INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

After its discovery almost 40 years ago, the Escherichia coli lactose operon, encoding enzymes of lactose metabolism, became the first model for gene regulation (reviewed in reference 4). The key component of this system is the lac repressor (LacI), the product of the lacI gene. The lac operon contains a primary operator (O₁), which is the major element of repression by LacI, and two pseudo-operators, which enhance repressor binding to O₁ by cooperativity. Control of the lac operon also involves activation by the cyclic AMP receptor protein. Many other paradigm systems of negative control have since been described, including GalR, one of the two repressors of the gal regulon encoding enzymes of galactose transport and metabolism in E. coli. Regulation of the gal regulon is mediated through GalR, GalS (Gal isorepressor), and the cyclic AMP receptor protein. GalR and GalS negatively regulate transcription of the two promoters of the gal operon, although GalS is not as efficient as GalR (57).

The bioconversion of lactose, which is the primary carbon and energy source in milk, into lactic acid is an essential process in industrial dairy fermentations carried out by lactic acid bacteria. Genetic studies of the metabolic pathways for lactose utilization in these gram-positive bacteria have revealed a variety of lac operons that differ from the paradigm known in E. coli (13). The thermophilic yogurt bacteria Streptococcus thermophilus and Lactobacillus bulgaricus contain a highly homologous lacSZ operon in which the -galactosidase (lacZ) gene is located downstream from the lacS gene encoding a lactose permease (LacS), which belongs to the galactoside-pentose-hexuronide translocators (27, 41, 48, 49).

Although lactose is efficiently transported and hydrolyzed intracellularly, many strains of S. thermophilus and L. bulgaricus do not grow on galactose and ferment only the glucose portion of lactose, while the galactose is excreted into the medium in amounts stoichiometric with the uptake of lactose (20, 22). Kinetic studies indicated that LacS mediates both galactoside exchange (e.g., lactose-galactose) and movement of galactosides and protons (15). The exchange reaction is highly favored with excess galactosides on either side of the membrane and may account for the galactose-negative (Gal) phenotype of S. thermophilus in milk which contains an excess of lactose (40). Another explanation for the Gal phenotype may be the absence of functional Leloir pathway enzymes, including galactokinase (GalK), galactose-1-phosphate uridylyltransferase (GalT), and UDPglucose 4-epimerase (GalE), products of the galK, galT, and galE genes, respectively. Remarkably, under appropriate selective conditions, such as limiting lactose and excess galactose, Gal⁺ derivatives of S. thermophilus were obtained which fermented galactose and contained Leloir enzyme activities (21, 50). However, no molecular explanation was given, and the genetics of the Leloir pathway has only been poorly investigated in S. thermophilus. The lacSZ operon of strain A147 was found to be preceded by galE and galM (42). The galM gene appeared to be constitutively expressed and could encode a mutarotase that, similar to the homologous enzyme of E. coli, forms the galactokinase substrate -D-galactose from -D-galactose pyranose generated from lactose by -galactosidase (LacZ) (7). However, S. thermophilus A147 is not a Gal⁺ strain.

The present study was undertaken to gain insight into the presence and regulation of the gal genes of S. thermophilus and the mechanism by which the genes, in particular the galK gene, are prevented from being expressed. Here we describe the characterization of the gal operon, consisting of the galK, galT and galE genes, and its promoter from S. thermophilus CNRZ 302, for which galactose-fermenting (Gal⁺) revertants have been reported (5). A regulatory gene, galR, was identified that is divergently transcribed from the gal operon. Analysis of mRNA for the gal metabolic genes from a Gal⁺ fermenting derivative of CNRZ 302 indicated that regulation occurred at the transcriptional level. In contrast, the gal metabolic genes of the original Gal strain were not sufficiently transcribed to allow galactose metabolism. Furthermore, we demonstrate that GalR acts as a transcriptional activator of both the gal and lac operons and negatively regulates its own expression. To the best of our knowledge, this is the first report describing the mechanisms regulating galactose utilization in S. thermophilus at the molecular level.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. The bacterial strains and plasmids used in this study are described in Table 1. S. thermophilus strains were subcultured in M17 broth (Oxoid, Basingstoke, England), containing either 1% lactose, glucose, or galactose as necessary, at 42°C unless stated otherwise. The taxonomic position of strain CNRZ 302 was confirmed by 16S rRNA sequence analysis and corresponded to Streptococcus thermophilus (GenBank accession number X68418). E. coli strains HB101 or LE392 and TG1 were used for the isolation of pACYC184- and pUC19-derived plasmids and for the propagation of bacteriophage M13 chimeras, respectively. E. coli was routinely grown in TY medium (45) or brain heart infusion (Difco) broth with aeration at 37°C. MacConkey agar base (Difco Laboratories) supplemented with 1% galactose was used to detect galactose-positive (Gal⁺) E. coli strains. Agar media were prepared by adding 1.5% agar to broth. The antibiotics used for selection in media were chloramphenicol at 4 µg/ml and erythromycin at 2.5 µg/ml for S. thermophilus and chloramphenicol at 15 µg/ml, tetracycline at 12 µg/ml, and ampicillin at 100 µg/ml for E. coli. Em^r E. coli strains were selected on brain heart infusion agar containing 150 µg of erythromycin per ml.

