INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

We have detailed knowledge about the regulation of expression of the pur genes, which encode enzymes of the purine biosynthetic pathway in bacteria (24, 25). In Escherichia coli, the pur genes are scattered on the chromosome and are found as single genes or small operons. A regulatory protein, the PurR repressor, of the LacI type of regulatory proteins, regulates the expression of the pur genes or operons. When E. coli grows in the presence of guanine or hypoxanthine, these compounds are taken up and salvaged and at the same time they bind to the PurR repressor. PurR binds to a 16-bp palindromic sequence that overlaps the 35 promoter region of the pur genes (11) and inhibits transcription of the pur genes.

In B. subtilis, the genes encoding the biosynthesis of IMP are located in the pur operon. Three other genes (purA, guaA, and guaB) required for AMP and GMP synthesis are located as single genes. Expression of the pur operon is subject to dual regulation of transcription termination and transcription initiation. Termination of transcription is regulated by a termination-antitermination mechanism in a 242-nucleotide mRNA leader region preceding the first gene of the pur operon (5). The termination mechanism is triggered by guanine or hypoxanthine; however, the molecular mechanism has not been clarified. Initiation of transcription of the pur operon, and also of the purA and purR genes, is repressed in response to the presence of adenine in the culture medium (5). Addition of adenine results in lowering of the cellular pool of the low-molecular-weight effector molecule phosphoribosylpyrophosphate (PRPP) (15). Two regulatory elements are required for this regulation, the PurR repressor and a DNA operator site for repressor binding. PurR binding to the operator site is blocked by PRPP. The PurR protein is a 62-kDa homodimer (17, 21) thatas judged by footprinting analysisinteracts with a region between 149 and 29 relative to the transcriptional start site of the pur operon (17). A second protein encoded by yabJ, which is located in an operon with purR, has been suggested to act together with the PurR repressor (10).

Recently, a regulatory protein, also named PurR, that activates pur gene transcription was identified in Lactococcus lactis. L. lactis PurR shows extensive amino acid sequence identity with the B. subtilis PurR repressor (7). The L. lactis PurR protein binds upstream of the promoter region of pur genes (6). Based on genetic analysis and sequence comparison between the nucleotide sequences upstream of genes in B. subtilis and L. lactis, Kilstrup and coworkers were able to suggest a possible cis-acting sequence (5'-AWWWCCGAACWWTH-3'), named the PurBox, which is required for PurR-mediated control of gene expression (6).

In the present work, we provide evidence that an operator site comprised of two PurBoxes is required for PurR control in B. subtilis and that other genes of importance for purine synthesis are also regulated by PurR.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and media. The bacterial strains and plasmids used in this work are listed in Table 1. DNA primers used for PCR amplification and primer extension are listed in Table 2. B. subtilis was grown in Spizizen minimal salt medium supplemented with 0.2% L-glutamate, 40 mg of L-tryptophan per liter, and 1 mg of thiamine per liter and with 0.4% glucose as a carbon source. Purine compounds (adenine and guanosine) were added to a final concentration of 1 mM. For selection of antibiotic resistance, antibiotics were used at the following concentrations: ampicillin, 100 mg/liter; neomycin, 5 mg/liter; erythromycin, 1 mg/liter; lincomycin, 25 mg/liter; chloramphenicol, 6 mg/liter.

Nucleic acid manipulation and genetic techniques. Isolation of DNA and RNA and basic molecular biology techniques were performed as previously described (12, 13, 26).

Construction of transcriptional lacZ fusions. Different promoter-containing PCR products were generated by using the primer combinations listed in Tables 1 and 2. The various DNA fragments were digested with restriction enzymes and ligated into pDG268cat or pDG268neo digested with the same enzymes and transformed into E. coli MC1061 selecting for Ap^r. Plasmids extracted from E. coli were integrated into the B. subtilis chromosome as described before (12). The yumD'-lacZ fusion was constructed by amplifying an internal segment of the yumD gene and cloning it in front of the lacZ gene of pMutin4 (20). The resulting plasmid was transformed into the B. subtilis yumD locus by selecting for Er^r as described by Vagner and coworkers (20).

