Evaluation of secondary structure of OxlT, the oxalate transporter of Oxalobacter formigenes, by circular dichroism spectroscopy.

OxlT, the oxalate/formate exchange transporter of Oxalobacter formigenes, was purified as a histidine-tagged variant, OxlTHis, using Ni2+-linked affinity chromatography. OxlTHis was readily obtained in high purity (≥95%) and reasonable yield (≥60%), and showed kinetic and biochemical features characteristic of its parent, OxlT, including an unusually high maximal velocity (60 μmol/min per mg of protein at 4°C). Circular dichroism spectroscopy of purified OxlTHis identified the α-helix as its dominant secondary structural unit, encompassing 60-70% of OxlTHis residues and consistent with a model suggesting 60% of OxlT (OxlTHis) residues are involved in the construction of 12 transmembrane α-helices (Abe, K., Ruan, Z.-S., and Maloney, P. C. (1996) J. Biol. Chem. 271, 6789-6793). In either octyl glucoside/lipid or dodecylmaltoside/lipid micelles, solubilized OxlTHis showed a striking substrate-induced stabilization of function, and at saturating levels of substrate (1000 × KD) activity recoverable by reconstitution disappeared with a half-life of 7 days at 23°C. Measurement of changes of ellipticity at 222 nm as a function of time and substrate concentration showed that maintenance of function was attributable to a substrate-induced stabilization of the α-helical ensemble with a KD of 10 μM for the 1:1 binding of oxalate to OxlTHis.

The anion transporter, OxlT, carries out the exchange of divalent oxalate with monovalent formate at the plasma membrane of the anaerobe, Oxalobacter formigenes (1,2), and in doing so catalyzes the vectorial portion of a proton-motive metabolic cycle (3) that drives ATP synthesis in this organism (1,3). While OxlT is of interest for this novel position in cell biology, it also attracts attention for certain of its biochemical features. For example, studies of OxlT purified from O. formigenes suggest this carrier operates with the highest velocity known among transporters of organic molecules (4). Moreover, examination of crude extracts of the solubilized protein suggests that substrate binding energy contributes substantially (Ն3.5 kcal/mol) to stabilization of the OxlT-substrate complex (5). As well as showing the soluble protein retains information relevant to substrate binding, this last observation suggests that structural transformation(s) accompanying substrate binding and transport might be investigated by suitable in vitro experiments with solubilized protein.
From analysis of the OxlT amino acid sequence (6) one infers a broad architectural similarity between this protein and other transporters, in both prokaryotes and eukaryotes, that display the reactions of antiport, uniport, or symport (3,7,8). This large collection of membrane proteins, now numbering in the several hundreds (3,4), is presently exemplified by a few well studied cases, such as the erythrocyte Cl/HCO 3 exchange carrier, Band 3 (9), the red cell glucose facilitator, GLUT1 (10), and the Escherichia coli H ϩ /lactose symporter, LacY (11,12). In these three examples, biophysical tests indicate the ␣-helix as the main secondary structural element (9,13,14), and this direct experimental result supports inferences made from more widely available (but indirect) biochemical and genetic studies that point to the presence of 10 -12 transmembrane segments in most examples in this group (3,7,8,13,15,29). Indeed, the transmembrane ␣-helix is often assumed as the dominant structural element in transporters of this type (16,17), but except for the cases noted this supposition has not been tested directly. For this reason, and because of the unusual kinetic features of OxlT, the initial objective of work reported here was to assess secondary structure in this transporter. To do this, we installed a histidine tag at the OxlT C terminus and purified the tagged protein, OxlTHis, by affinity chromatography using a Ni 2ϩ -linked adsorbent. The CD spectrum of solubilized, purified OxlTHis shows that the majority of its residues are organized in an ␣-helical configuration, supporting a generic model in which this and similar transporters contain 10 -12 transmembrane ␣-helices. Our second objective was to use the purified protein to correlate structure and function. Such studies show that the loss of activity observed in the absence substrate reflects a spontaneous breakdown of the ␣-helical ensemble.

