on mixing experiments in which inactive subunits or subunits with modified properties were mixed with the wild-type protein at various ratios. The results suggested the formation of oligomers and allowed for an estimate of its size.
In this work, evidence is presented that EmrE functions as a homooligomer. The effect of several inactive mutants on the activity of the wild-type protein was tested in a mixing approach both in vitro and in vivo. In the in vivo experiments, various mutants were coexpressed with the wild-type protein and shown to significantly decrease the ability of the latter to confer resistance to various toxicants. To further substantiate this finding, the mutant proteins were partially purified and co-reconstituted with wild-type EmrE. The nonfunctional protein has a negative dominant effect over the functional one manifested by a full inhibition of transport activity. The results demonstrate that EmrE functions as a homooligomer.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-Escherichia coli JM109 (15) and TA15 (16) are used throughout this work. The plasmids used are pKK56 (10), pT7-32 (10), pSN1, 2 and pGP1 (17). In plasmid pKK56, EmrE is cloned into EcoRI and HindIII sites of pKK223-3 (Pharmacia Biotech Inc.), and in plasmid pT7-32, EmrE is cloned into NdeI and HindIII sites of pT7-7. pSN1 is a derivative of pACYC184 in which the BamHI-HindIII insert of pKK56 was transferred into the corresponding sites of the original vector. pSN1 and pKK56 share the same promoter, but they are compatible so that they can be used to co-express two different proteins in the same cell.
Mutagenesis-Mutants were obtained by polymerase chain reaction mutagenesis using the overlap extension procedure described by Ho et al. (18). For most of the mutations, a set of two overlapping oligonucleotide primers containing the desired mutation were constructed. The outside primers were those used for the wild-type EmrE (10). In one of the mutations (E14C), only two outside primers were used, and one of them contained the desired mutation.
Mutagenic oligonucleotides were prepared incorporating a unique * This research was supported by a grant from the Israel Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 972-2-6585992; Fax: 972-2-634625; Email: shimons@leonardo.ls.huji.ac.il. restriction site to facilitate mutant identification, and in several cases, this required an additional conservative mutation that did not affect EmrE expression. Mutated DNA was identified by the acquisition of the new restriction site and sequenced to ensure that no other mutations occurred during the amplification process (T7 Sequencing TM Kit, Pharmacia).
The primer 5Ј-CGGAATTCATATGAACCCTTATATTTATCT TGGT-GGTGCAATACTTGCATGTGTCATTGG was used to obtain the E14C mutant. In this case, the unique MnlI site in the wild type is abolished.
5Ј-GCTTATGCTATCTTCTCAGGAGTCGG with the unique MboII site yielded W63F mutant; 5Ј-CAGGGATTGCATTCGCTATCTGGTC, with the unique BsmI site and a conservative change in Ala-59, was used to give Y60F mutant. In all the mutations, wild-type DNA was used as a template.
Resistance to Toxic Compounds-For testing resistance to toxic compounds, cells were grown at 37°C in LB medium with different concentrations of the compound, and cell density was estimated from absorbance at 600 nm (A 600 ).
Overexpression and Specific Labeling of EmrE-pT7-32, which contains the T7 polymerase promoter 10 and the translation start site for the T7 gene 10 protein, was used for labeling EmrE with [ 35 S]methionine essentially as described previously (10). pT7-32 was transformed into TA15 carrying pGP1-2 (17). Transformants were grown at 30°C in minimal medium supplemented with thiamin (2.5 g/ml), ampicillin and kanamycin (50 g/ml), and 0.5% glucose to a cell density of 0.6 A 600 . The temperature was then increased to 42°C to induce the T7 polymerase; 15 min later, rifampicin (200 g/ml) was added and incubation continued for an additional 10 min. Then the culture was shifted back to 30°C for 60 min. [ 35 S]methionine (specific activity of 1350 Ci/mmol) was added to the cell suspension (10 Ci/ml) and incubation continued for an additional 40 min. Cells were collected by centrifugation, washed with a solution containing 20 mM Tris-Cl, pH 7.5, and 150 mM NaCl, and sonicated 6 times for 10 s using a probe-type sonicator. Undisrupted cells were removed by centrifugation, and the membranes were then collected by further centrifugation at 435,000 ϫ g for 15 min. The membrane pellet was resuspended in the above buffer, frozen in liquid air, and stored at Ϫ70°C.
