Glycolysis is one of the main pathways of carbon catabolism in Bacillus subtilis. Although the biochemical activity of glycolytic enzymes has been studied in detail, no information about the expression of glycolytic genes has so far been available in this organism. Therefore, transcriptional analysis of all glycolytic genes was performed. The genes cggR, gapA, pgk, tpi, pgm and eno, encoding the enzymes required for the interconversion of triose phosphates, are transcribed as a hexacistronic operon as demonstrated by Northern analysis. This gapA operon is repressed by the regulator CggR. The presence of sugars and amino acids synergistically results in the induction of the gapA operon. The transcriptional start site upstream of cggR was mapped by primer extension. Transcripts originating upstream of cggR are processed near the 3′ end of cggR. This endonucleolytic cleavage leads to differential stability of the resulting processing products: the monocistronic cggR message is very rapidly degraded, whereas the mRNA species encoding glycolytic enzymes exhibit much higher stability. An additional internal constitutive promoter was identified upstream of pgk. Thus, gapA is the most strongly regulated gene of this operon. The pfk pyk operon encoding phosphofructokinase and pyruvate kinase is weakly induced by glucose. In contrast, the genes pgi and fbaA, coding for phosphoglucoisomerase and fructose-1,6-bisphosphate aldolase, are constitutively expressed.
Bacillus subtilis is capable of using numerous carbohydrates as single sources of carbon and energy. Although substrate-specific enzymes are required for the uptake of the individual carbohydrates and the generation of phosphorylated metabolic intermediates (for a review, see Stülke and Hillen, 2000), only a few central catabolic pathways are involved in the final reactions to provide energy and carbon backbones for anabolism. In B. subtilis, these pathways include glycolysis and the pentose-phosphate shunt as well as the tricarboxylic acid cycle (Fortnagel, 1993; Hederstedt, 1993;Sauer et al., 1996; de Wulf, 1998). Among these pathways, glycolysis is of central importance, because it not only generates precursors for anabolism but also allows energy conservation by substrate-level phosphorylation.
In addition to its important catabolic role, glycolysis is also directly linked to one of the major regulatory systems of the B. subtilis cell: the carbon and energy status of the cell is sensed by a bifunctional kinase/phosphatase, HPrK/P. At high ATP concentrations, HPrK/P phosphorylates HPr, a phosphocarrier of the phosphoenolpyruvate phosphotransferase system (PTS) and its regulatory paralogue Crh at their conserved residues Ser-46. At intermediate ATP concentrations, kinase activity of HPrK/P is triggered by glycolytic intermediates such as fructose-1,6-bisphosphate (Galinier et al., 1997; 1998; Reizer et al., 1998). HPr-Ser-P and Crh-Ser-P are cofactors for the pleiotropic regulator of carbon catabolism in B. subtilis, CcpA (Deutscher et al., 1995; Stülke and Hillen, 2000). Thus, at high glycolytic activity, CcpA can bind to its target sites and repress or activate catabolic operons. Although CcpA is a repressor of genes and operons involved in the utilization of secondary carbon sources, it directly or indirectly activates the expression of genes involved in overflow metabolism, ammonium assimilation and glycolysis itself (Grundy et al., 1993; Faires et al., 1999; Presecan-Siedel et al., 1999; Tobisch et al., 1999).
The regulation of genes and operons of the specific pathways of carbon catabolism by substrate-specific induction and the general mechanism of carbon catabolite repression has been extensively studied in B. subtilis (Stülke and Hillen, 2000). Similarly, the control of activity of glycolytic enzymes has been the subject of intensive studies, and the phenotypes of mutants affecting glycolytic genes have been analysed (Fortnagel, 1993; Nakano et al., 1999; Fry et al., 2000). In contrast, not much work has been devoted to the regulation of synthesis of glycolytic enzymes in B. subtilis. The genes encoding glycolytic enzymes were all identified. With the exception of pgi and fbaA encoding phosphoglucoisomerase and fructose-1,6-bisphosphate aldolase, respectively, all glycolytic genes are clustered in functionally related units. (Leyva-Vazquez and Setlow, 1994; Kunst et al., 1997; Tobisch et al., 1999; see Fig. 1). Recently, we have provided evidence that synthesis of the enzymes encoded by the gapA pgk tpi pgm eno gene cluster is inducible by glucose (Tobisch et al., 1999). Moreover, the presence of two genes encoding glyceraldehyde-3-phosphate dehydrogenase, gapA and gapB, was demonstrated. The enzyme encoded by gapA is active in glycolysis, whereas the gapB gene product acts in gluconeogenesis. Expression of gapA is negatively controlled by a novel repressor, CggR, and is induced by glucose (Fillinger et al., 2000). In contrast, gapB expression is strongly repressed by glucose (Fillinger et al., 2000; Yoshida et al., 2001). The first step in glycolysis, the phosphorylation of glucose, is performed concomitantly with its transport by the PTS. The gene encoding the glucose-specific component of the PTS, ptsG, forms an operon with the ptsH and ptsI genes encoding the general proteins of the PTS (Stülke et al., 1997). Expression of the ptsGHI operon is induced by glucose and depends on a transcriptional antiterminator (Bachem and Stülke, 1998).