EXPERIMENTAL PROCEDURES
Expression Plasmids and Bacterial Strains-pBKOxlT was constructed by insertion into pBluescript II SKϩ (Amp r ) of a 1.4-kilobase XbaI-HindIII fragment encoding OxlT (6). In this construct, the first of two in-frame OxlT UAA stop codons are flanked by BsrGI and NheI sites. To make an expression vector encoding histidine-tagged OxlT, we digested pBKOxlT with both BsrGI and NheI and then ligated the products with a synthetic BsrGI-NheI bridging oligonucleotide specifying nine consecutive histidine that would extend from the OxlT Cterminal histidine. The sequence of the resulting vector, pBDOxlTHis, was confirmed by sequencing of double stranded DNA using the dideoxy chain termination method of Sanger (18). pBKOxlT and pBDOxlTHis were expressed in XL3, a strain that also carries pMS421 (spec r , LacI q ) to ensure repression of OxlT and OxlTHis in the absence of induction (6).
Expression of OxlT and OxlTHis-A single colony of XL3 carrying either pBKOxlT or pBDOxlTHis was dispersed in 5 ml of LB broth with antibiotics (100 g/ml ampicillin, 50 g/ml spectinomycin); after overnight growth with vigorous shaking, cells were added to 0.5 liter of fresh media and grown until A 600 reached 0.2-0.3, at which point 1 mM isopropyl-1-thio-␤-D-galactopyranoside was added to induce expression of OxlT or OxlTHis. Four hours later, cells were harvested by a 10-min centrifugation at 4,000 ϫ g and lysed by incubation at 37°C for 15 min in 50 ml of lysis solution (300 g/ml lysozyme, 40 g/ml DNase, 0.5 mM freshly dissolved phenylmethylsulfonyl fluoride) (19). The resulting * This work was supported by Research Grant MCB-9220823 from the National Science Foundation and United States Public Health Service Grant GM24195 from the National Institutes of Health. 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. ghosts were washed twice with iced distilled water, and membrane proteins were extracted at pH 7 by incubation for 60 min on ice with 25 ml of solubilization buffer (20 mM potassium phosphate, 6 mM ␤-mercaptoethanol, 20% glycerol, 0.42% acetone/ether purified E. coli phospholipid, 1.5% octyl glucoside) containing 10 mM potassium oxalate. The mixture was cleared of cell debris and unextractable material by a 30-min centrifugation at 145,000 ϫ g; the remaining crude extract, usually about 300 g of protein/ml, was stored at Ϫ80°C.
Purification of Histidine-tagged OxlT-To purify OxlTHis, a 5-ml Quik-Sep column (Isolab) was packed with 0.2 ml of Ni 2ϩ -NTA resin (Qiagen), and the resin was washed at 4°C with 20 ml of distilled water. The column outlet was sealed with Parafilm, 3 ml of crude detergent extract was added, and after sealing the column inlet, the resin and extract were mixed by gentle rotation for 4 h at 4°C. The extract, depleted of OxlTHis, was allowed to drain from the column, and nonspecifically bound residual material was removed by a 20-ml wash at 1 ml/min, using solubilization buffer supplemented with 200 mM sodium fluoride and 50 mM imidazole, along with OxlTHis substrates (10 mM potassium oxalate or 100 mM potassium formate) as necessary. OxlTHis was eluted by centrifugation after a 20 -40 min incubation with 0.2 ml of solubilization solution containing either 500 mM imidazole or 200 mM EDTA, each neutralized by potassium hydroxide, with or without OxlT substrates.
Reconstitution of OxlTHis and OxlT-To compare the transport activities of OxlTHis and OxlT, we first set the two proteins at approximately equal concentrations by mixing 1 volume of purified OxlTHis or a crude cell extract containing OxlT with 149 or 9 volumes of solubilization buffer, respectively. Reconstitution from these mixtures then followed earlier procedures (1,4,20). Briefly, the proteins were dispersed in lipid/detergent micelles by a sequential mixing of the following components: 14.6 l of 100 mM potassium phosphate (pH 7.0); 4.5 l of 15% octyl glucoside; 35 l of 45 mg/ml bath-sonicated E. coli phospholipid; and 200 l of the OxlTHis or OxlT mixtures. The suspensions were kept on ice for 20 min, after which oxalate-loaded proteoliposomes were formed by adding 5 ml of 23°C loading buffer (100 mM potassium oxalate, 50 mM potassium phosphate, pH 7) to bring the detergent level below its critical micellar concentration. After an additional 20 min, proteoliposomes were used to determine OxlTHis or OxlT transport activity in either of the two assays outlined below.