For overexpression, E. coli JM109/pKK56 was grown in minimal medium A with 0.5% glycerol, thiamin, and ampicillin as above. When the culture reached an A 600 ϭ 1.0, isopropyl-1-thio-␤-D-galactoside was added to 0.5 mM; 2 h later, the cells were harvested by centrifugation. After washing, cells were resuspended in buffer containing 10 mM Tris-Cl, pH 7.5, 250 mM sucrose, 150 mM choline chloride, 0.5 mM dithiothreitol, 2.5 mM MgSO 4 , and 15 g/ml DNase I and sonicated 7 times for 10 s using a probe-type sonicator. Undisrupted cells were removed by centrifugation, and the membranes were then collected by further centrifugation at 435,000 ϫ g for 15 min. The membrane pellet was resuspended in the above buffer, frozen in liquid air, and stored at Ϫ70°C.
Purification and Reconstitution of EmrE-EmrE was extracted essentially as described (10) from 1 ml of membrane (10 mg of protein) with 23 ml of chloroform:methanol, 1:1. After an incubation of 20 min on ice, 4.6 ml of water were added for phase separation, and the suspension was centrifuged. The upper phase and the interphase were removed, and the lower phase with the enriched EmrE was stored at Ϫ70°C. For analysis in SDS-polyacrylamide gel electrophoresis, a sample was dried, resuspended in sample buffer, and analyzed in 16% tricine gels as described (19).
For reconstitution, protein (0.6 -7 g) in the organic solvent was mixed with 18 l of E. coli phospholipids (50 g/ml) and 180 l chloroform:methanol, 1:2. The suspension was dried under argon and resuspended in a solution (60 l) containing 0.19 M NH 4 Cl, 0.015 M Tris-HCl, pH 6.9. The suspension was frozen and kept at Ϫ70°C. Before the assay, the proteoliposome suspension was thawed and sonicated in a bath-type sonicator for a few seconds until clear.
Transport Assay-Transport of [ 14 C]methyl viologen into proteoliposomes was assayed by dilution of 3 l of the ammonium chloride containing proteoliposomes into 200 l of an ammonium-free medium containing 25 M [ 14 C]methyl viologen (60 nCi/assay), 140 mM KCl, 10 mM tricine, and 5 mM MgCl 2 , pH 8.5. At given times, the reaction was stopped by dilution with 2 ml of the same ice-cold solution without the radioactive substrate, filtering through Schleicher & Schuell filters (0.2 m), and washing with an additional 2 ml of solution. The radioactivity on the filters was estimated by liquid scintillation. In each experiment, the background value was obtained in a control reaction with 5 M nigericin.
In Vivo Co-expression-pKK56 plasmids that contained mutated EmrE were transformed into JM109 cells carrying pSN1, a derivative of the pACYC184 vector with wild-type EmrE. Resistance of the transformants to toxic compounds was tested as described above.
In Vitro Co-reconstitution-Proteoliposomes were prepared from wild-type and mutated EmrE using different protein ratios as indicated in the specific experiments. The amount of EmrE is quantitated by visualization in Coomassie Blue-stained SDS gels in which known amounts of pure EmrE are used for calibration. Concentration of pure EmrE is determined by a modification of Peterson (20). Transport of [ 14 C]methyl viologen into the proteoliposomes was assayed as described before.

Generation of Inactive Mutants of EmrE-
To generate inactive mutants, residues conserved in all the MiniTEXANS were mutagenized. In addition, the residues chosen were previously shown to play an important role in the homologous Smr protein, as judged from the loss of activity upon mutagenesis (8). Two types of residues were chosen: Glu 14, a charged residue in TMS 1; and Tyr 60 and Trp 63, two aromatics in TMS 3. As expected, the three mutants are inactive. Function was tested in intact cells as the ability to confer resistance to methyl viologen (0.2 mM) in solid media (not shown) and ethidium (600 g/ml) and acriflavin (100 g/ml) in liquid media. Cells transformed with wild-type EmrE are capable of growing under the above conditions (Fig. 1, B and C) at rates comparable with those in the absence of toxicants (Fig. 1A). However, none of the mutant proteins can confer resistance to either of the toxicants mentioned above at the given concentrations (Fig. 1, B and C).
To test whether the mutant proteins are synthesized to similar levels, they were extracted with organic solvents and visualized with Coomasie Blue after SDS-polyacrylamide gel electrophoresis. E14C and Y60F are expressed to levels similar to those of the wild type, as judged from this criterion. The ex- coli JM109 cells were transformed with pSN1 (a pACYC184 derivative containing wild-type EmrE, f) and with pKK56 with each of the mutants as above: E14C (q), W63F (å), or Y60F (ࡗ). For reference, cells transformed with pKK56 mutant E14C alone are also shown (ૺ). The cells were grown in LB medium with 100 g/ml acriflavin. Although not shown, growth of all the strains was identical in the absence of the toxicant. The experiment was repeated twice.
pression of W63F was about three times lower (data not shown). The purified proteins were then reconstituted into proteoliposomes, and ⌬pH-driven [ 14 C]methyl viologen uptake activity was measured. While proteoliposomes reconstituted with the wild-type protein accumulated [ 14 C]methyl viologen against its concentration gradient in a time-dependent process, none of the mutants displayed any activity (Fig. 2). In addition, none of the mutants can catalyze downhill efflux of [ 14 C]methyl viologen from preloaded liposomes (data not shown).