DNA isolation and manipulations. Transformation and isolation of plasmid DNA from E. coli were performed essentially by established protocols (46). Chromosomal DNA was extracted from exponentially growing cells of S. thermophilus by the procedure of Hayes et al. (19). The Anderson and McKay (3) lysis procedure was used to detect plasmid DNA in S. thermophilus. Restriction enzymes, T4 DNA ligase, and other DNA-modifying enzymes were used as recommended by the supplier (Gibco-BRL, Life Technologies, Gaithersburg, Md.). DNA fragments were recovered from agarose gels with the GlassMatrix DNA isolation system (Gibco-BRL). Electroporation of S. thermophilus was performed by the procedure of Mollet et al. (32) with the modification that the harvested cells were incubated in the electroporation buffer at 4°C for at least 4 h prior to electroporation. PCR was performed under the conditions described previously (25) using Taq polymerse (Life Technologies) or Pwo polymerase (Boehringer Mannheim). Oligonucleotides were synthesized by Eurogentech (Gent, Belgium).

Cloning of gal genes. S. thermophilus CNRZ 302 total genomic DNA was digested with EcoRI, and the DNA fragments were separated by agarose gel electrophoresis (0.7% agarose). The DNA was transferred to a GeneScreen Plus (Dupont, Boston, Mass.) membrane by established methods (46). The membrane was hybridized with a 700-bp AccI fragment, containing part of the S. thermophilus F140 galK gene kindly provided by B. Hutkins (34). The labeling of this fragment with horseradish peroxidase and the hybridization and detection methods were as described in the manufacturer's manual for the ECL system (Amersham, Little Chalfont, United Kingdom). Fragments of approximately 5 kb were recovered, ligated with EcoRI-linearized calf intestinal alkaline phosphatase-treated pACYC184, and used to transform E. coli HB101 and LE392. Gal⁺ clones were selected as red Tc^r colonies on McConkey galactose agar at a frequency of approximately 1%. Analysis of the plasmid content of all 10 Gal⁺ colonies indicated that they contained a recombinant plasmid with a 4.9-kb insert that showed an identical restriction pattern. One of the clones, designated HB101(pNZ680), was used in further experiments.

Nucleotide sequence analysis. DNA fragments were subcloned in the phage vectors M13mp18 and M13mp19 with TG1 as a host by using standard techniques (46). Nucleotide sequences of both strands were determined by the dideoxy-chain termination method (47), adapted for Sequenase version 2.0 (U.S. Biochemical Corp., Cleveland, Ohio) with either the M13 universal primer or specifically synthesized primers. The gal promoter regions of S. thermophilus CNRZ 302 and its 10 Gal⁺ derivatives were isolated as 350-bp PCR fragments from agarose gels. The purified fragments and primers were annealed by boiling for 5 min and rapidly freezing in liquid nitrogen, and sequencing proceeded as described above. The sequence data were assembled and analyzed with PC/GENE version 6.6 (Genofit, Geneva, Switzerland). Amino acid sequence comparisons were performed with the EMBL (release 31.0), SwissProt (release 28.0), and NBRF/PIR (release 40.0) databases using the FASTA program (36), through the facilities of the CAOS/CAMM Center (Nijmegen, The Netherlands). The curvature of DNA was predicted as described by Munteau et al. (33).

