Transmembrane Segment 11 of UhpT, the Sugar Phosphate Carrier ofEscherichia coli, Is an α-Helix That Carries Determinants of Substrate Selectivity*

In Escherichia coli, transport of hexose 6-phosphates is mediated by the Pi-linked antiport carrier, UhpT, a member of the major facilitator superfamily. We showed earlier that Lys391, a member of an intrahelical salt bridge (Asp388/Lys391) in the eleventh transmembrane segment (TM11) of this transporter, can function as a determinant of substrate selectivity (Hall, J. A., Fann, M.-C., and Maloney, P. C. (1999) J. Biol. Chem. 274, 6148–6153). Here, we examine in detail the role of TM11 in setting substrate preference. Derivatives having an uncompensated cationic charge at either position 388 or 391 (the D388C, D388V, or D388K/K391C variants) are gain-of-function mutants in which phosphoenolpyruvate, not sugar 6-phosphate, is the preferred organic substrate. By contrast, when an uncompensated anionic charge is placed at position 388 (K391C), we observed behavior consistent with an increased preference for monovalent rather than divalent sugar 6-phosphate. Because positions 388 and 391 lie deep within the UhpT hydrophobic sector, these findings suggested that an extended length of TM11 may be accessible to external substrates and probes. To explore this issue, we used a panel of TM11 single cysteine variants to examine the transport of glucose 6-phosphate in the presence and absence of the membrane-impermeant, thiol-reactive agentp-chloromercuribenzosulfonate (PCMBS). Accessibility to PCMBS, together with the pattern of substrate protection against PCMBS inhibition, leads us to conclude that TM11 spans the membrane as an α-helix, with approximately two-thirds of its surface lining a substrate translocation pathway. We suggest that this feature is a general property of carrier proteins in the major facilitator superfamily and that for this reason residues in TM11 will serve to carry determinants of substrate selectivity.

Secondary transport systems use the chemiosmotic energy generated by the movement of ions down their electrochemical gradients to facilitate the accumulation of small solutes (1)(2)(3). The best-studied secondary transporters of this sort belong to the MFS 1 (4,5), an evolutionarily related collection that ac-counts for a large fraction of known solute transporters (6). This taxonomic group is comprised of single-polypeptide carriers that show great diversity in both substrate specificity and kinetic mechanism. Despite this heterogeneity, most members of the MFS share a common structural theme, one characterized by the presence of 12 transmembrane segments believed to transverse the membrane in ␣-helical conformation. In select cases, a high preponderance of ␣-helix has been confirmed by circular dichroism or electron spin resonance spectroscopies (7)(8)(9). Similar tests suggest that an unrelated transporter, the Na ϩ /H ϩ antiporter, NhaA, also has 12 transmembrane helices, and in this case two-dimensional crystallography has confirmed the inference (10). In no case, however, has the structure of a secondary transporter been solved to a resolution affording molecular analysis.
In the absence of detailed information concerning the structure of such transport proteins, helix relationships and helix function in members of the MFS have been analyzed by less direct genetic approaches, such as second-site suppressor, sitedirected, and cysteine-scanning mutagenesis. Application of these techniques has clarified structural features for several important model systems. For example, second-site suppressor and site-directed mutagenesis has been used to identify intraand interhelical salt bridges in the H ϩ /lactose cotransporter LacY, the P i :sugar 6-phosphate antiporter UhpT, and the H ϩ : metal-tetracycline exchanger TetA(B), as well as in the unrelated Na ϩ /melibiose carrier MelB (11)(12)(13)(14)(15)(16)(17)(18). Cysteine-scanning mutagenesis, along with agents that exploit sulfhydral chemistry, has provided a way to probe the functional and structural significance of specific amino acid residues in LacY, UhpT, and the oxalate:formate exchanger OxlT (12, 19 -22). Such approaches have also given insight concerning residues likely to interact with substrate and helices that line the substrate translocation pathway (19,20,22). Taken together, this information has led to formulation of rational models for helix packing in the MFS (12,14,23,24).
