Functional Symmetry of UhpT, the Sugar Phosphate Transporter of Escherichia coli *

UhpT, the sugar phosphate transporter coli , acts to exchange internal inorganic phosphate for external hexose 6-phosphate. Because of this opera-tional asymmetry, we studied variants in which right-side-out (RSO) or inside-out (ISO) orientations could be analyzed independently to ask whether the inward- and outward-facing UhpT surfaces have different substrate specificities. To study the RSO orientation, we constructed a histidine-tagged derivative, His 10 K291C/ K294N, in which the sole external tryptic cleavage site (Lys 294 ) had been removed. Functional assay as well as immunoblot analysis showed that trypsin treatment of proteoliposomes containing His 10 K291C/K294N led to loss of about 50% of the original population, reflecting retention of only the RSO population. To study the ISO orientation, we used a His 10 V284C derivative, in which a newly inserted external cysteine (Cys 284 ) conferred sensitivity to the thiol-reactive agent, 3-( N -maleimidylpro-pionyl)biocytin. In this case, 3-( N -maleimidylpropio-nyl)biocytin treatment of proteoliposomes containing His 10 V284C gave about a 60% loss of activity, and immu- nodetection of biotin showed parallel modification of an equivalent fraction of the original population. Together, such findings indicate that the UhpT RSO and ISO orientations are in about equal proportion in proteoliposomes and that a single population can be generated by exposure of these derivatives to the appropriate agent. This allowed us to study proteoliposomes with UhpT functioning in RSO orientation (His 10 K291C/K294N) or ISO orientation (His To detect modification by MPB, we used low SDS ( , 0.15%) during sample preparation and electrophoresis to preserve the UhpT-MPB-streptavidin complex (9). Chemicals— Detergents were purchased from Calbiochem-Novabio-chem. E. coli phospholipid was from Avanti Polar Lipids, Inc., trypsin (sequencing grade, tosylphenylalanyl chloromethyl ketone-treated) was from Worthington Biochemicals, and MPB was from Molecular Probes, Inc. D -1-[ 14 C]G6P (54 mCi/mmol) was obtained from NEN Life Science Products.

In Escherichia coli, the uhp locus (for uptake of hexose phosphates) coordinates the expression of four proteins responsible for incorporation of external sugar phosphate. UhpA and UhpB together form a two-component regulatory system that activates expression of the transporter, UhpT, after extracellular G6P binds to the membrane receptor, UhpC (1). UhpT then acts to move sugar 6-phosphate inward in exchange for internal inorganic phosphate (2) in an electrically neutral antiport reaction (2,3). Hydropathy analysis of the UhpT amino acid sequence (4), the properties of UhpT-PhoA fusions (5,6), and comparisons with other members of the major facilitator superfamily (7,8) all argue that UhpT has 12 transmembrane ␣-helices and that its N and C termini lie in the cytoplasm (see Fig. 1). Accordingly, the structure of UhpT (and related transporters) is strongly asymmetric along an axis perpendicular to the membrane surface. This fact, along with the biochemical asymmetry of the in vivo reaction, in which external sugar phosphate can exchanged for internal phosphate, raises the question of whether UhpT has an inherent bias in substrate preference at its extracellular and intracellular surfaces.
To address this issue, we used an in vitro preparation in which purified protein can be studied in proteoliposomes loaded with inorganic phosphate. Because proteoliposomes might have UhpT in either (or both) a right-side-out (RSO) 1 or inside-out (ISO) orientation, we also designed UhpT derivatives in which one or the other of these orientations could be eliminated by exposure to an appropriate blocking agent. For example, in one case we reasoned that the high proportion of lysine and arginine residues at the UhpT inner surface would confer sensitivity to trypsin (see Fig. 1). To exploit any such side-specific trypsin sensitivity, we constructed a variant lacking the sole tryptic cleavage site at the UhpT extracellular surface (Lys 294 ), so that only RSO molecules would survive exposure to protease. Alternatively, we inserted cysteine (Cys 284 ) into the same extracellular loop (see Fig. 1), creating a mutant inhibitable by low concentrations of MPB, a thiol-reactive agent without effect on wild type UhpT (9); in this case, only ISO molecules would remain active after MPB exposure.
