Archaeal Binding Protein-Dependent ABC Transporter: Molecular and Biochemical Analysis of the Trehalose/Maltose Transport System of the Hyperthermophilic Archaeon Thermococcus litoralis

Abstract

We report the cloning and sequencing of a gene cluster encoding a maltose/trehalose transport system of the hyperthermophilic archaeon Thermococcus litoralis that is homologous to the malEFG cluster encoding the Escherichia coli maltose transport system. The deduced amino acid sequence of the malE product, the trehalose/maltose-binding protein (TMBP), shows at its N terminus a signal sequence typical for bacterial secreted proteins containing a glyceride lipid modification at the N-terminal cysteine. The T. litoralis malE gene was expressed in E. coli under control of an inducible promoter with and without its natural signal sequence. In addition, in one construct the endogenous signal sequence was replaced by the E. coli MalE signal sequence. The secreted, soluble recombinant protein was analyzed for its binding activity towards trehalose and maltose. The protein bound both sugars at 85°C with a K_d of 0.16 μM. Antibodies raised against the recombinant soluble TMBP recognized the detergent-soluble TMBP isolated from T. litoralis membranes as well as the products from all other DNA constructs expressed in E. coli. Transmembrane segments 1 and 2 as well as the N-terminal portion of the large periplasmic loop of the E. coli MalF protein are missing in the T. litoralis MalF. MalG is homologous throughout the entire sequence, including the six transmembrane segments. The conserved EAA loop is present in both proteins. The strong homology found between the components of this archaeal transport system and the bacterial systems is evidence for the evolutionary conservation of the binding protein-dependent ABC transport systems in these two phylogenetic branches.

High-affinity binding protein-dependent ABC transporters were originally discovered in gram-negative bacteria. They consist of a high-affinity substrate-binding protein located in the periplasmic space as their major substrate recognition site, two hydrophobic membrane proteins forming the translocation pore, and two additional subunits peripherally associated with the membrane proteins at the inner face of the membrane. By ATP hydrolysis the last two subunits provide the energy for the accumulation of substrate against the concentration gradient (7). In the case of the Escherichia coli maltose/maltodextrin transport system, the periplasmic binding protein (maltose-binding protein or MalE) is encoded by malE, the membrane components MalF and MalG are encoded by the malF and malG genes, and the two ATP-hydrolyzing subunits of MalK are encoded by malK. These genes form a cluster on the E. coli chromosome in which malE, malF, and malG constitute an operon that is oriented divergently to malK (8). Recently, it has been recognized that binding protein-dependent ABC transporters are also present in gram-positive bacteria (20). In these cases, the soluble periplasmic binding proteins are anchored in the membrane by an N-terminal lipid modification consisting of a diglyceride connected to the N-terminal cysteine via a thioether bond (51). Binding protein-dependent ABC transporters have also been found in thermophilic bacteria (25, 41). Despite the large amount of information available on this type of transport system in bacteria, only one study of an archaeal ABC system, that of the hyperthermophile Thermococcus litoralis, has been reported so far (52). This transport system has several unusual properties: it shows an extremely high affinity (K_m of about 20 nM) at 85°C, the optimum growth temperature of this organism; it recognizes with equal affinity its very different substrates, maltose and trehalose; and it is not inhibited by maltodextrins. We undertook to further characterize this newly discovered transport system. Here we report on the purification of the native trehalose/maltose-binding protein (TMBP), the cloning and sequencing of the malEFG gene cluster, and the expression of the malE gene in E. coli as well as the purification and characterization of its encoded binding protein. The rationale for analyzing a binding protein-dependent transport system from a hyperthermophilic organism whose function is optimal at 85°C but is less than 5% at room temperature is the expectation that is conformation will be more rigid at room temperature and will become accessible to structural analysis under these conditions. In addition, evolutionary aspects and its unusual substrate specificity make it attractive for study.

MATERIALS AND METHODS

Cloning and sequencing.

A DNA clone from Pyrococcus furiosus was sequenced and shown to have high homology to the malG gene from Mycobacterium leprae by BLASTX analysis (9). PCR primers for the gene were designed from the DNA sequence and were used to amplify a 500-bp malG fragment from P. furiosus genomic DNA. A Lambda Zap mixed partial EcoRI genomic library of T. litoralis was screened by using this PCR fragment, which was labeled with [α-³²P]dATP by random priming. Several positive plaques were rescued into the pBluescript KS⁺ plasmid (Stratagene, La Jolla, Calif.) and were purified with cesium chloride gradients (2). The positive clones were sequenced by the dideoxy chain termination method with primer-walking methodology (2). Computer analysis of the DNA sequences was done with programs of the Wisconsin Package, version 9.0 (Genetics Computer Group, Madison, Wis.) (15).

Organism and growth conditions.

T. litoralis DSM5473 was obtained from the Deutsche Sammlung von Mikroorganismen und Zellkultur GmbH (Braunschweig, Germany). Cells were cultured as previously described (52) with yeast extract (inducing conditions) and peptone as carbon sources. At the end of the exponential phase and at an optical density at 600 nm of 0.4, cells were harvested by centrifugation (5,000 × g for 15 min at 27°C) and washed once with a solution of the same composition as the growth medium (pH 6.5) but without an added carbon source. The cells were then frozen and stored at −70°C until used.

Purification of TMBP from membranes of T. litoralis.

