Mapping, cloning, expression, and sequencing of the rhaT gene, which encodes a novel L-rhamnose-H+ transport protein in Salmonella typhimurium and Escherichia coli.

A L-rhamnose transport-negative strain of Escherichia coli was generated by Mu d(ApR,lac)I mutagenesis. This strain was used to isolate a clone of Salmonella typhimurium DNA that encoded L-rhamnose-H+ transport activity, the gene for which, rhaT, was sequenced. The rhaT gene was mapped on the E. coli chromosome between rhaR and sodA at 87.9 min, initially by Southern blot analysis and then by the isolation, expression, and sequencing of the rhaT gene. Both rhaT genes encoded a hydrophobic protein of 344 amino acids (91% identical) that contained 10 putative transmembrane regions. The RhaT protein represents a novel class of sugar transport protein.

T h e rhaT genes of S. typhimurium and E. coli L-rhamnose, 5 mM lactose, 80 pg/ml histidine, 100 pg/ml ampicillin to select for insertion of the Mu phage into the rhu operon. PI transductions were carried out as described by Miller (1972). The E. coli strain used for the preparation of P1 phage to produce an fdp+ strain was CSH25, and strain PB13 was used to generate a recAstrain.
@-Galactosidase Assays-Quantitative &galactosidase assays were performed as described by Miller (1972); the method of Davis et al. (1984) was followed when performing plate assays for @-galactosidase.
Screening a Cosmid Library-The cosmid library was made by ligating a Sau3A partial digest of S. typhimurium C5 DNA into a BamHI-digested plasmid pHC79 (Hohn and Collins, 1980) and was a kind gift from Dr. C. Hormaeche (Department of Pathology, University of Cambridge). The cosmid library was amplified using standard protocols (Maniatis et al., 1982).
DNA Sequencing-The S. typhimurium rhuT gene was sequenced by generating partial restriction digests of the 3-kb2 EcoRV/PuuII fragment from plasmid pJAR6 with either AluI, HaeIII, or Sau3A. In addition, a number of specific fragments were generated by digestion with combinations of HincII, PuuII, EcoRV, and BglII. All the DNA fragments were ligated into bacteriophage M13mp18 and sequenced by the dideoxynucleotide chain termination method (Sanger et al., 1980). In addition, two oligonucleotides were synthesized to the rhuT gene sequence and were used as primers to obtain complete coverage of the DNA sequence. Single-stranded DNA was produced (Messing, 1983) and sequenced from universal primers using Sequenase (United States Biochemicals Ltd.).
The E. coli rhaT gene was sequenced by cloning the 1.6-kb SmaI/ XmnI fragment from plasmid pCGT6 into EcoRV-digested Bluescript plasmid (Stratagene), to make plasmid pCGT12. The sequencing strategy was to make unidirectional deletions in the fragment by using exonuclease 111 (Nested Deletion Kit, Pharmacia) followed by double-stranded sequencing of the product. To make deletions in the XrnnIISrnaI fragment, to be sequenced with the M13 universal primer, plasmid pCGT12 was digested with BarnHI and Sac1 prior to treatment with exonuclease 111. Deletions were made in the opposite direction, to be sequenced using the M13 reverse primer, by digesting plasmid pCGT12 with Hind111 and KpnI followed by digestion with exonuclease 111. Preparations of double strandedplasmid DNA (Johnson, 1990) were sequenced using Sequenase (United States Biochemicals Ltd.). The double-stranded DNA sequencing protocol was exactly as described in the Sequenase protocol manual except that NaOH was removed after the denaturation step using a spun column (0.3 ml of Sepharose CL-GB 200 in a 0.5-ml microcentrifuge tube with a small hole in the bottom) rather than by precipitation. Additional DNA sequence was obtained from two oligonucleotide primers synthesized to the E. coli rhuT DNA sequence; oligonucleotides were synthesized on a Milligen "Cyclone" machine operated by M. Wheldon, Biochemistry Department, Cambridge University. Electrophoresis of sequencing gels was performed on a Pharmacia 2010 Macrophor sequencing apparatus at a constant voltage of 1200 V. DNA sequence data was analyzed using the Staden DNA sequencing programs (Bishop and Rawlings, 1987) processed by a DEC Micro Vax 3100 operated by the University of Cambridge Departments of Biological sciences. DNA Manipulations-Cosmid and plasmid DNA was prepared by the alkaline lysis method (Maniatis et al., 1982). Restriction digests were carried out according to the Manufacturers' recommendations (Amersham, Pharmacia, or New England Biolabs), and ligations were performed according to Maniatis et al. (1982). The CaC12 method for preparing competent cells was routinely used (Maniatis et al., 1982). Restriction fragments were isolated from agarose gels either by electrophoresis onto Whatman DE81 ion exchange paper (Dretzen et al., 1981) or by excising the desired band and isolating the DNA by the "glass milk" method (Vogelstein and Gillespie, 1979).