In work described here, we examined UhpT, the P i -linked hexose phosphate antiport carrier of Escherichia coli (25)(26)(27), with an emphasis on the study of TM11. We focused on this target because earlier work showed that one of its residues, Lys 391 , can play a direct role in determining substrate selectivity (15), suggesting that TM11 lines the substrate translocation pathway. In the present analysis, we used both cysteinescanning and site-directed mutagenesis to broadly probe the role(s) of TM11 and to ask whether other residues in this segment might also influence substrate selectivity. Our analysis implicates positions 388 and 391 in TM11 as crucial determinants of substrate selectivity and suggests that TM11 is an ␣-helix, with roughly two-thirds of its surface facing a waterfilled translocation pathway.
Mutagenesis-Site-directed mutagenesis was performed by using the sequential polymerase chain reaction (29). To confirm the desired mutation and to rule out the presence of other changes, mutant alleles were sequenced either in the laboratory, using the Sequenase (version 2.0) reaction (Amersham Pharmacia Biotech), or at the Biosynthesis and Sequencing Facility of Johns Hopkins Medical School.
Immunoblot Analysis-SDS-polyacrylamide gel electrophoresis was performed using cell extracts without preheating in sample buffer, as described (30). Protein was transferred to nitrocellulose and probed with a peptide-directed rabbit antibody reactive to a UhpT C-terminal epitope (31,32). Western blots were developed using chemiluminescence (Amersham Pharmacia Biotech), and expression of UhpT and its derivatives was quantitated by densitometry of digitized images (32,33). Expression levels for each mutant were determined in at least three independent trials; mean values were used to normalize G6P transport for calculation of specific activity.
Transport Assays-Unless otherwise indicated, overnight cultures were diluted 200-fold into LB broth plus antibiotics (10 g/ml streptomycin, 100 g/ml ampicillin), grown at 37°C to a density of 2-5 ϫ 10 8 cells/ml, and harvested by centrifugation. Cells were washed twice and resuspended in buffer A (50 mM MOPS/K ϩ , 100 mM K 2 SO 4 , 1 mM MgSO 4 , pH 7) at OD 660 ϭ 1.4, equivalent to about 2 ϫ 10 9 cells/ml. Cell suspensions were allowed to equilibrate at room temperature, and tests of G6P, F6P, or PEP transport were initiated by adding a one-twentieth volume of labeled substrate to a final concentration of 50 M. At the indicated times, aliquots were removed for filtration on Millipore filters (0.45-m pore size), followed by two washes with 5 ml of buffer A lacking MgSO 4 . To monitor PCMBS inhibition of G6P and PEP transport, cells were first incubated for 10 min with PCMBS (0.05-200 M) at room temperature. Aliquots were then filtered to remove the inhibitor, and after two washes with 5 ml of buffer A, tests of transport were initiated by overlaying cells (on the filter) with buffer A containing 50 M labeled substrate. To assess substrate protection, the same procedure was used except that cells were preincubated with either 1 mM unlabeled substrate or 1 mM unlabeled substrate plus PCMBS.
For assays of UhpT-mediated P i transport, cells were grown in M63 minimal medium (34) (pH 7) containing thiamine (2 g/ml), required amino acids (50 g/ml), necessary antibiotics, and 0.2% (w/v) glucose as carbon source. The high P i content of M63 ensured maximal repression of other P i transporters (Pst and Pit) present in broth-grown cells, so that most P i transport occurred via UhpT (15,35). (Broth-and M63grown cells had comparable behavior with regard to the organophosphates transported by UhpT.) The maximal velocity (V max ) and Michaelis constant (K m ) of P i transport were estimated using the method of Hofstee (36). The inhibition constants (K i ) for 2-dG6P, F6P, or PEP as inhibitors of P i transport were determined using linear Dixon plots (37), with the assumption of competitive inhibition between P i and the test substrate. In these latter assays, the unlabeled test substrates (G6P, 2-dG6P, F6P, PEP) contained 0.5-1.0% P i (38), and because this was ignored in calculation of kinetic constants, the derived K i values should be considered minimal estimates.