Assays of Oxalate Transport-In routine work, oxalate transport was determined at 23°C by a simplified assay (4) in which replicate 0.1-ml portions of the proteoliposome suspension were applied to the center of a pre-soaked Millipore GS filters (0.22 m pore size). After removing external loading buffer by a wash with 5 ml of assay buffer (100 mM potassium sulfate, 50 mM potassium phosphate, pH 7), the vacuum was interrupted, and the filter with entrapped proteoliposomes was overlaid with 0.3 ml of assay buffer containing 0.1 mM [ 14 C]oxalate. The oxalate exchange reaction was complete by 4 min, at which time the assay was terminated by washing the filter twice with 5 ml of assay buffer. For this fixed-time assay, activity is reported as micromoles of [ 14 C]oxalate incorporated per mg of protein during the 4-min incubation.
To measure the kinetic parameters of oxalate transport, proteoliposomes were isolated by centrifugation (20), washed once with assay buffer, and after resuspension in assay buffer, the time course of oxalate transport was measured at 4°C by incubation with [ 14 C]oxalate of the indicated concentrations. Reactions were terminated by filtration and washing (1,18).
Electrophoresis and Western Blotting-Samples of fractions obtained during purification of OxlTHis were subjected to SDS-PAGE 1 using 12% acrylamide, as described (21), and protein content was evaluated by staining with Coomassie Brilliant Blue. The position of OxlTHis was verified by Western blot of a duplicate gel in which proteins were transferred to nitrocellulose (22) and exposed to a polyclonal antiserum selective for the OxlT N terminus (6). Binding of the primary antibody was visualized by chemiluminescence (Amersham) (6).
CD Spectroscopy-CD spectra were measured for OxlTHis which had been passed over a Centri-Spin20 desalting column (Princeton Separation) to remove the EDTA used for elution. Unless noted otherwise, the column was prehydrated with 0.5 ml of solubilization buffer (with or without added substrate, according to the experimental design), and residual buffer was removed by a 2-min centrifugation at 750 ϫ g. EDTA-eluted OxlTHis (80 l) was placed on the column, and the desalted product was collected by a second centrifugation; as a blank, a solution containing all components except OxlTHis was prepared in parallel.
CD spectra were collected using an AVIV 60DS circular dichroism spectropolarimeter; calibration with (ϩ)-10-camphorsulfonic acid gave a ratio of Ϫ2.25 for ellipticities at 192.5 and 290.5 nm. The CD spectrum of OxlTHis (0.2-0.5 mg/ml in protein) was measured in a 0.01-cm path length quartz cell, with temperature set to 25°C unless otherwise noted. Data were obtained at constant slit width (1.5 nm), and each spectrum was acquired as the result of five repeated scans from 260 to 190 nm, using scan intervals of 0.5 nm and an integration time of 1 s at each wavelength. The time for data collection was about 20 min for each set of five scans. Blank and experimental samples showed identical responses above 250 nm, corresponding to ϩ1 millidegree, suggesting the absence of significant differential scattering of left or right circularly polarized light by detergent/lipid micelles (23); blank values remained at ϩ1 millidegree as the wavelength was reduced to about 215 nm and then rose to ϩ3 millidegrees by 190 -195 nm. Experimental samples showed peak negative and positive deflections from the blank at 220 -222 nm and 190 -195 nm, respectively; depending on protein concentration, at these wavelengths experimental samples showed uncorrected signals of between Ϫ4.5 to Ϫ12 millidegrees (220 -222 nm) and ϩ20 to ϩ50 millidegrees (190 -195 nm), indicating adequate signalto-noise ratios within the range investigated. Raw spectra were corrected by subtraction of baseline spectra obtained using the blank solution, and after smoothing, the net spectra were used for estimation of OxlTHis ␣-helical content.