In this study, we investigated the regulation of genes encoding glycolytic enzymes at two levels: (i) transcription of the individual gene loci was studied by Northern blot analysis; and (ii) activity and regulation of the promoters of the glycolytic transcription units were determined using transcriptional fusions. Our results indicate that synthesis of the enzymes that catalyse irreversible reactions is induced by glucose, whereas enzymes that are involved in both glycolysis and gluconeogenesis are synthesized in significant amounts in both the absence and the presence of glucose. Expression of the gapA operon is controlled by two different regulators, CcpA and CggR. Moreover, this operon is synthesized as a precursor containing the promoter-proximal cggR repressor gene and is subject to mRNA processing. Our findings indicate that the transcripts specifying glycolytic genes are stable, whereas the cleaved-off monocistronic cggR mRNA is highly unstable. Most probably, this differential segmental mRNA stability ensures the synthesis of larger amounts of glycolytic enzymes compared with the CggR repressor protein.
Transcriptional organization of glycolytic genes encoding enzymes required for hexose phosphate interconversion
Glucose phosphorylation, the first step in glycolysis, is mediated by the glucose-specific PTS encoded by the ptsGHI operon. Transcription of the ptsGHI operon was the subject of a previous study (Stülke et al., 1997). As observed previously, a > 30-fold induction of ptsG expression by glucose was found (Table 1).
Table 1. Expression of glycolytic genes and operons in B. subtilis as judged from transcriptional fusions to the lacZ gene.
a. Bacteria were grown in CSE minimal medium. All strains contain transcriptional fusions of the presumed promoter regions of the indicated genes to a promoterless lacZ gene. β-Galactosidase activities are given as units mg−1 protein. Experiments were carried out at least three times. Representative results from one series are shown.
The pgi gene is strongly transcribed as a monocistronic transcription unit with a transcript size of 1.5 kb (Figs 1 and 2A). This correlates well with the size of the pgi open reading frame (ORF) of 1.35 kb and the presence of a putative transcriptional terminator downstream of pgi. The pgi transcript is present in similar amounts in cultures grown with or without glucose. In addition to the monocistronic pgi transcript, a minor transcript of 2.4 kb was identified. This transcript was less abundant in cells grown in the presence of glucose. According to the gene arrangement, this transcript may correspond to a pgi yugMN transcriptional unit. To study the activity of the pgi promoter, a 288 bp region (221 bp upstream of the ATG translational start codon and 67 bp of the pgi coding region) was transcriptionally fused to a promoterless lacZ gene. As shown in Table 1, this fusion was expressed irrespective of the presence of glucose in the medium, thus confirming constitutive expression of pgi.
The pfk gene encoding phosphofructokinase is clustered with pykA coding for pyruvate kinase. Transcription of this gene cluster was studied using a riboprobe specific for pfk and revealed the presence of a 2.8 kb mRNA (Fig. 2B). This size fits with the idea of a pfk pykA operon and with the presence of transcriptional terminators immediately upstream and downstream of the operon (Moszer et al., 1995). As judged from Northern analysis, expression of the pfk pykA operon may be weakly inducible by glucose. Under inducing conditions, a weakly expressed larger transcript became visible. This transcript may encompass the ytzA gene (Fig. 1). A transcriptional pfk–lacZ fusion was constructed by cloning a 364 bp fragment of the presumptive promoter region (263 bp upstream of the ATG translational start codon and 101 bp of the pfk coding region) upstream of the lacZ gene. The resulting strain, GP315, was grown in CSE medium with or without glucose. The lacZ expression driven from the pfk promoter was detectable under both conditions; however, a twofold elevation in expression was observed in the presence of glucose (Table 1). These results confirm the weak induction of the pfk pykA transcript found in Northern analysis.
The fbaA gene encoding fructose-1,6-bisphosphate aldolase was part of two transcripts. Among these, a monocistronic mRNA of 1.0 kb was dominant (Fig. 2C). A minor transcript of 1.7 kb may contain the downstream gene ywjH in addition to fbaA (see Fig. 1). This conclusion is derived from the finding that the upstream spo0F gene is transcribed monocistronically (Trach et al., 1988). The constitutive expression of fbaA seen in Northern analysis was confirmed using an fbaA–lacZ fusion (Table 1). The promoter was present in a 436 bp fragment directly upstream of the ATG start codon.
Transcriptional organization of the triose phosphate interconversion gene cluster
The gapA, pgk, tpi, pgm and eno genes encoding glyceraldehyde-3-phosphate dehydrogenase, phosphoglycerate kinase, triose phosphate isomerase, phosphoglycerate mutase and enolase, respectively, form a gene cluster. Three transcriptional units (cggR gapA, gapA and pgk tpi pgm eno) were previously proposed for this gene cluster (Tobisch et al., 1999). RNA isolated using the modified ‘mechanical disruption protocol’ (Hauser et al., 1998) was used to refine the transcriptional analysis. Four riboprobes specific for cggR, gapA, pgk and eno were applied (Figs 1 and 3).
