Nucleosides as a carbon source in Bacillus subtilis: characterization of the drm-pupG operon

Nucleoside phosphorylases are involved in the intracellular metabolism of nucleosides obtained from the external environment or arising from the intracellular breakdown of nucleotides. However, the regulation and the components of this system are not well characterized in Gram-positive bacteria. Two purine-nucleoside phosphorylases exist in Bacillus subtilis (Jensen, 1978 ): one specific for adenosine (and deoxyadenosine), PupA, and another specific for guanosine (and deoxyguanosine) and inosine (and deoxyinosine), PupG. Based on uptake studies, an adenosine-specific transport system and a transport system for guanosine and inosine have been suggested for B. subtilis (Beaman et al., 1983 ). The nucleoside phosphorylases catalyse the cleavage of ribonucleosides and deoxyribonucleosides to the free base plus ribose 1-phosphate or deoxyribose 1-phosphate, respectively (Fig. 1). The bases serve anabolic (re-utilization in nucleotide synthesis) or catabolic (use as nitrogen sources) functions. However, the degradation of the purine bases in B. subtilis occurs only when cells are growing on poor nitrogen sources (Christiansen et al., 1997 ). When nucleosides are added as the sole nitrogen source, the growth rate is reduced sixfold compared with that seen when glutamate plus ammonia serve as the nitrogen source (Nygaard et al., 1996 ). The pathways for the degradation of nucleosides in B. subtilis are shown in Fig. 1. The ribose 1-phosphate formed from ribonucleosides can be converted to ribose 5-phosphate by phosphopentomutase and then further catabolized or converted to phosphoribosylpyrophosphate which is utilized in the synthesis of nucleotides. Deoxyribose 1-phosphate, formed from deoxyribonucleosides, is also a substrate for phosphopentomutase and is converted to deoxyribose 5-phosphate. Deoxyribose 5-phosphate cannot be rescued for deoxynucleotide synthesis and is degraded by deoxyriboaldolase encoded by the dra gene. The dra gene is located in the dranupCpdp operon. The nupC locus encodes a pyrimidine-nucleoside transporter that mediates the transport of uridine, thymidine and deoxyuridine. The pdp gene encodes the only pyrimidine-nucleoside phosphorylase in B. subtilis and can phosphorylize uridine, thymidine and deoxyuridine to the free base. Uracil can be rescued for nucleotide synthesis, but cannot serve as a nitrogen source. The dranupCpdp operon is negatively regulated by the deoR gene product (Saxild et al., 1996 ; Zeng & Saxild, 1999 ).

(13K): Fig. 1. Nucleoside catabolic pathways in B. subtilis. The enzymes are indicated by their gene symbols: cdd, cytidine deaminase; pupA, adenosine phosphorylase; pupG, guanosine (inosine) phosphorylase; pdp, pyrimidine nucleoside phosphorylase; drm, phosphopentomutase, dra, deoxyriboaldolase, drsK, deoxyribokinase and rbsK, ribokinase. Purine base indicates adenine, guanine or hypoxanthine.

In Escherichia coli and Salmonella typhimurium, a single enzyme catalyses the phosphorolytic cleavage of adenosine, guanosine, inosine and the corresponding deoxyribonucleosides (Jensen & Nygaard, 1975 ). E. coli, in addition, contains a purine-nucleoside phosphorylase with specificity towards xanthosine, guanosine and inosine (Seeger et al., 1995 ). The synthesis of these phosphorylases and other proteins involved in nucleoside catabolism and transport in enterobacteria is induced several-fold by nucleosides added to the growth medium. As a result of this induction, the nucleosides are rapidly catabolized and serve as excellent carbon sources in E. coli (Hammer-Jespersen, 1983 ; Munch-Petersen & Mygind, 1983 ).