CD spectral measurements were expressed as mean residue ellipticity, in units of degree cm 2 dmol Ϫ1 , according to: mean residue ellipticity ϭ (100 ⅐ ⅐ M)/(c ⅐ d ⅐ n), where is measured ellipticity (deg), M is molecular weight, c is protein concentration (mg/ml), d is path length (cm), and n is the number of residues in OxlTHis (n ϭ 427). Spectral decomposition was done by least-squares fit (Sigma plot) of mean residue ellipticity values at 200 -260 nm to the contributions of four components (␣-helix, ␤-strand, ␤-turn, and unordered) taken from the basis set tabulated by Yang et al. (24) or Park et al. (25), constraining the sum of contributions to equal 1. Because measurements at wavelengths below 200 nm were not considered, only estimates of ␣-helix content are reported (26). For the basis set of Yang et al. (24), an average helix length of 24 residues was used to account for the dependence of ellipticity on helix length (24).
Protein Estimation-Protein was estimated by a modification of the method of Schaffner and Weissman (27).

RESULTS
Purification and Characterization of OxlTHis-To enable convenient purification of OxlT, we exploited Ni 2ϩ -linked affinity chromatography. The principle underlying this approach is that a sequence of six or more tandem histidine residues can form a relatively tight binding pocket for the divalent nickel cation, allowing a suitably tagged protein to be retained by a Ni 2ϩ -containing resin (28). Accordingly, we prepared OxlT variants having 10 consecutive histidine residues at either the N terminus (HisOxlT, not described) or the C terminus (OxlTHis, see "Experimental Procedures"). Because preliminary trials (not given) indicated that the latter showed somewhat tighter binding to the Ni 2ϩ -linked resin, we focused on purification and characterization of OxlT having the C-terminal histidine tag.
To purify OxlTHis, we first prepared membrane ghosts by lysis of cells overexpressing this protein and then solubilized membranes using octyl glucoside in the presence of excess E. coli phospholipid, with glycerol added as the osmolyte stabilant (20). As anticipated (4, 6), the SDS-PAGE profile of the detergent extract showed OxlTHis as a protein with an apparent mass near 35-kDa (Fig. 1, left). OxlTHis was depleted from the extract by incubation with the Ni 2ϩ -linked adsorbent, and reappeared only when wash conditions were altered so as to release Ni 2ϩ -linked materials (Fig. 1, left). Western blot analysis verified these findings. Thus, only the crude extract and EDTA-eluate contained material reactive with an antibody directed to the OxlT N terminus (Fig. 1, right). In each case, the major element moved as a 35-kDa protein, corresponding to monomeric OxlTHis; dimeric and trimeric OxlTHis (4, 6) were evident in the Western blot (about 75 and 125 kDa, respective-ly), but were not seen in the Coomassie-stained gel due to their low abundance. 2 Since proteins other than OxlTHis were not detected ( Fig. 1, left), we concluded that OxlTHis is readily prepared at high purity (Ն95%) by Ni 2ϩ -linked affinity chromatography.
The function of purified OxlTHis was assessed by [ 14 C]oxalate transport into oxalate-loaded proteoliposomes, and in the abbreviated assay used to monitor the purification, EDTAeluted OxlTHis showed an activity of 92 mol/mg of protein ( Table I); material of equivalent activity (114 mol/mg of protein) was found in experiments that used 500 mM imidazole for elution (Table I). Because these values compare favorably with that observed after purification of OxlT by conventional methods from O. formigenes (60 mol/mg of protein) (4), we conclude that heterologous expression of OxlT (OxlTHis) in E. coli has no effect on the intrinsic properties of this transporter.