Isolation of Gal⁺ S. thermophilus strains. S. thermophilus CNRZ 302 cultures grown in M17 broth supplemented with 1% lactose were diluted 100-fold into M17 broth containing 1% galactose and 0.01% glucose and incubated for 24 h. Cultures that exhibited growth were transferred to M17 broth containing 1% galactose (Gal-M17) and incubated for 16 to 20 h. Ten Gal⁺ single-colony isolates were obtained by plating 10 cultures treated as described above on M17 agar with 1% galactose, and these were designated NZ302G and SS1 to SS9.

RNA isolation, Northern blotting, and primer extension analysis. S. thermophilus strains were grown in M17 broth (50 ml) containing 1% lactose, glucose, or galactose to an optical density (600 nm) of 0.6 to 1.0. Total RNA was isolated from the harvested cells as described by Kuipers et al. (26) with the following modification: before being subjected to bead beating, the cells were treated with 2 mg of lysozyme per ml for 2 min on ice, which increased the RNA yield. The RNA was either fractionated on a 1.0% formaldehyde gel (46) or glyoxylated and fractionated on a 1.2% agarose gel as described previously (52). RNA size markers were obtained from Bethesda Research Laboratories. RNA was transferred to GeneScreen Plus membranes by following the protocols outlined by the manufacturers. Hybridizations were performed at 65°C in a 0.5 M sodium phosphate buffer (pH 7.2) containing 1.0% bovine serum albumin (fraction V), 1.0 mM EDTA, and 7.0% sodium dodecyl sulfate, and the blots were washed at 55 to 65°C in 1.0 to 0.1× SSC buffer (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) containing 1% sodium dodecyl sulfate (46).

Construction of plasmids for analysis of galK promoters. The promoters from the CNRZ 302 and a class II Gal⁺ mutant, strain SS2, were amplified by PCR using primers 5'-CGG GAT CCT GCT AAT TTT GCG ATA TCT G-3' and 5'-CGG AAT TCC TTT AAA CTT TTC TCT TAA C-3', with built-in BamHI and EcoRI sites (underlined), respectively. The 210-bp products were cloned into BamHI-EcoRI-digested pUC19, generating pNZ680.1 and pNZ680.2. A 1.5-kb NsiI-EcoRV fragment from pNZ680 containing the CNRZ 302 galR gene and a potential transcription terminator (Fig. 1A) was attached in frame to the galR promoters (this step was necessary for the stability of further constructs), using the PstI and EcoRV sites in the pUC19 derivatives, generating plasmids pNZ6861 and pNZ6862. The "gal promoter and galR gene cassettes" were removed as 1.7-kb EcoRI-HindIII fragments (the 3' recessed terminus of the HindIII sites were first filled) and subsequently ligated into the EcoRI-ScaI-digested pNZ273. Plasmid pNZ273 contains the gusA reporter gene that encodes the -glucuronidase enzyme. The plasmids, designated pNZ6871 and pNZ6872 for the CNRZ 302 and SS2 mutants, respectively, contain the galK promoter fused to the gusA gene and the galR gene under its own promoter. The integrity of the amplified promoter regions was confirmed by sequence analysis. The constructs were initially made in E. coli and were subsequently used to transform S. thermophilus ST11 and selected on M17 agar containing 1% sucrose and chloramphenicol. Histochemical screening for -glucuronidase activity by selecting for blue colonies with 5-bromo-4-chloro-3-indolyl--D-glucuronide (X-Gluc) (Research Organics Inc., Cleveland, Ohio) was performed as previously described (11).

View larger version (23K): [in this window] [in a new window]   FIG. 1.   Organization of the gal and lac genes in their operons, and construction of a galR gene disruption. (A) Illustration of the gal genes on the 4.9-kb EcoRI fragment of S. thermophilus CNRZ 302 cloned in pNZ680. The positions of relevant restriction enzyme cleavage sites are indicated, and the frameshift mutation in the KpnI site is marked with a vertical arrow. The galK and galR promoters are indicated by arrows P1 and P2, respectively. The black lines below the genes represent the DNA probes used in the Northern blot experiments. (B) Organization of the gal and lac genes in the chromosome of S. thermophilus CNRZ 302. The mRNA transcripts for the gal and lac genes are indicated by arrows. Positions of promoters and potential terminators are indicated. (C) Plasmid pNZ684 carrying the thermosensitive replicon (Ts repA) of pG+host9 (small arrow) is shown in a linear form for convenience; the directions of the sense strand in the 750-bp PmlI-NlaIV galR fragment in pNZ684 and the Emr gene (erm) are indicated. (D) Orientation of integration of pNZ684 in the galR gene. The two partially deleted copies of galR are designated galR', and the location of the primer GALRAS is indicated by the bar.