Transport Assays-Transport of "C-labeled sugar and pH measurements of sugar-H+ symport activities were carried out as described elsewhere (Henderson et al., 1977;Henderson and Macpherson, 1986). Initial rates of sugar uptake were determined from the 15-s time point.
Restriction digests of h clone 4B6 (Kohara et al., 1987) were probed with a gel-purified 32P-labeled oligonucleotide (Maniatis et al., 1982) and with the 400-bp HincIIIBalI restriction fragment STRHAl. The oligonucleotide (ORHA1) corresponded to nucleotides 2135-2152 of the DNA sequence downstream from the rhuR gene sequence (Tobin and Schleif, 1987). The restriction-digested h DNA was immobilized in the agarose gel and then it was probed with the 32P-labeled DNA. The preparation and drying of the agarose gel were as described by Thein and Wallace (1986). The dried agarose gel probed with oligonucleotide ORHAl was washed at it's T, (52 "C) in 5 X SSPE (20 X SSPE: 3 M NaCl, 180 mM NaH2P04.2H20, 20 mM EDTA, pH 7.4). The dried agarose gel probed with restriction fragment STRHAl was washed at high stringency in 0.3 X SSC at 50 "C.

RESULTS AND DISCUSSION
Construction of a L-Rhamnose Transport-negative Strain of E. coli-The strategy employed to produce a L-rhamnose transport-negative strain of E. coli was to use Mu d(ApR,lac)I mutagenesis in a fdp host strain. The Mu d(ApR,lac)I phage inserts randomly into the host chromosome, inactivating the gene it inserts into. In addition, the lac operon in Mu d(Ap',lac)I, which lacks its own promotor, can be expressed from a suitably positioned promoter in the host chromosome. An E. coli host strain with the fdp lesion was used, because the presence of gluconeogenic carbon sources such as Lrhamnose or L-fucose results in the accumulation of toxic intermediates due to the absence of fructose 1,6-bisphosphate-1-phosphatase. Therefore a Mu d(ApR,lac)I insertion into a gene required for L-rhamnose utilization will make the fdp strain resistant to L-rhamnose. E. coli strain JM2463 (Table I) was used as the parent strain for mutagenesis; the strain was fdp and could not grow on L-rhamnose, L-fucose, or glycerol. The ability to grow on L-rhamnose and L-fucose was restored by P1 phage-mediated transduction to fdp+ (strain JAR3, Table I), which showed that strain JM2463 contained functional rha and f u c operons. A Mu d(Ap',lac)I phage lysate was prepared from the Mu lysogen E. coli strain MAL103 (Table I) and was used immediately to infect strain JM2463. Ampicillin-resistant colonies were screened for L-rhamnose inducible growth on lactose. A single isolate, strain JAR1 (Table I) had the phenotypes ApR, Lac-, Rha-, and (Rha + Lac)+. Plate assays for @-galactosidase activity indicated that lac2 was induced by 1 mM Lrhamnose in strain JARl but not in strain JM2463 or strain JAR3. Therefore, strain JARl appeared to have a Mu d(Ap',lac)I insertion in a rha gene.