Northern analysis using the cggR probe revealed three mRNAs of 1.0 kb, 2.2 kb and about 7.4 kb. However, the largest transcript is detectable only as a very faint band (Fig. 3A). All three mRNAs are strongly inducible by glucose, and induction depends on a functional CcpA gene product, as all transcripts are only weakly expressed in the ccpA mutant. Based on transcript sizes and Northern analyses using the other riboprobes, the 2.2 kb mRNA corresponds to a bicistronic cggR gapA transcript that is terminated at the terminator between gapA and pgk(Fig. 1). The 7.4 kb mRNA represents a hexacistronic cggR gapA pgk tpi pgm eno full-length transcript of the operon terminating at the terminator structure downstream of eno(Fig. 1). The origin of the 1.0 kb mRNA remained unclear at first. To localize the putative promoter upstream of cggR, a transcriptional cggR–lacZ fusion was constructed using a 351 bp fragment (313 bp upstream of the ATG start codon and 38 bp of the coding sequence). This fragment contains a potential catabolite-responsive element cre (see Fig. 4B). Synthesis of β-galactosidase was assayed in strain GP313 after growth in CSE minimal medium with or without glucose. A fourfold increase in expression driven by the cggR promoter was observed in the presence of glucose (Table 1). This indicates that regulated promoter activity is present in the DNA fragment used to construct the cggR–lacZ fusion. To determine the exact position of the transcriptional start site of cggR, a primer extension experiment was carried out. Primer HL27 hybridized to nucleotides −67 to −48 with respect to the translational start point (see Fig. 4B). Transcription was initiated at a single A located 95 or 96 bp upstream of the translational start point of cggR(Fig. 4A). The start site is preceded by a reasonably conserved promoter sequence (5′-TTGAAT-3′ separated by 17 bp from the sequence motif 5′-TAAAAT-3′; see Fig. 4B). As observed by Northern analysis and the cggR–lacZ fusion, transcription was induced by glucose (Fig. 4A).
Northern analysis with a gapA-specific riboprobe allowed the identification of three mRNAs (Fig. 3B). The strongest signal was observed for a monocistronic gapA transcript of 1.2 kb. In addition, a 2.2 kb mRNA was detected in very low amounts compared with the monocistronic transcript. This mRNA represents the bicistronic cggR gapA transcript already detected using the cggR probe. The third transcript of about 6.2 kb corresponds to a pentacistronic message gapA pgk tpi pgm eno. This conclusion can be drawn, because the same transcript was detected with riboprobes specific for pgk and eno (see Fig. 3C and D). In the wild-type strain, all gapA-specific transcripts were inducible by glucose. As seen for cggR, induction depended on a functional ccpA gene product. The presence of the monocistronic and pentacistronic transcripts suggested the existence of a promoter upstream of gapA. Therefore, a 378 bp DNA fragment encompassing the 3′ end of cggR, the cggR gapA intergenic region and the 5′ end of gapA (325 bp upstream of the gapA ATG start codon and 53 bp of the gapA coding region) was cloned in front of a lacZ gene. The gapA–lacZ fusion was introduced into the chromosome of B. subtilis, and the β-galactosidase activity of the resulting strain GP136 was determined after growth in CSE in the presence or absence of glucose. Under both conditions, no expression of the fusion was observed. Transcription of gapA may thus originate from a promoter located upstream of cggR (see below).
Northern blots performed with riboprobes specific for pgk and eno gave identical results (Figs 3C and D), suggesting that both genes are part of the same transcriptional units. As the tpi and pgm genes are located between pgk and eno, we may suppose that all four genes are co-transcribed (see Fig. 1). This was reflected by the detection of a 4.8 kb transcript corresponding to a tetracistronic mRNA initiated upstream of pgk. A second larger transcript of 6.2 kb results from an mRNA spanning from gapA to eno (see Fig. 1 and 3). As observed with the gapA riboprobe, the 6.2 kb mRNA was strongly inducible by glucose, whereas the 4.8 kb transcript was constitutively synthesized. The promoter region of pgk was studied by a fusion to the lacZ gene spanning from −381 to +171 relative to the start codon of pgk. The expression was studied in B. subtilis GP317 after growth in CSE minimal medium. The promoter was active at similar levels in both the presence and the absence of glucose, confirming the existence of an internal constitutive promoter upstream of pgk. The precise position of this promoter was determined by primer extension analysis. Primer FMD10 hybridized to nucleotides +39 to +13 with respect to the translational start point (see Fig. 5B). Transcription was initiated at a single A located 255 bp upstream of the translational start point of pgk(Fig. 5A). The start site is preceded by a reasonably conserved promoter sequence (5′-TTGGTT-3′ separated by 17 bp from the sequence motif 5′-TATAAT-3′; see Fig. 5B). Interestingly, the promoter is located immediately upstream of the transcriptional terminator between gapA and pgk(Fig. 5B). As observed by Northern analysis and the pgk–lacZ fusion, transcription was constitutive and not affected by the cggR deletion (Fig. 5A).
