Topology of OxlT, the Oxalate Transporter of Oxalobacter formigenes, Determined by Site-Directed Fluorescence Labeling

Abstract

The topology of OxlT, the oxalate:formate exchange protein of Oxalobacter formigenes, was established by site-directed fluorescence labeling, a simple strategy that generates topological information in the context of the intact protein. Accessibility of cysteine to the fluorescent thiol-directed probe Oregon green maleimide (OGM) was examined for a panel of 34 single-cysteine variants, each generated in a His₉-tagged cysteine-less host. The reaction with OGM was readily scored by examining the fluorescence profile after sodium dodecyl sulfate-polyacrylamide gel electrophoresis of material purified by Ni²⁺-linked affinity chromatography. A position was assigned an external location if its single-cysteine derivative reacted with OGM added to intact cells; a position was designated internal if OGM labeling required cell lysis. We also showed that labeling of external, but not internal, positions was blocked by prior exposure of cells to the impermeable and nonfluorescent thiol-specific agent ethyltrimethylammonium methanethiosulfonate. Of the 34 positions examined in this way, 29 were assigned unambiguously to either an internal or external location; 5 positions could not be assigned, since the target cysteine failed to react with OGM. There was no evidence of false-positive assignment. Our findings document a simple and rapid method for establishing the topology of a membrane protein and show that OxlT has 12 transmembrane segments, confirming inferences from hydropathy analysis.

The gram-negative bacterium Oxalobacter formigenes sustains a proton motive force by utilizing a “virtual” proton pump based on the transport and metabolism of oxalate. An electric potential (negative inside) arises from action of the antiporter, OxlT, which links inward transport of divalent oxalate to the outward flow of monovalent formate, the product of oxalate decarboxylation. The net inflow of a single negative charge is then phenomenologically linked to generation of a pH gradient (alkaline inside), because decarboxylation of oxalate consumes a single cytosolic proton. Together, these elements comprise the proton motive force used to drive ATP synthesis in this obligate anaerobe (3, 14, 20, 32). Virtual pumps of equivalent construction have now been observed in a number of microorganisms (14, 22, 25).

It is evident that OxlT occupies a central position in the cell biology of O. formigenes and that study of this transporter is relevant to several aspects of microbial physiology. Added interest in OxlT stems from recent work (8, 9, 26) suggesting that this protein may also serve as a useful model for biochemical study of other transporters. Accordingly, OxlT may contribute to an understanding of membrane transport at both a functional level and a mechanistic level.

Hydropathy analysis of the OxlT amino acid sequence, along with other considerations, suggests the presence of 12 transmembrane segments (TM1 to TM12) (1), consistent with the presumed structure of most other members of the major facilitator superfamily (MFS) (30), the superfamily of related transporters to which OxlT belongs. On the other hand, this prediction placed a lone charge (K355) at the center of TM11, something rarely encountered in transmembrane helices of known structure (9). In much the same way, this model incorporated eight polar residues (N47, S51, Q56, T57, T60, S62, Q63, and Q66) into TM2, reducing its hydropathy to a level not usually associated with a transmembrane helix (35). Together, these two unexpected features lowered overall confidence in the validity of the model and raised the possibility that OxlT might contain 10 or perhaps 11 transmembrane segments, rather than 12.

After the initial cloning and analysis of the OxlT sequence (1), two efforts have sought to address such uncertainties in the presumed structure of this transporter. Initial work (9) focused on the orientation and organization of TM11, showing that this transmembrane segment indeed does have a charged residue (K355) at its center. The work reported here now examines the orientations of the remaining transmembrane segments in OxlT by using a simple approach that generates topological information in the context of the intact, functional protein. Our findings establish that OxlT has 12 transmembrane helices; these results prompt a model for substrate binding during transport.

MATERIALS AND METHODS

Plasmid construction.

OxlT, carrying a C-terminal extension of nine tandem histidine residues to facilitate protein purification, was encoded within pBluescript II SK(+) (Amp^r) under plac control. Inappropriate basal expression of OxlT and its derivatives was limited by housing all plasmids together with the middle-copy-number plasmid pMS421 (Spec^r LacI^q) in Escherichia coli strain XL-3 (1). For use as reagents to establish OxlT topology, we generated 73 single-cysteine derivatives by oligonucleotide-directed mutagenesis (Chameleon; Stratagene), using a cysteine-less variant (C28G C271A) as the template (9); all mutants were confirmed by DNA sequencing (9).