Solubilized membrane extracts from T. litoralis cells were prepared as reported previously (52). After cells were harvested, all manipulations were done under aerobic conditions. The cell paste was mixed with an equal volume of 50 mM Tris-HCl (pH 7.5) containing 1 mM MgCl₂ and homogenized, and a small amount of DNase I was added. Ten-milliliter aliquots of the cell suspension were passed through a French pressure cell at 16,000 lb/in² to break the cells. The suspension (10 ml) was clarified by centrifugation at 8,000 × g for 20 min at 4°C. The supernatant was then centrifuged at 100,000 × g for 1 h at 4°C. The pellet was washed twice with 10 ml of 50 mM Tris-HCl (pH 7.5) and resuspended in 10 ml of the same buffer containing 1% octyl-β-glucoside. This suspension was stirred for 1 h at 4°C. Insoluble material was removed by centrifugation (100,000 × g for 1 h). Typically, the detergent was added to a solution with a protein concentration of 2 to 4 mg/ml, and a clear solution containing 75% of the initial protein was obtained. One hundred microliters of 10% dodecyl-β-maltoside was added, and the solution was dialyzed twice against 3 liters of 50 mM Tris-HCl (pH 7.5) containing 0.01% dodecyl-β-maltoside. The solution was then applied to an HR5/5 MonoQ anion-exchange column (Pharmacia) that had been equilibrated with 50 mM Tris-HCl (pH 7.5) containing 0.01% dodecyl-β-maltoside. The column was washed with the same buffer until the eluate (15 ml) was protein free, and then 12.5 ml of 50 mM Tris-HCl (pH 7.5) containing 0.6% lauryldimethylamine oxide (LDAO) was added. Fractions were assayed at 85°C for binding to [¹⁴C]trehalose or [¹⁴C]maltose by using binding assays with saturated ammonium sulfate as described previously (52). Protein fractions were routinely analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) as described previously (29). The TMBP was eluted under these conditions in one peak without the application of a salt gradient. The fractions showing high binding activity were pooled and used for SDS-PAGE analysis as well as fluorescence spectroscopy.

Fluorescence spectroscopy.

All spectra were obtained at 55°C with an SPEX Fluorolog 2002 spectrofluorometer equipped with a thermostated cuvette. Spectra were recorded at an excitation wavelength of 280 nm and an emission wavelength of 328 nm. Emission scans were done from 280 to 400 nm. Measurements were done with 600 μl of 50 mM Tris-HCl (pH 7.5) containing 0.01% dodecyl-β-maltoside. Three microliters of the protein solution (0.2 mg/ml) was added to the prewarmed buffer, and the solution was mixed by inversion. The substrate was added in 3-μl samples of concentrated solutions. The fluorescence intensity was measured after 10 min of incubation to reach the desired temperature. The temperature was constant within ±0.2°C.

Construction of plasmids.

Chromosomal DNA from T. litoralis was isolated as described previously (2). To replace the N-terminal part of TMBP with the signal sequence of E. coli MalE, PCR was performed with the isolated T. litoralis chromosomal DNA and primers introducing a DraI site at the fusion point (Fig. 1A) and a HindIII site after the coding sequence of malE. The E. coli part of the recombinant sequence (corresponding to the N terminus of the recombinant protein) was amplified by PCR with plasmid pmalP2 (New England Biolabs) as the template and primers introducing a StuI site at the fusion point (Fig. 1A) and an MluI site upstream of the promoter sequence. After digestion of the PCR fragments with the corresponding enzymes, they were ligated in one step into pmalP2 opened with MluI and HindIII, yielding plasmid pRHo1000. To clone the entire T. litoralis malE gene (Fig. 1C), the gene was amplified from chromosomal DNA by PCR with primers introducing a BspHI site at the start codon and an SphI site downstream of the stop codon. After digestion with both restriction enzymes, the fragment was ligated into pJLA502 (43) opened with NcoI and SphI, yielding plasmid pRHo1001. The gene fusion with the signal sequence truncated (Fig. 1B) was constructed by PCR with pRHo1000 as the template and the following two primers: the 5′ primer changed AAA (encoding K [Lys]) to ATG (encoding M [Met]) and introduced a BspHI site; the 3′ primer introduced an SphI site after the stop codon of malE. The PCR fragment was digested with the corresponding enzymes and ligated into pJLA502 previously digested with NcoI and SphI, yielding plasmid pRHo1002. After all cloning steps for the PCR products, the correctness of the sequence was confirmed by sequencing the cloned fragment.

FIG. 1.

Constructs for the expression of malE from T. litoralis in E. coli. (A) pRHo1000. The cleavable signal sequence of E. coli MalE replaced the lipid anchor sequence of TMBP. Arrowhead 2 indicates where E. coli MalE is cleaved during secretion; arrowhead 1 shows the N terminus (determined by amino acid sequencing) of the periplasmic hybrid protein expressed in E. coli. The shaded sequence is that of the hybrid protein. (B) pRHo1002. The sequence of the hybrid protein from panel A has been shortened; its N terminus now is after the expected E. coli cleavage site SASALA. The N-terminal lysine (K) has been changed to methionine (M). The protein was soluble and remained in the cytoplasm. The shaded sequence is that of the hybrid protein. (C) pRHo1001. The intact malE gene of T. litoralis, including its lipid anchor-encoding sequence, was cloned after the inducible E. coli promoter. The protein expressed in E. coli (shaded sequence) was lipid modified and tightly membrane bound. The putative site of lipidation and its mature N terminus (cysteine) is indicated by arrowhead 3. Asterisks indicate identity; dots indicate homologous exchanges. Ec, E. coli; Tl, T. litoralis.