Enzyme assays and measurement of purine base uptake. Cell extracts were made as described before (15). Serine hydroxymethyltransferase (SHMT) activity was determined in a coupled assay using L-allo-threonine as the substrate (16). GMP reductase activity was determined by measuring the formation of [¹⁴C]IMP from [¹⁴C]GMP. A 50-µl volume of assay buffer contained 0.2 mM NADPH, 0.1 mM [¹⁴C]GMP (50 mCi/mmol), and glucose-6-phosphate dehydrogenase (10 U), as well as 2 mM glucose-6-phosphate to regenerate NADPH. Cell extract was added, and after 1, 2, 4, and 8 min, 5-µl samples were removed and spotted on a polyethyleneimine-impregnated thin-layer chromatography plate (Merck, Darmstadt, Germany). The chromatogram was dried and developed in 0.4 M phosphate buffer (pH 3.4) to separate IMP from GMP. The plate was dried, and radioactivity was measured in an InstantImager (Packard). -Galactosidase activity was determined as described previously (2). All enzyme determinations were repeated at least three times. Enzyme activity is given as nanomoles of product formed per minute (equals 1 U). Total protein was determined by the method of Lowry et al. Uptake of purine bases was performed as described by Saxild and Nygaard (14).

Bioinformatic tools. Searches for specific nucleotide sequences in the B. subtilis genome were performed by using the WinSeq computer software developed by Flemming Hansen (unpublished). The machine-learning algorithm ANN-Spec, which was designed to discover ungapped patterns in DNA sequences (23), was used to analyze the B. subtilis genome sequence for the presence of the tandem-PurBox sequence. The computer program for UNIX systems is available from us.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Initial search for potential PurR binding sites on the B. subtilis chromosome. The PurR binding sequence (PurBox) 5'-AWWWCCGAACWWTH-3' (6) was used as the query sequence in a computerized search of the B. subtilis genome using the WinSeq computer software. An alignment of the B. subtilis PurBox sequences upstream of the pur operon, purA and purR, revealed that the purR and pur operon PurBoxes diverge from the consensus sequence at one and two positions, respectively. These three positions were therefore considered less important and were assigned a low-importance weight, whereas all other positions were assigned a high-importance weight. Using these search parameters, we found 249 potential PurBoxes with zero, one, or two low-weight mismatches or one high-weight mismatch. Because PurR is reported to bind to a regulatory region upstream of the affected genes, the locations of the 249 potential PurBoxes were examined. Those PurBoxes that are located 0 to 350 nucleotides upstream of the start codon of the downstream open reading frame (ORF) were selected. This assortment resulted in 46 ORFs.

Test for possible PurR control of the expression of six selected operons. Among the 46 ORFs, six operons or genes, including xpt-pbuX, yebB, glyA, yumD, yqhZ-folD, and rapB, were selected for further analysis. The upstream control regions containing the putative PurBox sequence were cloned in front of lacZ in plasmid pDG268(cat) or pMutin4 and inserted into the chromosome. A fusion of the pur operon promoter was also constructed and used as a positive control (amyE::purE'-lacZ). An isogenic series of strains was constructed that contains the respective lacZ fusions in a purR genetic background. All strains were grown in minimal medium with or without adenine, and the -galactosidase level was determined. The expression of all of the genes, except rapB, was repressed two- to threefold in the presence of adenine, and the levels were increased in a purR genetic background (Table 3). When the DNA sequences of the upstream regulatory regions of the identified PurR-controlled genes were compared, it became evident that they all contain two divergently oriented PurBox-like sequences separated by 16 or 17 nucleotides (Fig. 1). From the alignment of putative PurBoxes upstream of the six selected operons or genes, it appears that of all the PurR-regulated genes are preceded by one PurBox sequence with relatively high sequence similarity to the consensus sequence and by another PurBox sequence that has a more degenerated sequence. The upstream region of rapB contains only one PurBox sequence.

View larger version (42K): [in this window] [in a new window]   FIG. 1.   Alignment of the tandem-PurBox motif located upstream of nine PurR-regulated genes or operons. Only one DNA strand is shown. Boxed sequences are individual PurBox sequences. Shaded positions indicate nucleotides (nt) that diverge from the 5'-AWWWCCGAACWWTH-3' consensus sequence defined by Kilstrup and coworkers (6). Letters in the two bottom boxes show nucleotides that are conserved in the tandem PurBox motif. Lightface letters indicate nucleotides that are conserved in eight of the nine PurBoxes, and boldface letters indicate nucleotides that are conserved in all nine PurBoxes.