Previously, we cloned and sequenced two adjacent B. subtilis genes, encoding products similar to phosphopentomutase and purine-nucleoside phosphorylase. These loci were designated drm and pnp (now referred to as pupG), and their sequences were submitted to GenBank in 1995 (accession number U32685). Here, we report the characterization of the products of these loci with respect to the regulation of their expression and their functions. These studies identify distinct differences between the B. subtilis nucleoside-catabolizing pathway and that of E. coli.

Bacterial strains.
The bacterial strains and plasmids used are described in Table 1. Strains constructed by transformations with plasmids as donors were confirmed by Southern-blot analysis (Sambrook et al., 1989 ). Strains SL6250 and SL6550 contain single-copy chromosomal insertions of plasmid pPP400 and pPP424, in the pupG and drm gene, respectively. Strain SL6479 bears a single copy drmlacZ transcriptional fusion inserted by double crossover recombination at the amyE locus (Table 1).

Table 1. Bacterial strains and plasmids

Growth and media.
Cells were grown at 37 °C as described previously (Jochimsen et al., 1975 ; Saxild et al., 1995 ; Schuch & Piggot, 1994 ). As minimal medium for B. subtilis the Spizizen salt-buffered medium was used; L-broth was used as rich medium. All plasmids were maintained in E. coli DH5α. When appropriate, bacteria were grown in the presence of antibiotics at the following concentrations: chloramphenicol, 5 µg ml^-1; erythromycin, 1 µg ml^-1; ampicillin, 50 µg ml^-1. Spore germination was assessed by the method of Nicholson & Setlow (1990) .

Sporulation.
Sporulation was induced as described by Piggot & Curtis (1987) . Time is indicated in hours after the end of exponential growth. At t₂₀t₂₄, the extent of heat-resistant-spore formation was determined by plating appropriate diluted aliquots of the cultures on L-broth; the remainder was heated at 90 °C for 20 min, and aliquots were plated on L-broth. Colonies arising from heat-treated and non-heat-treated samples were enumerated to determine the proportion of the starting population of cells that were heat-resistant spores.

Enzyme assays.
ß-Galactosidase was assayed at 30 °C as described by Nicholson & Setlow (1990) . Activities of nucleoside-catabolizing enzymes were determined at 37 °C as described previously (Hammer-Jespersen et al., 1971 ; Jensen, 1978 ). Enzyme activity is given as nmol product formed min^-1 (=1 unit).

DNA preparation and sequencing.
Methods used for transformation and for chromosomal and plasmid DNA isolation have been described previously (Wu et al., 1989 ; Saxild et al., 1995 ). DNA sequencing was using a Sequenase kit (US Biochemical). All sequencing analysis was done on double-stranded plasmid DNA templates. Using plasmids pHM2 and pKE5, we determined the nucleotide sequence of the drmpupG operon.

RNA analysis.
RNA preparation, Northern-blot analysis and primer extensions were performed as described previously (Penn et al., 1984 ; Wu et al., 1989 ; Nygaard et al., 1996 ). For Northern analyses, RNA samples (20 µg per sample) and size markers (Promega) were separated on 1·2% agarose gels, transferred to nylon membranes (Stratagene) and probed as described by Sambrook et al. (1989) . The pupG-specific probe was the 364 bp EcoRI fragment (nt 24457642446128), and the drmpupG-specific probe was the 823 bp HindIII fragment (nt 24456932446516) (Kunst et al., 1997 ).