Characterization of a processing site at the 3′ end of cggR
Northern blot experiments with the gapA-specific riboprobe resulted in the identification of transcripts of 1.2 kb and 6.2 kb. These transcripts correspond to a monocistronic gapA and a pentacistronic gapA pgk tpi pgm eno mRNA respectively (see Fig. 6A). However, there was no promoter activity detectable immediately upstream of gapA (see Table 1). These data suggest a post-transcriptional cleavage event of the mRNA between cggR and gapA. Primer extension was applied to confirm a processing site upstream of gapA. First, the 5′ end of gapA transcript was mapped using total RNA isolated from B. subtilis 168 grown in CSE with or without glucose. As shown in Fig. 6B, two 5′ ends were detected, a G and a U residue at positions −65 and −64, respectively, upstream of the start AUG of gapA. To rule out the possibility that the mapped 5′ ends represent a pausing site of the reverse transcriptase, we used a cggR gapA RNA synthesized in vitro. Reverse transcription with this RNA template did not result in any specific signal, indicating that the signals observed with the in vivo RNA indeed correspond to the real 5′ end of the monocistronic 1.2 kb gapA as well as of the pentacistronic 6.2 kb gapA pgk tpi pgm eno mRNA (Fig. 6B). As no promoter is present in the corresponding DNA region, these mRNAs may represent the 3′-proximal processing products generated by endoribonucleolytic cleavage of the bicistronic 2.2 kb cggR gapA and the 7.4 kb full-length primary transcripts respectively. Thus, the monocistronic 1.0 kb mRNA detected by the cggR probe can be interpreted as the 5′-proximal processing product.
The inspection of the region of the presumptive processing site revealed that the cleavage may occur at the end of the cggR gene interfering with the integrity of the cggR message. The potential folding of the RNA around the processing site according to the Zuker algorithm (Zuker, 1989) is shown in Fig. 6C. Interestingly, the cleavage site is located in a short AU-rich single-stranded region between two reasonably stable stem–loop structures. These features are typical of RNase E processing sites in Escherichia coli (McDowall et al., 1994; Condon et al., 1997).
Differential stabilities of mRNAs specified by the gapA operon
The processing event at the 3′ end of cggR generates the 1.0 kb cggR mRNA lacking the very 3′ end of the gene (see Fig. 6C) and the two mRNA species of 1.2 kb and 6.2 kb. We therefore asked for the function of the processing. It can be assumed that the glycolytic enzymes are required in much higher amounts compared with the CggR repressor. This idea was supported by the fact that the endonucleolytic cleavage occurs in the coding sequence of cggR, most probably initiating inactivation of the mRNA by exonucleolytic degradation. According to this model, the processing reaction might generate products exhibiting differential mRNA stabilities, resulting in varying protein amounts. This assumption was supported by the observation that all five glycolytic enzymes encoded by the gapA operon were identified in proteome analyses, whereas the CggR repressor was not detected (Tobisch et al., 1999).
To check for the postulated differential mRNA stabilities, the decay of the mRNAs specified by the gapA operon was compared directly. Northern analyses using RNA prepared after the addition of rifampicin, which prevents initiation of transcription, were carried out (Fig. 7). To allow direct comparison of the three images, we took advantage of the fact that there is always one mRNA species detectable by two different probes in each case. The bicistronic 2.2 kb cggR gapA primary transcript is detectable using cggR and gapA probes, and the 6.2 kb processing product can be detected by the gapA and eno probes. Luminographs were adjusted to achieve comparable signal intensities of two corresponding mRNAs in different images. The use of a LumiImager (Roche) for image analysis ensured that all signal intensities were in the dynamic range and not yet saturated.
Using the cggR probe, weak signals representing the 1.0 kb monocistronic cggR processing product and the bicistronic 2.2 kb cggR gapA primary transcript were detected under control conditions. Already, 2 min after the addition of rifampicin, both mRNA species were completely absent (Fig. 7A).
Using the gapA probe, the 2.2 kb cggR gapA primary transcript showed a signal intensity comparable with the cggR Northern before rifampicin addition. Again, the signal had disappeared 2 min after inhibition of transcription. The 1.2 kb gapA processing product was detected under control conditions as a very strong band and slowly decayed after rifampicin addition (Fig. 7B). As the bicistronic precursor showed a similar signal intensity in the cggR and the gapA Northern analyses, the bands representing the two processing products could be compared directly. This demonstrated clearly that, under steady-state conditions, the 3′gapA processing product was present in the cell in much larger amounts compared with the 5′cggR product, supporting the model of the regulation of protein amounts by differential segmental mRNA stability. In addition, the low stability of the 5′cggR product is underlined by the fact that the weak band representing this mRNA is actually a mixture of the product cleaved off from the bicistronic and hexacistronic primary transcripts. The pentacistronic 6.2 kb processing product was present as a distinct band decaying with kinetics comparable with the monocistronic gapA mRNA (Fig. 7B).
Using the eno probe, a result similar to that with the gapA riboprobe was obtained for the pentacistronic 6.2 kb processing product. Furthermore, the 4.8 kb mRNA initiated upstream of pgk decayed with related kinetics, but at a slightly slower rate (Fig. 7C).
We determined the half-lives for the 1.2 kb and the 6.2 kb processing products as well as for the 4.8 kb primary transcript (two independent RNA preparations). The calculated half-lives of 3.5, 3.0 and 4.5 min, respectively, turned out to be similar. Owing to the fact that the monocistronic cggR processing product was not detectable 2 min after the addition of rifampicin, it was not possible to perform a similar analysis in this case.
Regulation of the cggR promoter activity
Transcription of the hexacistronic gapA operon driven by the promoter in front of cggR is induced by glucose (see above). Similar results were obtained with a translational cggR–lacZ fusion present in B. subtilis GP303 (see Table 2). As glycolysis is a common pathway for the catabolism of many sugars in B. subtilis (Stülke and Hillen, 2000), we tested whether the cggR promoter was also inducible by other sugars (Table 2). Expression was most strongly inducible with glucose, the preferred carbon source for B. subtilis. However, the fusion was also induced with other glycolytically metabolized sugars such as fructose, mannose, mannitol, glucitol and glycerol. Interestingly, gluconate and arabinose, but not ribose, which are catabolized via the pentose phosphate pathway, also did induce the cggR–lacZ fusion.