Purification of the recombinant protein.

E. coli SF120 (3) was transformed with plasmid pRHo1000 (Fig. 1A) and cultivated at 28°C in 10 liters of NZA medium (10 g of NZ-amine A [Sheffield Products, Inc.], 5 g of yeast extract, and 5 g of NaCl per liter) containing 200 μg of ampicillin per ml. After an optical density at 600 nm of 0.7 was reached, the temperature was increased to 37°C and expression of the hybrid malE gene was induced by adding IPTG (isopropyl-β-d-thiogalactopyranoside) to a final concentration of 100 μM. After 3 h of incubation, the culture was harvested and cold osmotic shock was performed by a standard protocol (34). The lyophilized periplasmic proteins were resuspended in 4 ml of 30 mM Tris-HCl (pH 7.5) and dialyzed against the same buffer. Heat-labile E. coli proteins were denatured by heating the solution for 10 min to 80°C. After centrifugation for 10 min at 18,000 × g, aliquots of the clarified protein solution were applied to a MonoQ column and eluted with a linear gradient of 0 to 1 M NaCl in 30 mM Tris-HCl (pH 7.5). Fractions containing purified recombinant hybrid TMBP were pooled and stored at −20°C. Routinely, about 3 mg of periplasmic hybrid protein was obtained from 10 liters of culture.

For purification of the cytoplasmic form of the hybrid TMBP (Fig. 1A), the cells collected after cold osmotic shock were ruptured in a French pressure cell at 16,000 lb/in² and centrifuged for 30 min at 35,000 × g. Again, the heat-labile proteins were removed by heating the solution for 10 min to 80°C. After centrifugation (10 min at 18,000 × g), the protein appeared to be homogeneous on SDS-PAGE (see Fig. 7B). The recombinant protein showed a tendency to aggregate, which was most pronounced in samples kept at 4°C. Even though the protein appeared to be homogeneous on SDS-PAGE, the solution showed a large absorption at 260 nm. After treatment with DNase I and RNase followed by dialysis and DEAE chromatography, the protein was free of material absorbing at 260 nm. Simultaneously, it had lost the tendency to aggregate.

FIG. 7.

SDS-PAGE analysis of TMBP after expression in E. coli. (A) TMBP with the N-terminal cleavable signal sequence of E. coli (construct A in Fig. 1). Lanes: 1, uninduced E. coli cells harboring pRHo1000; 2, induced cells; 3, cytoplasmic extract of induced cells that were treated by cold osmotic shock; 4, same as lane 3 but the sample was heated to 80°C for 10 min; 5, periplasmic shock proteins of induced cells; 6, same as lane 5 but the sample was heated to 80°C for 10 min; 7, TMBP purified from the sample shown in lane 6; 8, TMBP isolated from T. litoralis membranes; St, molecular mass standards (from top to bottom: 94, 67, 43, and 30 kDa). (B) Cytoplasmic TMBP; its N terminus is the expected cleavage site after the E. coli signal sequence (construct B in Fig. 1). Lanes: 1, uninduced E. coli cells harboring pRHo1002; 2, induced E. coli cells; 3, cytoplasmic extract of induced cells; 4, same as lane 3 but the sample was heated to 80°C for 10 min; 5, purified periplasmic TMBP (construct A in Fig. 1); 6, TMBP isolated from T. litoralis; St, as for panel A. (C) Native precursor TMBP from T. litoralis. The protein was lipid modified and membrane bound (construct C in Fig. 1). Lanes: 1, membranes from E. coli harboring pRHo1001 solubilized in octyl-β-glucoside and induced for the expression of the native precursor TMBP; 2, same as lane 1 but the sample was heated for 10 min to 80°C; 3, purified TMBP from T. litoralis; 4, purified periplasmic TMBP endowed with the E. coli signal sequence; St, as for panel A. (D) Western blot of TMBP from T. litoralis membranes and from the different constructs expressed in E. coli. Lanes: 1, whole E. coli cells harboring pRHo1000 encoding the hybrid protein with the E. coli signal sequence; 2, whole cells harboring pRHo1002 encoding the hybrid protein without the E. coli signal sequence; 3, whole cells harboring pRHo1001 encoding the intact T. litoralis protein; 4, membrane preparations of T. litoralis uninduced for the transport system; 5, membrane preparations of T. litoralis induced for the transport system; St, protein standards; 6, same as lane 1 (whole cells harboring pRHo1000); 7, French pressure cell extract of cells harboring pRHo1000; 8, periplasmic proteins isolated from cells harboring pRHo1000.

Preparation of antibodies and Western blot analysis.

The recombinant hybrid TMBP (Fig. 1A) was separated from other proteins by SDS-PAGE (29) and eluted from the gel as described previously (23). A chicken was immunized five times with 50 μg of protein each. At 14 days after the last immunization, antibodies were prepared from 10 eggs (37). Western blot analysis was done as described previously (23, 50), using the primary antibody (17 mg/ml) in a dilution of 1:10,000.

Determination of the binding affinity (K_d).