Organization and sequence of the drmpupG gene cluster
Sequence analysis of the region immediately upstream of the dacFspoIIA operon was performed using plasmids pHM2 and pKE5 (Table 1). The ORF immediately upstream of dacF encoded a putative product of 271 amino acids with extensive similarity to guanosine (inosine) nucleoside phosphorylases of a broad range of organisms. On this basis, we refer to the ORF as pupG (formerly called pnp). The second ORF was identified immediately upstream of the pupG RBS, and found to encode a putative protein of 396 residues. A BLAST search (Altschul et al., 1990 ) revealed that this product was 47% identical over its entire sequence to phosphopentomutase (phosphodeoxyribomutase) of E. coli (Hammer-Jespersen, 1983 ; Valentin-Hansen et al., 1984 ). We refer to this ORF herein as drm. Upstream of the drm RBS was a 161 bp intergenic region, followed by the 3' end of another ORF, now called ripX (Kunst et al., 1997 ). In the ripXdrm intergenic region we identified a putative rho-independent transcription terminator. The energy of binding for this stemloop structure was calculated to be -19·8 kcal mol^-1 (-82·8 kJ mol^-1) using the rules of Tinoco et al. (1973) . These sequences may serve to stop transcription from ripX into drm. The remaining region between the stemloop and drm (roughly 130 bp) could serve as a promoter region directing the transcription of drm and possibly pupG. That drm and pupG may form an operon is supported by the proximity of the drm and pupG translation stop and start codons, respectively.

Complementation of E. coli purine-nucleoside catabolism defects
To confirm the identity of drm and pupG, plasmids pKE5 (expressing drm from its native promoter) and pHM2 (expressing pupG from a vector-encoded promoter) were transformed into E. coli strains HO1077 (deoB) and SØ446 (purE deoD), respectively, and scored for complementation. Strain HO1077 is a phosphopentomutase-deficient mutant and cannot use thymidine as its sole carbon source (Hammer-Jespersen, 1983 ). Introduction of pKE5 into HO1077 restored the ability of this strain to grow on minimal medium containing inosine as a carbon source. Additionally, phosphopentomutase activity increased from <1 unit (mg protein)^-1 in HO1077 to 8 units (mg protein)^-1 in HO1077/pKE5. Strain SØ446 is purine auxotrophic and is unable to use thymidine as its sole purine source. The transformed strain SØ446/pHM2 did grow with inosine as the sole purine source, indicating that the B. subtilis pupG gene could complement the salvage-pathway defect arising from mutation of the E. coli deoD gene. Inosine phosphorylase activity increased from <1 unit (mg protein)^-1 in SØ446 to 85 units (mg protein)^-1 in SØ446/pHM2. The B. subtilis drm and pupG loci, therefore, do encode phosphopentomutase and purine-nucleoside phosphorylase activity, respectively.

Transcription of the drmpupG locus
Transcription of drm and pupG during both late-exponential and early-stationary-phase L-broth cultures of strain SL4 was examined by Northern-blot analysis. Using a pupG-specific probe, we detected a predominant 2·1 kb transcript (Fig. 2a, lanes 1 and 2). With a probe containing both drm and pupG sequences, a similar 2·1 kb message was visualized (Fig. 2b, lanes 1 and 2). The transcript size observed in each case strongly suggests that drm (a 1·2 kb gene) and pupG (a 0·8 kb gene) are cotranscribed as an operon. A minor 0·9 kb band was also noted. This band may reflect a low level of expression of a specific pupG transcript or a processing/breakdown product of the 2·1 kb transcript. Expression of the drmpupG transcript was barely detectable in mRNA samples isolated at t₂ (Fig. 2, lanes 3), suggesting that it may be repressed in later stationary phase. To confirm the operon structure of drmpupG, we also analysed expression of single-copy drmpupG transcriptional fusions to lacZ integrated at the chromosomal amyE locus. Fusion to lacZ of several distinct DNA fragments extending from within pupG to at least 300 bp upstream of it yielded background levels of ß-galactosidase activity (data not shown), indicating that pupG is not likely to be transcribed from its own promoter. In contrast, a 1·4 kb PvuIIBglII fragment containing the ripXdrm intergenic region and the 5' end of drm fused to lacZ in SL6479 does support high-level ß-galactosidase activity in L-broth cultures: 135 units (mg protein)^-1 at t_-1, 176 at t₀ and 141 at t₂. pupG is, therefore, probably cotranscribed with drm from a promoter in the ripXdrm intergenic region. The lower enzyme level observed at t₂ reflects the fact that the drmpupG operon is no longer transcribed at this stage of growth (Fig. 2).