Study of these variants suggests that the RSO and ISO forms are present in about equal proportions in proteoliposomes. This finding, in turn, made it possible to ask whether the two orientations showed differential substrate specificity. For this purpose, we conducted a simple kinetic study of G6P transport by proteoliposomes containing UhpT in mixed, RSO, and ISO orientations and also evaluated the inhibitions exerted by alternative UhpT substrates. We found no substantial differences in the behavior of UhpT in its different orientations, indicating that the transporter possesses functional symmetry. the His 10 V284C derivative, a single mutation (V284C) removing an AccI site was introduced. The nature of each mutant was verified by sequencing (Core Facility, Johns Hopkins University). As host for plasmids encoding His 10 UhpT and its derivatives, we used E. coli strain RK5000, which carries a deletion within the chromosomal uhp locus (4). Expression of His 10 UhpT and its derivatives was under the control trc promoter (10).
Purification and Reconstitution-To purify His 10 UhpT and its derivatives (10), membrane vesicles (20 -40 mg of protein) (11) were solubilized at 4°C by addition of 8 ml of buffer A (200 mM NaCl, 100 mM potassium phosphate, pH 8, 50 mM G6P, 10% (v/v) glycerol, 1.5% (w/v) n-dodecyl-␤-maltoside, 0.2% (w/v) E. coli phospholipid, 5 mM 2-mercaptoethanol) and mixed with 0.4 ml of nickel-nitrilotriacetic acid-agarose equilibrated with buffer A. After overnight incubation, the mixture was placed in a Poly-PreP® column (Bio-Rad) and washed with 15 ml of buffer B (buffer A prepared at pH 7 containing 50 mM imidazole and 1.5% n-octyl-␤-D-glucopyranoside in place of 1.5% n-dodecyl-␤-maltoside). Before elution, the column was washed with 0.5 ml of buffer B in which n-dodecyl-␤-maltoside replaced n-octyl-␤-D-glucopyranoside; bound protein (0.3-0.5 mg) was eluted in 0.5 ml of buffer B with 200 mM imidazole added in place of 200 mM NaCl. Purified protein (5-10 g) was reconstituted by detergent dilution into proteoliposomes loaded with 100 mM potassium phosphate (pH 7) as described (10), isolated and washed by centrifugation, and finally resuspended (4°C) in 1 ml of assay buffer (100 mM potassium sulfate, 50 mM MOPS/K , pH 7). To study the UhpT RSO orientation, proteoliposomes with His 10 K291C/ K294N were resuspended with or without 1 mg/ml trypsin and then incubated at 33°C for 1 h before adding 0.3 mM phenylmethylsulfonyl fluoride to stop the reaction. To examine the ISO orientation, proteoliposomes with His 10 V284C were resuspended with or without 0.2 mM MPB before a quench with 20 mM 2-mercaptoethanol. After either treatment (trypsin or MPB), the suspension was expanded to 20 ml with iced assay buffer and centrifuged, and proteoliposomes were finally resuspended in 1 ml of assay buffer.
Transport Assay-To assay transport, washed proteoliposomes were diluted 10-fold with assay buffer and preincubated at 23°C for 3 min before adding 50 M [ 14 C]G6P. At the indicated times, 100-l aliquots were filtered on 0.22-m GSTF Millipore filters and washed twice with 5 ml of assay buffer. To assess competition between G6P and alternate substrates, the test substrate was added together with 50 M [ 14 C]G6P for 2 min before filtration.
SDS-PAGE and Immunoblot Analysis-After SDS-PAGE (12), protein was visualized by systems directed to either the UhpT N or C terminus. In one case, we used horseradish peroxidase-conjugated nickel-nitrilotriacetic acid (Qiagen) to monitor the N-terminal polyhistidine; alternatively, we used a rabbit polyclonal antibody directed against the UhpT C terminus (9,10). Signals were developed by enhanced chemiluminescence (Amersham Pharmacia Biotech) and quantitated as described (8). To detect modification by MPB, we used low SDS (Ͻ0.15%) during sample preparation and electrophoresis to preserve the UhpT-MPB-streptavidin complex (9).