Computerized search for regulatory regions containing the tandem PurBox motif. We then wanted to test whether a refined computer search using the novel information about the PurR binding motif could identify the expected PurR-regulated genes and perhaps predict new genes that did not appear in the initial search. Potential promoter-containing DNA sequences in the B. subtilis genome were organized in a list of 4,222 entries, each containing a 400-nucleotide sequence upstream of one of the 4,222 predicted ORFs in the B. subtilis genome. This list was searched for sequences having the tandem PurBox motif. Using the ANN-Spec bioinformatics software (23), a weight matrix for the PurBox sequence was calculated on the basis of the sequence of a total of 16 PurBoxes located pairwise upstream of purR, purA, yqhZ-folD, yumD, purE, glyA, xpt-pbuX, and yebB, respectively. The program calculates an arbitrary score for each of the potential PurBox sequences. The file containing the 4,222 potential promoter regions was searched for sequences having two potential PurBoxes separated by no less than 11 and no more than 21 nucleotides. A total of 129 sequences were found, and Table 4 lists the 10 top-ranked loci for which the upstream 400-bp sequence contains the tandem PurBox motif having one or two PurBox sequences with a high score. For the remaining 116 loci, the scores for one or both potential PurBoxes were below the level of significance. As expected, the program identified all of the genes that were used to calculate the weight matrix. The upstream region of ytiP and ytjP was also found to contain a potential tandem PurBox motif with the correct spacing of 16 or 17 nucleotides between the PurBox sequences. ytiP and ytjP are divergently oriented on the chromosome and are separated by a 96-bp intercistronic region. The two potential PurBoxes are located closest to the ytiP reading frame. The 432-amino-acid primary sequence of YtiP is 47% identical to yebB of B. subtilis. The 463-amino-acid primary sequence of YtjP is 40% identical to a dipeptidase from L. lactis (472 amino acids, accession no. AAC45369). In order to analyze whether the genes are subject to purine control, BFA2025 (ytiP) and BFA2026 (ytjP) were grown in minimal medium supplemented with adenine or guanosine. The basal level in BFA2025 was 22 U/mg of protein, and the expression was repressed by adenine (to 9 U/mg of protein) and induced by guanosine (to 49 U/mg of protein). ytjP (BFA2026) expression did not respond to addition of purines (data not shown). Inactivation of purR in BFA2025 (strain ED453) resulted in derepression of ytiP expression both in the presence of adenine (66 U/mg of protein) and in the absence of adenine (81 U/mg of protein). We therefore concluded that ytiP, but not ytjP, belongs to the PurR regulon.

cis-acting elements involved in PurR repression of glyA expression. The cis-acting requirements for PurR control of glyA expression was studied in more detail. The glyA transcriptional start site was determined in a primer extension experiment (data not shown), and the site is indicated in Fig. 2. Putative ^A 10 and 35 regions are located at suitable distances upstream of the +1 position. The DNA fragment covering the region from 120 to +118 (Fig. 2) directed PurR-regulated transcription when fused to lacZ in a wild-type genetic background (PEH06, Table 3). The same fusion was constitutively expressed in a purR genetic background (PEH03, Table 3). A fusion with a DNA fragment with nucleotides 120 to 99 deleted was also constitutively expressed (PEH08, Table 5), indicating that the deduced PurBox (nucleotides 116 to 103 in Fig. 2) is required for PurR control. A G₊₁₁₀C substitution was introduced into the 120 to +118 fragment, and this also leads to constitutive expression. This observation demonstrates the essential role of the central CG pair of the promoter-distal PurBox in mediating the negative control of gene expression by PurR. The promoter-proximal PurBox sequence was altered in two ways. T₊₇₆ was replaced with a G, and in theory, this should create a more consensus-like PurBox sequence. C₊₇₈ was replaced with a G, and in theory, this should result in a less consensus-like PurBox sequence. When fused to lacZ, the fragment containing the T₊₇₆G mutation mediated a stronger repression by PurR in medium with adenine present whereas a fusion with the fragment containing the C₊₇₈G mutation reduced repression by adenine to 1.3-fold, compared to 2.7-fold repression in the wild type.

View larger version (14K): [in this window] [in a new window]   FIG. 2.   Organization of the glyA regulatory region. Italic boldface letters indicate nucleotides constituting the tandem-PurBox motif. Boldface roman letters indicate the translational start codon of the glyA reading frame. Arrows and letters above the PurBox sequences indicate base pair substitutions in the various strains described in Table 5. The  symbol surrounded by dashed lines indicates the extent of the PurBox deletion in strain PEH08 (Table 5). Lines above the sequence indicate the locations of the putative 10 and 35 regions of the glyA promoter. The designation +1 indicates the transcriptional start site determined by primer extension analysis of glyA mRNA from cells grown in glucose minimal medium using primer 7 (Table 2).