Table 2. Regulation of cggR expression by the carbon source.
a. Bacteria were grown in CSE minimal medium. Carbon sources were added to a final concentration of 0.5%. β-Galactosidase activities are given as units mg−1 protein. Experiments were carried out at least three times. Representative results from one series are shown.
In a previous work, the cggR gene product was proposed to be the repressor for the gapA operon based on data with a cggR disruption mutant (Fillinger et al., 2000). However, polar effects of that mutation could not be ruled out. Therefore, we constructed a cggR in frame deletion mutant. To confirm the absence of polar effects of the cggR deletion, we performed a Northern analysis with the mutant strain GP311. In this strain, the pentacistronic gapA pgk tpi pgm eno mRNA showed strong constitutive expression, whereas synthesis of the tetracistronic pgk tpi pgm eno transcript was not affected (data not shown). Based on the promoter analysis of the gapA operon, we hypothesized that the strong transcription in the cggR deletion mutant originated from the promoter located upstream of cggR, which was not affected by the mutation. This idea was tested by determining the activity of the cggR–lacZ fusion in the ΔcggR mutant strain GP311. High constitutive expression of β-galactosidase derived from the cggR promoter was observed in the mutant strain (Table 2). These findings confirm the idea that CggR acts as a repressor at its own promoter.
The experiments described above indicated that different sugars that may all yield glyceraldehyde-3-phosphate induce transcription of the gapA operon. As CggR is the negative regulator of the gapA operon, we may suppose that these sugars or a derived signal molecule interfere(s) with repression by CggR. However, although strong constitutive expression was found in the ΔcggR mutant strain, none of the sugars tested induced the operon to a similar level. This apparent lack of complete induction was addressed by analysing the effect of different glucose concentrations on the expression of the cggR–lacZ fusion. In the range of concentrations tested (2.75–82.5 mM), no effect on cggR induction was observed (data not shown). This obvious contradiction led us to speculate that full glycolytic activity may be required under conditions in which glycolysis is the only pathway to provide the cell with energy. This may be the case if the cells are supplied with amino acids in addition to glucose. The TCA cycle is inactive under such conditions (Hederstedt, 1993). We tested whether the induction of cggR transcription by glucose could be stimulated further in the presence of casein hydrolysate as a source of amino acids (Fig. 8). Indeed, the addition of casein hydrolysate increased cggR expression to a level similar to that seen in the ΔcggR mutant. Moreover, the presence of casein hydrolysate alone in CSE resulted in significant derepression of the cggR promoter. Stimulation of cggR expression by amino acids may depend on CggR or may result from an additional regulatory event. To distinguish between these possibilities, the effect of casein hydrolysate on cggR promoter activity was assayed in the ΔcggR mutant GP311. Only a weak further increase in β-galactosidase synthesis was observed (see Fig. 8). To verify that the amino acids present in the casein hydrolysate caused the complete derepression of cggR transcription, a synthetic mixture of amino acids that resembled the amino acid composition of casein hydrolysate was tested for its effect on the cggR promoter (Fig. 8) The addition of this amino acid mixture to CSE medium containing glucose resulted in full induction of the cggR–lacZ fusion, as observed with casein hydrolysate. We may therefore propose that the amino acids, a metabolic derivative of them or a signalling molecule may act in addition to the sugar signal to yield full induction of the CggR-regulated gapA operon.
Induction of the gapA operon by glucose depends on a functional ccpA gene (Tobisch et al., 1999; Fillinger et al., 2000; Fig. 3A and B). Moreover, a potential catabolite-responsive element cre located upstream of the cggR promoter may be a target for CcpA. We therefore attempted to verify the role of CcpA in expression of the gapA operon and to study the role of the presumptive cre site. Expression of the cggR–lacZ fusion in a wild-type (GP303) and a ccpA mutant strain (GP307) was compared (Table 3). Although glucose induced the fusion in the wild-type background, almost no induction occurred in the ccpA mutant strain. This finding confirms a role for ccpA in the control of transcription from the cggR promoter. A mutation in the proposed cre that affected bases known to be important for cre function was introduced into the cggR promoter fragment fused to the lacZ gene in GP304 (see Fig. 4B). This mutation had no effect on glucose regulation (Table 3). Therefore, the proposed element is not a target for positive regulation of the cggR promoter by CcpA. Thus, there might be another target for CcpA, or the effect of CcpA might be indirect. This question was addressed using a ccpA cggR double mutant. If CcpA were required directly for glucose induction, one would expect loss of full derepression in the double mutant strain. However, this mutant (GP312) exhibited high constitutive expression of the cggR–lacZ fusion similar to the cggR mutant strain (see Table 3). This result is consistent with the idea that the effect of the ccpA mutation on the expression of the gapA operon is indirect.
Table 3. Regulation of the cggR promoter by glucose and the effect of different cis- and trans-acting mutations.
a. Bacteria were grown in CSE minimal medium with or without 0.5% glucose. β-Galactosidase activities are given as units mg−1 protein. Experiments were carried out at least three times. Representative results from one series are shown.