To monitor in vivo function of UhpT and its derivatives, cells from an overnight broth culture were diluted 1000-fold into M63 containing 0.15% F6P or 0.15% PEP as carbon source. Cultures were placed at 37°C with continuous shaking, and growth was monitored by changes in optical density at 660 nm.

RESULTS
TM11 as a Determinant of Substrate Specificity-In previous work, we had identified Asp 388 and Lys 391 of UhpT TM11 as participants in an intrahelical salt bridge (15). Further study of this ion pair showed that if the anionic partner is replaced by cysteine, the derivative protein has a substrate selectivity biased toward compounds, such as PEP, that carry an additional anionic charge (15). Because PgtP, the related PEP transporter of Salmonella typhimurium, has arginine at this position, we argued that the gain-of-function phenotype reflects the presence of an uncompensated positive charge (lysine) at position 391 and not the presence of cysteine, a proton donor, at position 388. In the present study, this supposition is confirmed by the finding of an equivalent result after replacement of Asp 388 with Val, the amino acid occupying the corresponding position in PgtP. Thus, although phosphate transport by both the D388C and D388V variants shows a kinetic profile resembling the parent protein (Table I), each mutant has a 20 -25-fold reduced affinity for hexose 6-phosphate, coupled with a 10-fold increase in the affinity for PEP (Table I). Tests of both growth and transport confirmed that each mutant, but not their parent, could use PEP as sole carbon source for growth ( Fig. 1).
Such findings confirm that an uncompensated positive charge at position 391 alters selectivity so as to favor substrates with an increased anionic character. With this in mind, we next asked whether the location of such charge is also a critical determinant of specificity. To do this, we constructed the D388C/K391C variant to remove both partners in the natural salt bridge. This variant, like its parent, is able to transport sugar 6-phosphates but not PEP (Fig. 1). We then scanned this derivative with lysine placed at positions in a registration (i, i Ϯ 3 or 4) that would span four turns of an ␣-helix. In most Rates of P i self-exchange were measured as described under "Experimental Procedures" using the method of Hofstee (36). b Kinetic constants are means Ϯ S.E. for three independent experiments. c Inhibition constants (K i ) were determined for each substrate as an inhibitor of P i self-exchange using linear Dixon plots (37), with the assumption of competitive inhibition between P i and the test substrate. The concentration range of inhibitor used was 2.5 M to 5 mM. d nd, not determined. K i values for PEP were determined for those variants exhibiting PEP, but not G6P, transport (Fig. 1), and K i values for F6P were determined only for those variants capable of transporting G6P but unable to grow on F6P at pH 7 (Figs. 1 and 2; data not shown; see Ref. 15). cases (positions 384, 387, 395, and 398), the introduction of lysine into the D388C/K319C background yielded a non-functional protein (data not shown). It was evident, however, that placement of Lys at either position 388 or 391 altered UhpT substrate selectivity so as to favor PEP. Thus, the D388K/ K391C variant showed a 10-fold decrease in affinity for 2-dG6P at the same time its affinity for PEP increased 40-fold (Table I).
As before, this gain-of-function derivative was able to both transport and use PEP as sole carbon source for growth (Fig. 1).