For the experiment illustrated by Fig. 1, densitometry of the Coomassie-stained gel indicated OxlTHis comprised 8.4% of the initial crude detergent extract. Since purified OxlTHis was recovered at about 10 -20 g of protein/ml crude extract, or about 5% of the starting material (4.8 Ϯ 1.2% (mean Ϯ S.E.) in six experiments), overall recovery was near 60% (new work indicates that yield may be increased by extending the time of exposure to EDTA during elution). Given OxlTHis as 5-10% of the crude extract, one expects a 10 -20-fold increased specific activity if the purified protein is fully functional. Since the mean specific activity of the pure product was 15-19-fold increased over the crude extract (Table I), it seemed likely that OxlTHis was obtained with little if any loss of function.
In additional work, we conducted a more detailed comparison of the transport activities of purified OxlTHis and its parent, OxlT; to obtain quantitative information, kinetic behavior was monitored at 4°C, since OxlT-mediated transport is unusually rapid (maximal velocity of about 1 mmol/min per mg of protein at 23°C) (4). Such tests established that for the usual conditions of assay the time course of oxalate exchange was the same for OxlTHis and OxlT (half-times of about 5 min) and that, as expected for simple antiport, the reaction could be described as an exponential approach to equilibrium (Fig. 2A). The two proteins also had similar Michaelis constants for [ 14 C]oxalate transport (135 and 145 M, respectively, for Ox-lTHis and OxlT) (Fig. 2B), although the maximal velocities of the two preparations differed considerably (60 versus 4.3 mol/ min per mg of protein, respectively). This last finding was expected, since OxlTHis function had been assessed with purified material, whereas OxlT activity was measured in a crude extract (e.g. Table I). These observations, together with the information noted above, lead us to conclude that the histidinetagged OxlTHis retains the essential aspects of catalytic activity of its parent, OxlT.
Stability of OxlTHis-OxlTHis had been purified to evaluate its secondary structure by CD spectroscopy, and because this would require extended incubation of solubilized protein at 25°C (or higher), it was essential to document the stability of OxlTHis for these general conditions. For this reason, solubilized (and desalted) OxlTHis was dispersed in mixed lipid-detergent micelles in the presence of 10 mM oxalate (to take advantage of substrate-stabilization (5)), and at intervals during incubation at room temperature samples were withdrawn to measure residual activity by reconstitution. The analysis was done with both octyl glucoside and dodecylmaltoside as solubilizing agents, since these are the detergents most commonly used for solubilization of bacterial membrane proteins. For each of these preparations, 50 -60% of the initial activity was recovered after 7 days incubation (Fig. 3), indicating that purified OxlTHis was sufficiently stable for spectral analysis.
CD Spectroscopy of OxlTHis-Early work showed that solubilized OxlT retains a capacity to bind its substrates (5), implying that the solubilized material has structural information relevant to the overall process of membrane transport. Consequently, it seemed reasonable to use CD spectroscopy to probe secondary structure of the solubilized protein.
The CD spectra recorded for solubilized OxlTHis   (Fig. 4, panel 2) showed features expected for a protein with a predominantly helical structure, that is, negative deflections in mean residue ellipticity above about 200 nm, with a minimum near 222 nm, and stronger positive deflections below 200 nm, with a maximum near 195 nm. Accordingly, one may conclude that OxlTHis (and OxlT) contains the ␣-helix as its main structural element. This overall structure was not influenced by the presence of bulk lipid, since an equivalent spectrum was re-corded for protein eluted from Ni 2ϩ -agarose and further processed with lipid-free buffer (Fig. 4, panel 3). It was also evident that the spectra of OxlTHis bound to oxalate or formate were nearly superimposable (Fig. 4, panel 2), while the spectrum taken in the absence of substrate (Fig. 4, panel 4) had both its minimum (near 222 nm) and maximum (near 195 nm) reduced in absolute value. Since there had been gradual decline of these peak values during the repeated scans required for signal averaging (not shown), it appeared that the helical nature of OxlTHis became progressively less prominent with continued incubation in the absence of substrate. This behavior was not studied further in these experiments (but see below), except to note that it correlated with loss of function, as shown by a separate experiment in which activity was monitored subsequent to the recording of CD spectra. In those cases, there was essentially complete recovery of [ 14 C]oxalate transport activity for samples incubated with oxalate or formate (95 or 80% recovery, respectively), but only a 30% recovery for the sample processed without substrate.