Construction and use of integrating and complementing plasmids. A 750-bp PmlI-NlaIV fragment of the galR gene from pNZ680 was ligated into the calf intestinal alkaline phosphatase-treated EcoRV site of the thermosensitive pG⁺host9 vector, generating pNZ684 (Fig. 1C). Electrotransformation of pNZ684 into S. thermophilus NZ302G resulted in four Em^r transformants, all of which contained the expected plasmid at 30°C. To obtain integration of pNZ684, cultures grown overnight in M17 sucrose broth with erythromycin at 28°C were diluted 100-fold into fresh medium and reincubated at 28°C to allow the exponential phase of growth to resume. The cultures were shifted to 42°C and grown until they reached stationary phase. Dilutions of the cultures were plated at 42°C, and integrants appeared as Em^r colonies after 24 to 48 h of incubation. Correct integration within the galR gene in the chromosome (Fig. 1D) was confirmed by both Southern hybridization and PCR for integrant NZ302GR (data not shown). The galR gene of strain ST11 was disrupted in the same manner.

Enzyme and protein assays and chromatography. The S. thermophilus strains were grown in M17 broth containing either 1% lactose, galactose, glucose, or sucrose with the appropriate antibiotics to an optical density at 600 nm (OD₆₀₀) of 1.0. For the preparation of extracts, cells were disrupted with zirconium glass beads in a Bead Beater (Biospec Products, Bartlesville, Okla.) for 3-min treatments with intervals of 1 min on ice between treatments and cellular debris was removed by centrifugation. The extracts were kept on ice, and enzyme assays were performed within 4 h. Galactose 1-phosphate uridylyltransferase activity was assayed in the resultant extracts with 30 to 350 µg of protein per assay by the spectrophotometric method of Isselbacher (23). -Galactosidase was assayed at 37°C by the method of Miller (31) using 1 to 6 µg of protein per assay and galactokinase assays by the method of Ajdic et al. (2). All enzyme activity measurements presented were the mean of at least two independent experiments. Proteins concentrations were estimated by a dye binding assay (8).

Nucleotide sequence accession number. The GenBank accession number assigned to the nucleotide sequence encoding S. thermophilus galR, galK, galT, and the partial galE gene is U61402.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation and localization of the S. thermophilus galK and galT genes. Southern hybridizations of genomic DNA of S. thermophilus strain CNRZ 302 identified an EcoRI fragment of approximately 5 kb that hybridized with a probe consisting of a 0.7-kb internal fragment of the galK gene of S. thermophilus F140 (34). A minibank of fragments including the hybridizing fragment was constructed in the chloramphenicol resistance (cat) gene of pACYC184, and the putative galK gene of S. thermophilus CNRZ 302 was isolated by functional complementation of the galK2 mutation of E. coli HB101. The complementing plasmid, designated pNZ680 (Fig. 1A), contained a 4.9-kb insert that allowed HB101 to form red colonies on McConkey galactose agar and utilize galactose as the sole carbon source in minimal M9 medium. Introduction of a frameshift mutation in the KpnI site on pNZ680 resulted in a plasmid which could not restore a Gal⁺ phenotype to HB101, indicating that the galK gene was overlapping this site (data not shown). Moreover, pNZ680 also complemented both the galK2 and galT22 mutations of E. coli LE392, indicating that it also contained the galT gene of S. thermophilus. Although these gal genes are hardly expressed in S. thermophilus, the promoter upstream of the cat gene in pACYC184 is likely to be responsible for their expression in E. coli. The presence of a functional galT gene was further confirmed by assaying for GalT enzyme activity. Cell extracts of LE392(pNZ680) contained 119 nmol of GalT activity per min per mg, whereas no activity was detected for the LE392 strain alone.

Organization and similarity studies of the S. thermophilus gal region. Commencing at the KpnI site on pNZ680, nucleotide sequence analysis in both directions revealed the galK open reading frame (ORF), 1,164 bp in length (Fig. 1A). The deduced galK sequence had the strongest similarity to the GalK proteins of several gram-positive bacterial species including Streptococcus mutans (83%) and Lactobacillus casei (70%) (2, 6). The S. thermophilus CNRZ 302 GalK was also 79% similar to the GalK of the Gal⁺ S. thermophilus F410 strain (34). A potential ribosome binding site (5'-GAGA-3'), complementary to the 3' end of the 16S rRNA of lactic acid bacteria (12), was located 8 bp upstream from the first translational initiation codon at nucleotide (nt) 1483.