The finding that UhpT selectivity is biased in favor of more anionic substrates upon placement of a lone cationic residue at position 388 or 391 led us to ask if a change in selectivity would also arise on introduction of a negative charge at either of these positions. The K391D/D388C variant, in which aspartate is moved to position 391, was too poorly expressed for such study, so that only the K391C variant was examined. We considered two possible models. On the one hand, an uncompensated anionic residue at position 388 might bias selectivity to favor substrates carrying an extra cationic group. Equivalently, altered selectivity might favor substrates with one less negative charge. That the K391C variant had near parental activity, with only a 2-5-fold decrease in affinity for both G6P and F6P (Tables I and II; Figs. 1A, 2C, and 3B) did not favor the first model. In the same way, tests (data not shown) using either glucosamine 6-phosphate and O-phosphorylethanolamine as inhibitors of P i transport failed to support the idea.
Our next tests addressed the idea that the uncompensated negative charge at position 388 biased selectivity toward substrates carrying one negative charge rather than two (i.e. HG6P 1Ϫ rather than G6P 2Ϫ ). If so, one might expect enhanced transport and growth under more acidic conditions where monovalent sugar phosphate is enriched. To test this possibility, we measured transport and growth at external pH values from pH 5.5 to 8.25, a range that spans the pK 1 of sugar phosphate (pK 1 ϭ 6.1). In such work, we found that maximal transport by the mutant (tested at 50 M substrate) occurred at a pH approximately 0.5 pH units more acidic than found for the parent protein (pH 6.3 and 6.9, respectively) (Fig. 2C). We also found that the K391C variant, unlike its parent, showed a marked decrease in function above pH 7. Both observations are consistent with the idea that the mutant works best when monovalent sugar phosphate is the predominant species. This idea was further supported by kinetic work showing that the K m for sugar 6-phosphate transport by the mutant increased 4 -5-fold at alkaline pH (125-640 M), whereas that of the parental protein remained constant throughout the entire pH range studied (Fig. 3). In similar fashion, growth of the K391C derivative was pH-dependent. Thus, although the parent protein supports growth equally well between pH 5.5 and 7.0, the K391C mutant grows at progressively slower rates as the environment becomes more alkaline, with complete growth stasis at pH 7 (Fig. 2, A and B). This finding, coupled with the kinetic observations (Fig. 3), leads us to suggest that in this variant the capacity to handle divalent sugar phosphate is compromised. This interpretation further emphasizes the region in and around these positions as essential determinants in substrate specificity.
UhpT TM11 Lies on the Substrate Transport Pathway-Because the topology of UhpT shows the Lys 391 -Asp 388 salt bridge to be deep within the hydrophobic sector (15, 19, 39, 40), it seemed plausible that a water-filled pathway would extend FIG. 1. Growth on, and transport of, G6P and PEP. Transport of G6P (A) and PEP (B) by UhpT and its derivatives. Cells were grown in M63 minimal medium containing 0.2% (w/v) glucose as carbon source, and rates of transport were monitored by a 5-min incubation of cells with labeled substrate, as described under "Experimental Procedures." Values shown are means Ϯ S.E. for three independent experiments. C, RK5000 without (f) and with plasmids expressing cysteine-less UhpT (q) and its D388C (OE), D388V (‚), and D388K/K391C (E) derivatives were grown in M63 minimal medium with PEP as sole carbon source. Growth was monitored by changes in optical density at 660 nm. These data, from a single experiment, are representative of findings made in three independent trials. inward from the periplasm to at least position 388. If so, we reasoned that one might identify the part of TM11 bordering this pathway by use of a suitable probe. Accordingly, we next analyzed a library of single-cysteine variants encompassing the whole of TM11 (residues 383-404), so that individual positions along TM11 could be probed by PCMBS, an impermeant, thioldirected agent with the same molecular volume and charge as the normal substrates of UhpT (20). An initial screen of TM11 single-cysteine variants showed that most were expressed at a level sufficient for functional analysis (usually 30%) (Table II). In the preceding work (Table  I, Fig. 1; Ref. 15), we found that gain-of-function variants in which Lys 391 acts as a selectivity determinant retained UhpTmediated P i transport despite a defect in transport of G6P. By contrast, of the six TM11 single-cysteine derivatives with substantially reduced G6P transport (ϳ10% or less) (Table II), all but one had comparable reductions in P i transport (data not shown); the exceptional case was the D388C variant, which retained normal levels of P i transport, as described in the earlier study (15). Thus, generation of a gain-of-function phenotype by cysteine substitution may be a rare event.