To measure the K_d of binding, a method based on the retention of ligand by the binding proteins (1, 46) was used. A small dialysis tubing (Medicell International, Ltd., London, United Kingdom) (diameter, 1 cm) open on one end was tightly fit with its open end onto a bluntly cut plastic pipetting tip of a 1-ml automatic pipette. Four hundred microliters of pure binding protein solution (0.41 μM in 50 mM Tris-HCl, pH 7.0) was introduced into the dialysis tubing and immersed in a 2-liter Erlenmeyer flask filled with 50 mM Tris-HCl, pH 7.0. A final concentration of 135 nM [¹⁴C]maltose (630 mCi/mmol) or 303 nM [¹⁴C]trehalose (500 mCi/mmol) (26) was added in 20 μl. The buffer in the Erlenmeyer flask was kept at 85°C and gently stirred during the assay. Aliquots of 20 μl were removed from the bag at different time intervals, and the radioactivity in 6 ml of toluene-based scintillation fluid was counted. The same procedure was repeated for both substrates but in the absence of protein. The time to release half of the substrate from the dialysis bag is greater by a factor of 1 + (P/K_d) in the presence of molar concentrations of binding protein P (assuming one binding site) than in its absence. K_d is obtained in molar concentration.

Detection of binding activity in nondenaturing polyacrylamide gels.

SDS-free gels (12% polyacrylamide) were prepared as described previously (53) except that no pyridoxal phosphate was present in the gel solution and no activity stain was added to the gel. Prior to electrophoresis 1 to 2 μl of [¹⁴C]trehalose (500 mCi/mmol; about 30,000 cpm) (26) was added to the individual protein samples and incubated for 5 min at 85°C. After electrophoresis at room temperature, the gel pockets were carefully rinsed, and the gel was dried and subjected to autoradiography.

Nucleotide sequence accession number.

The sequences reported in this paper have been deposited in the GenBank database and assigned accession no. AF012836.

RESULTS

Purification of TMBP from membranes of T. litoralis.

Membranes from trehalose-induced T. litoralis cells (with yeast extract in the growth medium) were isolated by disruption in a French pressure cell and solubilized in 1% octyl-β-glucoside at a ratio of 0.2 to 0.4 mg of protein per mg of detergent. Initial attempts to purify the protein by affinity chromatography through an amylose column were unsuccessful, since the protein failed to bind to this material. Therefore, we used anion-exchange chromatography. In the first attempts, the membrane extract (100 mg of total protein) in 1% octyl glucoside was applied to a Source 30Q anion-exchange column after equilibrating the column with 50 mM Tris-HCl (pH 7.5) and 0.8% octyl-β-glucoside. Under these conditions, more than 75% of the binding protein did not adsorb to the column and was eluted even before the application of an NaCl gradient. Increasing the pH and reducing the ionic strength did not improve the adsorption properties. The protein remaining on the column in some experiments could be eluted by a salt gradient (0 to 500 mM) and appeared at 150 mM NaCl but was not free of contaminants. Adsorption of the protein on the column was greatly improved by replacing octyl-β-glucoside (after solubilization of the proteins from the membrane in this detergent) with dodecyl-β-maltoside, yet the protein could not be eluted from the column with a 0 to 500 mM NaCl gradient. We found, however, that a major portion of the protein eluted as a rather homogeneous preparation when the column was washed with 50 mM Tris-HCl (pH 7.5) containing 0.6% of the detergent LDAO, even in the absence of a salt gradient. The protein sample eluted in this way was analyzed by SDS-PAGE (Fig. 2, lane 3); its apparent molecular weight was 47,000. The protein was highly active as judged by a binding test involving precipitation with ice-cold saturated ammonium sulfate of a sample heated to 85°C in the presence of [¹⁴C]trehalose or [¹⁴C]maltose, followed by filtration and measurement of the radioactivity of the filter (52). This test cannot be used to determine the binding characteristics of the protein quantitatively even though it is very useful in qualitative assays during purification. Also, equilibrium dialysis at 85°C in the presence of detergents could not be done, since the detergents precipitated at this temperature and blocked the diffusion pores of the dialysis tubing. In addition, we found that dodecyl-β-maltoside, the detergent that allowed adsorption of the protein to the ion-exchange column, acted as a substrate, interfering with the binding assay.

FIG. 2.

SDS-PAGE of the native binding protein as isolated from the membrane of T. litoralis. Lanes: 1, membrane preparation from the uninduced strain (no yeast extract in the medium); 2, membrane preparation from the induced strain (with yeast extract in the medium); 3, purified protein containing the lipid anchor; 4, water-soluble hybrid protein (the construct carries the cleavable signal sequence of E. coli at its N terminus) isolated from the E. coli periplasm (as a reference); St, molecular mass standards (from top to bottom: 66, 45, 36, 29, 24, 20.1, and 14.2 kDa).

We used the change in the intrinsic fluorescence of the protein upon binding the substrate as a method to monitor a possible conformational change resulting from substrate binding. At 55°C the detergents did not precipitate, allowing the reading of the fluorescence spectra (excitation at 280 nm). Figure 3 shows that the emission decreased in the presence of trehalose and increased in the presence of maltose. An excess of one substrate competes with the effect of the other. This demonstrates that substrate binding elicits a conformational change of the protein and that the protein binds maltose and trehalose in different conformations. Using different trehalose concentrations, we found that the half-maximal fluorescence change occurred at concentrations of above 5 μM. From the K_m of transport in intact cells, which is around 20 nM, we had expected a much better K_d for binding. Possibly, the presence of dodecyl maltoside, which cannot be removed by dialysis, interferes by competitive binding. Therefore, binding assays had to be carried out with a soluble derivative of the binding protein with which dodecyl-β-maltoside would not interfere.