Transcriptional analysis of the gal genes. The transcription of the gal genes was analyzed in the wild-type S. thermophilus strain CNRZ 302, and in strain NZ302G, an isogenic spontaneous Gal⁺ mutant strain. Strain NZ302G has a doubling time of 58 min in M17 medium containing 1% galactose at 42°C, in contrast to the wild-type CNRZ 302 parental strain which does not grow at all on galactose. The Gal⁺ phenotype of NZ302G was stably maintained even after several subcultures in M17 containing lactose.

View larger version (71K): [in this window] [in a new window]   FIG. 2.   Northern blot analysis of RNA isolated from S. thermophilus strains CNRZ 302 (Gal) and NZ302G (Gal+) grown on glucose (Gl), lactose (L), or galactose (Ga). The probes used for the hybridization are indicated below the blot. The 23S and 16S rRNAs and the mRNA transcript sizes are indicated by arrows.

Mapping and characterization of the galR and galK promoters. The transcriptional start points of the galK and galR genes were determined by primer extension analysis. Total RNA was isolated from lactose- and glucose-grown S. thermophilus CNRZ 302 cells and from lactose-, glucose-, and galactose-grown NZ302G cells. A major transcriptional start site was observed for the galK gene of strain NZ302G that mapped at an A residue (nt 1452), 6 bp downstream of the inferred 10 sequence (TACGAT) (Fig. 3A). The latter was separated by 17 bp from a 35 sequence (TTGATT) that conforms well to the E. coli and S. thermophilus promoter consensus sequences (Fig. 4A). Essentially no clear signal for the galK gene of CNRZ 302 was detected, in accordance with its Gal phenotype (Fig. 3A).

View larger version (54K): [in this window] [in a new window]   FIG. 3.   Primer extension analysis of the 5' ends of the RNA transcripts for the galK (A) and galR (B) genes of S. thermophilus CNRZ 302 (two left-hand lanes) and NZ302G (three right-hand lanes) grown in lactose (L), glucose (Gl) or galactose (Ga). The ladder obtained from pNZ680 sequenced with the relevant primer is in the middle of each panel. Relevant nucleotide sequences of the promoter regions are presented on either side of the figure. The 10 region is denoted by a vertical bar, and the transcription initiation sites are indicated by +1. The arrows point to the major primer-extended products.

View larger version (24K): [in this window] [in a new window]   FIG. 4.   (A) Intergenic region containing the promoter sequences for the galK and galR genes. The 10 and 35 regions are underlined, and the transcriptional start sites are indicated as +1. The point mutations for the three classes (I, II, and III) of S. thermophilus Gal+ mutants are as described in the text. Inverted repeats are indicated by arrows. (B) Alignment of similar regions in the promoters of lac and/or gal genes of S. thermophilus and S. mutans. Inverted repeats in the S. thermophilus galK promoter are indicated by arrows. Nucleotides that are complementary in each arm of the inverted repeats are in boldface type.

Effect of the galR gene disruption on galactose utilization. To determine the function of the galR gene in the Gal⁺ S. thermophilus NZ302G, the gene was disrupted using pNZ684, consisting of the temperature-sensitive pG⁺ host9 vector that carried an internal fragment of the galR gene (Fig. 1C). The disruption caused by pNZ684 resulted in two partially deleted copies of the galR gene, one of which lacks the DNA for the 17 N-terminal amino acids including most of the conserved DNA binding motif while the other suffers from a 200-bp deletion that can encode 67 amino acids of the C-terminal region (Fig. 1D). In contrast to the NZ302G strain, the isogenic NZ302GR integrant could no longer grow in M17 broth containing 1% galactose. To determine whether NZ302GR could utilize the galactose moiety of lactose, its growth was compared with that of the parental strain NZ302G in medium containing 0.4% lactose. Although NZ302G is Gal⁺, when the strain is grown in medium containing excess lactose (1%), the glucose moiety is preferentially metabolized while galactose is excreted, and presumably acid inhibits growth and prevents subsequent metabolism of the galactose portion of lactose. Reduction of the concentration of lactose to 0.4% eliminates this imbalance while supporting normal growth of strain NZ302G (Table 2). High-performance liquid chromatography analysis of the spent medium indicated that the galactose moiety of lactose was completely metabolized by strain NZ302G while, in contrast, a substantial amount of galactose (72% of the amount that could be hydrolyzed from lactose) was not utilized by the NZ302GR integrant. Moreover, the doubling time of NZ302GR on lactose increased to approximately 1 h in comparison to that of NZ302G, which is 25 min, and the final OD₆₀₀ was less than half that of NZ302G (Table 2).