Screens for expression and function.

To ensure adequate expression of functional protein, we conducted preliminary screens of all mutants. A few colonies from a plating of freshly transformed cells were grown overnight at 37°C with shaking in 5 ml of Luria-Bertani medium containing antibiotics. Cells were diluted 100-fold into 40 ml of Luria-Bertani medium and grown until cell density reached an optical density at 600 nm of 0.09, at which point OxlT expression was induced by addition of 1 mM isopropyl-1-thio-β-d-galactopyranoside. After 3.5 h of growth, a small aliquot was processed for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (1, 9), and OxlT expression was evaluated by immunoblot analysis (1, 9), using a mouse monoclonal antibody directed against tetrahistidine (Qiagen). Reactivity was detected by chemiluminescence and quantitated using Fujifilm MacBAS v2.2, and mutant expression was evaluated with reference to the cysteine-less parent processed in parallel (9). To assess function, the remainder of the culture (ca. 35 ml) was collected by centrifugation, suspended in 10 ml of a lysis solution (300 μg of lysozyme per ml, 40 μg of DNase per ml, 5 mM EDTA, 10 mM Tris-HCl [pH 7.5]), and incubated at room temperature for 15 min to initiate cell rupture. Osmotic lysis was completed by a 10-fold dilution with iced distilled water, after which released cytoplasmic proteins were removed by three cycles of centrifugation and washing with iced distilled water (35). The membrane pellet was taken up in 0.5 to 0.75 ml of solubilization buffer (20 mM MOPS [morpholinepropanesulfonic acid]-K, 20% [vol/vol] glycerol, 0.42% [wt/vol] E. coli phospholipid, 6 mM β-mercaptoethanol, 10 mM potassium oxalate, 1.5% [wt/vol] octyl-β-d-glucopyranoside [pH 7]) (2). After 1 h at 4°C, the detergent extract was clarified by centrifugation and used for assays of transport by reconstitution (below).

Site-directed fluorescence labeling.

To evaluate OxlT topology, we selected mutants that displayed both expression and activity that were ≥30% of those of the cysteine-less parent; later assays with purified protein confirmed that these variants had ≥20% of the parental specific activity (see below). For fluorescence labeling, a 100-ml induced culture was divided into three equal parts. After harvesting, cells from the first aliquot were resuspended in 5 ml of buffer A (100 mM potassium sulfate, 50 mM potassium phosphate [pH 8]) and given 40 μM freshly prepared Oregon green 488 maleimide carboxylic acid (OGM) (9). After 15 min at 23°C, the reaction was quenched with 6 mM β-mercaptoethanol and by three cycles of centrifugation and washing with buffer B (100 mM potassium sulfate, 50 mM MOPS-K [pH 7]). Membranes from this aliquot, prepared as described above, were taken up in 2.5 ml of solubilization buffer, and the crude extract was used for purification of OxlT by Ni²⁺-nitrilotriacetic acid (Ni²⁺-NTA) affinity chromatography (see below). The second aliquot was treated in the same fashion except that cells received 2 mM freshly prepared methanethiosulfonate etheyltrimethylammonium (MTSET) (16) for 10 min at room temperature rather than OGM. After removal of MTSET by three cycles of centrifugation and washing with buffer B, membranes were prepared (as described above) and resuspended in 5 ml of buffer C (20 mM potassium phosphate [pH 8]) with 40 μM OGM at 23°C for 15 min. The reaction was quenched by addition of 6 mM β-mercaptoethanol, and after three cycles of washing with distilled water, the membranes were solubilized in 2.5 ml of solubilization buffer in preparation for OxlT purification. The third aliquot was treated in identical fashion, except there was no pretreatment with MTSET preceding exposure of membranes to OGM.

Purification of OxlT.