(43K): Fig. 2. Northern-blot analysis of the drmpupG region of the B. subtilis chromosome. RNA samples were obtained from L-broth cultures at t-1 (lane 1), t0 (lane 2) and t2 (lane 3). The numbers are hours relative to the end of exponential growth (t0). The arrow indicates the major message. Probes used were either specific to pupG (a) or to both drm and pupG (b).

Role of Drm and PupG in sporulation and spore germination
Several observations show that the intracellular levels of a purine compound are important for the initiation of sporulation (Beaman et al., 1983 ) and for the germination of spores (Gardner & Kornberg, 1967 ; Engelbrecht, 1972 ; Senesi et al., 1991 ). Based on these findings, we investigated the requirements for drmpupG in spore formation. Strains SL6250 (pupG) and SL6550 (drm) displayed 65% and 61% sporulation, respectively, after overnight growth in liquid sporulation medium (a time at which the wild-type parental strain SL4 yielded 70% sporulation). Drm and PupG are, therefore, not required for efficient spore formation. Similarly, Drm and PupG are also dispensable for spore germination (data not shown).

Identification of the low-molecular-mass effector of induction
Preliminary analysis had indicated that the levels of Drm and PupG, but not that of PupA, were increased by the addition of ribonucleosides and deoxyribonucleosides to the growth medium. The level of PupA was uninducible and remained between 27 and 34 units (mg protein)^-1. The activities of Drm and PupG in wild-type cells grown in the presence of thymidine, inosine and adenosine are shown in Table 2. No induction was observed in the presence of any of the naturally occurring purine and pyrimidine bases (data not shown). To investigate whether a nucleoside must be degraded to induce, experiments were conducted with mutants defective in nucleoside catabolism. In a pupG mutant defective in inosine degradation, inosine no longer induced Drm synthesis, while increased Drm activity was seen when adenosine was added (Table 2). This indicates that a nucleoside must be phosphorylized to act as an inducer. To determine whether the first products of nucleoside degradation, ribose 1-phosphate or deoxyribose 1-phosphate, were the low-molecular-mass effector molecules, induction experiments were performed in a drm mutant strain. This mutant is unable to convert ribose 1-phosphate and deoxyribose 1-phosphate, formed from inosine and thymidine, respectively, to ribose 5-phosphate and deoxyribose 5-phosphate. The level of PupG and ß-galactosidase activity from a drmlacZ transcriptional fusion was measured in the drm knock-out mutant HH245 grown in the presence of either inosine or thymidine. The PupG level was low due to the interruption of the drm gene and this level was not affected by addition of nucleosides to the growth medium. However, the ß-galactosidase activity, driven by the drm promoter, was also unaffected. This indicates that ribose 1-phosphate and deoxyribose 1-phosphate must be further metabolized to affect the enzyme levels. In the dra mutant HH234, which can degrade deoxyribose 1-phosphate one step more to deoxyribose 5-phosphate and no further, addition of thymidine results in increased levels of Drm and PupG, indicating that the deoxyribose 5-phosphate formed may act as a low-molecular-mass effector molecule. Since no induction is observed in the drm mutant unable to convert ribose 1-phosphate to ribose 5-phosphate, ribose 5-phosphate is a likely low-molecular-mass effector molecule too. An alternative way of forming ribose 5-phosphate and deoxyribose 5-phosphate is through phosphorylation of ribose and deoxyribose (Fig. 1). Induction experiments with ribose and deoxyribose were performed with cultures grown on succinate as the carbon source, because neither ribokinase (OReilly et al., 1994 ) nor deoxyribokinase (X. Zeng, personal communication) is synthesized when glucose serves as carbon source. When added to cultures growing on succinate both ribose and deoxyribose caused slightly increased enzyme levels in both wild-type and the drm mutant strain, supporting the suggestion that both ribose 5-phosphate and deoxyribose 5-phosphate are low-molecular-mass effector molecules. Addition of thymidine or inosine to a deoR mutant (HH232) resulted in the same level of induction as in wild-type cells (Table 2). This indicates that the B. subtilis deoxynucleoside regulator protein DeoR, which negatively regulates the expression of the dranupCpdp operon, does not appear to be involved in the regulation of expression of the drmpupG operon.