To facilitate further investigation of the Mu insertion in strain JAR1, the strain was cured of the lesion in fdp by P1mediated transduction, using phage P1 propagated on strain CSH25 (fdp+). Transductants were selected on minimal medium containing glycerol as sole carbon source; all the Gly' colonies also grew on L-fucose but were negative for growth on 10 mM L-rhamnose (strain JARZ, Table I). The Mu d(ApR,lac)I phage could have inserted into any of the rha genes to give the phenotype of strain JAR1; therefore, Lrhamnose transport assays were performed to determine whether strain JAR2 was indeed impaired in L-rhamnose transport.
Characterization of the L-Rhamnose Transport-negative E. coli Strain JAR2-The transport of 14C-labeled L-rhamnose or L-fucose into strain JAR2 was compared with transport into strain JAR3 (no Mu d(ApR,lac)I insertion) and with strain JM2513 (Table 11). E. coli strain JM2513 contained a Xplac Mu I insertion in a rha gene encoding a metabolic enzyme? Cultures were grown on minimal medium with glyc-M. C. Jones-Mortimer, unpublished data.

TABLE I1
Sugar transport assays in a control strain (JAR3) and two Mu lysogen strains (JM2513 and JAR2) Uninduced cultures were grown on minimal medium with glycerol as carbon source. Cultures induced with either L-rhamnose or Lfucose were grown in minimal medium with glycerol and 10 mM inducer. Cultures were grown overnight at 30 "C; preparation of the cell suspension for the uptake experiments and @-galactosidase assays are as described in Henderson et al. (1977) and Miller (1972 erol as carbon source and were induced with either L-rhamnose or L-fucose, or not induced. The positive control strain JAR3 exhibited L-rhamnose-inducible uptake of both 14Clabeled L-rhamnose and 14C-labeled L-fucose (Table 11); a common product of L-rhamnose and L-fucose metabolism is the true inducer of the fuc operon (Chen et aZ., 1987). Strain JAR2 showed no L-rhamnose-inducible transport of either Lrhamnose or L-fucose, despite having a functional fuc operon as evidenced by the presence of fucose-inducible L-fucose transport. A single lesion in the L-rhamnose transport gene would be expected to give this phenotype. In contrast, strain JM2513 had a L-rhamnose-inducible L-rhamnose transport activity, but L-rhamnose did not induce the fucose transporter; this is consistent with the Mu insertion being in a rha gene encoding a metabolic enzyme rather than a transporter. @-Galactosidase assays carried out in parallel to the sugar uptake assays indicated that both strains JAR2 and JM2513 contained Mu insertions in the rha operon and that L-rham-nose induction had taken place despite the low transport activities observed (Table 11). Neither L-rhamnose nor Lfucose could be transported by any of the uninduced strains. Additional evidence that the Mu d(ApR,Zuc)I phage in strain JAR2 was inserted in the L-rhamnose-H+ symporter gene was obtained by measuring the sugar-H+ symport activity of strain JAR2 after induction with L-rhamnose ( Fig. 1). On addition of L-rhamnose to de-energized JAR3 cells (positive control) a rapid alkalinization of the external medium was observed followed by a rapid acidification (Fig. 1). This is characteristic of a sugar being imported into the cell via a sugar-H+ symporter followed by metabolism of the sugar to acidic end products that are excreted into the external medium (Henderson and Macpherson, 1986). In contrast, L-rhamnose-induced JAR2 cells did not elicit a rapid alkalinization on addition of L-rhamnose, but only a slow acidification. This apparent metabolism was at a reduced rate compared with that observed for strain JAR3 and would have been unlikely to obscure sugar-H+ symport activity if it had been present. L-Rhamnose induction of the rha operon in strain JAR2 could be observed by measuring the activity of Lacy, encoded by the Mu phage and expressed from a rha promoter. On addition of isopropyl-1-thio-B-D-galactopyranoside (a substrate for Lacy) to JAR2 cells induced with L-rhamnose, a rapid alkalinization of the medium was observed (Fig. 1). This is not followed by acidification because isopropyl-1-thio-(3-D-galactopyranoside cannot be metabolized (Fig. 1). This experiment showed that the absence of L-rhamnose-H+ symport activity from strain JAR2 was not due to leakiness of the membrane.