Negative Dominance in Vivo-To test dominance of the various mutations, each of them was co-expressed with wild type in JM109 cells. In these experiments, wild-type EmrE was cloned in pACYC184 to obtain pSN1, and the mutant genes were expressed in pKK56. Cells carrying either the wild type alone or the two plasmids grew at similar rates in the absence of toxicants (data not shown). However, in the presence of acriflavine (100 g/ml; Fig. 1D), the cells carrying the two plasmids grew much slower than the strain carrying only the wild type. Thus, cells co-expressing either one of the mutants together with the wild type grew with a doubling time of about 2-2.5 h, compared with slightly less than an hour for the wild type alone.
Negative Dominance in Vitro-Because the conclusions from in vivo experiments may be ambiguous and difficult to quantitate, EmrE and the various mutant proteins were purified and mixed in vitro as well. The proteins were overexpressed and extracted with organic solvents from membrane preparations of the corresponding strains. The purified proteins were then mixed and reconstituted as described previously (12), and transport of [ 14 C]methyl viologen into the proteoliposomes was then assayed. To minimize possible artifacts the mixing was done in two different ways. In the first one, given amounts of inactive mutant protein were added to fixed amounts of wild type (Fig. 3). In another, the total amount of protein was kept constant, and the ratio between the wild type and the mutants was modified (Fig. 4).
The results of experiments in which the concentration of wild-type protein in the reconstituted proteoliposomes was always constant (30 ng/l) are described in Fig. 3. The effect of the addition of increasing amounts of each of the three inactive mutant proteins is shown in panels A for mutant W63F, B for E14C, and C for Y60F. In the experiments shown in Fig. 3, A-C, three different concentrations of the mutant proteins were added (8,30, and 120 ng/ml) so that their fractional concentration (mutant protein over total amount of protein) is 0.2, 0.5, and 0.8, respectively. Practically no inhibition was observed at the low mutant concentration. Inhibition was maximal at the highest concentration used and intermediate when the fraction of mutant protein was 0.5. It is concluded that each of the mutants has a deleterious effect on the transport catalyzed by the wild type, and the dose dependence of this inhibition is, in essence, quite similar.
The inactive proteoliposomes, reconstituted with 0.8 fractional concentration of the mutants, were also found incapable of catalyzing downhill efflux in experiments in which they were previously preloaded with [ 14 C]methyl viologen (data not shown). These experiments demonstrate that the inhibition of transport is not due to the generation of some nonspecific leak across the membrane.
The simplest interpretation of the results described in this work is that the mixing brings about the formation of composite oligomers between the wild type and the mutant proteins. The results can be analyzed quantitatively assuming that the pres-FIG. 2. Transport activity in proteoliposomes reconstituted with purified EmrE. Wild-type and mutant proteins were purified and reconstituted essentially as described under "Experimental Procedures." Ammonium-loaded proteoliposomes (3 l) were diluted into an ammonium-free medium containing [ 14 C]methyl viologen, and radioactivity incorporated at various time periods was measured. Protein concentration in the proteoliposomes was 75 ng/l. f, wild type; q, E14C; å, W63F; and ࡗ, Y60F. The experiment was repeated three times.

FIG. 3. Negative dominance of three mutations over the wildtype activity.