Table 2. Effects of mutations in genes encoding enzymes involved in the catabolism of nucleosides on the expression of the pupG gene, the drm gene and the transcriptional LacZ fusion (drmlacZ) in B. subtilis

Expression of the drmpupG operon during growth in minimal medium
The apparent difference in enzyme levels observed when nucleosides were added to glucose- and succinate-grown cells was studied in more detail. Expression of the drmpupG operon was followed by determining both Drm and PupG activity after addition of thymidine and inosine. We also investigated whether the expression was co-regulated with the dranupCpdp operon by measuring pyrimidine nucleoside phosphorylase (Pdp) activity. The level of all three enzymes was increased in thymidine-supplemented cultures; however, a more dramatic increase was observed when succinate served as the carbon source (Table 3). Addition of inosine only affected the level of Drm and PupG. There was no additive effect when both thymidine and inosine were added to the growth medium. The addition of nucleosides to the cultures in most situations resulted in an increased growth rate, most significantly in succinate-grown cultures. From these experiments, we conclude that the expression of the drmpupG operon appears to be subject to catabolite repression.

Table 3. Effects of nucleosides on growth and induction of the synthesis of phosphopentomutase (Drm), guanosine (inosine) phosphorylase (PupG) and pyrimidine nucleoside phosphorylase (Pdp) in B. subtilis (168)

Identification of two drmpupG transcription start sites
From the Northern blot and lacZ fusion studies, drmpupG appeared to be co-transcribed from a promoter in the ripXdrm intergenic region. The induction experiments indicated that transcription was increased when either a ribonucleoside or a deoxyribonucleoside was present in the growth medium. The transcription start points were identified by primer-extension mapping (Fig. 3). Total RNA was prepared from strain 168 grown in minimal succinate medium (Fig. 3, lanes 1 and 2) and in the same medium to which inosine (Fig. 3, lanes 3 and 4) or thymidine (Fig. 3, lanes 5 and 6) was added. Primers complementary to the nucleotide sequence from position 24475962447613 or 24475122447529 were used for both primer extension and to generate a sequence ladder. Only results obtained with the upstream primer 24475962447613 are shown. Two transcriptional-start sites are inferred from the 5' end of the mRNA and are positioned at nucleotide numbers 2447716 and 2447687 (53 and 24 nt upstream of the putative translational-start signal of drm); they were detected with both primers. The same two transcriptional-start sites were seen when strain MB24 was grown in minimal glucose medium (data not shown). The distal promoter is referred to as drmP1 and the proximal as drmP2. The -10 and -35 sequences for both transcription-start sites are very similar to the σ^A consensus recognition sequences (TATAAT for the -10 site and TTGACA for the -35 site). As judged by band intensity, both promoters were active under all conditions, with the highest activities in the inosine- and thymidine-supplemented cultures. The increase in transcription resulting from inosine or thymidine addition was reflected by the increased levels of Drm and PupG in these cultures (Table 3). Furthermore, the drm proximal promoter (drmP2) appears to be favoured over the distal promoter (drmP1) by a ratio of approximately 2:1.

