Topology of AspT, the Aspartate:Alanine Antiporter of Tetragenococcus halophilus, Determined by Site-Directed Fluorescence Labeling

Abstract

The gram-positive lactic acid bacterium Tetragenococcus halophilus catalyzes the decarboxylation of l-aspartate (Asp) with release of l-alanine (Ala) and CO₂. The decarboxylation reaction consists of two steps: electrogenic exchange of Asp for Ala catalyzed by an aspartate:alanine antiporter (AspT) and intracellular decarboxylation of the transported Asp catalyzed by an l-aspartate-β-decarboxylase (AspD). AspT belongs to the newly classified aspartate:alanine exchanger family (transporter classification no. 2.A.81) of transporters. In this study, we were interested in the relationship between the structure and function of AspT and thus analyzed the topology by means of the substituted-cysteine accessibility method using the impermeant, fluorescent, thiol-specific probe Oregon Green 488 maleimide (OGM) and the impermeant, nonfluorescent, thiol-specific probe [2-(trimethylammonium)ethyl]methanethiosulfonate bromide. We generated 23 single-cysteine variants from a six-histidine-tagged cysteineless AspT template. A cysteine position was assigned an external location if the corresponding single-cysteine variant reacted with OGM added to intact cells, and a position was assigned an internal location if OGM labeling required cell lysis. The topology analyses revealed that AspT has a unique topology; the protein has 10 transmembrane helices (TMs), a large hydrophilic cytoplasmic loop (about 180 amino acids) between TM5 and TM6, N and C termini that face the periplasm, and a positively charged residue (arginine 76) within TM3. Moreover, the three-dimensional structure constructed by means of the full automatic modeling system indicates that the large hydrophilic cytoplasmic loop of AspT possesses a TrkA_C domain and a TrkA_C-like domain and that the three-dimensional structures of these domains are similar to each other even though their amino acid sequences show low similarity.

In some strains of the lactic acid bacterium Tetragenococcus halophilus, a proton motive force (PMF) is generated by the combined action of an intracellular l-aspartate decarboxylation catalyzed by an l-aspartate-β-decarboxylase (AspD) and an electrogenic aspartate¹⁻:alanine⁰ exchange reaction catalyzed by an aspartate:alanine antiporter (AspT; transporter classification [TC] no. 2.A.81.1.1). The generated PMF is sufficiently high to drive ATP synthesis via the bacterial F₀F₁ ATPase. This combination of PMF and ATP synthesis has been proposed as a proton motive metabolic cycle, and the prototype is found in Oxalobacter formigenes (3, 8, 32).

In previous work using proteoliposomes, we found that the aspartate:alanine exchange catalyzed by AspT is electrogenic (1, 2). The biochemical features of substrate transport by AspT indicate that this protein can be classified as a conventional secondary transport protein and that it is an electrogenic antiporter similar to the prototype precursor:product exchanger OxlT, the oxalate:formate antiporter, which is a member of the major facilitator superfamily, from O. formigenes (3, 8, 32, 37).

Although AspT and OxlT exhibit similar transport properties, such as electrogenic precursor:product antiport behavior and electroneutral self-exchange, their putative amino acid sequences show no similarity. In addition, AspT belongs to the newly classified aspartate:alanine exchanger (AAE) family (TC no. 2.A.81) of transporters in the TC system (http://www.tcdb.org/index.php) developed by M. H. Saier, Jr. et al. The TC system has been approved by the International Union of Biochemistry and Molecular Biology (http://www.chem.qmul.ac.uk/iubmb/) as a classification system for membrane transport proteins.

Recently, the results of a BLAST (http://www.ncbi.nlm.nih.gov/BLAST/ [5]) search of the nucleotide sequence of the aspT gene and the amino acid sequence of the AspT protein against current nucleotide and protein databases have suggested that AAE family transporters are conserved in many bacterial species (11, 20). AspT is the only functionally characterized protein belonging to the AAE family (2). The putative broad distribution of AspT orthologues and paralogues in bacteria suggests that additional biochemical study of AspT can be a valuable part of the ongoing investigation of membrane transport.

We have studied the topology of AspT by using alkaline phosphatase (PhoA) and β-lactamase (BlaM) fusion methods (34), and our results suggested that AspT has a unique topology. However, the structure of the C-terminal half of AspT remains ambiguous because fusion methods that use reporter proteins having high molecular masses do not yet have a resolution that is high enough to provide a detailed topology of the transporter.

Over the past decade, the membrane topology of primary and secondary transporters has been studied extensively by means of the substituted-cysteine accessibility method (SCAM), which involves cysteine-scanning mutagenesis in combination with thiol-specific labeling of single cysteines artificially introduced into the target membrane proteins. This method has been used to elucidate the membrane topologies of, for example, OxlT (18, 19, 45, 46), the tetracycline:H⁺ transporter (41), the Na⁺:H⁺ antiporter (29), lactose permease (27), the multidrug transporter P-glycoprotein (31), and the glutamate transporter GLT-1 (21). Therefore, we used this method to elucidate the detailed membrane topology of AspT.

First, we used SCAM to analyze the orientation of transmembrane helix 3 (TM3). Our PhoA and BlaM fusion studies indicated that there is probably an isolated charged residue (arginine 76) in the middle of TM3. The presence of an uncompensated charge within a hydrophobic transmembrane region is generally thought to destabilize the structure of membrane proteins (15, 30). We next examined the orientations of the remaining transmembrane segments in AspT by using SCAM to elucidate the unknown structure of the C-terminal half of AspT. The topology analyses revealed that AspT does indeed have a unique topology; the protein has 10 TMs, a large hydrophilic cytoplasmic loop (about 180 amino acids) between TM5 and TM6, and N and C termini that face the periplasm. Computations using recent genome data have predicted that the AAE family is composed of membrane proteins possessing a unique membrane topology (11), and our topology analyses support this prediction.