In other experiments, we examined each single-cysteine variant with regard to its response to the impermeant thiol-reactive probe PCMBS. The first trials monitored the effects of a 10-min exposure to 200 M PCMBS. As judged by inhibition of G6P transport, we concluded that eight positions were accessible to the probe (positions 391, 393, 395, 397, 398, 401, 402, and 404); study of PEP transport by the D388C derivative showed that position 388 was also accessible to PCMBS (Table III). For each of these PCMBS-responsive mutants, we then estimated the PCMBS concentration yielding 50% inhibition (K 0.5 ) (Table  III). This effort suggested two classes of derivatives. Most mutants (seven of nine) showed a relatively high sensitivity, with K 0.5 values in the range of 0.05-3 M, whereas two variants showed relatively low sensitivity (K 0.5 of 20 -40 M) (Table III). Finally, for each PCMBS-responsive protein, we evaluated whether protection was afforded by coincubation with substrate (1 mM G6P or PEP), using a probe concentration near its K 0.5 value. Neither of the two variants of low sensitivity to PCMBS showed evidence of substrate protection, whereas six of the seven highly sensitive mutants benefited from the presence of substrate (Table III).
We conclude from these experiments that positions accessible to PCMBS (Table III) may be found along nearly the full length of TM11 (positions 388 -402 versus positions 383-404), suggesting that a hydrophilic pathway extends inward along a similar length. That substrate protection is observed for many of these positions, especially those deep within the hydrophobic sector, indicates that this pathway is closed by the structural changes that accompany substrate binding or transport. DISCUSSION For a number of membrane transport proteins, use of sitedirected and scanning mutagenesis has identified charged residues that contribute to a substrate-binding domain. For example, within the MFS two charged residues (Glu 126 and Arg 144 ) coordinate substrate binding by LacY (24), several cationic residues (Arg 46 , Arg 275 , and Lys 391 ) function as recognition elements in UhpT (15,33), and a lysine (Lys 355 ) facilitates substrate binding by OxlT, the oxalate:formate antiporter of Oxalobacter formigenes (22,41). The MFS also contains examples in which one or more transmembrane helices have been associated with the permeation pathway. Thus, TM5 of the GLUT1 glucose uniporter and TM7 of UhpT are known to line the sugar/sugar phosphate transport pathway (20,42), as do residues on TM11 of OxlT (41). Similarly, several helices in LacY have been shown to line the pathway taken by lactose (12).
In the present study, we exploited site-directed and cysteine- FIG. 2. Growth on, and transport of, F6P at varying pH. RK5000 without (OE) and with plasmids expressing the parental cysteine-less UhpT (q) and its K391C derivative (f) were grown in M63 minimal medium with F6P as a carbon source at pH 7 (A) and pH 5.5 (B). Growth was monitored as described in the legend to Fig. 1. C, transport of F6P at varying pH was examined for the parental cysteine-less UhpT (E) and its K391C derivative (Ⅺ). Data from three separate trials were normalized to peak values (9.1-13 nmol/min/mg of protein for cysteine-less UhpT; 2-4.6 nmol/min/mg of protein for its K391C variant) and are shown as means Ϯ S.E. Transport was measured as described under "Experimental Procedures" except that assay and wash buffers used PIPES rather than MOPS.