A-C, chloroform:methanol extracts of the purified mutant proteins W63F (A), E14C (B) and Y60F (C) were added to the wild-type protein prior to reconstitution. The concentration of the wildtype protein in the proteoliposomes was always 30 ng/l. The mutant proteins were added to yield 8 (q), 30 (å), and 120 (ࡗ) ng/l final concentrations. No mutant protein was added to (f). Control liposomes reconstituted with 120 ng/l mutant protein (") are shown in every panel. D, the fractional rate (wild type ϭ 1) of uptake detected in the different proteoliposome preparations is plotted against the fraction of mutant protein in each case. The data shown are a composite obtained by averaging results from experiments performed with five different proteoliposome batches; each of the batches was assayed for transport at least twice. For each experiment, initial rates were calculated from the linear part of the uptake. The theoretical curves (broken lines) were plotted according to the analysis by MacKinnon (13) of a binomial distribution of oligomeric channels and assuming that oligomers with one inactive subunit have no activity (13). The function used is based on the equation: measured rate ϭ f WT n R WT , where f WT is the fraction of wild-type subunits, n is the subunit stoichiometry, and R WT is the rate measured when all the subunits are wild type. For clarity, only theoretical curves for n ϭ 2 and n ϭ 3 are shown. The plotted curve ( full line) was fit to the experimental points using a polynomial fit. ence of one inactive monomer suffices to fully inactivate the oligomer and assuming a binomial distribution of the array of the oligomers as described by MacKinnon (13). Such an analysis is shown in Fig. 3D in which the fractional rate of transport is plotted against the fraction of mutant EmrE protein in the experiment. For clarity, only the theoretical curves for dimer and trimer formation are shown. The experimental points are average values of several independent experiments. The fit to either one of the theoretical curves obtained, assuming two (Fig. 3D, n ϭ 2) or three (n ϭ 3) subunits, is not very good. Our ability to analyze these results quantitatively is hindered by several factors including the untested basic assumption that an oligomer containing one inactive subunit is fully inactive. In addition, a certain innate variability exists in the experimental design because the reproducibility of the reconstitution experiments that are performed for every single point in the curve is within 15%. This reproducibility is quite good for this type of experiment but not accurate enough for the quantitative analysis since the differences between the theoretical curves is not large enough. Therefore, we are unable at present to decide what is the size of the oligomer.
Another experiment was performed to discard the possibility that the inhibition of activity observed could be due to competition for reconstitution of the wild-type protein by the mutants. For this purpose the negative dominance has been demonstrated also in an experiment such as the one described in Fig. 4. The dependence of the rate of transport on the amount of protein added to the reconstitution mix is linear in a very wide range of concentrations up to about 150 ng of protein/l of proteoliposome. For the sake of clarity, only part of the range is shown in Fig. 4, A and B. The results in Fig. 4A demonstrate the time course of uptake of [ 14 C]methyl viologen into proteoliposomes reconstituted with various amounts of EmrE so that its concentration is 5, 10, 15, and 20 ng/l. In Fig. 4B, the initial rates of uptake are plotted against the protein concentration, and a linear dependence is observed. Therefore, if the total EmrE concentration is always kept at 20 ng/l and only the fraction of the mutant to wild type is modified, the transport rates measured will reflect the activity of the mixed oligomers. The results in Fig. 4C demonstrate the effect of the addition of W63F protein to wild-type protein when the total concentration is always 20 ng/l. Increasing the fraction of the mutant protein inhibits the transport activity displayed by the wild type at a given concentration (Fig. 4A). Addition of 5 ng/l W63F to the wild-type at 15 ng/l decreases the activity by 65% (from 84 to 30 pmol/min). Mixing of equal amounts of wild type and W63F yields an activity almost 80% lower than that of the liposomes containing the same amount of wild-type protein (from 40 to 9 pmol/min). Proteoliposomes containing 15 ng/l W63F and 5 ng/l wild type showed practically no activity. A quantitative analysis of the latter results, as performed above, is shown in Fig. 4D. The theoretical curves for dimer, trimer, and tetramer formation are shown. The experimental points are average values of two independent experiments. The fit to either one of the theoretical curves is not very good. As discussed above, with present technology, we are unable to decide what is the size of the oligomer. DISCUSSION In this work, the oligomeric nature of EmrE has been demonstrated. EmrE is the smallest ion-coupled transporter known with only four putative transmembrane ␣-helical segments. Since most ion-coupled transporters are 12 TMS proteins, it has been claimed that this must be their minimal functional size, and it was therefore especially interesting to study the oligomeric nature of a 4 TMS protein.
One particularly difficult problem to resolve with hydrophobic membrane transport proteins is their functional oligomeric state. Because of their nature and the need to use detergents, biophysical and biochemical techniques do not necessarily provide us with information on the functional size of the transporter. Therefore, a genetic approach was used here.
In the in vivo studies, a clear and reproducible negative dominance effect was observed. However, these studies are limited because they may also reflect effects on plasmid copy number and protein expression and/or targeting to the membrane and because of the difficulty in manipulating the dose of each of the proteins. Since we have previously shown that a decrease of even 80% in the activity of some mutants is not necessarily reflected in the phenotype (12), it became necessary to develop an in vitro approach as well.