To purify OxlT, the crude detergent extract was clarified by centrifugation and mixed with 0.1 ml of Ni²⁺-NTA resin (Qiagen). After overnight incubation at 4°C, the resin was packed into a spin minicolum (Bio-Rad) and washed with 8 to 10 ml of solubilization buffer containing 200 mM NaF and 50 mM imidazole. Samples exposed to OGM or MTSET were eluted in a 50-μl volume of 0.5 M imidazole–2% SDS–10% glycerol–50 mM Tris-HCl (pH 7). For SDS-PAGE, 20 μl of this sample was used, and after electrophoresis in a 12% acylamide matrix, a fluorescence profile was obtained by scanning with a Molecular Dynamics STORM fluorescence imaging system (blue fluorescent chemifluorescence mode, excitation wavelength of 450 ± 30 nm). The same gel was then stained with Coomassie brilliant blue, and densitometry was recorded using a Microtek ScanMaker 5.0. We observed an unusual degree of dimerization and oligomerization in these experiments. This is attributed to elution from Ni²⁺-NTA under denaturing conditions as well as to the increased tendency of cysteine-less OxlT and its derivatives to show aggregation during SDS-PAGE (9). When OxlT was purified for purposes of functional tests, membranes from a 200-ml culture were taken up in 10 ml of solubilization buffer. The crude extract was incubated overnight with 0.3 ml of Ni²⁺-NTA resin overnight, washed as described above, and eluted in 0.2 ml of solubilization buffer containing 500 mM imidazole. Protein concentration was measured as described previously (1, 9).

Oxalate transport.

In a total volume of 200 μl, solubilized protein (200 μg of crude extract; 2 μg of purified protein) was mixed at 4°C with 1.75 mg of bath-sonicated liposomes (2) and sufficient detergent to maintain a 1.5% concentration. The mixture was then diluted at room temperature with 5 ml of loading buffer containing 100 mM potassium oxalate and 50 mM MOPS-K (pH 7). After 20 min at room temperature, proteoliposomes were chilled and initial rates (1 min) of [¹⁴C]oxalate transport were measured, in duplicate, by the abbreviated filtration assay described earlier (1, 9). Because of the unusually high turnover number of OxlT (9), these transport assays were performed at 4°C, as before (9).

Chemicals.

[¹⁴C]Oxalate was from New England Nuclear Life Science Products, OGM was purchased from Molecular Probes Inc., and MTSET was obtained from Toronto Research Chemicals Inc. The octyl-β-d-glucopyranoside was from Boeheringer-Calbiochem Corp.; the E. coli phospholipid came from Avanti Polar Lipids, Inc.

RESULTS

Single-cysteine derivatives.

Table 1 describes the single-cysteine derivatives that served as targets in the study of OxlT topology. In total, we considered 73 variants in which cysteines were placed either at the ends of putative transmembrane segments or in the loops that connect these segments. Of these mutants, about half were discarded because they displayed either low expression (assessed by immunoblots or recovery of protein after purification), low specific activity (assessed by reconstitution of crude extracts or of purified material), or both. In a few other cases, there was a surplus of mutants in a target region, and the excess was not analyzed. Altogether, attention centered on 34 single-cysteine variants as reagents to establish OxlT topology. Because these variants retained normal or near-normal (≥20%) specific activity after purification and reconstitution (Table 1), we assumed that the targeted cysteines did not significantly perturb the structure of OxlT. Thus, the topology of these targets reflects the location of the original residues in the parental (cysteine-less) and wild-type proteins.

TABLE 1.