Additional manipulations were required before strain JAR2 could be used to screen a genomic library to isolate the rhuT gene. It was first made recA-by phage P1-mediated transduction; the P1 lysate was obtained from P1-infected strain PB13 ( Table I). The derived recA-strain JAR62 subsequently had the Mu phage excised from the L-rhamnose transport gene to produce strain JAR66 with a stable genetic background for the screening of a genomic library using ampicillin for plasmid selection. During excision of the Mu phage there is often deletion of genomic DNA which was adjacent to the phage, thus producing a stable deletion in the E. coli chromosome.

9.
and no longer had Lrhamnose-inducible &galactosidase activity. Interestingly, strain JAR66 would not grow at even high concentrations (100 mM) of L-rhamnose, whereas strain JAR2 could grow on L-rhamnose as sole carbon source if the concentration was greater than 40 mM. This suggested that additional genes had been deleted during the excision of Mu d(ApR,lac)I.
Screening a S. typhimurium Genomic Library to Isolate the rhuT Gene-A genomic library prepared from S. typhimurium C5 DNA (see "Experimental Procedures") was amplified in strain HU835 and then used to infect the L-rhamnose transport-negative strain JAR66. 3000 ApR colonies were screened by replica plating for growth on 10 mM L-rhamnose as sole carbon source. Three colonies grew on L-rhamnose. Restriction analysis of the plasmids from the three Rha' colonies indicated that they were identical (pJAR1).
Plasmid pJARl (Fig. 2) fully compensated for the deletion at the rhu locus in strain JAR66. Strain JAR66(pJARl) grew on minimal medium with L-rhamnose as sole carbon source and effected the uptake of I4C-labeled L-rhamnose following induction by L-rhamnose. Plasmid pJAR2 (SalI-digested pBR322 ligated to the 9.8-kb SalI fragment from plasmid pJAR1) fully complemented the deletion in pJAR66; L-rhamnose-H+ symport activity was also observed (Fig. 1). Plasmid pJAR2 was used as a source of DNA for further subcloning of the rhaT gene. Expression of the rhaT Gene from a X Promoter and the Localization of the rhaT Gene-Plasmid pAD2M4 contains the X P L promoter on a 2.4-kb HindII/BamHI X DNA fragment ligated into plasmid pBR322. Plasmid constructs derived from pAD284 can be maintained in either of two X lysogenic E. coli strains which produce cI repressor to prevent potentially lethal levels of expression from h P L during routine growth of strains. E. coli strain AR120 contained wild type X as lysogen; the X PL promoter could be induced by nalidixic acid (Mott et al., 1985). E. coli strain AD5827 was a X lysogen that expressed the cIs57 gene product; increasing the cell culture temperature from 33 to 42 "C resulted in the inactivation of the thermolabile cIS7 protein, thus allowing expression from the X PL promoter. The 2.4-kb HindIII/BamHI fragment from plasmid pAD284 containing the X PI, promoter was ligated to HindIII/BamHI-digested plasmid pJAR2, producing plasmid pJAR3. Only one orientation of the 9.8-kb SalI insert in relation to the X PL promoter was obtained with this method. To obtain the opposite orientation (pJAR4), plasmid pJAR3 was digested with SalI to remove the 9.8-kb  pJAR2 ( a and b ) both complemented the rha-lesion in JAR66, as indicated by growth of strains JAR66(pJARl) and JAR66(pJAR2) on minimal medium that contained L-rhamnose as sole carbon source. Uptake activity and L-rhamnose-H+ symport assays were performed on strains JAR66(pJARl) and JAR66(pJAR2) after growth in minimal media that contained L-rhamnose as sole carbon source. Plasmids pJAR3-10 (c-h) contained partial deletions of the cloned S. typhirnuriurn genomic DNA downstream from the X promoter PL in plasmid pAD284. Induction of the X PL promoter upstream from the rhaT gene was achieved using both nalidixic acid induction for the plasmids in E. coli strain AR120 and by heat shock for the plasmids in E. coli strain AD5827.