In the second part of the experiment shown by Fig. 4, each sample was exposed to elevated temperature (55°C) before recording its CD spectrum in a single 3-min scan. Elevated temperature had little apparent effect on samples in which substrate was present, even after a 50 -55 min incubation (Fig.  4, panels 6 and 7), but in the absence of substrate, the signal became so severely degraded that heating was discontinued after only 15 min (Fig. 4, panel 8). This maintenance of an organized helical structure at high temperature reinforces the idea that the OxlTHis-substrate complex has an unusual stability (Ref. 5, and see below).
CD spectra collected in this (Fig. 4) and two other experiments (not shown) were used for estimation of secondary structure using basis spectra describing the responses of the ␣-helix, ␤-sheet, ␤-turn, and random coil configurations, constraining the sum of the fractional contributions to equal 1. To reconstruct experimental spectra, a least-squares analysis was performed using the basis set of Yang et al. (24) (Fig. 4, panel 1) or   FIG. 3. Stability of OxlTHis. OxlTHis was purified by elution with EDTA in the presence of 10 mM potassium oxalate, as described under "Experimental Procedures," except that elution was performed with either octyl glucoside (⅜) or dodecylmaltoside (Ⅺ) as detergent, each at 1.5%. After desalting, the purified proteins were kept at 23°C in sealed Microfuge tubes, and at the indicated times aliquots were removed and frozen at Ϫ80°C for later analysis. Frozen aliquots were thawed, diluted 400-fold with solubilization buffer, and reconstituted for measurement of remaining [ 14 C]oxalate transport activity as described under "Experimental Procedures." Values shown are the means Ϯ S.D. of triplicate determinations.  (24). The corresponding helix content is indicated in each panel.
Park et al. (25) (Fig. 4, panel 5), both of which derive from analysis of globular proteins of defined structure. The OxlTHis spectra were not well accommodated over their entire span (190 -260 nm) by either basis set, largely because of deviations at wavelengths below 200 nm (not shown). Since the background signal increased significantly in this region, poor fit was attributed in part to non-protein chiral signals that were not removed by background subtraction. To improve overall fit, therefore, the experimental spectra were truncated at 200 nm, and with this modification it became possible to obtain consistent findings with regard to the ␣-helical composition of Ox-lTHis. When calculations were based on the reference set of Yang et al. (24), OxlTHis was best described as containing 62-67% ␣-helix in the presence of substrates and 52% ␣-helix in their absence (at 25°C) (Table II, Method I), whereas the Park et al. (25) basis set gave helix contents as 72-74% (with substrates) or 65% (without substrate) (Table II, Method II).
While each method of calculation gave a similar estimate of OxlTHis helical content (Table II), the calculated contributions by ␤-sheet, ␤-turn, and unordered regions varied significantly (not shown). This had been anticipated, since different assumptions are embodied by the different basis sets (e.g., compare Fig. 4, panels 1 and 5), and because estimates of non-helical structures are inaccurate unless the experimental spectra extend to 184 nm or below (26). As a result, while it was not possible to assess the contributions of ␤-sheet, ␤-turn, or random coil, there was consistent evidence that the ␣-helix comprises the dominant structural element, encompassing about 60 -70% of OxlTHis residues.
Correlation between Structure and Function-Prior studies had noted a relatively rapid loss of OxlT activity (half-life of 6 -7 min at 37°C) when crude extracts were not supplemented with substrate (5). An explanation for this phenomenon was suggested by the present observations that incubation in the absence of substrate was associated with loss of both function and helical structure ( Fig. 4 and text). Thus, it was feasible that substrate binding to OxlTHis might act to stabilize the helical ensemble itself. To test this idea, we measured the time course of changes in ellipticity at 222 nm as a function of oxalate concentration, using temperature (37°C) and substrate levels (0 -1000 M) that gave measurable changes in ellipticity over a time interval (Յ5 min) minimizing baseline drift. This work (Fig. 5) confirmed the strong relationship between substrate binding and preservation of helical structure. If one assumes the binding of oxalate and OxlTHis is associated with a change in protein structure from a conformation that decays spontaneously (half-life of 4 -5 min at 37°C) to one that is stable on the time scale relevant to the experiment (e.g. Fig. 3), these data are accommodated by the 1:1 binding of oxalate and OxlTHis with a dissociation constant (K D ) of 10 M (Fig. 5,  inset). Such findings establish a direct correlation between the function of OxlTHis and its helical content.