Single-cysteine mutants used to determine OxlT topology

Variant	Sp acta (μmol/min/mg of protein)	Deduced location
Q5C	10.5	Cytoplasm
T6C	6.3	Cytoplasm
T10C	11.5	Cytoplasm
L39C	25.8	Periplasm
Y40C	22.0	Periplasm
A41C	23.6	Periplasm
N42C	13.3	Periplasm
P43C	28.9	Periplasm
D46C	23.9	Periplasm
G49C	25.6	Periplasm
V50C	14.8	Periplasm
S51C	41.5	Periplasm
Q71C	10.5	Cytoplasm
F80C	11.2	Cytoplasm
P82C	8.6	Cytoplasm
V103C	24.6	Indeterminate
D104C	25.2	Periplasm
A131C	13.5	Indeterminate
N132C	15.6	Cytoplasm
G171C	19.1	Periplasm
A172C	10.8	Periplasm
Q203C	34.0	Cytoplasm
P220C	6.5	Cytoplasm
P245C	7.4	Periplasm
I282C	10.0	Cytoplasm
R284C	12.8	Cytoplasm
L308C	11.9	Indeterminate
G309C	17.1	Periplasm
D310C	19.7	Periplasm
V311C	33.2	Indeterminate
A312C	22.4	Indeterminate
A345C	21.8	Cytoplasm
F373C	41.6	Periplasm
G401C	14.6	Cytoplasm
Parent	30.0 ± 2.6b	Not done.

^aFrom assays of purified and reconstituted protein. 

^bMean ± standard deviation. 

Site-directed fluorescence labeling.

To probe OxlT topology, we expanded on a strategy that had proven useful in establishing the orientation and organization of TM11 (9). This approach relies on the labeling of either intact cells or membrane fragments by the generally impermeable, fluorescent, and thiol-directed agent OGM. Evidence that a targeted position was exposed to the periplasm was obtained when the single-cysteine variant was modified by exposure of intact cells to OGM. By contrast, cytoplasmic positions were identifiable if cysteine modification took place only after cell lysis. As a control for these interpretations, we also required that the labeling of an external position be blocked by prior treatment of cells with the impermeable but nonfluorescent MTSET (16) and that this maneuver have no effect on the labeling of positions assigned to the cytoplasmic surface.

To define operational parameters that meet these criteria, we studied the behavior of cysteines whose topology on TM11 was known from early work, one (A345C) exposed to the cytoplasm and the other (F373C) exposed to the periplasm (9). In these trials we suspended intact cells at pH 8 in a salts solution (50 mM potassium phosphate, 100 mM potassium sulfate [pH 8]) of sufficient osmolality (ca. 450 mOsm/liter) to avoid osmotic gradients that might induce unnecessary swelling. To these cells we added the fluorescent probe OGM, and we monitored the concentration dependence and the time course of labeling of periplasmic and cytoplasmic residues by measuring the fluorescence of purified OxlT after SDS-PAGE (Fig. 1). We concluded that a 15-min or longer incubation with 20 to 40 μM OGM yielded about a 10-fold discrimination between the external (A373C) and internal (A345C) targets (Fig. 1A and B), consistent with the restricted permeation of this probe under these conditions (9). In addition, we verified that for these same conditions prior addition of MTSET (0.4 to 2.8 mM) blocked labeling of the external residue without affecting subsequent reactivity of the internal marker in membrane preparations (Fig. 1C). We also showed that MTSET, if added to membranes, blocked labeling of the internal residue (Fig. 1D).

FIG. 1.

Site-directed fluorescence labeling of cysteine residues in OxlT. (A to C) Washed cells of the A345C (open symbols) or F373C (closed symbols) variant were suspended in buffer A together with OGM as indicated, without or with preincubation with MTSET. After SDS-PAGE of purified OxlT, a fluorescence profile was recorded (insets, lower panels), quantitated, and normalized to protein content as measured by densitometry after Coomassie brilliant blue staining of the same gel (insets, upper panels); for convenience in presentation, the A345C and F373C profiles were digitally separated from scans of each parent gel. Fluorescence is given in arbitrary units. (A) Cells incubated for 15 min with OGM at the indicated concentration. (B) Cells placed with 40 μM OGM for the specified time. (C) Effectiveness of the MTSET block. Cells were exposed to MTSET for 10 min. After removal of MTSET by washing, membranes prepared as described in Materials and Methods were exposed to 40 μM OGM for 15 min. (D) As indicated, membranes of the A345C mutant were pretreated with 2 mM MTSET as in panel C; washed membranes were then exposed to 40 μM OGM for 15 min.