The utilization of sugars and sugar derivatives in B. subtilis as a source of carbon and energy requires their conversion to a few common metabolic intermediates (glucose-6-phosphate, fructose-6-phosphate, fructose-1-phosphate, ribose-5-phosphate) that can be metabolized further by either glycolysis or the pentose-phosphate pathway. The mechanisms used to introduce the sugars into the central metabolism and the regulation of the sugar-specific catabolic genes and operons have been analysed in great detail (Stülke and Hillen, 2000). Here, we studied the regulation of one of the central catabolic pathways, glycolysis.
Glycolytic enzymes can be divided into two groups based on their reactions and the regulation of expression of the respective genes: the first group of enzymes catalyses irreversible reactions, whereas the second group catalyses reversible reactions that are part of both glycolysis and gluconeogenesis. Members of the former group are the glucose permease, phosphofructokinase, glyceraldehyde-3-phosphate dehydrogenase and pyruvate kinase, encoded by ptsG, pfk, gapA and pykA respectively. Expression of these genes is inducible by glucose, providing a means for regulating the direction of carbon flow. Glucose induction of these genes was also observed in E. coli and Lactococcus lactis (Kimata et al., 1997; Charpentier et al., 1998; Luesink et al., 1998;Plumbridge, 1998). In contrast, the genes encoding enzymes that are active in both directions are expressed in the presence or absence of glucose. The co-ordinated regulation of the direction of carbon flow was also observed in E. coli. The mechanisms of this regulation differ, however, from those described here for B. subtilis. The Mlc regulator, the glucose permease of the PTS and the Crp–cAMP complex seem positively to control expression of ptsG, gapA and gapB pgk (Kimata et al., 1997; Charpentier et al., 1998; Plumbridge, 1998; Nam et al., 2001). In addition, the pleiotropic regulatory protein Cra is crucial for the differential control of glycolysis, the Entner–Doudoroff pathway, gluconeogenesis and the TCA cycle in E. coli (Ramseier et al., 1995).
Among the glycolytic genes in B. subtilis induced by glucose, the strongest regulation was observed for ptsG, mediating the initial glucose transport and phosphorylation, and for gapA, encoding glyceraldehyde-3-phosphate dehydrogenase. It is interesting to note that at least three different mechanisms of glucose induction operate to control these genes. Expression of ptsG depends on the transcriptional antiterminator GlcT, which is in turn controlled by the ptsG gene product, the glucose permease (Stülke et al., 1997; Bachem and Stülke, 1998). In contrast, expression of the gapA operon involves two transcriptional regulators, CcpA and CggR. The data provided here and in a previous work (Fillinger et al., 2000) suggest that CggR is a direct repressor of the gapA operon, whereas the effect of CcpA may be indirect. The two regulators seem to sense the availability of sugars. CcpA is active in transcriptional regulation upon binding of one of its cofactors, HPr-Ser-P or Crh-Ser-P. These cofactors signal high glycolytic activity (Deutscher et al., 1995; Galinier et al., 1997; Stülke and Hillen, 2000). In contrast, CggR seems to respond synergistically to two different signals, a catabolic signal derived from the presence of sugars and an anabolic signal derived from amino acid metabolism. The presence of the individual signals results in partial derepression of the gapA operon. Full induction occurs only if both signals are present. Moreover, a weak additional stimulation of the cggR promoter by glucose occurs in the cggR mutant strain. It is therefore possible that another regulatory factor may be involved in the control of cggR promoter activity. This complex mode of regulation might integrate the needs of glycolysis and the TCA cycle under the different conditions. In the absence of carbohydrates, the cell needs to perform gluconeogenesis rather than glycolysis. If sugars are present, glycolysis serves to produce energy by substrate-level phosphorylation and pyruvate to generate the precursor for anabolism in the TCA cycle. In the presence of both sugars and amino acids, the TCA cycle is switched off (Hederstedt, 1993), and glycolysis is required exclusively for energy production. The availability of amino acids may allow very fast growth and thus require high energy production via high glycolytic activity. Under these conditions, the resulting pyruvate is converted to acetate and acetoin by the enzymes of overflow metabolism. Expression of these enzymes is activated in the presence of glucose by CcpA (Tobisch et al., 1999). Sensing of dual signals was also observed in the regulation of the TCA cycle, which is repressed by glucose. As found for glycolysis, CcpA seems to be involved in the glucose regulation of several TCA cycle enzymes. The actual mechanism is, however, unknown, and the role of CcpA may be indirect (Tobisch et al., 1999). In addition, the expression of two key enzymes of the TCA cycle, citrate synthase and aconitase, is repressed by the CcpC repressor (Jourlin-Castelli et al., 2000). Glucose repression of aconitase is even more severe in the presence of glutamate (Rosenkrantz et al., 1985). The observation that the same regulatory elements, the repressor CcpC and its target in the citB promoter region, are required for synergistic repression by both compounds (Fouet and Sonenshein, 1990; Jourlin-Castelli et al., 2000) suggests that CcpC activity might also be controlled by the combined action of a catabolic and an anabolic signal.