We used the full automatic modeling system (FAMS) (http://www.pharm.kitasato-u.ac.jp/fams/index.html [24, 35]) to carry out homology modeling of the three-dimensional structure of the large hydrophilic cytoplasmic loop of AspT. The model predicted that the loop possesses two alpha-beta complex structures (CATH Protein Structure Classification; http://cathwww.biochem.ucl.ac.uk/latest/index.html [36]). Furthermore, we produced and purified recombinant protein of the large hydrophilic cytoplasmic loop region to measure the circular dichroism spectrum of this protein. The circular dichroism analysis confirmed that the loop contained α-helices and β-structures.

MATERIALS AND METHODS

Materials, cells, and expression plasmids.

l-[2,3-³H]aspartic acid (1.07 GBq/mmol) was purchased from Amersham-Pharmacia Biotech (Piscataway, NJ). 1-O-n-Octyl-β-d-glucopyranoside (OG) was obtained from Nacalai Tesque (Kyoto, Japan). Escherichia coli phospholipid was provided by Avanti Polar Lipids (Alabaster, AL) (7). Oregon Green 488 maleimide (OGM) was purchased from Invitrogen Co. (Carlsbad, CA), and [2-(trimethylammonium)ethyl]methanethiosulfonate bromide (MTSET) was purchased from Toronto Research Chemicals (Toronto, Canada). E. coli strain XL1 Blue harboring pMS421 (Spec^r LacI^q), referred to as strain XL3 (3), was used for expression of the asp operon with pTrc99A (Amersham-Pharmacia Biotech).

Construction of histidine insertion variants.

AspT six-histidine insertion variants were constructed with a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) in plasmid pBluescript II KS(+) (Stratagene) including the gene for AspT (pBlueAspT). Table 1 lists the hydrophilic loop domains into which the six-histidine tag insertions were made. The oligonucleotides contained 5′-CATCATCATCATCATCAT-3′ sequences as six-histidine codons. For hydrophilic loop 7, which contains one endogenous histidine residue, the corresponding oligonucleotide utilized one endogenous histidine codon as part of the string of six-histidine codons.

TABLE 1.

Production levels and specific activities of histidine-tagged AspT variants relative to those of wild-type AspT

Variant	Production (%)a	Sp act (%)b
Wild-type AspT	100.0	100.0
N terminus	58.6 ± 1.7	54.5 ± 1.2
Loop 2 (63-64)c	79.5 ± 0.3	8.6 ± 1.8
Loop 5 (331-332)	63.9 ± 1.4	101.5 ± 10.0
Loop 7 (413-414)	68.7 ± 1.4	3.6 ± 0.3
Loop 9 (512-513)	50.5 ± 2.6	9.9 ± 0.8
C terminus	10.4 ± 4.3	32.3 ± 2.6

^aProduction is expressed relative to that of the wild-type AspT parent. The values are means ± standard deviations.

^bThe initial rates of l-[2,3-³H]aspartate transport, normalized for the levels of AspT production, are expressed relative to the rate for the wild-type AspT parent, whose activity was 10.6 nmol min⁻¹ mg of protein⁻¹. The data are the means ± standard deviations for three independent trials. For each histidine-tagged variant, the same amount of protein was used for reconstitution in l-aspartate-loaded proteoliposomes.

^cThe position of the histidine tag insertion is shown in parentheses.

Site-directed mutagenesis.

Site-directed mutations were introduced using a QuikChange site-directed mutagenesis kit or the oligodeoxyribonucleotide-directed dual amber (ODA) method (Takara Bio, Tokyo, Japan) (22). Cysteineless AspT-His₆ was constructed by replacing three endogenous cysteine residues with alanine (C19A, C110A, and C476A) in plasmid pBluescript II KS(+), including the gene for six-histidine-tagged AspT (pBlueAspT-His), using a QuikChange site-directed mutagenesis kit. DNA fragments encoding cysteineless AspT-His₆ were cloned into pTrcAsp instead of wild-type AspT. Single-cysteine variants were constructed by the ODA mutagenesis method. The pKF 19k/18k vector (Takara Bio) harboring the gene of cysteineless AspT-His₆ was used as a DNA template for the ODA mutagenesis method. DNA fragments encoding AspT into which a single cysteine had been introduced were ligated back into the corresponding site of pBlue cysteineless Asp-His. DNA fragments encoding single-cysteine AspT-His₆ were cloned into pTrcAsp instead of into wild-type AspT. The DNA sequences of all mutagenized AspT variants were verified by DNA sequencing as described previously (2).

Production of wild-type AspT, AspT-His₆, cysteineless AspT-His₆, and single-cysteine AspT-His₆.

A preculture of E. coli XL3 carrying pTrcAsp/pTrcAsp-His, pTrc cysteineless AspT-His, or pTrc single-cysteine AspT-His was diluted 100-fold in fresh Luria-Bertani (LB) medium containing 30 mM d-glucose, 30 μg/ml carbenicillin, and 30 μg/ml spectinomycin. The cells were grown for 2.5 h at 37°C with shaking, and then the culture was diluted twofold in fresh Luria broth containing 30 mM d-glucose, 60 mM l-aspartate, and 1 mM pyridoxal 5′-phosphate. The cell suspension was incubated statically for 13 h at 37°C. At 12 h prior to cell harvest, 200 μM isopropyl-β-d-thiogalactoside (IPTG) was added to the culture.

Preparation of membrane ghosts and solubilization and reconstitution of AspT.