RESULTS
Early work showed that addition of trypsin to sensitized intact cells leads to cleavage of wild type UhpT but not of a variant lacking Lys 294 (9). However, subsequent studies in proteoliposomes appeared to contradict this finding, because even mutants lacking Lys 294 had some deficit after exposure to protease (see below). A simple explanation for this discrepancy is that UhpT exists in mixed orientation in proteoliposomes, with some molecules inserted RSO and others ISO. Thus, the partial trypsin sensitivity of Lys 294 variants would reflect cleavage of the ISO orientation but not the RSO population. This hypothesis prompted us to develop two UhpT variants, His 10 K291C/K294N and His 10 V284C (see "Experimental Procedures"), whose properties would allow us to evaluate the proportion of RSO and ISO forms in proteoliposomes and to characterize the properties of each orientation.
Comparison of His 10 UhpT with His 10 K291C/K294N and His 10 V284C-The variants studied in this work each have a polyhistidine N-terminal extension to facilitate protein purification (10). The parental molecule, His 10 UhpT, was otherwise wild type in sequence. In one derivative, His 10 K291C/K294N, Lys 291 and Lys 294 were replaced by cysteine and asparagine, respectively, thereby eliminating tryptic cleavage site(s) 2 exposed in the extracellular loop between TM7 and TM8 ( Fig. 1).
Using this variant, we anticipated that only RSO molecules would survive exposure of proteoliposomes to trypsin. In a second derivative, His 10 V284C, we placed a cysteine in the same extracellular loop ( Fig. 1) to provide a target for the thiol-reactive agent, MPB (9); here, we expected MPB modification of Cys 284 to inhibit the RSO but spare ISO proteins. Initial experiments showed that each variant could be purified in high yield, and a simple kinetic study of [ 14 C]G6P transport showed comparable behavior for both wild type and mutant proteins (Fig. 2); maximal velocities were within a 2-fold range, and in each case we recorded a Michaelis constant (K m ) of about 50 M. We confirmed that treatment of His 10 K291C/K294N proteoliposomes with trypsin and His 10 V284C proteoliposomes with MPB gave partial inhibitions (about 50% in each case) as expected if both RSO and ISO populations were present (data not shown; see below). In those cases, we also found that the K m for G6P transport was largely unaffected by treatment with trypsin or MPB. Thus, with His 10 K291C/K294N we noted K m values of 64 and 67 M, respectively, for the control and trypsin-treated material, whereas with His 10 V284C, we found K m values of 32 and 39 M for control and MPB-treated preparations. These findings would suggest that the RSO and ISO orientations have comparable kinetic responses to G6P (K m values of 67 and 39 M, respectively), if it can be shown that trypsin and MPB act in the manner predicted. In the experiments described below, therefore, our first goal was to provide direct evidence that these probes function as expected.
RSO Molecules Remain after Exposure of His 10 K291C/ K294N Proteoliposomes to Trypsin-The tryptic cleavage site at the UhpT extracellular surface is absent in His 10 K291C/ K294N, leading one to expect that only the lysine/arginine-rich intracellular surface of ISO forms would be sensitive to external trypsin. To test this prediction we studied trypsin sensitiv-FIG. 1. Topology of UhpT. The presumed topology of UhpT is shown (4,8,28). Vertical columns represent the 12 transmembrane ␣-helices (TM1-TM12). The locations of possible tryptic cleavage sites (arginine and lysine) are shown as black circles. The extracellular loop between TM7 and TM8 contains two lysines that have been altered to asparagine and cysteine in the His 10 K291C/K294N variant. The cysteine inserted at position 284 in the His 10 V284C variant is also shown. Previous work (9) showed these residues each lie in the extracellular loop connecting TM7 and TM8.