Another interesting feature in the control of the glycolytic gapA operon is its transcriptional organization. Three primary transcripts are specified by this operon: the bicistronic cggR gapA transcript; the hexacistronic cggR gapA pgk tpi pgm eno transcript; and the tetracistronic pgk tpi pgm eno transcript. The two former mRNAs are initiated at a well-conserved σA-dependent regulated promoter upstream of cggR, whereas the third transcript starts at an internal constitutive promoter located upstream of pgk. This arrangement allows highly regulated synthesis of gapA, whereas expression of the downstream genes is only weakly regulated. Although the promoter of the gapA operon was mapped upstream of cggR, the most abundant transcripts do not contain the cggR message. Therefore, an mRNA processing event was postulated, and the processing site mapped upstream of gapA. Interestingly, the cleavage resembles processing sites used by RNase E in E. coli. However, no homologue of RNase E is present in B. subtilis. An enzymatic activity similar to that of RNase E was proposed to be involved in processing of the thrS leader mRNA in B. subtilis. As the processing occurs at the very 3′ end in the open reading frame (ORF) of cggR, we may suppose that no functional CggR protein can be formed after processing. In addition, the cleaved-off cggR transcript is highly unstable, whereas the 3′ processing products decay at a significantly lower rate. This mechanism may reflect the different needs for the products of the gapA operon. We may suppose that CggR is needed in the cell only in very small amounts. Moreover, its presumed site of action in the cggR promoter region is very close to the site at which it is produced, allowing the establishment of a sufficient local concentration of CggR at a small total concentration of the protein. In contrast, the glycolytic enzymes encoded by the gapA operon are required at high concentrations, as they are crucial for the production of energy in the cell. The differential segmental stability is thus a mechanism ensuring that the individual proteins are synthesized in the required amounts even if their genes are transcribed as a single primary transcript. mRNA processing to cleave off a promoter-proximal repressor gene was observed previously for the primary transcript of the B. subtilis dnaK operon, which is processed after the hrcA repressor gene (Homuth et al., 1999). Moreover, the glycolytic gap pgk operon of Zymomonas mobilis is processed between the two genes, and the resulting gap transcript is very stable. This suggests that more glyceraldehyde-3-phosphate dehydrogenase is produced compared with phosphoglycerate kinase, thus compensating for the lower catalytic efficiency of the former enzyme (Eddy et al., 1991; Burchhardt et al., 1993). The requirement for different amounts of individual mRNAs from a primary transcript may be a common physiological problem and the differential segmental mRNA stability a common strategy to cope with it. Similarly, the gnt operon and the presumptive fruRBA operon of B. subtilis have promoter-proximal repressor genes (Fujita and Fujita, 1987; Reizer et al., 1999). The exact transcriptional organization of these and other operons remains to be studied in more detail.
Bacterial strains and growth conditions
The B. subtilis strains used in this work are listed in Table 4. E. coli DH5α (Sambrook et al., 1989) was used for cloning experiments. B. subtilis was grown in CSE or ASM minimal media as indicated (Stülke et al., 1993;Faires et al., 1999). The media were supplemented with auxotrophic requirements (at 50 mg l−1) and carbon sources as indicated. Casein hydrolysate (CAA) and a synthetic amino acid mixture (AA) were used as additional nitrogen sources. The AA solution that mimics CAA (1% solution) was composed as follows: alanine 0.02%, arginine 0.025%, cysteine 0.0002%, glycine 0.01%, histidine 0.01%, isoleucine 0.025%, leucine 0.05%, methionine 0.02%, phenylalanine 0.025%, proline 0.06%, serine 0.035%, threonine 0.02%, tyrosine 0.005%, valine 0.04%, lysine 0.05%, and aspartic acid 0.05%. E. coli was grown in LB medium, and transformants were selected on plates containing ampicillin (100 µg ml−1). LB, SP and CSE plates were prepared by the addition of 17 g l−1 Bacto agar (Difco) to LB, SP or C medium respectively.
Transformation of E. coli and plasmid DNA extraction was performed using standard procedures (Sambrook et al., 1989). Restriction enzymes, T4 DNA ligase and DNA polymerases were used as recommended by the manufacturers. DNA fragments were purified from agarose gels using the Nucleospin Extract kit (Macherey and Nagel). Pfu DNA polymerase was used for the polymerase chain reaction (PCR) as recommended by the manufacturer. DNA sequences were determined using the dideoxy chain termination method (Sambrook et al., 1989).
Transformation and characterization of phenotype
Bacillus subtilis was transformed with plasmid or chromosomal DNA according to the two-step protocol described previously (Kunst and Rapoport, 1995). Transformants were selected on SP plates containing spectinomycin (Spc; 100 µg ml−1), chloramphenicol (Cm; 5 µg ml−1) or kanamycin (10 µg ml−1). Quantitative studies of lacZ expression in B. subtilis in liquid medium were performed with cell extracts as described previously (Kunst and Rapoport, 1995).
Transcriptional fusions of promoter fragments to a promoterless lacZ gene were constructed using the vector pAC6 (Stülke et al., 1997), which allows the introduction of the fusion into the amyE locus of B. subtilis. Plasmids pGP510, pGP511, pGP601, pGP509, pGP604 and pGP514 containing pgi–lacZ, pfk–lacZ, fbaA–lacZ, cggR–lacZ, gapA–lacZ and pgk–lacZ transcriptional fusions, respectively, were constructed as follows. Fragments containing the promoter regions were amplified by PCR using specific primer sequences (see Supplementary material). The PCR products were digested with EcoRI and BamHI and ligated with pAC6 linearized with the same enzymes.