Membrane ghosts were prepared by an osmotic shock procedure (7). Harvested cells were suspended in 5 ml of a lysis solution (500 μg/ml lysozyme [Seikagaku Co., Tokyo, Japan], 40 μg/ml DNase I [Sigma, St. Louis, MO], 10 mM Tris-HCl [pH 7.5], 1 mM phenylmethylsulfonyl fluoride [PMSF]) and incubated at 37°C for 30 min. Cells were disrupted by 10-fold dilution into 45 ml of iced distilled water. After cell disruption, the released cytoplasmic proteins were removed by two cycles of centrifugation and washing with iced distilled water (42). Membrane ghosts were solubilized (7) using 1.25% (wt/vol) OG in the presence of 0.4% (wt/vol) E. coli phospholipid, 100 mM KH₂PO₄ (pH 7) as the potassium salt, and 20% glycerol. Control extracts were prepared in the same way, but without added protein. The solubilized membrane proteins were reconstituted in a final volume of 1 ml with 800 μl of detergent extracts (1.2 mg of protein) (or control lipid extract), 130 μl of bath-sonicated liposomes (5.9 mg of E. coli phospholipid), and 18 μl of 15% OG, with the balance consisting of 100 mM KH₂PO₄ (pH 7) as the potassium salt. After incubation for 20 min on ice, proteoliposomes (or control liposomes) were formed at room temperature by rapid injection into 20 ml of a loading buffer containing 100 mM KH₂PO₄ (pH 7) and 100 mM l-aspartate as the potassium salt. The substrate-loaded proteoliposomes (or liposomes) were kept at room temperature for 20 min.

Assay of aspartate transport.

Unless otherwise noted, initial rates of l-[2,3-³H]aspartate entry were measured in duplicate at 25°C by means of a filtration assay (43, 46). Proteoliposomes were applied directly to the center of a 0.22-μm-pore-size GSTF Millipore filter (Millipore Co., Billerica, MA) and washed twice with 5 ml of chilled assay buffer (100 mM KH₂PO₄ [pH 7], 100 mM K₂SO₄). Upon release of the vacuum, proteoliposomes were covered with preincubated assay buffer containing 0.1 mM l-[2,3-³H]aspartate, and the reaction was terminated after 1 min by filtration and washing. AspT function is usually reported as relative specific activity by normalization of observed rates to levels of AspT production as determined by immunoblot analysis (described below).

Immunoblot analysis.

Detergent extracts of membrane ghosts of each of the single-cysteine variants and the cysteineless parent were analyzed by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE). For SDS-PAGE, 45 μg of proteins was used, and after electrophoresis, proteins were transferred to a polyvinylidene difluoride membrane (Nippon Genetics, Tokyo, Japan) by semidry electrophoretic blotting (Bio-Rad Laboratories, Hercules, CA) and probed with an anti-AspT rabbit polyclonal antibody, which was prepared by Operon Biotechnologies (Tokyo, Japan) using synthetic polypeptides (SKLPISDHLKTLYSNQ and NDVSERVGSDASPF) as antigens. AspT production was detected by chemiluminescence with a LAS-3000 imaging system (FUJIFILM, Tokyo, Japan), and signals were quantified using NIH Image (v. 1.63). Antisera were used at the following dilutions: anti-AspT rabbit polyclonal, 1:2,000; and anti-rabbit immunoglobulin G goat polyclonal antibody horseradish peroxidase conjugate (StressGen Bio. Co., Victoria, Canada), 1:2,000. The production of each AspT variant was normalized to that of the cysteineless parental control on the same gel.

Site-directed labeling.

Site-specific fluorescent labeling was designed to selectively attack either cysteines exposed to the extracellular medium or cysteines exposed at the intracellular surface. To label cysteines exposed at the extracellular surface, intact cells were harvested by centrifugation (6,000 × g, 30 min) and suspended in buffer A (100 mM K₂SO₄, 50 mM KH₂PO₄ [pH 8]), and the absorbance at 530 nm was adjusted to 8.5. Then 40 μM (final concentration) OGM was added to 5 ml of the cell suspension, and the mixture was incubated for 20 min at 25°C. The labeling reaction was quenched by addition of 6 mM (final concentration) β-mercaptoethanol. The cells were collected immediately and washed three times with buffer B [100 mM K₂SO₄, 50 mM 3-(N-morpholino)propanesulfonic acid (pH 7) as the potassium salt]. Labeled cells were then resuspended in 5 ml of lysis solution (500 μg/ml lysozyme, 5 mM EDTA, 10 mM Tris-HCl [pH 7.5]) and incubated at 37°C for 30 min to initiate cell rupture. Membrane ghosts were obtained by 10-fold dilution in cold distilled water, and cytoplasmic proteins were removed by three cycles of centrifugation (10,000 × g, 10 min). The membranes served as the source of AspT for solubilization and purification (see below).

Cysteines exposed at the intracellular surface were identified by a two-step protocol. External cysteines were blocked by an initial 15-min preincubation of cells at 30°C in buffer A containing 2 mM MTSET. MTSET was removed by three cycles of centrifugation and washing using buffer B without β-mercaptoethanol, and preblocked cells were used for preparation of membrane ghosts, as described above. This procedure exposed unmodified cysteines that had faced the cytoplasm, and these cysteines were labeled by incubation of membrane ghosts for 20 min at 25°C in buffer C (20 mM KH₂PO₄ [pH 8] as the potassium salt) containing 40 μM OGM. The reaction was quenched with 2 mM (final concentration) β-mercaptoethanol, and the quenching was followed immediately by three cycles of centrifugation and resuspension, using distilled water, and by protein solubilization and AspT purification (see below).

In some cases, we used OGM labeling under conditions that did not discriminate between internal and external locations. In these tests, membrane vesicles were prepared directly, without blocking of the external cysteines. The vesicles were incubated for 20 min at 25°C in buffer C containing 40 μM OGM, and then the reaction was quenched as described above.

AspT purification.