Purification and Characterization of OxlTHis-The
OxlT amino acid sequence (6) indicates it has a global structure resembling any one of a large number of transport systems that mediate the reactions of symport, uniport, or antiport (3). The transmembrane ␣-helix is usually taken as the main structural element for these and many other transporters, a presumption that is rarely supported by experiment. For this reason, and because early work with crude extracts had shown OxlT to have unusual stability as a solubilized protein (5), we reasoned that direct and useful information could be gained by CD spectroscopy of OxlT if a suitably purified preparation were available.
OxlT had been purified earlier using traditional techniques (4), but as is often the case with membrane proteins, the tendency of OxlT to associate in non-stoichiometric fashion with detergents and phospholipids leads to a heterogeneity in both size and charge that makes it difficult to apply conventional methods of purification in predictable fashion. It is possible to simplify and accelerate this process by affinity methods (30 -32), and we elected to use Ni 2ϩ -linked chromatography of a histidine-tagged variant (OxlTHis), a tactic that enabled the convenient production of highly purified material on a routine basis. Purified OxlTHis has the specific activity expected of the fully functional protein (Table I and text), as one might expect from the finding that (with substrate present) OxlTHis lifetime is long compared to the time required for its purification. OxlTHis also retains the kinetic features of its parent, including its unusually high maximal velocity (Fig. 2); this corresponds to a turnover number of about 100/s at 4°C, since only half the protein used for reconstitution is incorporated by proteoliposomes (4). Clearly, analysis of OxlTHis is relevant to understanding its parent, OxlT.
CD Spectroscopy of OxlTHis-In extracting structural infor-  5. Substrate stabilization of OxlTHis helical structure. OxlTHis was purified in the absence of substrate. OxlTHis (9 M) was mixed with oxalate of the designated concentrations, and ellipticity at 222 nm was monitored at 30-s intervals at 37°C. Inset, given that ellipticity of unliganded OxlTHis decays spontaneously, while that of liganded OxlTHis does not, the rate constant for decay in the presence of substrate (K app ) equals (1-Y)K, where K is the rate constant for decay in the absence of substrate, and Y is determined by the dissociation constant for the liganded complex (K D ), the concentration of oxalate (S), and binding stoichiometry (n) (Y ϭ S n /[S n ϩ K D n ]). The inset graph expresses these relationships by giving 1-Y (ϭK app /K) as a function of oxalate concentration. The line is a least squares fit obtained by nonlinear regression, with a derived K D of 9.7 Ϯ 1.7 M and an apparent stoichiometry of 1.1 Ϯ 0.2 (mean Ϯ S.D.). In calculating occupancy at 3-10 M oxalate, the concentration of OxlTHis itself was taken into account. mation from CD spectroscopy, one makes four general assumptions (24 -26, 33): (i) that the CD spectrum arises as the linear combination of its component spectra (i.e. those of the ␣-helix, ␤-sheet, ␤-turn, unordered regions, etc.); (ii) that these component, or basis, spectra may be deduced from the behavior of reference proteins whose structures are understood; (iii) that an appropriate set of reference proteins is available; and (iv) that the experimental spectrum extends over a sufficiently wide range to provide statistically significant information about the component spectra. Of these assumptions, the last two are at issue in the work described here. It is acknowledged, for example, that basis spectra deduced for the ␤-pleated sheet, ␤-turn, and random coil configurations are sensitive to the choice of the reference set (33) (e.g. Fig. 4, panels 1 and 5), and an inability to obtain accurate spectral information below 195-200 nm (as in these experiments) further limits one's confidence in attempts to evaluate such components (26). These are the likely explanations for failure to derive consistent estimates for these structures in OxlTHis, as well as for the relatively poor fit by theoretical reconstructions when helix content was significantly diminished (e.g. Fig. 4, panel 8). By contrast, it is generally held that the ␣-helix yields a signal of relative uniformity in different environments (but see below), and since the helix chiral signal dominates in the more readily detected wavelengths, estimates