The ability of induced strains to exhibit uptake activity and L-rhamnose-H+ symport activity is indicated by "positive" or "negative" in the column labeled "rhamnose transport." The rate of [14C]~-rhamnose uptake in positive strains varied between 1.7 and 14.7 nmol/min/mg dry mass, whereas negative strains typically showed uptake rates between 0.0 and 1.1 nmol/min/mg dry mass. a and b, linear diagrams of plasmid constructs pJARl and pJAR2. The hatched box in the diagram of plasmid pJARl represents part of plasmid pHC79, and the stippled box in plasmid pJAR2 represents part of plasmid pBR322. c-h, linear diagrams of plasmid constructs pJAR3 to pJAR10. Filled boxes represent part of plasmid pAD284 (Fig. 5). insert followed by re-ligation of the insert with phosphatasetreated vector. Nalidixic acid induction of strains AR120(pJAR3) and AR120(pJAR4) resulted in expression of L-rhamnose transport activity (4.2 f 0.3 nmol/mg/min) only in strain AR120 (pJAR3). The L-rhamnose transport activity in nalidixic acid induced strain AR120(pJAR4) was 0.5 f 0.4 nmol/mg/min. This defined the direction of the transcription of the rhaT gene as BglII to AuaI (Fig. 2). To locate the rhaT gene within the 9.8-kb SalI fragment, a series of plasmids were constructed from plasmid pJAR3 with deletions in the insert DNA. Each construct was transformed into E. coli strains AR120 and AD5827 and tested for induction of L-rhamnose transport activity. The results (Fig. 2) suggested the presence of the rhaT gene between the EcoRV site (proximal to a BglII site) and the AuaI site, with transcription starting somewhere between the EcoRV site and the BgZII site. Thus, we have located the rhaT gene to a 2.4-kb portion of DNA within the original 9.8-kb fragment (Fig. 2i).
Sequencing of the S. typhimurium rhaT Gene-A 3-kb EcoRV-PuuII fragment from plasmids pJAR6 and pJAR10, which contained the complete rhaT gene, was sequenced (see "Experimental Procedures"). A contiguous sequence of 3052 bp was assembled; this comprised 2377 bp of S. typhimurium DNA (Fig. 3) between the EcoRV and AuaI sites in plasmid pJARlO (Fig. 2i) and 675 bp of plasmid pBR322 DNA sequence (data not shown) between the AuaI and PuuII restriction sites. The S. typhimurium DNA sequence was determined a.  RhaT protein sequence is given in the single-letter amino acid code above the DNA sequence. A putative Shine-Delgarno sequence is shown by a line above the DNA sequence. Possible -10 and -35 regions are indicated by asterisks, and a potential CRP binding site is shown by a box. DNA sequences predicted to form hairpin loop structures, or palindromic repeats, are shown by arrows. b, the DNA sequence was determined on both strands of the DNA except for nucleotides 1-54. The filled box represents the coding region of the r h T gene. A partial restriction map is shown below. Arrows represent a sequence of DNA derived from one sequencing reaction. If a particular sequence was obtained more than once, this is indicated by a number at the start of the arrow.

CATGAGCTTCTATGTGCTGTGCGGGGGGCTTGTCGGTCTGGTGCTAAAACGATAGGAGTGG~TGCT~CGCCGTCCCGTTGCTGTATTMGCCTCGGCTGCGTGGTMTTATTATCGCGGC M S F Y V L C G G L V G L V L K E W K N A G R R P V A V L S L G C V V I I I A
on both strands of the DNA, except for nucleotides 1-54 which were upstream from the rhaT gene (Fig. 3). A single open reading frame was identified by measuring the codon usage in all of the possible reading frames. The single open reading frame was in the 5' to 3' orientation, as predicted from the L-rhamnose-transport data derived from the expression of the rhaT gene in pJAR3 and pJAR4 (Fig. 2). However, the reading frame started much further downstream from the BgDI site than expected. The reading frame is 1032 nucleotides long, which corresponded to a protein of 344 amino acids (Fig. 3). Upstream from the putative initiation codon of the rhaT gene there was a good match to the consensus Shine-Delgarno sequence (nucleotides 1025-1029), potential -10 and -35 regions (nucleotides 980-985 and 960-966, respec-tively) and a possible CRP binding site (nucleotides 888-908). The poor match of the potential -10 and -35 regions to the consensus is typical of bacterial genes whose expression is controlled by a positive transcriptional activator, such as RhaR controlling expression of the rhaS/R genes Schleif, 1987, 1990b). In addition, there was a short palindromic repeat (nucleotides 920-926 and 944-950) adjacent to and between the potential ribosome and CRP binding sites. This palindromic repeat could be the binding site for a positive acting transcriptional element (RhaR or RhaS?), but it is not homologous to the RhaR binding site in E. coli (Tobin and Schleif, 1990b). At the 3' end of the rhaT gene, 16 bp after the termination codon, there was an inverted repeat (nucleotides 2088-2096 and 2105-2113) which could form a stem loop structure comprising a 9-bp stem and eight-nucleotide loop. It is unlikely that this stem loop represents a rho-independent terminator, because there are no contiguous thymine residues in the DNA sequence downstream from the loop, but it could be a rho-dependent terminator (reviewed by Platt, 1986).