Protein was solubilized by resuspending the membrane vesicles in 1 ml of solubilization buffer (50 mM Tris-HCl [pH 7], 0.75 mM PMSF, 1% Triton X-100, 0.1% SDS). After incubation at 4°C for 2 h on a rotary platform shaker, insoluble debris was removed by centrifugation in the cold (17,500 × g, 30 min), and AspT was purified by a one-step affinity procedure (17). In brief, six-histidine-tagged AspT was precipitated with 50 μl of Ni-nitrilotriacetic acid (NTA) Superflow resin (QIAGEN, Valencia, CA) by means of overnight batch incubation in the cold on a rotary platform shaker. The resin, with bound AspT, was washed on ice with 3 ml (total volume) of wash buffer (solubilization buffer supplemented with 200 mM sodium fluoride and 30 mM imidazole). AspT was eluted by a brief centrifugation in the cold using 30 μl of elution buffer (50 mM Tris-HCl [pH 7], 2% SDS, 0.5 M imidazole).

Detection of OGM-labeled protein.

Protein was subjected to SDS-PAGE using a 10% polyacrylamide gel matrix. For SDS-PAGE, 15 μl of an elution sample was used, and after electrophoresis, the gel was rinsed briefly with a destaining solution (10% glacial acetic acid, 15% methanol). Fluorescence profiles were recorded with a LAS-3000 imaging system. After the fluorescence profiles were recorded, the protein content of each lane was evaluated by staining the gel with Coomassie brilliant blue.

Modeling the structure of the large hydrophilic loop region.

A homology model of the large hydrophilic loop of AspT was constructed using SKE-CHIMERA (40) and FAMS Ligand&Complex (39). We used the amino acid sequence of the AspT from T. halophilus (accession no. Q8L3K8) as a target, and we constructed and examined a number of alignments derived from an eight-sequence search method using SKE-CHIMERA. Considering the sequence identities, the E-values, and the degree of agreement between the secondary structure elements of the target predicted by PSI-PRED (26) and the structure of the reference protein predicted by STRIDE (16), we selected the crystal structures of the putative potassium-channel-related proteins from Methanothermobacter thermautotrophicus (PDB code 1LNQ) and from Pyrococcus horikoshii (PDB code 1VCT) as templates. Alignments of the target and the template were generated by using PSI-BLAST (6) and PSF-BLAST, respectively. PSF-BLAST is a variation of PSI-BLAST in which a position-specific scoring matrix construction process is revised using reference protein family sequences from the nonredundant sequence database (ftp://ftp.ncbi.nih.gov/BLAST/db/FASTA/nr.gz) at NCBI. The structure of the target was predicted by FAMS Ligand&Complex based on these alignments.

Construction of the histidine-tagged large hydrophilic loop expression vector.

To clone the DNA fragment encoding the AspT large hydrophilic loop region (arginine 178 to asparagine 361) into the pCold II vector (Takara Bio, Tokyo, Japan), the plasmid pCold II vector was digested with restriction endonucleases NdeI and XbaI, yielding a 4.33-kbp DNA fragment. The DNA fragment of the loop region was amplified from pBlueAspT by using a pair of mutagenic oligonucleotide primers. The amplified DNA fragment obtained with the primer set was designed to introduce an NdeI site at the 5′ end and an XbaI site at the 3′ end to promote subsequent ligation. The PCR product (570 bp) was digested with the reaction endonucleases NdeI and XbaI. Each fragment was isolated from an agarose gel. Ligation reactions were performed by using the NdeI-XbaI-digested PCR product and the pCold II fragment at various vector/insert molar ratios. After the ligated product was transformed into E. coli DH5α, recombinant clones were selected on LB medium plates containing carbenicillin. Plasmid DNA was prepared from carbenicillin-resistant colonies, and the DNA sequence of the insert in the new plasmid was confirmed by DNA sequencing, as described previously (2). The plasmid was then transformed into E. coli strain BL21(DE3) for expression studies.

Expression and purification of the histidine-tagged AspT large hydrophilic loop region.

An overnight culture of E. coli BL21(DE3) carrying pCold II vectors harboring the gene for the AspT large hydrophilic loop region (arginine 178 to asparagine 361) was inoculated into 200 ml of LB medium, and the culture was allowed to grow at 37°C with shaking until an optical density at 600 nm of 0.6 was reached. The culture was cooled immediately under static conditions at 15°C for approximately 30 min. The cells were then induced with 0.4 mM IPTG and grown at 15°C with shaking for approximately 24 h. We used SDS-PAGE to confirm production of recombinant protein of the AspT large hydrophilic loop region.

The recombinant AspT large hydrophilic loop region was purified at 4°C. Cells were collected by centrifugation at 6,000 × g for 20 min and resuspended in 10 ml of lysis buffer (50 mM Tris-HCl buffer [pH 7.0], 200 mM NaCl, 10 mM PMSF). The sample was then lysed by sonication on ice by using 15-s pulses separated by 30-s intervals until the solution became transparent. The resultant lysate was centrifuged at 10,000 × g at 4°C for 20 min, and the supernatant was collected and placed on ice. Ni-NTA Superflow resin (QIAGEN, Valencia, CA) was equilibrated with 10 ml of wash buffer (50 mM Tris-HCl buffer [pH 7.0], 200 mM NaCl, 10 mM PMSF, 30 mM imidazole), and the supernatant containing the AspT large hydrophilic loop region was applied to the equilibrated resin. After five washes with 20 ml of wash buffer, six-histidine-tagged protein was eluted with 1 ml of elution buffer containing 500 mM imidazole. The N-terminal amino acid sequence of the purified protein was analyzed with an Applied Biosystems 491HT protein sequencer (Applied Biosystems, Foster City, CA).

Circular dichroism measurement of the AspT large hydrophilic loop region.