FIG. 3.

Fluorescence changes of TMBP in the presence of trehalose and maltose. The recordings were done consecutively. First, the middle trace was obtained after temperature equilibration without addition of substrate. The upper trace was then recorded after the addition of 1 μM (final concentration) maltose, followed by the lower trace after the addition of 100 μM trehalose. Similar tracings in the reverse order were obtained when the order of the additions was reversed, i.e., first 1 μM trehalose and then 100 μM maltose (data not shown). Conditions were as follows: temperature, 55°C; excitation, 280 nm; TMBP concentration, 16 μg/ml; buffer, 50 mM Tris-HCl (pH 7.5) containing 0.6% LDAO.

Cloning and sequencing of the maltose/trehalose operon of T. litoralis.

A fragment of the P. furiosus malG gene has been previously identified by random sequencing of a genomic DNA clone (9). A PCR product derived from this clone was used to screen a lambda Zap library of T. litoralis. Several clones were identified by plaque hybridization and rescued into pBluescript KS⁺. Sequencing of two of these clones, pJDR1.12 and pJDR6.22, gave the complete sequences of the T. litoralis malE, -F, and -G genes and their flanking regions. Part of this sequence, including the promoter region and the intergenic regions between malE and malF as well as between malF and malG, is shown in Fig. 4. The sequence of malE consists of 1,353 nucleotides; hence, the putative binding protein is composed of 450 residues, i.e., is slightly larger than the E. coli MalE protein, which contains 396 residues. The deduced TMBP amino acid sequence shows 28 and 26% identity with the maltose-binding protein sequences of E. coli (16) and Streptococcus pneumoniae (39), respectively, and 24% identity with the maltodextrin-binding protein of Thermoanaerobacterium thermosulfurigenes (41). The T. litoralis sequence is also 31% identical with the product of a gene from M. leprae which is located upstream of malF on the chromosome of this organism (GenBank accession no. U1756V). In addition, the TMBP amino acid sequence showed high homology with a family of periplasmic binding proteins, named cluster 1 binding proteins, for malto-oligosaccharides, multiple sugars, sn-glycerol-3-phosphate, and iron transport systems (Fig. 5) (48). This short segment of sequence was derived from the alignment of the eight proteins constituting cluster 1 binding proteins and is specific for this family of proteins (48). In gram-negative bacteria, the binding proteins are located in the periplasm between the inner and outer membranes and are water soluble. In gram-positive bacteria, which lack an outer membrane, binding proteins are soluble lipoproteins with an N-terminal glyceride-cysteine (51) which allows the anchorage of the binding proteins to the external surface of the cell membrane. Archaeal membrane lipids are based on isopranyl glycerol ethers (di- and tetraethers), and in hyperthermophiles the membrane is organized as a monolayer of lipids (27) with an S-layer on the outside (36). This is a situation similar to that of gram-positive bacteria with the absence of a restricting outer membrane. We found that the first 27 amino acids of the putative T. litoralis TMBP show characteristics typical of signal peptides of secretory precursor proteins: (i) a hydrophobic core adjacent to the N terminus and (ii) the sequence VASGCIG corresponding to the consensus of lipoprotein signal peptidase cleavage sites (LAAGCSS) (48).

FIG. 4.

Partial nucleotide sequences of the T. litoralis malE, -F, and -G genes and flanking regions. The deduced amino acid sequences are given in one-letter code below the nucleotide sequences, and the nucleotide and amino acid numbers are given on the right. 300/13 indicates the last amino acid of the malF gene product (no. 300) and last amino acid of MalG in this line (no. 13). In the nucleotide sequence, a putative boxA promoter element is boxed, the putative ribosome binding site is in boldface, and a putative terminator sequence is underlined. In the deduced amino acid sequence of TMBP (the malE gene product), the conserved recognition sequence for the cleavage sites of lipoprotein signal peptidases is underlined and the presumably formed N-terminal cysteine of the mature protein is in boldface and marked with an arrowhead. Most of the sequence within the structural genes is omitted, as indicated by points and deletion signs. The complete sequence is deposited in GenBank under accession no. AF012836.

FIG. 5.

Alignment of the signature sequences of cluster 1 binding proteins with TMBP of T. litoralis. The different amino acids possible for each position are indicated below the signature sequence. Numbers in parentheses indicate positions. The highly conserved lysine residue (K) is boxed. Residues of T. litoralis that match the signature sequence are in boldface, and residues conserved in other sequences are underlined. The signature sequence is from reference 48. Tl, T. litoralis; Ec, E. coli (16); St, Salmonella typhimurium (12); Sp, S. pneumoniae (39); Tt, T. thermosulfurigenes (41); MalE, MalX, and AmyE, maltose/maltodextrin-binding proteins; UgpB (35), glycerol-3-phosphate-binding protein.

Sequences of the inner membrane proteins.

In E. coli, MalF and MalG are hydrophobic inner membrane components mediating the energy-dependent translocation of substrate into the cytoplasm. Two T. litoralis genes for inner membrane components have been sequenced, i.e., malF and malG, consisting of 903 and 837 bp, respectively. The deduced amino acid sequences of the corresponding proteins are homologous to the MalF and MalG protein sequences of gram-positive and gram-negative bacteria (Table 1). Both the T. litoralis MalF and MalG proteins contain the sequence EAAX₂DGAX₈IXLP, which is homologous to the consensus sequence EAAX₂L/DGAX₈IXLP found in membrane components of ABC transporters (7, 42). In this conserved region, the amino acids are predominantly hydrophobic, with a consistent location at the C terminus. The exact function of the EAA loop is not yet clear; it might be involved in binding the ATP-hydrolyzing subunit of the transport system and thus be connected to the energy transduction process (33).