FIG. 3. Effect of pH on kinetic constants for F6P transport. K m (A) and
V max (B) values for parental cysteine-less UhpT (E) and its K391C derivative (Ⅺ). Assays were performed as described under "Experimental Procedures" except that assay and wash buffers used MES/ MOPS rather than MOPS. Kinetic constants are means Ϯ S.E. for three independent experiments.
scanning mutagenesis, together with the use of thiol-directed probes, as tools to analyze TM11 of UhpT. This focus was based on early work showing that Lys 391 , normally part of an intrahelical TM11 salt bridge, acts as a determinant of substrate specificity when present without its normal partner, Asp 388 (15). This finding showed that at least one residue on TM11 must intersect with the UhpT transport pathway. New evidence presented here both reinforces this idea and extends the argument to implicate a substantial portion of TM11 as lining the pathway. Thus, it is now clear that positive charge placed (as Lys) at either position 388 or 391 can influence substrate selectivity (see also below), as does a lone negative center at position 388, strengthening the argument that residues on TM11 have strategic value. Moreover, study of a single-cysteine panel now shows that the impermeant and anionic probe PCMBS has access to a large tract along TM11, encompassing nearly two-thirds of its length, from position 404 to position 388 (Table III; Fig. 4). The periodicity of PCMBS accessibility (Fig.  4) suggests that TM11 is an ␣-helix having roughly two-thirds of its surface facing a water-filled channel. Because of their relatively high K 0.5 values relative to other PCMBS-sensitive positions (Table III), we suggest that Thr 393 and Lys 404 lie at the boundary of this surface. We also note that this region lies adjacent to a stripe containing the uncharged residues that show low specific activity as single-cysteine variants (Gly 389 , Gly 392 , Tyr 396 , Gly 399 ) ( Table II), suggesting that this second stripe abuts a neighboring helix. The significance of these observations is highlighted by the added finding that PCMBSresponsive residues are unaffected by the probe when treatment is in the presence of excess substrate. At the least, such substrate protection indicates that access to this region is blocked by structural events that accompany substrate binding or transport, consistent with the idea that this portion of TM11 forms part of the translocation pathway.
This structural model (Fig. 4) offers one way to view this part of the transport pathway, but it does not address the mecha-nism by which altered substrate selectivity becomes associated with positions 388 and 391. To understand this phenomenon, it is helpful to recall the elements believed to contribute to maintenance of substrate selectivity by the wild type protein. Because UhpT transports sugar phosphates but not sugars, there must be a mechanism that identifies the presence of phosphate at the substrate C6 (or C5) position. Crystallography of enzymes and receptors that interact with P i or organophosphates shows that arginine residues almost always take part in the recognition of the anionic phosphoryl group (44 -48; summarized in Ref. 33). (A prominent exception, triose phosphate isomerase, uses lysine rather than arginine (49)). We believe that in UhpT, this role is played by two arginines (Arg 46 and Arg 275 ) at the periplasmic poles of TM1 and TM7, about 30 residues prior to the internal duplication characteristic of all members of the MFS (33,50); only these arginines are both essential and conserved throughout the UhpT family (33). Much less is known about the interaction between UhpT and the polyol ring of its ligands, but crystallography of carbohydrate-protein complexes suggests that this may occur via aromatic amino acid stacking interactions with the furanose/pyranose ring and by hydrogen bonding with at least some of the offshooting hydroxyls (51)(52)(53). These interactions are presumed to occur within the hydrophobic core of UhpT, because the arginines required for recognition of the phosphoryl group are located near the periphery. Certainly, the finding that an uncompensated positive charge at position 388/391 allows UhpT to process substrates carrying an additional anionic charge (e.g. PEP) is consistent with the idea that a substrate orients within UhpT with its anionic phosphate pointed toward the periplasm and the C1 position toward the cytoplasm (15).