It is worth noting that in these and other cases, we monitored the reactivity of the internal marker using a solution of low ionic strength (20 mM potassium phosphate [pH 8]), since membranes tended to aggregate if placed in the solution used for labeling of intact cells. Although labeling of the internal marker could be observed for the latter conditions, the accompanying aggregation made it more difficult to solubilize OxlT, and the effectiveness of affinity chromatography was compromised, leading to a reduced yield (<20%) (not shown). For convenience, then, we chose to assess reactivity in membrane preparations using the lower-ionic-strength buffer. With one prominent exception (see below), targets that were reactive in the intact cell (e.g., externally exposed) appeared to be comparably reactive when membrane fragments were tested in this low-ionic-strength buffer (see below).

Topology of TM12.

Using these general conditions, we moved to establish the orientation of TM12 by study of a presumptive cytoplasmic residue (G401C) in an experiment that included the two single-cysteine derivatives of known topology on TM11, A345C and F373C (Fig. 1). In Fig. 2, which outlines this work, the scans at the left show the fluorescence profile of the SDS-polyacrylamide gel of purified material (bottom panel) and the total protein revealed by later Coomassie brilliant blue staining (top panel). These profiles form the experimental basis for deducing the topological arrangement given at the right. Thus, when OGM was added to intact cells, fluorescent protein was recovered only for the F373C variant, yet when membrane preparations were analyzed, all three variants scored positive for OGM modification (Fig. 2, compare lanes A and C). Note also that when MTSET was used to pretreat intact cells, subsequent OGM labeling of membranes was blocked in the F373C protein but was unaffected in the A345C and G401C variants (Fig. 2, compare lanes B and C). From these data it is evident that the labeling pattern of cysteine in the G401C mutant is compatible only with an internal location.

FIG. 2.

Orientation of TM12. Using conditions outlined in Materials and Methods, OGM accessibility for the G401C variant was studied. In the same experiment, cysteines of known location served as controls for assignment of a position to the cytoplasmic (A345C) or periplasmic (F373C) surface. (Left) After SDS-PAGE of purified OxlT, a fluorescence profile was recorded (bottom panel) before the same gel was stained with Coomassie brilliant blue to reveal total protein (top panel). The OxlT monomer, dimer, and higher multimers are evident (8, 9). Lanes A, OGM added to intact cells; lanes B, OGM added to membranes after pretreatment of cells with MTSET; lanes C, OGM added to membranes without pretreatment with MTSET. (Right) Deduced orientations of TM11 and TM12. Black squares represent the cytoplasmic targets, Q345C and G401C; the gray square is the periplasmic F373C.

This same approach was used to probe the topologies of other OxlT transmembrane segments, with each experiment including variants of known internal or external location, as described above (Fig. 2). These studies gave generally satisfactory results, although we did find some derivatives that were uninformative as to topology (see below), perhaps reflecting a generally compact structure for OxlT. Our conclusions rely only on those residues that could be positively analyzed as to location.

Cysteines of unusual accessibility or reactivity.

In the collection of 34 single-cysteine variants examined, we found five examples in which the location of cysteine could not be assigned (Table 1) (see below) because the target was not labeled by OGM. Nevertheless, in each of these cases, the presence of cysteine could be documented by labeling of denatured protein (not shown here; see reference 9). These examples were therefore classified as uninformative, and their topology was considered indeterminate (Table 1).

In one other instance, we noted that tests with intact cells gave efficient labeling of an external target (G309C) but that an equivalent response was not found on probing membranes, for reasons that are unclear. As a result, in this instance the effectiveness of the MTSET block could not be confirmed in the usual way, as a negative response to labeling of membranes after exposure of cells to MTSET (Fig. 2). To avoid a misclassification of this position, we asked instead whether preexposure of intact cells to MTSET could block subsequent labeling of those same cells by OGM. In fact, we found that the impermeable MTSET did block OGM labeling of this protein in intact cells (Fig. 3), and for this reason, position 309 could be assigned to the periplasmic surface.

FIG. 3.

External MTSET prevents OGM labeling of G309C. Intact cells expressing the single-cysteine variant P245C (known external location) or G309C (unknown location) were incubated with 40 μM OGM as described in the legend to Fig. 2, either before or 10 min after exposure to 2 mM MTSET.