Effect of galR disruption on transcription of the gal and lac genes. Northern hybridizations were performed to determine the effect of the galR disruption on the transcription of the galR gene and of the gal and lac operons. The primer GALRAS, which is complementary to the 5' end of the sense strand of galR, was chosen since it can hybridize to the single copy of the 3'-deleted galR (galR') in the chromosome that is under control of the galR promoter (Fig. 1D). The 1.2-kb galR transcript, which was only weakly visible for NZ302G growing on glucose, gave a signal of much greater intensity for strain NZ302GR (Fig. 5A). The 5'-deleted copy of galR did not appear to be transcribed since the same result was obtained by probing with a 600-bp fragment internal to the galR gene (Fig. 5B). This constitutive overexpression of the 3'-truncated galR' was also observed for lactose-grown cells (data not shown) and suggests that the product of the galR gene is a negative regulator of its own expression. It should be noted that termination of transcription of galR' in NZ302GR is located at an unknown point within the integration plasmid pNZ684, and therefore it is coincidental that the galR and galR' transcripts appear to have similar sizes.

View larger version (43K): [in this window] [in a new window]   FIG. 5.   (A and B) Northern blot analysis of RNA from S. thermophilus NZ302G and NZ302GR grown on glucose hybridized with the GALRAS primer (A) and galR probe (B). (C and D) Northern blot analysis of RNA from S. thermophilus NZ302G and NZ302R grown on glucose (Gl), lactose (L), or galactose (Ga) hybridized with the galK probe (C) and the lacS probe (D). The mRNA transcript sizes are indicated by arrows.

Effect of galR disruption on GalT and LacZ enzyme activities. To further substantiate the effect of the galR gene disruption on the expression of the gal and the lac operons, enzyme assays were performed. Since the galT gene is the central one of the three genes in the gal operon, GalT activity was used as a measure of the activity of the Leloir pathway enzymes (Table 3). In strain NZ302G, GalT activity is induced three- to fourfold in lactose- and galactose-grown cultures in comparison to growth in glucose-containing medium. In contrast, very low activity was detected in the galR disruption strain NZ302GR when grown on glucose and no significant increase was observed following growth on lactose, indicating a lack of induction of the enzyme. However, the levels of activity of the NZ302GR mutant complemented by pNZ6811 were similar to those of NZ302G on all the sugars tested and thus were also induced on lactose and galactose.

Characterization of the galK promoters for the Gal⁺ mutants. To gain an understanding of the ability of NZ302G to transcribe the gal metabolic genes, in contrast to the parent strain CNRZ 302, a series of hybridization experiments using specific galR and galK probes on the gal region and PCR amplification of the galKTE gene cluster were performed on the Gal⁺ mutant. However, no DNA structural rearrangements were detected within galR or the metabolic gal gene cluster compared to the parent CNRZ 302 strain (data not shown). The galR and galK promoter regions of CNRZ 302 and NZ302G were amplified by PCR in duplicate with the same primers as those used in the primer extension experiments, and both strands of each product were sequenced. The analysis revealed that an extra A residue was inserted in a stretch of adenines preceding the 35 region (nt 1410 to 1414 [Fig. 4A]) of the galK promoter in the NZ302G DNA sequence, resulting in 6 A residues in the mutant in comparison to 5 in the Gal parent. To determine whether other Gal⁺ mutants of CNRZ 302 contained similar mutations in the promoter region, nine more independently isolated Gal⁺ mutants of CNRZ 302 were investigated. DNA sequence analysis of these nine promoters showed that the galK promoter of each Gal⁺ mutant also contained a point mutation. The mutants could be divided into three classes based on their mutations: class I consisted of five mutants with an A insertion, as described above for NZ302G; in class II, the three mutants contained a G-to-T substitution 3 bp preceding the 10 box; and the one mutant in the third class had a G-to-A substitution in the 10 box (Fig. 4A).