Is the yabJ gene product involved in the regulation of expression of PurR-controlled genes? The yabJ gene located downstream of the purR gene has been suggested to encode a protein involved in the adenine-mediated repression of purA gene expression (10), although this was not observed when the purR-yabJ operon was first identified (21). To investigate whether the expression of the glyA gene was altered in a yabJ mutant, we determined the effects of adenine and guanosine on glyA expression in both the wild type and a yabJ mutant strain. As a control, we determined purA gene expression (Table 6). However, we found that adenine repression and guanosine induction of both genes were similar in wild-type strains and yabJ mutant strains. This finding favors the view that the yabJ gene product has no effect on glyA and purA gene expression.

Function of PurR-controlled genes glyA, yumD, yqhZ-folD, yebB, and ytiP The derived amino acid sequence of glyA has high amino acid sequence similarity to SHMT from E. coli (accession no. P00477). In agreement with this, glyA mutant strain HH413 required glycine for growth. The SHMT levels were determined in cultures grown in the presence of 1 mM guanosine to induce the expression of the enzyme. The SHMT activity was found to be 3.2 U/mg of protein in strain 168 and <0.2 U/mg of protein in strain HH413. This indicates that the glyA gene actually encodes SHMT activity.

View larger version (27K): [in this window] [in a new window]   FIG. 3.   Map of metabolic pathways in B. subtilis that are regulated by PurR. The different enzymatic steps are represented by the corresponding gene designations. Gene designations in a large font and in boldface represent genes that are regulated by PurR, while gene designations in a small font represent genes that are not regulated by PurR. Abbreviations: GAR, phosphoribosylglycinamide; FGAR, phosphoribosylformylglycinamide; AICAR, phosphoribosylaminoimidazole carboxamide; FAICAR, phosphoribosylformamidoimidazole carboxamide; SAMP, adenylosuccinate. Gene designations: purF, glutamine PRPP amidotransferase; purD, phosphoribosylglycinamide synthetase; purN, THFA-dependent phosphoribosylglycinamide transformylases; purQLS, phosphoribosylformylglycinamidine synthetases I, II, and III; purM, phosphoribosylaminoimidazole synthetase; purEK, phosphoribosylaminoimidazole carboxylases I and II; purC, phosphoribosylaminoimidazolesuccinocarboxamide synthetase; purB, adenylosuccinate lyase; purH, phosphoribosylaminoimidazole carboxamide formyltransferase and IMP cyclohydrolase; purA, adenylosuccinate synthetase; guaB, IMP dehydrogenase; guaA, GMP synthetase; apt, adenine phosphoribosyltransferase; hpt, hypoxanthine-guanine phosphoribosyltransferase; xpt, xanthine phosphoribosyltransferase; guaC, GMP reductase; ade, adenine deaminase; pbuG, hypoxanthine-guanine permease; pbuX, xanthine permease; pbuO, guanine permease; glyA, SHMT; folD, methylenetetrahydrofolate dehydrogenase-methenyltetrahydrofolate cyclohydrolase. Dashed lines indicate multiple enzyme-catalyzed steps.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Based on the experimental results presented in this work, we were able to expand the B. subtilis PurR regulon with six mono- or dicistronic operons. The function and expression of the xpt-pbuX operon have been previously reported (2), while the functions of the genes yumD (guaC), yebB (pbuG), glyA, yqhZ-folD, and ytiP (pbuO) are described in this work. Two genes were shown to encode purine base permeases. yebB encodes a high-affinity hypoxanthine-guanine permease already known as pbuG (14). A pbuO (formerly ytiP) mutant was shown to be impaired in guanine and hypoxanthine uptake. The purine base concentration used in the uptake assay was low (1 µM). At this concentration, PbuG has been shown to be the major transport system for guanine because pbuG deficiency results in a low level of guanine uptake (14). The residual guanine uptake at 1 µM guanine in the pbuG mutant strain could be due to transport through PbuO. PbuG deficiency has no effect on the growth of a purine-requiring mutant strain when guanine or hypoxanthine is present at a concentration higher than 100 µM (14). Most likely, pbuO encodes a guanine-hypoxanthine permease working at purine concentrations higher than 100 µM.

Two genes, glyA and folD, encode enzymes involved in N¹⁰-formyl-THFA formation. Based on genetic data and on growth analysis of a glyA mutant, Dartois and coworkers suggested that glyA encodes SHMT (4). By measuring SHMT activity in a glyA knockout mutant, we have finally established the function of this gene in B. subtilis. folD was the only gene whose function was only indirectly demonstrated. The gene appears not to be essential as long as IMP can be synthesized from an external purine source. Finally, yumD (guaC) was identified as the gene encoding GMP reductase activity.