Identification of the rhaT Gene in E. coli-With the publication of a complete restriction map for the E. coli K12 chromosome (Kohara et al., 1987), it is comparatively easy to map the location of a new gene of known DNA sequence by Southern blot analysis. This technique was used to map the position of the rhuT gene. E. coli K10 genomic DNA was digested singly and in combination with various restriction enzymes. The DNA fragments were separated by agarose gel electrophoresis and blotted onto a membrane. The Southern blots were probed with a 32P-labeled 400-bp HincIIIBalI restriction fragment (STRHA1) corresponding to the C-terminal portion of the RhaT protein. A single band in each lane was seen on autoradiography of the Southern blot (results not shown). The sizes of the restriction fragments produced by BamHI, EcoRI, PstI, HindIII, PuuII, and KpnI digestions were consistent with only one region of the E. coli DNA restriction map (Fig. 4), adjacent to the rhu locus. However, fragments derived by EcoRV digestion did not concur with the published data. The X clone 4B6 which encompassed this region was obtained from Dr. Y. Kohara (Kohara et al., 1987) was digested with restriction enzymes and the fragments were separated by gel electrophoresis. The gel was probed with a "P-labeled HincIIIBalI restriction fragment of the S. typhimurium rhaT gene and a "P-labeled oligonucleotide complementary to the sequence downstream of the rhaR gene (nucleotides 2135-2152 (Tobin and Schleif, 1987)). A single agarose gel that contained two identical sets of restriction digests was electrophoresed and stained with ethidium bromide (a, lanes 1-8).
Half of the gel was probed with the S. typhimurium rhaT gene fragment (b, lanes 1-7), the other half was probed with the oligonucleotide (c, lanes 1-7). Japan. To define the position of the rhaT gene in relation to the sequenced rhaSIR genes (Tobin and Schleif, 1987), an oligonucleotide (ORHA2) was synthesized to a region downstream of the rhaR gene. A dried agarose gel containing restriction digests of the bacteriophage clone was probed with 32P-labeled oligonucleotide ORHA2 and the S. typhimurium rhuT gene fragment, STRHA1, also labeled with 32P (see "Experimental Procedures"). Autoradiography of the probed gel indicated that the rhaT gene was on the same 3-kb PuuII fragment as the region downstream from the rhaR gene.