of this parameter are the more likely to be accurate in the range considered here (Ն200 nm), especially if the target has a substantial helical content (26,33). In fact, the CD spectrum of solubilized OxlTHis has the profile expected of a largely helical protein (Fig. 4), and calculations using two different basis sets gave similar estimates of Ox-lTHis helical content (Table II). One might also note that as structural information on membrane proteins is limited, analysis of the OxlTHis CD spectrum relies on component spectra derived from globular proteins, and this may carry potential risk. As an example, it has been proposed that the transmembrane ␣-helix differs from the ␣-helix exposed to water in having a somewhat stronger signal than expected (33), and this idea, still under discussion, raises the possibility that standard calculations (e.g. Table II) may overestimate the helix content of membrane proteins. Nevertheless, for the photosynthetic reaction centers of Rhodobacter viridis and Rhodopseudomonas sphaeroides, whose structures are known from crystallography, such error appeared to be small, and Method I (Table II) overestimated helix content by only about 10% (39% observed versus 42.5% calculated and 51% observed versus 56% calculated, respectively) (33). Similarly, Method I provided a helical content for the nicotinic acetylcholine receptor (23%) (34) consistent with electron diffraction data (35).
It appears probable that the structure of solubilized OxlTHis resembles that adopted by this protein in the lipid bilayer.
Although this assumption was not tested here in any specific way, it has been verified for the acetylcholine receptor (34) and the water channel, CHIP28 (36). And for the LacY symporter, presumed to resemble OxlT in general structure, overall helix content was not different for the solubilized and reconstituted protein (cited in Ref. 13). In addition, since tests with crude preparations of OxlT indicate the solubilized protein binds its substrates (5), we take the information reported here as relevant to understanding structure-function relations in both solubilized and membrane-bound OxlT (OxlTHis). Accordingly, given a helical content of 60 -70% (Table II), one expects that 256 -295 of the 427 OxlTHis residues are organized in this way, suggesting that OxTHis could have 10 -13 ␣-helices of sufficient length (23-25 residues) to span the lipid bilayer. The OxlTHis CD spectrum is therefore consistent with models derived from hydropathy analysis, which predict 11 or 12 trans-membrane ␣-helices (6). Inasmuch as hydropathy analysis and other indirect methods generate similar models for a large number of transporters (3,7,11), our findings, together with those of the LacY symporter (13), the band 3 anion exchanger (9), and the glucose carrier (14) reinforce this as a generic feature of such transporting systems.
More important than an overall consistency between such models and experiment, this work now establishes a quantifiable relationship between structure and function for OxlTHis. This was first noted in a qualitative way by the parallel reduction of both activity and helical content during spectral analysis (Fig. 4). Subsequently, we explored this explicitly by measuring the kinetics of spontaneous decay of helix content and by showing that a K D for oxalate could be derived from the substrate dependence of this process (Fig. 5). We had earlier used reconstitution of activity as the method of analysis, and while that gave a similar estimate of K D (20 M [5] versus the 10 M recorded here), such observations remained a phenomenological description without clue as to underlying mechanism. The present work confirms the initial findings (5) and suggests as well that loss of function in the absence of substrate arises from a breakdown in helical structure. Such observations offer strong evidence that the solubilized protein binds substrate, likely in a 1:1 stoichiometry (Fig. 5, legend), and show how CD or other spectroscopic tools might assess secondary structure for the solubilized transporter.
We note in conclusion that purified OxlTHis can be maintained for an extended period in detergent/lipid micelles without loss of function, provided that substrate is present (Fig. 3).
Since it has been difficult to study membrane transporters after their extraction from the lipid bilayer, our work establishes OxlT as a robust model with which to investigate correlations between structure and function using the solubilized material.