Circular dichroism spectra of the AspT large hydrophilic loop region were measured by the method of Ichishima et al. (23). The circular dichroism spectra were recorded over the 190- to 250-nm wavelength region with a J-720 circular dichroism spectrometer (JASCO International, Tokyo, Japan) in a 1-mm quartz cuvette (GL Science, Tokyo, Japan). The temperature was kept at 25°C, and the sample compartment was continuously flushed with N₂ gas. Each spectrum was the average of 10 scans, using a bandwidth of 1 nm, a step width of 1 nm, and 4-s averaging per point. The spectra were corrected by using a reference solution without the protein. A protein concentration of 5.0 μM in distilled water was used. The results were expressed as the mean residue ellipticities in degrees cm² dmol⁻¹. The existence of the α-helix and β-structure of the AspT large hydrophilic loop region was confirmed by using the SSE-338 program (44).

RESULTS

Creation of a fully functional histidine-tagged variant of AspT.

To construct six-histidine-tagged AspT, we fused the six-histidine tag to the N terminus or to the C terminus, as has been done for many other membrane proteins whose topologies have been experimentally determined. Immunoblot and functional analyses of the six-histidine-tagged AspT variants indicated that the production levels and activities of the variants were significantly reduced relative to those of the wild type and that the variants were therefore not suitable for cysteine-scanning experiments (Table 1).

Therefore, we constructed four six-histidine-tagged insertion variants with the tag inserted in putative cytoplasmic and periplasmic loop regions predicted by the results of topology analysis based on the fusion methods described previously. Immunoblot and functional analyses of the insertion variants showed that only one variant, in which the six-histidine tag was inserted into putative cytoplasmic loop 5, maintained production and activity levels that were nearly the same as those of the wild-type AspT (Table 1). In addition, the results of the functional analysis of this insertion variant using proteoliposomes suggested that this variant exhibits aspartate:aspartate self-exchange and electrogenic aspartate:alanine heterologous exchange activities (see Fig. S1 in the supplemental material).

We designed a purification scheme for this six-histidine-tagged variant by using Ni-NTA Superflow resin. The six-histidine-tagged variant was successfully purified by this method, which could also be used to purify single-cysteine variants of six-histidine-tagged AspT for cysteine-scanning mutagenesis as described below.

Membrane topology analyzed by site-directed fluorescence labeling.

Prior to cysteine-scanning mutagenesis of AspT, we constructed a cysteineless variant of AspT from the six-histidine-tagged AspT by substituting alanine for all three endogenous cysteine residues (cysteine 19, cysteine 110, and cysteine 476). Because the cysteineless AspT produced in E. coli strain XL3 (3) exhibited activities that were almost the same as those of the six-histidine-tagged AspT in proteoliposomes (Table 2), we used the cysteineless variant as starting material for the generation of a variant series with single cysteine substitutions. Table 3 describes the single-cysteine variants that served as targets in the study of AspT topology. To establish the topology of AspT, we examined 23 variants in which cysteines were placed either at the ends of putative transmembrane segments or in the loops that connect the transmembrane segments. Because these variants retained normal or near-normal (≥20% [45]) specific activities after reconstitution in proteoliposomes (Table 3), we assumed that the targeted cysteines did not significantly perturb the structure of AspT. Thus, we believe that the topology of each of these variants reflects the location of the original residues in the cysteineless variant and in wild-type AspT.

TABLE 2.

Production level and specific activity of cysteineless AspT variant relative to those of the AspT-His₆ variant

Variant	Production (%)a	Sp act (%)b
AspT-His6	100.0	100.0
Cysteineless AspT	55.7 ± 0.8	74.9 ± 6.7

^aProduction is expressed relative to that of the AspT-His₆ parent as the mean ± standard deviation.

^bThe initial rate of l-[2,3-³H]aspartate transport, normalized for the level of AspT production, is expressed relative to the rate for the AspT-His₆ parent, whose activity was 8.37 nmol min⁻¹ mg of protein⁻¹. The data are the mean ± standard deviation for three independent trials. For each variant, the same amount of protein was used for reconstitution in l-aspartate-loaded proteoliposomes.

TABLE 3.

Production levels and specific activities of single-cysteine variants relative to those of cysteineless AspT

Variant	Production (%)a	Sp act (%)b	Location
Cysteineless AspT	100.0	100.0
A3C	98.7 ± 2.1	101.6 ± 1.5	Periplasm
S33C	108.9 ± 3.2	90.8 ± 9.2	Cytoplasm
T59C	81.2 ± 8.3	60.2 ± 1.9	Periplasm
A90C	100.7 ± 0.7	70.5 ± 5.1	Cytoplasm
L151C	98.6 ± 0.4	106.6 ± 1.3	Periplasm
S156C	63.2 ± 1.3	261.2 ± 17.8	Periplasm
P160C	85.6 ± 2.2	66.6 ± 1.1	Periplasm
M182C	86.0 ± 1.4	96.5 ± 2.3	Cytoplasm
M186C	120.6 ± 4.3	33.3 ± 0.7	Cytoplasm
L190C	95.0 ± 1.8	34.4 ± 0.4	Cytoplasm
A194C	91.1 ± 1.5	51.8 ± 4.9	Cytoplasm
S291C	105.4 ± 1.7	78.8 ± 2.2	Cytoplasm
S351C	98.2 ± 0.5	86.0 ± 3.3	Cytoplasm
S386C	94.8 ± 1.3	127.5 ± 6.7	Periplasm
Y408C	48.4 ± 2.5	140.0 ± 4.3	Cytoplasm
H412C	129.5 ± 2.3	49.3 ± 0.9	Cytoplasm
S421C	84.7 ± 1.5	121.4 ± 6.8	Cytoplasm
S448C	99.1 ± 1.3	69.4 ± 4.0	Periplasm
T472C	109.7 ± 0.6	42.5 ± 2.7	Cytoplasm
M480C	88.2 ± 0.5	59.5 ± 2.9	Cytoplasm
S486C	108.1 ± 1.7	111.6 ± 2.9	Cytoplasm
S510C	94.4 ± 1.7	49.8 ± 2.9	Cytoplasm
S543C	84.3 ± 2.8	110.4 ± 11.0	Periplasm

^aProduction is expressed relative to that of the cysteineless parent. The values are means ± standard deviations.