Plasmid pAC7 (Weinrauch et al., 1991) was used to construct translational fusions with the lacZ gene. Plasmid pGP503 containing a cggR–lacZ fusion was created by cloning a 376 bp EcoRI–BamHI PCR fragment covering the region upstream of cggR and the first 14 codons of the cggR gene, into pAC7. Plasmid pGP504 containing a mutant variant of the suggested cre upstream of cggR (see Fig. 4B) was constructed by the same approach. Primer HL15 carrying the mutation was used in this case to generate the PCR product.
Plasmids used as templates for the primer extension experiments were constructed as follows. A 1452 bp EcoRI–XbaI PCR product (primers HL14 and HL24) containing the cggR gene with its promoter region was cloned into pBluescript-II SK (Stratagene) to yield pGP508. This plasmid was used to map the start point upstream of cggR. The DNA template for mapping the processing site upstream of gapA, a 432 bp EcoRI–BamHI PCR fragment (primer FMD11 and HL50), was introduced into pAC6 to create plasmid pGP512. For mapping the start site upstream of pgk, plasmid pGP513 was constructed by cloning a 568 bp EcoRI–BamHI PCR fragment (primers HL51 and HL52) into plasmid pAC7.
Construction of a cggR in frame deletion mutant
An in frame deletion of the cggR gene was generated using plasmid pGP507. This plasmid was constructed by cloning a 376 bp EcoRI–BamHI PCR fragment (primers HL14 and HL16) into pGP107 (Bachem and Stülke, 1998) yielding pGP502. Downstream of this fragment, a 392 bp BamHI–XbaI PCR fragment (primers HL23 and HL24) covering the last 29 codons of cggR, the downstream intergenic region and part of the gapA gene was inserted. The cggR deletion strain GP311 was obtained after co-transformation of B. subtilis 168 with pGP507 and chromosomal DNA of GP303. Transformants were selected for kanamycin resistance conferred by the aphA3 gene linked to the cggR–lacZ fusion and screened for expression of the fusion in the absence of glucose on CSE plates with kanamycin and Xgal. Because of the deleted repressor CggR, the cggR–lacZ fusion was expected to show a high expression level on these plates. Positive (blue) candidates were selected and verified by PCR.
Northern blot analysis
Preparation of total RNA from B. subtilis and Northern blot analysis were carried out as described by Homuth et al. (1997). To isolate high-quality RNA suited for the detection of long transcripts and their precursors, RNA was prepared by the modified ‘mechanical disruption protocol’ described by Hauser et al. (1998). The cells were harvested at the exponential phase. For RNA preparation, 50 ml of cells was used. After mechanical cell disruption, the frozen powder was instantly resuspended in 3 ml of lysis buffer [4 M guanidine isothiocyanate, 0.025 M sodium acetate, pH 5.2, 0.5% N-laurylsarcosine (w/v)]. Subsequently, total RNA extraction with acid–phenol solution was carried out as described by Homuth et al. (1997). Digoxigenin (DIG) RNA probes were obtained by in vitro transcription with T7 RNA polymerase (Roche Diagnostics) using PCR-generated fragments as templates. The primer sequences used for PCR are available at http://www.blackwell-science.com/products/journals/suppmat/mole/mole2523/mmi2523sm.htm. Reverse primers contained a T7 RNA polymerase recognition sequence. In vitro RNA labelling, hybridization and signal detection were carried out according to the manufacturer's instructions (DIG RNA labelling kit and detection chemicals; Roche Diagnostics). The sizes of the RNA molecular weight markers (Gibco BRL) were as follows: 9.49, 7.46, 4.40, 2.37, 1.35 and 0.24 kb.
Oligonucleotides (100 ng) were labelled with 10 U of polynucleotide kinase and 50 µCi of [γ-32P]-ATP (Amersham Pharmacia Biotech). To generate the respective cDNAs, Superscript II RNase H– reverse transcriptase (Gibco BRL) was used according to the manufacturer's instructions. The corresponding DNA sequencing reactions were carried out using the T7 sequencing kit (Amersham Pharmacia Biotech) and 10 µCi of [α-32P]-dATP according to the manufacturer's instructions with primers HL27, HL49 and FMD10 and plasmids pGP508, pGP512 and pGP513 as templates respectively. The primer extension samples were separated from their corresponding sequencing reactions on the same 6% acrylamide gel.
In vitro transcription
To obtain a template for the in vitro synthesis of the cggR gapA RNA, a 2.192 bp PCR product was generated using chromosomal DNA of B. subtilis as template and primers HL47 and HL48 (see Supplementary material). The presence of a T7 RNA polymerase recognition site on primer HL47 allowed the use of the PCR product as a template for in vitro transcription with T7 RNA polymerase (Roche Diagnostics). The synthesized cggR–gapA RNA was checked by Northern blot analysis with a specific cggR RNA probe (data not shown).
We are grateful to Wolfgang Hillen for helpful discussions and continuous encouragement. Julia Bandow, Hans-Matti Blencke and Christoph Meinken are acknowledged for help with some experiments. We are grateful to Daniela Zühlke for the generous gift of some primers. This work was supported by the DFG priority program ‘Regulatorische Netzwerke in Bakterien’ and by grants from the Fonds der Chemischen Industrie to J.S.