What determines whether substrates on the translocation pathway will be transported? Early work indicated that the exchange reactions mediated by UhpT are electrically neutral in nature (27,54), and the abrupt alteration in selectivity that accompanies lysine insertion at positions 388 or 391 reinforces the idea that maintenance of electrostatic neutrality is an essential criterion. We also believe that this same view can help interpret the behavior of the K391C variant, which shows a distinct acid shift in the pH optimum for growth and transport (Figs. 2 and 3). We speculate that in this mutant the uncompensated electronegative center, Asp 388 , acts as a resident fixed anion, so that preservation of an electrostatic neutrality would require a selectivity biased toward monovalent sugar phosphate. A consequence of this bias would be that, as observed, the pH optimum for sugar phosphate transport and growth would shift in the acid direction. However, at the sugar phosphate concentrations used for growth studies, the levels of monovalent substrate would exceed the K m for transport, even at pH 7. Therefore, we believe that the growth phenotype is best explained by changes in the protonation state of Asp 388 . At more alkaline pH, the anionic charge associated with this position would preclude transport of divalent sugar phosphate, because electrostatic neutrality could not be achieved. Protonation of the resident anion would be favored as external pH is lowered, thereby restoring the parental character of the translocation pathway. This interpretation would be unrealistic if the pK a of Asp 388 is as low as that found for aspartate in model compounds (pK a of 4) (55). On the other hand, it is known that the pK a of aspartate may take on a significantly higher value, depending on the nature of its local environment. It remains feasible, then, that the pK a of Asp 388 falls within the range of pH values we have tested (Figs. 2 and 3).
The assumption that substrate preference is influenced by the protonation state of Asp 388 also allows us to reconcile the contradictory findings that at pH 7 the K391C variant transports but does not grow on sugar phosphate (Fig. 2). UhpT, as do other P i -linked antiporters, carries out both heterologous (P i :sugar phosphate) and homologous (P i :P i , sugar phosphate: sugar phosphate) exchanges (25,27,38,56). One might conclude that heterologous exchange is the preferred reaction in vivo, because homologous exchanges do not usually lead to net substrate fluxes. However, the affinity of UhpT for sugar phosphate is 10 -40-fold higher than for P i , suggesting that homologous exchange must be preferred over heterologous reaction (56). Furthermore, because E. coli maintains a slightly alkaline cytoplasmic pH over a wide range of external pH values (57), one expects a reaction in which net sugar phosphate accumulation arises from the electroneutral exchange of two monova-lent sugar phosphates (external) for a single divalent species (internal) (25). We suggest that such an asymmetric exchange is not possible for the K391C mutant placed at relatively alkaline pH, because the deprotonated state of Asp 388 would not permit use of the divalent anionic substrate. Instead, this mutant would only be able to carry out exchange using monovalent sugar phosphate. This might then appear as near normal levels of transport, but the overall 1-for-1 stoichiometry of such a reaction would not support growth.
Our observations also indicate that a large tract of TM11 lines the UhpT translocation pathway. That UhpT TM7 also lies on the translocation pathway (20) is in agreement with the general helix-packing model proposed by Goswitz and Brooker (23), which suggests that both of these helices are within the core of potential pathway-lining segments in MFS transporters. This model, as well as mutagenesis studies in LacY and the H ϩ /sucrose symporter CscB (17,58,59), suggests that TM7 and TM11 may be neighbors in UhpT, an idea that can be tested by disulfide cross-linking using the single-cysteine libraries currently available. FIG. 4. ␣-Helical representation of TM11. Overhead (A) and side (B) view of TM11 (residues 383-404) modeled as an ␣-helix (3.6 residues per turn). The cytoplasmic (residue 383) and periplasmic (residue 404) boundaries are derived from the analysis of hydropathy (39) and the results of reporter gene fusions in UhpT and GlpT (40,43). Amino acid residues accessible to PCMBS are shaded; residues shaded black are protected from PCMBS attack by either G6P or PEP, whereas those shaded gray are not. Boxed residues in A refer to inactive variants (Table II). Dotted lines in A demarcate the likely boundary between that region of TM11 facing a water-filled translocation pathway and the region that contacts either lipid or protein.