TABLE 1.

Similarities and identities of inner membrane proteins of binding protein-dependent transport systems and the inner membrane proteins of T. litoralis

T. litoralis protein	% Similarity/% identity with homolog(s) ina:
	M. leprae	S. pneumoniae	E. coli	T. thermosulfurigenes
MalF	62/38	56/31	57/35, 57/33	59/26
MalG	61/34	56/28	57/31, 56/29	56/27

^aThe homologs for MalF and MalG, respectively, are as follows: M. leprae, MalF and MalG; S. pneumoniae, MalC and MalD; E. coli, MalF, UgpA and MalG, UgpE; T. thermosulfurigenes, AmyD and AmyC. 

The topology of T. litoralis inner membrane protein MalG is similar to that of E. coli MalG (11), with six membrane-spanning segments (MSS) (Fig. 6B). However, the T. litoralis MalF protein is missing MSS1 and MSS2 of E. coli MalF (10, 18), including the N-terminal portion of the large periplasmic loop that is found in E. coli MalF between MSS3 and MSS4 (Fig. 6A). T. litoralis MalF is composed of 300 amino acids, compared to 514 amino acids for the E. coli protein. The MSS of the T. litoralis proteins shown in Fig. 6 are based on computer analysis. Preliminary experiments using malF-phoA fusions to the N-terminal portion of MalF are in agreement with the model shown in Fig. 6A (21), implying that both protein termini are located in the cytoplasm. MalF homologs from bacteria other than E. coli, for instance, from gram-positive bacteria, lack the same portion of the corresponding MalF protein, in particular the large periplasmic loop. It seems as if this loop is a peculiarity of the maltose system in E. coli and its very close neighbors (12). The alignment in Fig. 6 shows no particular enrichment of identity along the sequence except for the EAA motif, which is common to all membrane components of binding protein-dependent ABC systems throughout the bacterial world.

FIG. 6.

Putative MSS in inner membrane proteins. (A) T. litoralis and E. coli MalF proteins. In contrast to E. coli MalF, the T. litoralis protein has only six putative MSS (there are eight in E. coli MalF) and no large periplasmic loop between MSS3 and MSS4. MSS1 and MSS2 from E. coli MalF are missing in T. litoralis MalF. Since the numbering of the MSS corresponds to the E. coli sequence, the most N-terminal MSS of TMBP is termed MSS3. (B) T. litoralis and E. coli MalG proteins. The positions of the MSS appear to be conserved; the numbering is identical to that of the E. coli sequence. Tl, T. litoralis; Ec, E. coli. Asterisks indicate identical residues; dots indicate homologous residues.

Expression of T. litoralis malE in E. coli.

In the first construct, pRHo1000 (Fig. 1A), the putative signal sequence of the T. litoralis malE gene was replaced with the signal sequence of the E. coli malE gene contained in the IPTG-inducible vector pmalP2 (New England Biolabs). Upon induction with IPTG, the hybrid protein was expressed in E. coli. After application of the classical cold osmotic shock procedure of Neu and Heppel (34), less than 5% of this protein were recovered in the periplasmic fraction (Fig. 7A, lanes 5 and 6); the bulk remained in the cytoplasm. The protein was purified from the periplasmic fraction by heating to 80°C for 10 min, removing the precipitate, and subjecting the supernatant to ion-exchange chromatography through a MonoQ column (Fig. 2, lane 4, and 7A, lane 7). The protein thus obtained was used for measuring the binding constant (see below).

When the cells harboring pRHo1000 were disrupted with a French pressure cell (after cold osmotic shock treatment), most of TMBP was found in the cellular extract. Heating the extract to 80°C for 10 min precipitated all other proteins and yielded homogeneous TMBP (Fig. 7A, lane 4). Thus, the contaminating proteins of heat-treated periplasmic extracts (Fig. 7A, lane 6) consisted of a few heat-resistant periplasmic E. coli proteins that were absent in heat-treated cytoplasmic extracts of cells from which the periplasmic proteins had been removed by the osmotic shock procedure (Fig. 7A, lane 4). In SDS-polyacrylamide gels, the periplasmic and the cytoplasmic forms of the protein were identical in size (45 kDa) and carried the same N-terminal amino acid sequence, SASALAKIEEGKIV…, indicating that they had not been processed at the expected site during secretion. The expected cleavage site observed in E. coli MalE was after the third amino acid in the ASA part of the sequence shown above (16). The hybrid protein therefore lacks the first 20 amino acids of the mature native protein as isolated from T. litoralis, which had been replaced by the E. coli N-terminal sequence SASALAKIEE. The yield of the periplasmic form of the protein was not increased when the construct was expressed in the prlA421 mutant WP794 (38), which allows the secretion in E. coli of MalE forms lacking a signal sequence (14). This may indicate that the folding process of TMBP in E. coli at 37°C is aberrant and interferes with the secretion process.