(23K): Fig. 3. Primer extension analysis of drmpupG mRNA from cells grown in minimal succinate medium and organization of the drmpupG regulatory region. (a) The primer used corresponds to nt 24475962447613. The sequence ladder (C, T, A, G) was obtained with the same primer used for cDNA synthesis. The positions that correspond to the start of transcription are marked with arrows. Lanes 1 and 2 contain RNA from cells grown in the absence of inducer. Lanes 3 and 4 contain RNA from cells grown with 1 mg inosine ml-1 and lanes 5 and 6 contain RNA from cells grown with 1 mg thymidine ml-1. Lanes 1, 3 and 5 contain 3 µl samples and lanes 2, 4 and 6 contain 4 µl samples. (b) Lines above the sequence indicate the location of the putative -35 and -10 regions of the drmP1 promoter and underlined sequences indicate the drmP2 promoter elements. The boxed sequence indicates the location of the putative cre sequence. The transcription-start points for drmP1 and drmP2 are indicated with a +1, above or under the sequence, respectively. Letters in bold indicate the putative translational start of the drm reading frame.

Catabolism of nucleosides
The ability to grow on inosine (0·1%) or thymidine (0·1%) as the carbon source was compared with growth on ribose (0·1%) and deoxyribose (0·1%). Except for the initial steps, the degradation of inosine and thymidine follow the same pathways as that of ribose and deoxyribose, respectively (Fig. 1). When wild-type cells were incubated in liquid media with ribose, deoxyribose, inosine or thymidine as the sole carbon source and ammonium ions as the nitrogen source, no growth on inosine or deoxyribose was observed after 36 h, while cells grew with a doubling time of 280 min on ribose and 210 min on thymidine. The doubling time on inosine plus deoxyribose was 8 h. For comparison, we found that E. coli (strain SØ199) grew on thymidine and inosine as the sole carbon source with a doubling time of 87 min and 115 min, respectively. When B. subtilis strains were tested on agar plates containing inosine or deoxyribose as carbon sources and ammonium ions as nitrogen source, colonies were visible after 2 d incubation; they were visible after 1 d when thymidine served as the sole carbon source. The colonies were significantly larger than those observed when no carbon source was added to the plate. To test whether Drm or PupG activities were required for growth on nucleosides and pentoses as the carbon source, strain SL6550 (drm) and strain SL6250 (pupG) were incubated with a mixture of inosine and deoxyribose, thymidine or ribose as carbon sources. Both strains grew on ribose, and strain SL6250 also on thymidine. These results indicate that Drm activity is required for the utilization of thymidine and inosine as a carbon source and that PupG also was required for inosine utilization. E. coli has served as a paradigm for the analysis and understanding of metabolic pathways in bacteria. However, often there are major differences with respect to how pathways from different bacteria are organized and regulated. The anabolic pathways responsible for purine-nucleoside salvage in E. coli and B. subtilis are quite different. Only E. coli is able to phosphorylate guanosine and inosine to GMP and IMP, and only B. subtilis can phosphorylate deoxyadenosine and deoxyguanosine, to dAMP and dGMP (Neuhard & Nygaard, 1987 ; Nygaard, 1993 ).