ity with proteoliposomes containing either His 10 UhpT or His 10 K291C/K294N (Fig. 3). Both proteins showed partial sensitivity at low levels of trypsin (Ͻ10 -100 g/ml), whereas at higher levels (Ͼ100 g/ml), although His 10 UhpT continued to display sensitivity, His 10 K291C/K294N never fell below about 55% of its initial activity (Fig. 3A). These findings are consistent with degradation of both RSO and ISO forms of His 10 UhpT but loss of only the ISO population of His 10 K291C/K294N, and direct tests fully support this interpretation. For example, immunoblots using a C-terminal antibody confirmed that trypsin had acted on the entire His 10 UhpT population but on only a subset of the His 10 K291C/K294N present (Fig. 3B, lanes 1-4). A similar conclusion came after examination of duplicate gels probed with the nickel-nitrilotriacetic acid-conjugated peroxidase designed to detect the polyhistidine N terminus (Fig. 3B, lanes [5][6][7][8]. Thus, cleavage at Lys 294 in the RSO population should generate an 18-kDa N-terminal fragment and a 33-kDa C-terminal peptide, and, in fact, both products are visualized in the appropriate blots (arrows in Fig. 3B, lanes 2 and 6). (The 18-kDa fragment shows poor recovery, perhaps because small hydrophobic proteins are not well retained on nitrocellulose during transfer from the parent gel.) In this same experiment, we also monitored transport of [ 14 C]G6P (Fig. 3C), and those data, too, supported the idea that trypsin targets both RSO and ISO forms of His 10 UhpT but only ISO molecules of His 10 K291C/K294N. Moreover, from the residual activity in His 10 K291C/K294N (55%), one would conclude that the RSO and ISO forms were present in equal proportion; as noted later, this reduction in activity is quantitatively matched by loss of immunoreactivity at 50 kDa (see below). Together, these data indicate that RSO and ISO orientations are present in about equal amounts after reconstitution of UhpT and that preparations containing only the RSO population can be generated by trypsin treatment of His 10 K291C/K294N. ISO Molecules Remain after Exposing His 10 V284C Proteoliposomes to MPB-Work with His 10 K291C/K294N (Fig. 3) gave positive evidence for RSO molecules in proteoliposomes. To document the activity of the remaining ISO population, we next studied His 10 V284C, in which a V284C substitution confers high sensitivity to the thiol-reactive agent, MPB (9). (Cys 291 in His 10 K291C/K294N is not a target for such agents (9).) Proteo-liposomes containing His 10 UhpT were not responsive to MPB at concentrations as high as 200 M, but we found a clear sensitivity on the part of His 10 V284C, whose activity was reduced by 60% (Fig. 4A), as anticipated if only the RSO population of His 10 V284C is inhibitable by the maleimide. With this assumption, the data indicate 50% inhibition of the RSO molecules at 10 -15 M MPB. Matos et al. (9) had found a considerably higher apparent sensitivity for this same mutation (50%

FIG. 3. Effects of trypsin on proteoliposomes containing
His 10 UhpT and His 10 K291C/K294N. Proteoliposomes were treated with trypsin as described under "Experimental Procedures." A, after exposure to trypsin at the indicated concentrations, the initial rate of G6P transport was estimated by a 5-min incubation with labeled substrate. Rates are expressed as a fraction of the initial value. B, proteoliposomes treated with 1 mg/ml trypsin were processed for SDS-PAGE, and in duplicate gels protein was visualized by either the C-terminalspecific antibody (lanes 1-4) or by the N-terminal polyhistidine detection system (lanes 5-9). In each gel, His 10 UhpT is shown at the left (lanes 1, 2, 5, and 6); His 10 K291C/K294N is shown at the right (lanes 3, 4, 7, and 8); control liposomes (no protein) treated with trypsin are shown in lane 9. The arrow and arrowhead indicate, respectively, the N-terminal 18-kDa fragment and C-terminal 33-kDa fragment expected of tryptic cleavage at Lys 294 in His 10 UhpT. C, in the same experiment, inhibition at 1 M MPB), but this may be expected because they exposed for both longer time (10 min versus 2 min) and at higher temperature (23°C versus 4°C). In this experiment (Fig. 4), we had also exposed MPB-treated and untreated proteoliposomes to the impermeant thiol-reactive probe, [2-(trimethylammonium)-ethyl]methane-thiosulfonate (13). In that case, inhibition of activity (by 55%) was found only for the preparation not previously exposed to MPB (data not shown), supporting the idea that MPB reactivity was confined to the RSO population.
To show MPB modification directly, we used streptavidin to generate a His 10 V284C-biotin-streptavidin complex whose formation is reflected by loss of immunoreactivity at 50 kDa and its partial recovery at 80 -90 kDa (9) (Fig. 4B). 3 We found no evidence of such a complex after MPB treatment of His 10 UhpT proteoliposomes (Fig. 4B, lanes 1 and 2) unless endogenous cysteines had been exposed by a preincubation with SDS. By contrast, denaturation was not required for formation of the complex in parallel trials with His 10 V284C (Fig. 4B, lanes 4  and 5). Assays of [ 14 C]G6P transport in this same experiment confirmed that MPB modification was, in fact, associated with inhibition of function ( Fig. 4C and see below).