The previously identified PurR-regulated genes (pur operon and purA [17]) plus the newly identified ones allowed us to construct a map of the PurR-affected pathways in B. subtilis. In Fig. 3, it can be seen that the majority of the genes involved in purine base, purine nucleoside, and purine nucleoside monophosphate metabolism are regulated by PurR. Figure 3 also illustrates the three steps of THFA metabolism that are regulated by PurR. In E. coli, the formation of N⁵,N¹⁰-methylene-THFA is regulated by purine levels and PurR through the repression of glyA expression (18). However, folD in E. coli appears not to be controlled by PurR. Among all of the PurR-regulated genes, yqhZ, which encodes a potential NusB-like factor involved in transcription termination, is the only gene without an obvious role in purine metabolism.

We have shown that all of the B. subtilis genes and operons that have been experimentally demonstrated to be regulated by PurR are preceded by a palindromic sequence composed of two divergently oriented PurBoxes separated by 16 or 17 nucleotides. We have compared our data with previously obtained footprinting data (17) in which purified PurR protein was found to protect an extended region upstream of the pur operon, purA and the purR-yabJ operon. From this comparison, it is evident that the common dyad symmetry 5'-GAAC-N_(24-25)-GTTC-3' motif identified by Shin and coworkers (17) is included in the tandem PurBox motif defined in this work (Fig. 1). Characteristic for the footprinting data are the large regions of 80 to 90 nucleotides that were protected by PurR protein. The extended protected regions reported by Shin and coworkers were found to be primarily on the 5' side in relation to the two PurBoxes and the 5'-GAAC-N_(24-25)-GTTC-3' motif. Analysis of the minimal regulatory sequence requirement for full PurR control of glyA expression revealed that no extended 5' region relative to the tandem PurBox motif was required. This leads us to suggest that the binding of PurR to sequences upstream of the twin PurBox sequences, as demonstrated by previous in vitro footprinting experiments, most likely plays no role in vivo. The tandem PurBox motif may be located at various positions both up- or downstream of the transcriptional start site (Fig. 1). In the case of ytiP (pbuO), yumD (guaC), and purR, the PurBoxes are located close to or overlapping the sequence encoding the potential ribosome binding site. In the case of the pur operon, the xpt-pbuX operon, and yebB (pbuG), the PurBoxes are located 230 to 274 nucleotides upstream of the coding region of the first gene of the operon. This long distance is due to the presence of a long untranslated leader sequence that, in the case of the pur and xpt-pbuX operons, has been shown to be the site for the hypoxanthine-and-guanine-controlled regulatory mechanism. The PurBoxes in front of the pur operon, xpt-pbuX and pbuG, are located 4 (pur operon and pbuG) and 14 (xpt-pbuX) nucleotides upstream of the promoter 35 elementsdistances that are consistent with the PurBoxes functioning as repressor binding sites. In the glyA regulatory region, the PurBoxes are located 35 nucleotides upstream of the 35 element. This may appear to be a rather long distance. However, as demonstrated by the published footprint analysis (17), PurR protects DNA sequences (approximately 20 nucleotides in length) located downstream of the 5'-GAAC-N_(24-25)-GTTC-3' motif, which coincides with the PurBoxes. We speculate that PurR represses glyA transcription by first binding to the PurBoxes and then multimerizes along the DNA as suggested previously (7).

Addition of adenine to B. subtilis results in a drop in the cellular PRPP pool, thereby increasing the binding of PurR to its operator sequence. This results in an average of 2.5- to 3-fold repression of gene expression (Table 3) (10, 15). In contrast, addition of guanosine increases PRPP pools, resulting in decreased PurR binding and two- to threefold induction of gene expression (Table 7) (10, 15). Rappu and coworkers have suggested that stronger binding of PurR to operator DNA when the PRPP pool is low requires the yabJ gene product, and it was suggested that a possible function of YabJ is to interact with PurR to form a multimeric PurR structure. This would result in the binding and protection of the extended operator sequence by PurR observed in footprinting experiments. We investigated the effect of YabJ deficiency on the repression of expression of one of the novel PurR-controlled gene glyA and of purA, for which the repression was shown by Rappu and coworkers to be YabJ dependent. We were not able to detect any changes in either glyA or purA expression in a yabJ mutant strain compared to that in the wild type (Table 6). The two yabJ mutations, however, were not identical. Rappu and coworkers constructed a 39-amino-acid deletion of the YabJ (125 amino acids long) N-terminal end, whereas the mutation analyzed in this report was a 42-amino-acid deletion of the C-terminal end. Even though it appears unlikely that the repressor auxiliary function of YabJ may be dependent on the N-terminal part, this might be a possibility. Until this has been analyzed in more detail, the role of YabJ in the process of PurR-controlled gene expression remains questionable.