^bThe initial rates of l-[2,3-³H]aspartate transport, normalized for the levels of AspT production, are expressed relative to the rate for the cysteineless parent, whose activity was 4.24 nmol min⁻¹ mg of protein⁻¹. The data are the means ± standard deviations for three independent trials. For each single-cysteine variant, the same amount of protein was used for reconstitution in l-aspartate-loaded proteoliposomes.

To probe AspT topology, we employed a strategy successfully used to establish the membrane topology of OxlT (18, 45). Fusion analysis of AspT suggested the presence of eight TMs (TM1 to TM8) and that arginine 76 may lie within TM3 (34). Because the peak hydropathy in the area surrounding TM3 is marginally lower than that in other TMs, our initial experiments were designed to establish whether this region actually traverses the membrane. We established this by examining the accessibility to an impermeant, fluorescent, thiol-specific probe (OGM) of single-cysteine targets predicted to be either at the extracellular surface or at the intracellular surface. 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 experiment to validate these interpretations, the external target cysteine residue in each AspT variant had to be protected from OGM by pretreatment of intact cells with the impermeant, nonfluorescent, thiol-specific probe MTSET prior to OGM treatment of cell membranes from disrupted cells. The protection procedure had no effect on the labeling of positions assigned to the cytoplasmic surface.

In the initial experiments, we chose two single-cysteine variants, T59C and A90C; the high levels of production and activity (Table 3) of these variants supported the idea that for the most part they retain the structure of wild-type AspT. Fusion methods predicted that residues threonine 59 (T59C) and alanine 90 (A90C) of AspT are located on opposite sides of the transmembrane segment previously assigned as TM3 (34). After the intact cells independently producing T59C and A90C were directly exposed to OGM, both the T59C and A90C proteins could be readily purified, but only the T59C protein contained the fluorescent label (Fig. 1A, lanes 1). In contrast, when extracellular cysteines were first protected with MTSET, subsequent exposure of isolated membranes to OGM yielded fluorescent labeling of the A90C protein but not the T59C protein (Fig. 1A, lanes 2). Moreover, the T59C protein that was not pretreated with MTSET was efficiently labeled with OGM, which suggests that MTSET can selectively label and consequently protect external cysteines (Fig. 1A, lanes 3).

FIG. 1.

Orientation of TM3. (A) After SDS-PAGE of purified proteins, a fluorescence profile was recorded (right panels) before the same gel was stained with Coomassie brilliant blue (CBB) to reveal total protein (left panels). Lanes 1, OGM added to intact cells; lanes 2, OGM added to membrane ghosts prepared after pretreatment of cells with MTSET; lanes 3, OGM added to membrane ghosts prepared without pretreatment with MTSET. (B) Deduced orientations of TM3. The black circle represents the cytoplasmic target, A90C; the gray circle represents the periplasmic target, T59C.

These observations led to the following two conclusions. First, positions 59 and 90 are clearly located on opposite membrane surfaces (Fig. 1B); the former faces the periplasm, and the latter is exposed to the cytoplasm. This result confirms the putative topology derived from fusion analysis of this region (34) and verifies the existence of TM3. Second, the OGM-labeling profiles of T59C and A90C confirm that the cysteine-scanning method employed in the experiments can be used to determine AspT topology.

Topology of AspT.

The topological analysis of AspT single-cysteine variants is summarized in Fig. 2 and 3, which show results for all the positions examined (Table 3) together with a representative OGM-labeling pattern for each cysteine targeted in an intracellular or extracellular loop and in the N or C terminus. In particular, positions assigned to the extracellular surface were readily labeled by the addition of OGM to the intact cells (Fig. 2, bottom panels, left lanes). This labeling pattern clearly contrasts with that of the putative intracellular cohorts, all of which were accessible by OGM only after cell lysis (Fig. 2, bottom panels, right lanes). The OGM-labeling efficiency of the T472C variant was markedly low (Fig. 2, bottom panels, T472C right and left lanes). However, OGM labeling of T472C was detected only after MTSET blocking, when the detection sensitivity of the LAS-3000 imaging system was increased (data not shown). The low labeling efficiency of T472C might be attributable to the possibility that T472C is located close to the cytoplasmic side of the membrane. We concluded that AspT has a minimum of 10 TMs, N and C termini that face the periplasm, and a large hydrophilic loop between the 10 TMs (that is, a five TM-large loop-five TM structure).

FIG. 2.

Labeling of single-cysteine variants of AspT with OGM and MTSET. After SDS-PAGE of purified proteins, a fluorescence profile was recorded (bottom panels) before the same gel was stained with Coomassie brilliant blue (CBB) to reveal total protein (top panels). Lanes −, OGM added to intact cells; lanes +, OGM added to membrane ghosts prepared after pretreatment of cells with MTSET. The cysteineless AspT variant was used as a negative control in the labeling reaction. For further details, see Materials and Methods.

FIG. 3.

Membrane topology of AspT and locations of introduced cysteines. The predicted topology of AspT is shown together with topological assignments determined from site-directed fluorescence labeling of 23 single-cysteine derivatives (Table 3). The residues replaced by a cysteine residue are indicated by letters in circles, and the numbers indicate the positions. Single-cysteine variants labeled by addition of OGM to the intact cells are indicated by gray circles. Single-cysteine variants labeled by addition of OGM to membrane ghosts prepared after pretreatment of intact cells with MTSET are indicated by black circles. Positively charged residues and negatively charged residues are indicated by open circles and open boxes, respectively. The three cysteine residues (cysteine 19, cysteine 110, and cysteine 476) that were replaced by alanine are indicated by letters in open circles. The large hydrophilic loop region is surrounded by a dashed circle. The alpha-beta complex structures (domain 1, valine 208 to phenylalanine 277; domain 2, threonine 283 to glutamine 352) predicted by means of homology modeling of the three-dimensional structure of the loop region using FAMS are indicated by dark gray and light gray shading, respectively, in the large hydrophilic loop region.