Distribution of RSO and ISO Orientations in Proteoliposomes-In several experiments, we assessed the relative loss of sugar phosphate transport caused by treatment with trypsin or MPB and compared this with the fraction of protein that remained unmodified as monitored by immunoblots. Both measurements showed parallel change (Table I), and use of either data set leads one to conclude that proteoliposomes contain RSO and ISO orientations in roughly equal proportion.
Functional Equivalence of RSO and ISO Orientations-In the preliminary work comparing His 10 UhpT with its derivatives, His 10 K291C/K294N and His 10 V284C, we noted no significant change of K m for sugar phosphate transport when proteoliposomes were treated to generate the single orientations (see above). This may now be interpreted as evidence that the two orientations have similar kinetic features during G6P transport. To extend this analysis, we measured apparent inhibition constants (K i values) derived by competition between [ 14 C]G6P and a number of alternate substrates. The experiments shown in Fig. 5 give examples of such work using a substrate of relatively high affinity (fructose 6-phosphate, K i Ϸ 0.2 mM) and one of relatively low affinity (phosphate, K i Ϸ 3-5 mM); a number of other experiments of this sort are summarized by Table II. Collectively, those data show that the two UhpT orientations respond in equivalent fashions to the five substrates tested and that the ratio of derived K i values (RSO/ISO) varies no more than 2-fold for substrates of physiological significance (14). Such tests were conducted using proteolipo- FIG. 4. Effects of MPB on proteoliposomes containing His 10 UhpT and His 10 V284C. Proteoliposomes were exposed to MPB as described under "Experimental Procedures." A, after exposure to MPB at the indicated concentrations, the initial rates of G6P transport were measured as described in the legend to Fig. 3. Rates are normalized to the untreated control value. B, proteoliposomes exposed to 200 M MPB were processed for SDS-PAGE, and samples containing His 10 UhpT (lanes 1-3) or His 10 V284C (lanes 4 -6) were analyzed by immunoblot in the presence (lanes 2, 3, 5, and 6) or absence (lanes 1 and 4) of 8 g/ml streptavidin (9). In two cases (lanes 3 and 6) proteoliposomes were solubilized with 0.5% SDS before labeling with MPB and streptavidin. C, in the same experiment,  somes loaded with inorganic phosphate, but in two additional experiments (not shown) we studied His 10 V284C (Ϯ MPB) in G6P-loaded proteoliposomes. In that case, apparent K i values for inorganic phosphate were the same as before (Fig. 5 and Table II). Together with the data noted in Fig. 5 and Table II, these findings lead us to conclude that the two orientations of UhpT are functionally equivalent with respect to processing of fructose 6-phosphate, mannose 6-phosphate, 2-deoxoxyG6P, galactose 6-phosphate, and inorganic phosphate.

DISCUSSION
The orientation of proteins incorporated into proteoliposomes has been analyzed in several cases (15)(16)(17)(18)(19)(20)(21), and those findings suggest a variable result depending not only on the method of reconstitution but perhaps also on the protein in question and the detergent or phospholipid employed (reviewed in Ref. 20; see also Ref. 21). A simple argument developed by Rigaud (20) and others (21) suggests that if reconstitution proceeds by insertion of protein into preformed detergent-destabilized liposomes, one might anticipate a population of uniform orientation due to preferential insertion of the hydrophobic sector into the lipid bilayer. Because the cytoplasmic surface of most membrane proteins is considerably more hydrophilic than the extracellular face, this approach predicts recovery of proteins in a largely ISO orientation, as observed with the Ca 2ϩ -ATPase (19), the sodium/proline symporter (22), and the chimeric LacS protein (21). By contrast, if reconstitution proceeds from mixed micelles, as in the method used here, the final population is more likely to be of mixed orientation; this latter expectation may also be favored if, as here, the use of high lipid/protein ratios avoids protein packing considerations by ensuring only a single transporter per proteoliposome (3). Despite such arguments, it is clear that direct tests must be used to evaluate each case.