We used the same abbreviated screen (Fig. 3) during analysis of the (putative) extracellular loop connecting TM1 and TM2, where we had generated a relatively large number of single-cysteine variants. For six of these, we conducted a full analysis of OGM accessibility (studies with both cells and membranes), and in each case it was possible to assign the target cysteine to a periplasmic location (see below). With this in mind, we chose to test the remaining three variants using only the labeling of intact cells, as described above (Fig. 3). In these cases, as before, we could document that the OGM accessibility observed with the intact cell was blocked by prior exposure to impermeative MTSET, indicating an extracellular location for these positions (see Fig. 4).

FIG. 4.

Topological assignment of cysteine substitutions in OxlT. The predicted topology of OxlT (10) is shown together with topological assignments from site-directed fluorescence labeling of 34 single-cysteine derivatives (Table 1). The 29 informative cases are shown as black (internal) or gray (external) squares; the 5 uninformative cases are shown as open squares (see text); gray squares with diagonals represent cysteines (Y40C, A41C, N42C, and G309C) assigned to the periplasm as described by Fig. 3. Representative panels describing fluorescence and Coomassie brilliant blue profiles after SDS-PAGE of purified OxlT (Fig. 2) are shown for each intracellular or extracellular loop and for the N and C termini. Polar residues now assignable to TM2 and TM11 are also indicated.

Topology of OxlT.

The topological analysis of OxlT single-cysteine variants is summarized in Fig. 4, which shows the findings for all positions analyzed (also see Table 1) together with a representative example of the OGM labeling pattern for cysteines targeted to intracellular or extracellular loops and to the N and C termini. These data are superimposed on the earlier representation of topology, which was derived from less direct considerations (1). In particular, positions assigned to the extracellular surface were readily labeled by addition of OGM to the intact cell (Fig. 4, bottom, left lanes). These same cysteines were also labeled by exposing membranes to OGM (Fig. 4, left bottom, right lanes), provided that the intact cell had not been exposed to MTSET (Fig. 4, bottom, compare center and right lanes). This labeling pattern sharply contrasts with that of the intracellular cohort, all of which were accessible only after cell lysis (Fig. 4, top, compare left and right lanes) and without regard to a prior exposure to MTSET (Fig. 4, top, compare center and right lanes). We conclude from this catalog that OxlT has a minimum of 12 transmembrane segments and that the locations of these segments are consistent with the topology predicted earlier, on the basis of purely indirect arguments.

DISCUSSION

It is common practice to derive a preliminary topological map of a membrane protein by viewing its amino acid sequence in at least three distinct ways. Overall topology is usually suggested by the analysis of hydropathy (21). The cytosolic and extracytosolic faces are then often distinguished by applying the positive-inside rule (36). Finally, these inferences may sometimes be supported if one can also identify motifs characteristic of internal or external landmarks (putative glycosylation sites and nucleotide binding sites, etc.). In any individual case, it is essential to verify a preliminary structure by direct experimentation. For bacterial or yeast proteins, this second phase is most often based on use of a reporter system in which a set of N-terminal fragments is fused to a C-terminal reporter whose location is easily determined by phenotypic tests (4, 6, 23, 27, 28, 36). While this experimental approach can often generate a satisfactory model, the mechanistic basis of such genetic methods is not completely understood, nor is it always clear why the topology of the intact, functional target should be replicated in the collection of its hybrid (usually inactive) derivatives.

These kinds of issues were faced in analysis of OxlT, the founding member of a family of anion transporters in the eubacteria (3), in the archaea (33), and in plants (19). Because of its assignment to the MFS, one expects OxlT to show a topological organization that includes 12 transmembrane segments arranged in two groups of six segments each (24, 25, 30), and because circular dichroism spectroscopy (8) shows OxlT to be about 65% α-helix, one also expects these segments to be structured as transmembrane α-helices. A topological model of OxlT had provisionally identified the expected transmembrane segments, based on the interpretive tools noted above and also by comparisons with the experimentally determined topologies of other bacterial examples in the MFS, including LacY (5), UhpT and GlpT (12, 15), and RhaT (34). In the latter cases, as in many others, topology had been extensively probed by use of fusions with a reporter protein, but it seemed possible that this approach could be less informative for OxlT, which has two (putative) transmembrane segments (TM2 and TM11) enriched for polar residues. In fact, an early version of an algorithm (31) widely used to detect transmembrane helices had predicted an 11-helix structure lacking TM2 (1). (The present version of this algorithm now generates a 12-helix model compatible with our findings.) In this circumstance, it was arguable that fusions with a reporter could misrepresent OxlT topology, especially for those constructs in which the relatively polar TM2 contributed a substantial component. As a result, we felt it necessary to examine topology in a way that preserved the normal structure of our target.