In the second construct, pRHo1002 (Fig. 1B), the DNA segment encoding the E. coli signal sequence of the above-described hybrid protein was removed in order to express the protein as a soluble cytoplasmic protein in E. coli. The ATG initiation codon was positioned in such a way that the hybrid protein shown in Fig. 1A would begin directly after the predicted cleavage site ALA. This construct, cloned in the temperature-inducible vector pJLA502 (43), formed soluble, active TMBP when expressed in E. coli. Heating crude extracts to 80°C for 10 min resulted in considerable purification as judged by SDS-PAGE analysis (Fig. 7B, lane 4). Heat treatment of the cell extracts again was more effective (and yielded homogeneous protein) if, prior to disruption, the cells were subjected to cold osmotic shock for the removal of heat-resistant periplasmic proteins (data not shown).

The cytoplasmic forms of the protein containing the E. coli signal sequence (Fig. 1A and 7A, lane 4) or lacking it (Fig. 1B and 7B, lane 4) both consisted of multimeric aggregates of various compositions, containing up to five identical polypeptide chains. These aggregates could be seen clearly when analyzed in nondenaturing polyacrylamide gels. When ¹⁴C-labeled trehalose was added (and incubated at 80°C for 10 min) prior to electrophoresis, the label remained bound to the different aggregates (Fig. 8, lane 1), demonstrating their binding activity. We observed that the tendency to aggregate was associated with the presence of nucleic acid in the heat-treated sample. Treatment with DNase I and RNase followed by extensive dialysis and anion-exchange chromatography removed the material absorbing at 260 nm and the tendency to form multimeric forms.

FIG. 8.

PAGE under nondenaturing conditions and in the presence of radioactively labeled trehalose. Lanes: 1, cytoplasmic extract (not heat treated) of cells containing pRHo1000 (construct A in Fig. 1); 2, purified periplasmic TMBP (construct A in Fig. 1); 3, heterologously expressed wild-type TMBP (construct C in Fig. 1) solubilized in 1% octyl-β-glucoside; 4, purified TMBP solubilized in 1% octyl-β-glucoside. The gel was dried and autoradiographed. Staining was indicative of trehalose binding. Lane 3 contains two forms of the protein, one of which hardly entered the gel.

In the last construct (Fig. 1C), the entire open reading frame of the T. litoralis malE gene was cloned behind the temperature-inducible promoter of plasmid pJLA502 (43). This construct (pRHo1001) was also expressed in E. coli and produced a membrane-bound protein. After solubilization in 1% octyl-β-glucoside, it was binding active. When tested in SDS-PAGE, the protein’s apparent molecular weight was similar to that of the protein which had been isolated from the membranes of T. litoralis, but it formed a double band on these gels (Fig. 7C, lane 2), indicating heterogeneous lipid modification when expressed in the heterologous host. Separation in nondenaturing gels supported this conclusion. On these gels the two forms were widely separated, but both were binding active (Fig. 8, lane 3).

All recombinant TMBPs produced by the different DNA constructs as well as the protein contained in T. litoralis membranes cross-reacted by Western blot analysis (Fig. 7D) with antibodies that were raised in chicken against the cytoplasmic form of the recombinant TMBP containing the E. coli MalE signal sequence (Fig. 1A). The most important point is that these antibodies recognize a trehalose-inducible protein in the membranes of T. litoralis (Fig. 7D, lanes 4 and 5) and the purified recombinant TMBP proteins produced from all T. litoralis malE-harboring constructs. Thus, it is clear that the gene malE encodes the protein purified from the T. litoralis membranes.

The hybrid protein carrying the cleavable signal sequence from E. coli was clearly synthesized as a precursor (Fig. 7D, lanes 1 and 6) which was no longer seen after preparation of a cellular extract (Fig. 7D, lane 7). Thus, the cytoplasmically localized precursor in E. coli apparently gained access to the signal sequence peptidase and matured mainly after rupturing the cells. The TMBP antiserum was raised in chicken and contained antibodies against a few E. coli proteins but not against any protein from T. litoralis other than TMBP. The antibody did not recognize E. coli MalE. Likewise, antibodies against E. coli MalE did not recognize TMBP (data not shown).

Binding affinity of the periplasmic form of TMBP.

Of the several possible ways to measure the K_d of binding, only the use of the retention behavior (1, 46) is insensitive to the presence of bound unlabeled substrate. Since the assay could not be done with the native protein isolated from the membranes of T. litoralis (see above), we used the recombinant protein from a DNA construct in which the natural signal sequence was replaced by the E. coli MalE signal sequence. The protein isolated from the periplasmic fraction was monomeric in solution (Fig. 8, lane 2). At a concentration of 0.2 mg/ml, it was dialyzed against 2 liters of Tris-HCl (pH 7.0) at 85°C. Small amounts of ¹⁴C-labeled trehalose or ¹⁴C-labeled maltose were added into the dialysis tubing, and the rate of substrate exit was measured over time. By comparing the rates of substrate exit in the presence and absence of binding protein (Fig. 9) and with the knowledge of the amount of binding protein present (assuming one binding site), the K_d of binding can be calculated. It was determined to be 160 nM for trehalose and maltose.

FIG. 9.