The in vitro studies are facilitated by the ease of purification and reconstitution of EmrE. They provide the advantage of a simple system that is easy to quantitate and control. To minimize possible artifacts due to competition of insertion into the proteoliposome, the experiments were performed in two modes. In one mode, inactive protein was added to fixed amounts of active EmrE, and a dose-dependent effect on activity was observed. These experiments were possible because of the linearity of the rate of transport observed under a wide range of protein concentrations (only part of the range was shown in Fig. 4B). This linearity was observed even in the very low protein range with a large excess of lipids over protein (1000:1, w/w) where it would have been possible, in theory, to observe an exponential relationship if more than a single monomer is required for active accumulation. The fact that an exponential relationship was not observed, however, can be due to the fact that the oligomers are maintained in the solvent system used or formed prior to reconstitution. Since the experiments described in this work demonstrate subunit mixing, we speculate that the oligomers form while concentrating the protein in the organic solvent, just prior to reconstitution.
In the other mode of the in vitro experiments, the total amount of protein was kept low and constant, and only the ratio between active and nonactive transporters was modified. In both modes, similar results were obtained.
An important aspect in the findings described is the fact that the negative dominance was observed with three independent mutants in which quite different replacements were made. In one, a negative charge in a putative TMS was replaced with an uncharged residue. In the others, two aromatic residues, tryptophan and tyrosine, were replaced. This fact increases the generality of the phenomenon and reduces the possibility of some nonspecific indirect inhibition. In this context, it should also be stressed that the inhibition of uptake is not due to the generation of some nonspecific leak across the membrane because the mixed oligomers were also incapable of catalyzing downhill efflux of substrate. We did not observe positive complementation with any of the possible mutant pairs either in vitro or in vivo. This finding may suggest that all the mutations are in the same part of the catalytic cycle and cannot supplant for each other. It may also hint that three copies of the appropriate residues must be present in order to have a functional oligomer.
The quantitative analysis to determine the size of the oligomer is confronted with two main problems. The technical one is due to the innate variability of the reconstitution assay of about 15%, which does not allow us to make a clear choice between two alternatives in which predicted activities do not differ by much more in most of the range. In addition, the theoretical analysis is based on the assumption that an oligomer containing one inactive subunit is fully inactive. Support for this contention can be assumed from the following facts: the experimental points fit reasonably well to the theoretical curves, full inhibition of activity is obtained when enough mutant protein is added, and no positive complementation has been observed. Although the quantitative analysis does not enable to make a strict distinction between the dimer and the trimer model, a reasonable bias points to the latter. This propensity to choose the trimer model is due to the fact that, in the range in which the activity is reasonably high, the fit is best to the trimer model (Fig. 3D). The deviation is more pronounced in the range of low transport activity tested (high mutant concentration). Our bias rests also on the overwhelming consensus that a vast majority of the transporters known contain 12 putative TMS. The MiniTEXANs are a notable exception to the rule with a few others being 10 or 14 TMS (1, 2). Another interesting exception to the 12 TMS consensus is the 6 TMS adenine nucleotide translocator from mitochondria, which has been proposed to function as a dimer (21). Since the 12 TMS proteins are the vast majority and it would seem that this is the basic functional unit, it is interesting to speculate that at some time in evolution and under given conditions there must have been some kind of advantage to the larger proteins as opposed to the small and more rare ones. If so, it may be interesting to find out why some of the small ones survived evolutionary pressures. It is possible to speculate that one advantage of forming oligomers as compared with a larger monomer is the feasibility of mixing various subunits with slightly different properties that may very much broaden the range of functional possibilities, as is seen in the case of receptors for glutamate and other neurotransmitters in the nervous system of higher organisms (for example, see Ref. 22). In the same vein, we should raise some caution about the generality of the theory that all 12 TMS transporters must have evolved from a gene duplication of two halves (6 ϩ 6 model (1)). This generalization may now require some further evaluation in light of the presence of 4 TMS transporters.
A genetic approach to study the oligomeric structure of ioncoupled transporters has been used in several cases. In the case of the ␤-galactoside transporter from E. coli, despite intensive studies and initial results that suggested dominance of some mutations (23), no genetic complementation has been found between point mutants (24). Complementation has been observed only between two transporters in which large deletions were engineered in different domains of the protein, as has been reported (25). This phenomenon has been interpreted as demonstrating that complementation can occur only in molecules in which the deletion leaves a gap in the structure that can be filled with another molecule containing the deleted segment (26). In the tetracycline resistance protein (TetB) two intracistronic complementation groups have been demonstrated corresponding to the N-and C-terminal halves of the transporter. The results have been interpreted as suggesting that the Tet proteins occur as oligomers (27). However, to the best of our knowledge, EmrE is the first ion-coupled transporter that has been shown to function as an oligomer using in vivo and in vitro approaches.