Modeling of the large hydrophilic loop of AspT.

We next modeled the large hydrophilic loop region of AspT. The crystal structures of the putative potassium-channel-related proteins from M. thermautotrophicus (PDB code 1LNQ[25]) and P. horikoshii (PDB code 1VCT) were selected as templates. Figure S2 in the supplemental material shows alignments of AspT and the reference proteins, as well as the putative secondary structure of AspT predicted by PSIPRED (26) and the secondary structure of the reference proteins assigned by STRIDE (16). The levels of sequence identity between AspT and the reference proteins are very low (AspT domain 1 and MthK, 15.7%; AspT domain 2 and PHO0236, 10.2%), but the E-values of the alignments (AspT domain 1 and MthK, 4e⁻²³; AspT domain 2 and PHO0236, 1e⁻³²) are low enough to have confidence in the modeling. In addition, the agreements between the secondary structure predicted for AspT and the secondary structures of the reference proteins are good. In the resulting model, we found two alpha-beta complex structures (domain 1, valine 208 to phenylalanine 277; domain 2, threonine 283 to glutamine 352) in the large hydrophilic loop region of AspT (Fig. 4A and B).

FIG. 4.

Modeling of the large hydrophilic loop region of AspT. (A) Superposition of the model of AspT loop domain 1 (valine 208 to phenylalanine 277) (green) and the structure of the MthK TrkA_C domain (methionine 264 to isoleucine 3340) (yellow). The root mean square deviation of these structures is 0.38Å. (B) Superposition of the model of AspT loop domain 2 (threonine 283 to glutamine 3520) (green) and the structure of the protein PH0236 TrkA_C domain (isoleucine 122 to isoleucine192) (yellow). The root mean square deviation of these structures is 0.54Å. (C) Superposition of the model of AspT loop domain 1 (green) and the model of AspT loop domain 2 (cyan). The structures of domain 1 and domain 2 were superposed by means of the combinatorial extension algorithm (38). The root mean square deviation of these structures is 1.29Å.

Circular dichroism measurements of the AspT large hydrophilic loop region.

To support the prediction that the large hydrophilic cytoplasmic loop contained alpha-beta complex structures, we measured the circular dichroism spectrum of a histidine-tagged recombinant protein of the loop region (arginine 178 to asparagine 361) after purification with an Ni-NTA column (Fig. 5A and B). The circular dichroism spectrum indicated that the loop region had a secondary structure that included α-helices and β-structures, as predicted by the FAMS.

FIG. 5.

SDS-PAGE profile and circular dichroism spectrum of the AspT large hydrophilic loop region. (A) SDS-PAGE profile obtained after purification of the AspT large hydrophilic loop region (arginine 178 to asparagine 361), as described in Materials and Methods. The right lane contained 2 μg of the purified loop region. The left lane contained standards having the indicated masses. (B) Circular dichroism spectrum recorded as described in Materials and Methods. The spectrum represents the average of 10 scans corrected by using a reference solution without the protein.

DISCUSSION

To understand the mechanism of substrate transport by AspT, functional and structural information about AspT is required. Various algorithms can be used to predict the likelihood that a polypeptide forms a membrane-spanning α-helix. The use of various programs to predict the AspT membrane topology from hydropathy profiles generated different models, and the segments having hydrophobicity in the middle range were ambiguously predicted.

To verify the in silico topology prediction for AspT, we carried out biochemical topology analyses of AspT (34). For bacterial, yeast, and mammalian proteins, this phase of analysis is most often based on the use of a reporter system in which a series of C-terminal deletions of a protein of interest are fused with a reporter protein at the C terminus of each deletion mutant. The localization of the reporter is easily determined by phenotypic tests (13, 14, 33). Although this experimental approach can often generate a satisfactory model, the general mechanisms of topology formation for membrane proteins are not well understood. For this reason, it is sometimes questionable whether fusion methods that use reporter proteins having high molecular masses correctly reflect the native protein topologies.

As an alternative to fusion methods, extensive studies have been performed over the past decade to elucidate the detailed membrane topology of primary and secondary transporters by means of the SCAM (12, 28, 47). SCAM was first introduced by Akabas et al. to study ion channel-lining residues of the acetylcholine receptor channel (4) and then was expanded to other multiple transmembrane proteins. By using single-cysteine mutagenesis in combination with a neutral or charged sulfhydryl reagent, SCAM can precisely locate the position of the introduced cysteine residue. This technique has proved to be applicable to high-resolution biochemical topology analyses of membrane proteins, because the introduction of cysteine residues is minimally invasive and has little effect on the structure and function of the target membrane protein.

We used affinity purification of histidine-tagged AspT along with cysteine-scanning mutagenesis and impermeant fluorescent and nonfluorescent thiol-specific probes. Together, these tools allowed us to document residue accessibility and therefore topology by examining the fluorescence of the purified protein. Topology analyses based on selective labeling of external and internal cysteines depend on the use of an impermeant fluorescent probe, and for this purpose we selected OGM, a maleimide-linked reporter having both acidic hydroxyl (pK_a, 4.7) and carboxyl (pK_a, <5) functional groups. The modest molecular size (460 Da) and the charged nature of this probe suggested that it would penetrate the outer membrane of E. coli but not the plasma membrane, and this suggestion was supported by the following two observations. First, when OGM was used to treat intact cells, a cysteine (T59C) expected to be exposed externally was readily labeled, whereas a cysteine (A90C) expected to be exposed internally was not labeled (Fig. 1A, lanes 1). Second, we could also show that the external cysteine (T59C) was protected by treatment of the cells with MTSET and that there was no effect on subsequent OGM reactivity of the internal target (A90C) (Fig. 1A, lanes 2). The reactivities of variants T59C and A90C to the impermeant probes clearly indicated that the two positions are located on opposite membrane surfaces (Fig. 1B); A90C faces the cytoplasm, and T59C faces the periplasm, as predicted by the initial analysis via fusion methods (34).