Computerized analyses of the DNA sequences for S. typhimurium rhaT and E. coli rhaR genes identified the region downstream from the E. coli rhaR gene as the C terminus of RhaT. However, there was not a single open reading frame in the region downstream from the rhuR gene (Tobin and Schleif, 1987) that might encode RhaT; this was ascribed to a sequencing error in this region of DNA (see below). Expression of the E. coli rhaT Gene under Control of a X Promoter-To confirm the presence of the rhuT gene on the 3-kb PuuII fragment, a restriction fragment that contained the r h T gene was ligated into plasmid pAD284 downstream of the X PL promoter (Fig. 5). The 3-kb E. coli genomic DNA PuuII fragment from X clone 4B6 (Kohara et al., 1987) was first subcloned into plasmid pBR322 (pCGT6); a derived 1.6kb XmnIISmaI restriction fragment predicted to contain the rhaT gene was subsequently cloned from plasmid pCGT6 into the HpaI site of plasmid pAD284 (Fig. 5). The SmaI site was predicted to be at the end of the rhuR gene (Tobin and Schleif, 1987), whereas the XmnI site was in the pBR322 DNA, 35 nucleotides from the PuuII site (Fig. 5). Two orientations of the insert were obtained; plasmid pCGTlO was predicted to contain the insert in the right orientation for its expression by the X PL promoter, whereas in plasmid pCGTll the rhaT gene was predicted to be in the wrong orientation for expression (Fig. 5). Plasmids pCGTlO and pCGTll were transferred into E. coli strain T G~(~C I~~~) , which expressed the temperature-sensitive CI gene product, cIeS7 (Remaut et al., 1983). Thus, at 33 "C the X PL promoter was inactive due to the cIs57 protein binding at the operator sites, whereas at 42 "C CIs57 protein was nonfunctional and the X P L promoter expressed the DNA inserted downstream from it. After induction at 42 "C, only strain TG2(pCGT10/pc1~,~~) produced a protein which could effect the uptake of radiolabeled L-rhamnose into cells (Fig. 5). This confirmed that there was a functional open reading frame in the XmnIISmaI fragment, despite the lack of an open reading frame in the region downstream from the rhaR gene (Tobin and Schleif, 1987). The XmnIISmaI fragment was also cloned into EcoRV-digested Bluescript plasmid (plasmid pCGT12) so that the E. coli rhuT gene could be sequenced.
DNA Sequence of the E. coli rhuT Gene-The 1.6-kb XmnIl SmaI fragment containing the E. coli rhaT gene was sequenced directly from plasmid pCGT12 (see "Experimental Procedures"). The DNA was sequenced at least once on both strands; Fig. 6 depicts the 1560 nucleotides of DNA sequenced between the PuuII and SmaI restriction sites. Three open reading frames were identified in the sequence. A single complete open reading frame that encoded a 344-amino acid protein was found; the orientation of the open reading frame corresponded to the orientation of the rhaT gene deduced from the induction of the X P L promoter in plasmid pCGT10. The protein was 91% identical to the RhaT protein predicted from the S. typhimurium gene sequence. Upstream from the putative initiator Met codon of the E. coli rhaT gene was a good match to the consensus Shine-Delgarno sequence (nucleotides 462-466), possible -35 and -10 regions (nucleotides a.
BamH I e.

a. CTGGTCCAGTTTGGTGATCAGCTCTTCMCCGGCAGGTTGGCAAATTCTGGCAGGCTTTC 60
PStl CAGCGCCGCGTTGGCGTTGTTTACGTRGGTCTGATGGTGTTTGGTGTGGT~TTTCCAT 120 HindIIIIBamHI-digested plasmid ~B R 3 2 2 .~ b, plasmid pCGT6 (7.36 kb) was constructed by ligating the 3-kb PuuII fragment from the X clone 4B6 into PuuII-digested plasmid pBR322. One PuuII site was not regenerated after the ligation; this was due to a 2-bp deletion at the PuuII site in the vector sequence (data not shown). c and d, plasmids pCGTlO (8.01 kb) and pCGTll (8.01 kb) were constructed by ligating the 1.6-kb SmaI/XmnI restriction fragment from pCGT6 into HpaI-digested pAD284. The AuaI site was used to determine the orientation of the insert. e, rate of 14C-labeled L-rhamnose uptake in heat-induced strains TG2(pCGT10/pcIBs7) and TGB(pCGTll/pcI%7). Plasmid pcIBs7 encoded the temperature sensitive cIg57 protein. Another open reading frame (nucleotides 1-189) at the start of the DNA sequence encoded the N terminus of a predicted protein identical to the published sequences of the manganese-containing superoxide dismutase SodA (Steinman, 1978). The DNA sequence of the sodA gene has also been published (Takeda and Avila, 1986); nucleotides 1-329 are identical to the published sequence except for a base change at nucleotide 16.

Q D L K T I L E E V P L N A F E P L S E L A A N A N N V Y T Q H H K T H H I E M T Q K D F H P E L A D Y
Nucleotides 1508-1560 were part of an open reading frame that encode the C terminus of RhaR identical to that published by Tobin and Schleif (1987). The DNA sequence of their rhaR gene included 325 nucleotides of sequence downstream from the rhaR gene, which should have encoded the rhaT gene. However, the presence of a compression resulted  shown; bracketed restriction enzyme sites were present in the original X clone but were destroyed during cloning.
in the omission of a single base (nucleotide 1306, Fig. 6), thereby disrupting the coding sequence of the rhaT gene.