Although several methods provide topological information in the context of an intact protein (7, 17, 29, 37, 38), we chose to capitalize on the targeted insertion of cysteine into an otherwise cysteine-less host. In earlier studies using this approach (18, 23), topology was inferred by asking if the labeling of a target cysteine by a membrane-permeable probe could be blocked by an impermeable thiol-specific agent. By contrast, for our work we selected impermeable probes and compared target reactivities using intact cells and membrane fragments. As a readily detectable reporter, we used OGM, whose presence was easily scored by monitoring the fluorescence of affinity-purified material (Fig. 1) (9). In most cases, it would have been enough to measure the fluorescence of purified material directly, but by recording the fluorescence profile after SDS-PAGE, we could also confirm the success of purification and evaluate the presence of impurities. (As long as aggregation of membrane preparations was limited, such impurities were of minor significance [Figs. 2 and 4].) It is also likely that we could have introduced our targets into wild-type OxlT, whose two cysteines do not react with OGM unless the protein is denatured (not shown); for simplicity, however, it seemed sensible to focus on variants having only a single modifiable residue. Where feasible, we believe this approach has distinct advantages over others, largely due to its speed and simplicity, its high signal-to-noise ratio (Fig. 1), and the low frequency with which false-positive findings are made (see below).

Using this general strategy, we analyzed 34 single-cysteine OxlT derivatives (Fig. 4 and above), and in 29 cases we were able to assign, without ambiguity, the target residue to either the internal or external surface. In five cases these tactics were uninformative because the target cysteine showed low reactivity to OGM (Fig. 4), perhaps reflecting that such positions lie on helices exposed to the lipid bilayer (9, 11), where cysteine would show a relatively high pK and a correspondingly low reactivity with maleimides. More significant, we found no positions of uncertain location (e.g., OGM labeling only on cell lysis yet blockable by pretreatment of cells with MTSET, or positions labeled from the outside by OGM yet not blocked by treating cells with MTSET). This suggests a low frequency of false positives, which strengthens confidence in the OxlT 12-helix model that emerges.

The present topological assignments define a minimum of 12 transmembrane segments in OxlT (Fig. 4). In particular, we devoted special attention to asking about the locations of the OxlT N terminus and of residues now assigned to the loop connecting TM1 and TM2. Designation of these positions as cytoplasmic and periplasmic, respectively, excludes the most likely 10-helix structure, in which the N and C termini would be extracellular and in which residues now placed in TM2 and TM11 would have been incorporated into cytoplasmic loops. The data also rule out the likely 11-helix model, whose N-terminal half would be organized as in the 10-helix variant. These findings are of considerable mechanistic significance, since OxlT now incorporates two transmembrane helices (TM2 and TM11) whose polar character suggests a simple model of how an anionic substrate(s) might be bound during transport. Thus, by virtue of its position at the center of TM11 (9) (Fig. 4), it is arguable that the positively charged K355 engages in ionic interaction with one of the two carboxylates of oxalate²⁻. Indeed, positive evidence in support of this latter view has been obtained (10). In addition, concrete evidence from UhpT, another member of the MFS, supports the idea that a charged residue at the center of TM11 can play this kind of role (13). In the same way, having confirmed the identify of TM2 in the present work, one may reasonably speculate that the second oxalyl carboxylate is accommodated in a network of hydrogen bonds made available by the eight polar residues on this helix (Fig. 4). With a clear view of OxlT topology now in hand, it is appropriate to begin testing the hypothesis that the polar residues of TM2, together with K355 on TM11, form a substrate-binding element. Owing to the small size of oxalate, this scenario also demands the close proximity of TM2 and TM11. For this reason, we believe it significant that these two helices are neighbors in a theoretical reconstruction of the helix array for members of the MFS (11).