The very efficient nucleoside transport and catabolism found in E. coli and the great number of genes encoding regulatory proteins, transport proteins and enzymes involved in nucleoside catabolism do not have counterparts in B. subtilis. In E. coli, the genes encoding most of the proteins involved in purine- and pyrimidine-nucleoside catabolism are organized in a regulon composed of several single genes: nupC, nupG, tsx, cytX (all encoding transport proteins), udp, (encoding uridine phosphorylase), cdd (encoding cytidine deaminase) and the deo operon deoABCD (Valentin-Hansen et al., 1996 ). This regulon is negatively regulated by the DeoR and CytR proteins. A second operon, xapAB (encoding a purine-nucleoside phosphorylase and a nucleoside transport protein), is not part of the regulon, but is positively regulated by the XapR protein (Seeger et al., 1995 ). The E. coli deoB and deoD genes (equivalent to the B. subtilis drm and pupG genes) are encoded in the deo operon, together with deoA (encoding thymidine phosphorylase) and deoC (encoding deoxyriboaldolase). The low-molecular-mass effector that reacts with the regulatory DeoR protein of E. coli is deoxyribose 5-phosphate, while ribose 5-phosphate is not active (Hammer-Jespersen, 1983 ). CytR, on the other hand, reacts with cytidine. Additionally, the CytR-controlled promoters are activated by the cAMP receptor protein (Valentin-Hansen et al., 1996 ). In B. subtilis, the genes identified corresponding to those of the E. coli regulon are located in two operons: the drmpupG operon and the dranupCpdp operon, in addition to the single genes cdd, encoding cytidine deaminase, and pupA. Expression of pupA and cdd has been found not to respond to nucleosides in the growth medium (Nygaard, 1993 ). The dranupCpdp operon is regulated by the DeoR protein (Saxild et al., 1996 ; Zeng & Saxild, 1999 ) and the drmpupG operon is not. However, the expression of both operons is increased by deoxyribonucleosides. Additionally, only the expression of drmpupG is affected by the presence of ribonucleosides. Our finding that the transcription of the drmpupG operon started at two different sites could suggest distinct ribose 5-phosphate- and deoxyribose 5-phosphate-controlled start sites. This turned out not to be the case, as similar increases in transcription from both the drmP1 and drmP2 promoter were detected in the presence of either thymidine or inosine (Fig. 3). Most likely there is a protein that negatively regulates the initiation of transcription, and which recognizes both ribose 5-phosphate and deoxyribose 5-phosphate. We observed a major difference between growth on inosine or thymidine as the carbon source, compared with growth on free ribose and deoxyribose. Significant growth was only observed when thymidine or ribose served as the sole carbon source. While the level of the nucleoside-catabolizing enzymes is comparable to that of E. coli, the nucleoside-transport activity is lower in B. subtilis. We therefore suggest that what limits the catabolism of nucleosides in this organism is the transport of nucleosides. In agreement with this is our finding that an E. coli strain carrying a plasmid (pHM2) with the cloned drm gene grew well on inosine as the carbon source although the Drm level was only 30% of that in wild-type B. subtilis, which cannot grow on inosine as the sole carbon source.

Synthesis of the nucleoside-catabolic enzymes in B. subtilis is subject to catabolite repression (Saxild et al., 1996 ). Catabolite repression can occur via a common regulatory mechanism that involves the cis-acting cre element, 5'-TGWAANCGNTNWCA-3', which is active whether located in the promoter region or within a gene (Miwa et al., 1997 ). Two such putative elements were found in the drmpupG operon. The first putative cre site (nucleotide 24477392447726) has 12 out of 14 matches with the common sequence, and overlaps with the putative -10 region of the drmP1 promoter and extends to within 2 nt of the putative -35 region of the drmP2 promoter. The second putative cre site (nucleotide 24461682446155) lies in the pupG coding region and matches 11 out of 14 positions. For both cre sites the mismatches are located at the 3' end. The presence of these sites offers an explanation for the observed induction pattern (Table 3): that the extent of induction of Drm and PupG in particular was lower in glucose-grown cultures than in succinate-grown cultures.

A few studies dealing with nucleoside catabolism in other bacilli have shown that the levels of phosphopentomutase and purine-nucleoside phosphorylase in Bacillus cereus and Bacillus thuringiensis are increased by nucleosides and nucleotides in the growth medium (Gardner & Kornberg, 1967 ; Ipata et al., 1983 ; Grigorieva & Sukhodolets, 1979 ). Another aspect of purine-nucleoside metabolism found in E. coli and incidentally also in B. cereus is the involvement of purine-nucleoside phosphorylases and adenosine deaminase in the interconversion of adenine compounds to guanine compounds. Such a pathway is not present in B. subtilis (Nygaard et al., 1996 ).

We wish to thank Jenny Steno Christensen for excellent technical assistance. This work was supported by Public Health Service grant GM43577 (PJP) and training grant 5 T32 AI07101 (RS) from the National Institutes of Health, USA and by the Danish Centre for Microbiology grants to PN and AG.