Using this strategy, we analyzed 23 single-cysteine AspT variants in which each target cysteine residue is located at either the internal or the external surface (Table 1 and Fig. 2 and 3). The present topological assignments suggested that the N and C termini of AspT face the periplasm and that AspT has a large cytoplasmic loop (about 180 amino acids) between TM5 and TM6 (Fig. 3). These results are consistent with the topology predicted by fusion methods (34). However, this study revealed that the C-terminal region—the topology of which was not unambiguously determined by fusion methods, as described above—forms five transmembrane α-helices. This result suggests that fusion of high-molecular-mass reporter proteins with target proteins disorders the native structure of AspT and interferes with topology prediction by the fusion methods.

In the present study, we demonstrated that AspT has a unique topology: N and C termini that face the periplasm and a large hydrophilic loop in the center of 10 TMs (that is, a five TM-large loop-five TM structure). To our knowledge, transporters in which the N and C termini face the periplasm are rare. Although we constructed six-histidine-tagged fusions at either the N terminus or the C terminus, we did not succeed in producing the two six-histidine-tagged fusions (Table 1). The positive charges in the six-histidine tag may have inhibited traversal of the two termini across the membrane and consequently may have been responsible for the failure to produce the two terminal histidine-tagged fusions. Therefore, we constructed variants with six-histidine tags inserted into putative loop regions, and we succeeded in obtaining one insertion variant with production and activity levels nearly the same as those of wild-type AspT (Table 1). By means of functional analysis using proteoliposomes, we verified that this variant had aspartate:aspartate self-exchange and electrogenic aspartate:alanine exchange activities (see Fig. S1 in the supplemental material). Moreover, we could purify this variant using Ni-NTA Superflow resin. We propose that this method will be useful for the study of membrane proteins for which histidine-tagged fusions at the N or C terminus are not active.

The new topological model of AspT based on our data (Fig. 3) represents a new class of secondary transporter in terms of topology. Interestingly, like OxlT, AspT catalyzes precursor:product exchange reactions, but the structure of AspT is entirely different from that of OxlT, which is a member of the major facilitator superfamily and has 12 TMs with cytoplasmic N and C termini (45). In addition, our results suggest the presence of a positively charged residue (arginine 76) within TM3. The presence of a charged residue in the hydrophobic sector implies that TM3 forms part of the substrate translocation pathway. To determine the substrate transport mechanism of AspT, ongoing studies are focusing on using cysteine-scanning mutagenesis to locate TMs that form putative substrate translocation pathways.

We found 90 aspT orthologues, such as yidE and ybjL of E. coli, in 60 bacterial or archaebacterial genomes (20). All the putative AspT orthologues possess the typical conserved AAE domains (PF06826 Asp-Al_Ex [http://www.sanger.ac.uk/cgi-bin/Pfam/getacc?PF06826] and IPR006512 YidE/YbjL duplication [http://www.ebi.ac.uk/interpro/IEntry?ac=IPR006512]) listed in the Pfam database (http://www.sanger.ac.uk/Software/Pfam/) and the InterPro database (http://www.ebi.ac.uk/interpro/index.html). Analyses of hydropathy and sequence alignment of the AspT orthologues suggested that most of them have 9 to 11 TMs and a large hydrophilic loop centered between these TMs. It was computationally supposed that these large hydrophilic loops of the AspT orthologues possess one or two TrkA_C domains (PF02080 TrkA_C domain [http://www.sanger.ac.uk//cgi-bin/Pfam/getacc?PF02080], IPR006037 TrkA_C [http://www.ebi.ac.uk/interpro/IEntry?ac=IPR006037]) (11) found in the regulatory subunit TrkA of the bacterial potassium uptake system (9).

We carried out homology modeling of the three-dimensional structure of the large hydrophilic loop region of AspT by means of FAMS (24, 35). We found that the loop possessed two alpha-beta complex structures (domain 1, valine 208 to phenylalanine 277 [Fig. 4A]; and domain 2, threonine 283 to glutamine 352 [Fig. 4B]). Furthermore, we produced and purified recombinant protein of the loop region (arginine 178 to asparagine 361) (Fig. 5A). The circular dichroism spectrum of the recombinant protein was analyzed, and the results suggested that the recombinant protein had secondary structures such as α-helices and β-structures (Fig. 5B). Although we should note that the structure of a partial peptide may not always be the same as that in a whole protein, these results appear to support our homology modeling of the large hydrophilic loop.

Interestingly, the two alpha-beta complex structures modeled by FAMS had similar three-dimensional structures (Fig. 4C). The root mean square deviation of these structures was 1.29Å, and the Z-score was 5.2, although the sequence identity of these structures was very low (10.3%) (see Fig. S3 in the supplemental material). Moreover, domain search analysis of the amino acid sequence homology by means of InterProScan Sequence Search (http://www.ebi.ac.uk/InterProScan/ [10]) suggested that AspT had only one TrkA_C domain, in the N-terminal half of the large hydrophilic loop (domain 1) (Fig. 4A), but the results of homology modeling suggested that AspT had a TrkA_C-like domain in the C-terminal half of the loop region (domain 2) (Fig. 4B). In brief, domain 2 had the specific three-dimensional structure of the TrkA_C domain.

TrkA_C domains have also been identified in members of the divalent anion:Na⁺ symporter family (TC no. 2.A.47) (11). However, TrkA_C domains are broadly distributed in membrane transport proteins of archaebacteria, bacteria, and eukaryotes, and the functions of these domains are not well defined (9, 11). Therefore, we believe studying the structure and the function of the TrkA_C domains is worthwhile and that AspT is a good example of these membrane transport proteins.