galK promoter activity. The GalK activity of the mutant strains was used as a reporter for comparing the expression of the mutated promoters with that of the wild-type. A low level of GalK activity, 10 nmol/min/mg, was detected for the Gal S. thermophilus CNRZ 302 grown on glucose-containing medium, and it increased to 37 nmol/min/mg on lactose. For the Gal⁺ NZ302G and the other five class I mutants, the GalK activity increased from an average 46 nmol/min/mg on glucose to 264 and 305 nmol/min/mg on lactose and galactose, respectively. The activity increased from 53 nmol/min/mg on glucose to 193 and 323 nmol/min/mg on lactose and galactose, respectively, for the class III mutant. The highest activity was observed with the three class II mutants, an average of 68 nmol/min/mg on glucose to 400 and 458 nmol/min/mg on lactose and galactose, respectively. Thus, it is likely that the basal GalK activity has increased in the mutant strains due to promoter-up mutations, which allows sufficient expression of the galK gene on induction for galactose utilization.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This study demonstrates conclusively that S. thermophilus CNRZ 302 possesses the full complement of genes necessary for galactose metabolism despite its Gal phenotype. In many organisms, the galK, galT, and galE genes, which constitute the Leloir pathway of galactose metabolism, may be clustered or organized in a single operon, and the order of these genes within the operon may be highly variable. While the similarity between the deduced primary sequences of the enzymes is very high within the lactic acid bacteria, the genomic organization of the gal clusters and gene order is species specific (17). The gene order for the S. thermophilus galKTE operon and the divergent galR gene is identical to that in S. mutans (2), which reflects the evolutionary relationship between these bacteria. A potential transcriptional regulatory gene, galR, has also been identified in L. casei and is transcribed in the L. casei gal operon galKETRM (6). In contrast, the putative homologues in E. coli, galR and galS, are not linked to the gal operon (56), and regulatory genes have as yet not been identified for other gal operons in lactic acid bacteria such as L. lactis, Leuconostoc lactis, and Lactobacillus helveticus.

The effect of glucose, lactose, and galactose on the galR mRNA levels strongly suggested that GalR was a transcriptional regulator of the S. thermophilus gal operon. Transcriptional analysis of the two galR copies generated following disruption of galR in the Gal⁺ NZ302G indicated that the copy resulting in a C-terminally truncated GalR protein is driven by the galR promoter. The second copy, which lacks most of the DNA binding motif, was not transcribed. This was expected based on the orientation of the galR gene fragment in pNZ684. Since the substantial deletion from the C terminus includes residues contributing to inducer binding and dimerization (55), the truncated S. thermophilus GalR proteins are no longer functional. Northern analysis of NZ302GR confirmed that GalR functions as an activator of transcription for the gal operon and explained the inability of the disruption mutant to use galactose or the galactose moiety of lactose as a carbon source. The very low activity still detected for the GalT enzyme in NZ302GR might be due to a basal transcription level in the absence of GalR. Alternatively, it might reflect a low level of reversion to wild type. When the galR gene was provided in trans, the Gal⁺ phenotype was restored.

The constitutive transcription of the nonfunctional galR gene in the absence of GalR indicates that the latter normally functions as a negative regulator of its own expression. Autoregulation is a common feature in prokaryotic gene regulation strategies. Negative autoregulation has been reported for other members of the LacI-GalR family such as GalS, PurR, and CytR (16, 30, 56). It is noteworthy that the majority of LacI-GalR members function as negative regulators while some positive regulators also belong to this family and some (like CcpA) perform both functions (43, 45, 51).

The expression of the lac operon genes of both S. thermophilus NZ302G and ST11 is induced by growth on lactose- or galactose-containing medium. In both strains, -galactosidase activity could no longer be induced to the usual level when the galR gene was disrupted, confirming that GalR also functions as a transcriptional activator of the lac operon. -Galactosidase is more strongly induced in NZ302G in galactose-containing medium than in lactose-containing medium. This differential gene expression of the lac genes may be due to catabolite repression by the glucose moiety of lactose on the lac operon promoter (51), which will result in repression of the lac operon by lactose but not by galactose. Furthermore, the excretion of galactose by LacS in lactose medium would effectively reduce the availability of inducer in the cell.