In our work with UhpT, a simple detergent-dilution protocol was used for reconstitution at a high lipid-to-protein ratio (3,23,24), with the expectation that this antiporter might be recovered in mixed orientation in proteoliposomes. The availability of specifically engineered variants has allowed experimental verification of this prediction. Thus, use of His 10 K291C/ K294N, which lacks the sole extracellular tryptic cleavage site (9), enabled an unambiguous demonstration that about half the recovered activity arises from molecules of a RSO orientation ( Fig. 3 and Table I). Access to the ISO population was offered by use of a second variant, His 10 V284C, whose MPB-reactive cysteine allowed inhibition of the RSO orientation without effect on ISO molecules. It is also likely that such maneuvers, especially use of trypsin, are of general utility. Most membrane proteins are enriched for tryptic cleavage sites (lysine and arginine) on their cytoplasmic surfaces (25), so this surface should be targeted preferentially by trypsin. By contrast, a rarity of lysine and arginine at the extracellular surface suggests this surface will have few, if any, tryptic cleavage sites; those that are present may well be susceptible to mutagenesis. For these reasons, trypsin may serve as a simple diagnostic tool for determining the orientation of reconstituted protein and perhaps also as the basis of a method for generating a preparation of uniformly RSO orientation as illustrated here and in unpublished work in the OxIT, the oxatate/formate antiporter of oxalobacter (26).
Use of His 10 K291C/K294N or His 10 V284C, along with appropriate treatments with trypsin or MPB, made it possible to characterize the UhpT RSO and ISO orientations in a direct way. Our key finding has been that the RSO and ISO forms are functionally equivalent in the handling of substrates of physiological relevance (Table II). On the one hand, as noted earlier, the absence of a K m shift for G6P after conversion of a population of mixed to single orientations argues that the two UhpT surfaces have roughly comparable kinetic features. We also complemented this observation with a more extended survey FIG. 5. Apparent K i values for fructose 6-phosphate and inorganic phosphate. [ 14 C]G6P transport was measured in the absence and presence of fructose 6-phosphate (top panels) or inorganic phosphate (bottom panels) using proteoliposomes containing UhpT of mixed orientation (open symbols) or uniform orientation (closed symbols). The protein used to obtain RSO or ISO orientations is indicated on the individual graphs. Dotted lines intersect the abscissa at the apparent K i value. Given a purely competitive interaction with G6P, for these assay conditions the apparent K i should be about 80% of the true K i . to measure the apparent inhibition constants for a variety of UhpT substrates (Table II). Neither approach revealed substantial differences in the behavior of the RSO or ISO orientations, and this distinguishes UhpT from several bacterial porters in which a clear kinetic bias is evident (21,22). Nevertheless, the finding of a functional symmetry is fully consistent with the biological role of UhpT. Although this anion exchanger may be thought of in the context of sugar phosphate accumulation, under some conditions UhpT must also function in the extrusion of excess phosphate (14). Parenthetically, we note this work did not examine the possibility that function might be affected by a lipid asymmetry. Although there is no reason to suspect that such tests would alter our basic conclusion, it is clear that for the LacY protein, phospholipid composition can affect the coupling between lactose and proton symport (27). The finding that UhpT has functional symmetry is compatible with the suggestion of a single substrate binding site alternately accessible from either membrane surface (28,29), as well as the proposal of a simple Ping Pong kinetic mechanism (28) in which rate-limiting events are not imposed by the pathways substrates use to move into and out of the binding site (29). This last point is of special interest. Current evidence suggests that the pathway is lined by several residues at the center of TM7 (29,30), and by two required arginine residues, Arg 46 and Arg 275 , located in equivalent positions with respect to the N-and C-terminal halves of the molecule (8). Because arginine is commonly found in binding sites that accept phosphate or organic phosphates (cf. Ref. 8), it is presumed that Arg 46 and Arg 275 contribute to the anionic selectivity mechanism that ensures an overall electroneutral exchange (14,28). If so, one might expect these residues to lie at the midpoint of the translocation pathway, yet present models of topology (5,6) place them at its external border (8). This, along with our present finding of functional symmetry of the RSO and ISO forms, suggests that the main rate-limiting events in anion exchange must occur after substrates interact with the binding site and/or selectivity mechanism; such events are presumably associated with changes of protein conformation required to close the pathway after substrate enters from one surface, trapping substrate in an occluded state and then open-ing the pathway to the other surface, enabling transmembrane passage (29).