A Comparison of the RhuT Protein and Gene Sequences from S. typhimurium and E. coli-An alignment of the E. coli and S. typhimurium RhaT protein sequences showed that the proteins are 91% identical (Fig. 7); out of 344 amino acids, there are 15 conservative changes, 10 semi-conservative RhaT protein sequence is on the bottom line. The alignment was obtained using the program BESTFIT (Devereux et al., 1984 FIG. 9. Hydropathy plot of the RhaT protein from S. typhimurium and a putative model of its orientation in the membrane. a, the hydropathy plot was derived using the algorithm of Kyte and Doolittle (1982) with a window size of 11. b, the orientation of the putative model of RhaT was determined by the overall charge on each hydrophilic loop. Asp and Glu residues are indicated by a negative sign, whereas Argand Lys residues are indicated by apositiue sign. frequency of 0.92 (Ikemura, 1981). The RhaT protein is extremely hydrophobic; the E. coli RhaT protein contains 73.3% hydrophobic amino acids with a hydropathic index of 0.82 (Kyte and Doolittle, 1982).
Alignment of the DNA sequences from the two organisms revealed a region of 1236 nucleotides with 81% identity (nucleotides 329-1560 of the E. coli sequence aligned with nucleotides 888-2123 of the S. typhimurium DNA sequence). The sequence upstream from the rhaT gene in S. typhimurium did not contain a sequence which corresponded to the s o d gene and was not homologous to any DNA sequence in computer databases. The region downstream from the rhaT gene in S. typhimurium was homologous to the rhaR gene from E. coli. However, two nucleotides in different locations were absent from the S. typhimurium rhaR gene, which resulted in the disruption of the reading frame for RhaR. This region was sequenced adequately on both strands of the DNA (Fig. 3), so it is unlikely that the differences are due to a sequencing error. Therefore these changes may be a cloning artifact or may really represent the true rhaR gene sequence in S. typhimurium C5. This strain of S. typhimurium cannot grow on Lrhamnose as sole carbon source and does not possess a Lrhamnose-inducible RhaT activity, despite having a fully functional rhaT gene (see Figs. 1 and 2); a nonfunctional rhaR gene in S. typhimurium C5 could explain these observations.

CONCLUSIONS
We have isolated and sequenced the rhaT genes from S. typhimurium and E. coli. Expression of the rhaT genes from a X P L promoter resulted in an E. coli strain that could effect the uptake of 14C-labeled L-rhamnose and showed sugar-H+ symport activity. The rhaT gene maps between the sodA and rhaR genes at 87.9 min (3605 kb) on the E. coli chromosome restriction map (Kohara et al., 1987). It encodes an extremely hydrophobic protein of 344 amino acids; the S. typhimurium and E. coli RhaT proteins are 91% identical. The RhaT protein sequence has been compared with protein sequences in computer databases, but no significant homology was detected. The model for RhaT in Fig. 9b was deduced from a hydropathy plot (Fig. 9a) which identified 10 clearly defined hydrophobic regions in the protein that could span a lipid bilayer. The orientation of RhaT in the membrane was deduced from the "positive inside" rule (von Heijne, 1986), which predicts that hydrophilic loops in the cytoplasm generally have a net positive charge. A model of RhaT (Fig. 9) with the N and C terminus on the periplasmic face of the membrane conforms to the positive inside rule; all the hydrophilic loops in the cytoplasm have a net positive charge, whereas all the loops in the periplasm have a net negative charge or are uncharged. The proposed topology of the RhaT protein is therefore completely different to the 12 transmembrane domain models proposed for other sugar-H+ symporters (reviewed by Henderson, 1990). In addition, the amino acid sequence of the RhaT protein is not homologous to any other protein. Thus the L-rhamnose-H+ symporter represents a novel type of sugar transport protein. Current work is focusing on the overexpression of rhaT and the use of P-lactamase as a topological reporter to define the structure of the RhaT protein.