The mutation to a Gal⁺ phenotype does not result in constitutive expression of the gal genes that are induced in the presence of lactose and galactose, which strongly suggests that S. thermophilus was Gal⁺ but became Gal in the recent past. While the advantages of the exchange reaction of the lactose transport protein offer a rationale for the observed excretion of galactose (40), the precise mechanism by which the enzymes of the Leloir pathway are suppressed has not been determined. Characterization of the 10 Gal⁺ mutants of CNRZ 302 revealed that a point mutation had occurred in the galK promoter region of every isolate, the majority of which were single-base insertions in a homopolymeric run of adenine residues. Interestingly, a study of the molecular basis for the adaptive response of E. coli populations to conditions of nonlethal selection such as nutrient deprivation also identified single-base variations mainly in short mononucleotide repeats (14, 44), and slipped-strand mispairing was proposed as the responsible mechanism. In the S. thermophilus galK promoter, the G-to-A substitution in the class III mutant results in a 10 box (TACAAT) with greater homology to the 10 consensus (TATAAT) sequence (29). In class II mutants, the G-to-T substitution gives a TG doublet 1 bp upstream of the 10 sequence, which is a feature present in the promoters of gram-positive bacteria (12). This may correspond to the "extended 10" sequence that functions as a 35-independent promoter and requires the TG motif for efficient initiation at such promoters (24). Thus, these substitutions that resemble promoter-up mutations may increase the level of transcription of the gal genes and allow metabolism of galactose. The A insertion in class I mutants may also be a promoter-up mutation, although the reason for the enhanced activity in this case is not so apparent. The extra A increases the size of the inverted repeat preceding the 35 box from 11 to 15 nt (see below). In particular, the intrinsic DNA curvature that is predicted in this region is enhanced by the A insertion (data not shown), and this may result in increased promoter strength. The presence of curved DNA upstream of promoters, of which A tracts appear to be a major determinant, is associated with increased transcription (37).

The CNRZ 302 and SS2 galK promoter fusions to the gusA gene support the hypothesis that mutations in the galK promoter of S. thermophilus CNRZ 302 suppress the expression of the gal genes. Although -glucuronidase activity was not expected from the ST11(pNZ6871) strain since galactokinase activity is barely detectable in CNRZ 302, factors such as the high copy number of this plasmid (12) and the gene dosage effect of the GalR activator are likely to be responsible. The only difference between the pNZ6871 and pNZ6872 plasmids was the G-to-T mutation in the galK promoter of the latter. Very high -glucuronidase activity as a consequence of this promoter-up mutation would result in lethal effects on the host (P. G. G. A. de Ruyter, personal communication). This would explain the reduced frequency of transformation for pNZ6872, the small size of the blue colonies on medium containing X-Gluc, and the instability observed for this plasmid in the S. thermophilus host.

The intergenic region between galR and galK of S. thermophilus consists of 142 bp, which contains the promoter sequences of both genes in a back-to-back configuration (Fig. 4A). An 11-bp inverted-repeat sequence (IR) (nt 1392 to 1413; 5'-TTTTACTA-3', 8 out of 11 matching nt) was detected in this region that could be an operator for GalR. The potentially global regulation by GalR prompted us to search for homologous sequences in the promoters of the lac operon and galR gene of CNRZ 302. Similar 11-bp IRs were found 13 bp upstream of the 35 box in the lac promoter and also overlapping the 10 box of the galR promoter (Fig. 4B). A consensus sequence could be deduced, with the 11-bp half of the IR consisting of a central 3 bp highly conserved portion, A(C/G)T, flanked on either side by four predominantly adenine and thymine bases. It is usually the 40 to 35 region of a promoter that is approached by an activator site; exceptions include the MerR family regulators (10, 35). In contrast, sites for repressors may be located from the 35 to 10 region. When activator proteins are used for repression, which generally occurs in cases of autoregulation, the operators often appear in positions for repression rather than activation. The potential operator sites for GalR conform to these general rules.

The gal genes of S. mutans are homologous to those of S. thermophilus and are organized in a similar divergent orientation (2). Alignment of the nucleotide sequences of the S. thermophilus and S. mutans promoter regions revealed homologous palindromic sequences, as described above (Fig. 4B). It is noteworthy, however, that a single-crossover disruption of the S. mutans galR gene resulted in constitutive expression of galactokinase, indicating that GalR functions as a repressor of the gal operon in this species. In contrast, GalR of S. thermophilus activates the gal and lac operons while repressing its own expression. The presence of these potential operators gives further credence to the hypothesis that repression of the gal operon was caused by recent mutations in the galK promoter.