Binding affinity of the periplasmic form of TMBP as measured by substrate retention. Exit of substrate (trehalose [A] or maltose [B]) from a dialysis bag containing purified periplasmic TMBP (0.41 μM) (closed symbols) or substrate only (open symbols) is shown. Samples of 20 μl were removed from the dialysis bag at different time intervals, and radioactivity was counted in a scintillation counter. The half-life of internal substrate was calculated after the rate of exit had become first order. The temperature was kept constant at 85°C. The half-life in the presence of TMBP (37 min in the case of trehalose and 41 min in the case of maltose) was larger by the factor 1 + (P/K_d) than that in the absence of the protein (10.1 and 11.5 min, respectively) (P is the concentration of TMBP in molar concentration of binding sites; one binding site per polypeptide chain was assumed).

DISCUSSION

The organization of the T. litoralis trehalose/maltose transport operon (Fig. 4) is very similar to that of E. coli and other bacterial binding protein-dependent ABC transport systems (7). These transport systems usually have an outer membrane diffusion pore, a soluble periplasmic substrate-binding protein of high affinity, two integral membrane proteins, and one or two energy-transducing and ATP-hydrolyzing polypeptides located at the cytoplasmic side of the membrane. In the archaeon T. litoralis, we found two genes, malF and malG, with strong homology to the genes of two membrane components of the maltose (13, 19) and glycerol-3-phosphate (35) transport systems of E. coli, both of which are members of the binding protein-dependent transport system family. Upstream of the malF and malG genes we found malE, with the characteristic signature sequences of binding proteins (48) (Fig. 5). These three genes constitute an operon with a putative archaeal promoter sequence (22, 24) and a putative prokaryotic ribosome-binding site upstream of malE (Fig. 4). Only 26 bp separates the malE and malF genes, and 1 bp, creating a frameshift, separates malF and malG (Fig. 4). An oligo(dT) sequence detected at the end of malE resembles transcription termination sequences described for archaea (49). The gene cluster containing malEFG did not contain the E. coli malK analog (4, 45). This has been observed with other binding protein-dependent ABC transport systems in gram-positive bacteria as well as in the thermophilic bacterium T. thermosulfurigenes (41). The possibility that T. litoralis has an ATP-hydrolyzing subunit for more than one transport system (44) should also be considered.

Binding protein-dependent transport systems must have appeared early in evolution. A malE-like gene has also been found in Thermotoga maritima, which, like T. litoralis for archaea, is one of the most deeply branched bacteria, with a maximum growth temperature of 90°C (31).

It was important to demonstrate that the malEFG gene cluster of T. litoralis was indeed encoding the transport system (including the membrane-bound binding protein) that we have discovered in this organism as a high-affinity (K_m = 20 nM) and trehalose-inducible system for the uptake of trehalose and maltose (52). This was corroborated by the observation that after expression in E. coli, the gene product of malE showed the same binding specificity as the binding protein isolated from trehalose-induced T. litoralis cells. Moreover, antibodies raised against the protein expressed in E. coli cross-reacted with the protein isolated from T. litoralis membranes, either in pure form or when present in detergent-solubilized membranes.

The finding that the T. litoralis system which is obviously related to the E. coli maltose system also recognizes trehalose (and is even induced by it) is rather surprising. In E. coli trehalose utilization depends on a different type of transport system belonging to the large group of sugar phosphotransferase systems (28). In addition, the enzymes metabolizing trehalose in E. coli are different from those which degrade maltose (40). Since trehalose can accumulate in T. litoralis under stress conditions (high salt) (30) but only when it is present in the medium, this sugar might possibly not be metabolized, and its acquisition as a substrate of the maltose system might be part of an osmoprotecting mechanism. Yet, the transport system described here is also peculiar with respect to maltose transport. In E. coli the maltose transport system is in fact a maltodextrin transport system geared for the utilization of maltose as well as larger maltodextrins. This can be deduced from the function of the outer membrane λ receptor as a diffusion pore for maltodextrins (17) and the binding specificity of maltose-binding protein, as well as the characteristics of the maltodextrin-degradative enzymes (8). In contrast, the T. litoralis uptake system recognizes only maltose and not longer maltodextrins. It will be interesting to analyze the maltose-degradative enzymes in T. litoralis and to determine whether these enzymes are related to trehalose metabolism as well.

The determination of the binding activity of the recombinant hybrid TMBP gave a K_d of binding of 0.16 μM. This is surprising in view of the fact that the apparent K_m of uptake in intact cells was 1 order of magnitude lower. One could argue that the protein without its lipid anchor exhibits a lower binding affinity than the native protein. This appears unlikely to us, since binding is brought about (at least in E. coli MalE) by the movement of the two soluble lobes forming the binding site between them (47). Thus, the lipid anchor should not influence binding affinity. Therefore, the transport in intact cells may exhibit a low K_m (20 nM) despite a higher K_d of its major substrate recognition site. Considering models for the mechanism of binding protein-dependent transport systems (5, 6), one may argue that in the case of the T. litoralis trehalose/maltose system, only the substrate-loaded binding protein is able to interact with the membrane components (model 1). This would be in contrast to the E. coli maltose system, where both the substrate-loaded and substrate-free binding protein can interact with the membrane components (model 2) (32). Mathematical analysis for model 1 predicts that the K_m of transport becomes infinitely low (increasing affinity) with increasing amounts of binding protein, whereas in model 2 the K_m of transport approaches the K_d of the binding protein at increasing binding protein concentrations. Clearly, the amount of binding protein in the cell envelope of T. litoralis after induction by trehalose is substantial, and the transport mechanism may in fact follow the predictions of model 1. The putative transcription termination site downstream of the malE gene (Fig. 4) indicates that the amount of TMBP is far larger than that of the membrane components.