The methionine salvage pathway in Bacillus subtilis

Background Polyamine synthesis produces methylthioadenosine, which has to be disposed of. The cell recycles it into methionine through methylthioribose (MTR). Very little was known about MTR recycling for methionine salvage in Bacillus subtilis. Results Using in silico genome analysis and transposon mutagenesis in B. subtilis we have experimentally uncovered the major steps of the dioxygen-dependent methionine salvage pathway, which, although similar to that found in Klebsiella pneumoniae, recruited for its implementation some entirely different proteins. The promoters of the genes have been identified by primer extension, and gene expression was analyzed by Northern blotting and lacZ reporter gene expression. Among the most remarkable discoveries in this pathway is the role of an analog of ribulose diphosphate carboxylase (Rubisco, the plant enzyme used in the Calvin cycle which recovers carbon dioxide from the atmosphere) as a major step in MTR recycling. Conclusions A complete methionine salvage pathway exists in B. subtilis. This pathway is chemically similar to that in K. pneumoniae, but recruited different proteins to this purpose. In particular, a paralogue or Rubisco, MtnW, is used at one of the steps in the pathway. A major observation is that in the absence of MtnW, MTR becomes extremely toxic to the cell, opening an unexpected target for new antimicrobial drugs. In addition to methionine salvage, this pathway protects B. subtilis against dioxygen produced by its natural biotope, the surface of leaves (phylloplane).


Background
The fate of methylthioribose (MTR), the end-product of spermidine and spermine metabolism, as well as of ethylene biosynthesis has not yet been fully explored in most organisms. In Escherichia coli this molecule is excreted in the medium [1] while in Klebsiella pneumoniae it constitutes the methionine salvage pathway, being metabolized back into methionine [2,3]. In eukaryotic parasites it is also recycled into methionine, presumably through a pathway similar to that in K. pneumoniae [4]. In Bacillus subtilis we found that MTR is an excellent sulfur source [5] and we unraveled some of the steps involved in its metabolism, which starts from phosphorylation of MTR, mediated by the MtnK protein [6].
It has been shown previously that the ykrW gene, annotated as similar to ribulose phosphate carboxylase/oxygenases (Rubisco) in the original sequence, has links with sulfur metabolism. Indeed, Henkin and co-workers found that the corresponding coding sequence (CDS) was preceded by a S-box typical of sulfur metabolism genes in B. subtilis [7] and Hanson and Tabita found that two classes of enzymes similar to Rubisco were associated with sulfur metabolism [8]. Interestingly, these authors, working with Chlorobium, which uses the reverse TCA cycle for CO 2 fixation, postulated these proteins as possible precursors of cyanobacteria, then plant Rubisco at an early time of the development of life on Earth.
This raised interesting questions about the origin of this pathway. In particular the YkrW gene origin could have been early in evolution, or resulting from lateral transfer from plants to bacilli. We demonstrate here that proteins YkrUWXYZ are needed for MTR recycling into methionine in B. subtilis, while YkrV, an aminotransferase, is probably more specific of methionine transamination, but is dispensable in the present conditions because of the present of a variety of izozymes (up to nine amino acid transaminases are present in B. subtilis).

Transposon insertion mutations and phenotype of inactivated mutants
The MTR analog trifluormethylthioribose (3F-MTR) is toxic if the methyl sulfur moiety of the molecule is recycled [9]. This molecule was therefore an excellent candidate to explore the steps needed for MTR recycling to methionine. Mutants were obtained by transformation of a wild type strain with a random transposon library, selecting for growth in the presence of 3F-MTR in the presence of sulfate as sulfur source. The mutants were subsequently tested for growth on plates lacking sulfur source but supplemented with MTR: only those that could not grow were retained for further study. In order to ascertain that the resistant phenotype was not coming from secondary mutations but was directly related to the trans-poson insert, the chromosome DNA was extracted from each putative mutant and back transformed into a wild type strain selecting for the transposon antibiotic marker. The 3F-MTR and MTR phenotypes were subsequently tested and only those mutants that passed the test were retained. The insertion positions of the transposons were then sequenced. As shown in Figure 1 we recovered mutants in several genes located in the close vicinity of each other. One mutant was located at the mtnK locus (previously named ykrT [6]), four were located into ykrW and six into the ykrY gene. One clone with transposon insertion into the ykrW gene (strain BSHP7064, insertion situated 353 bp downstream of ykrW translation start point) and one into the ykrY gene (strain BSHP7065, insertion situated 556 bp downstream of ykrY translation start point) were retained for further studies.
Using the collection of the lacZ transcriptional fusion mutants constructed during the Bacilllus subtilis functional analysis program ( [http://locus.jouy.inra.fr/cgi-bin/genmic/madbase/progs/madbase.operl] and [http://bacillus.genome.ad.jp] , [10]) and constructing mutants which were not available in the collection, we tested all genes in the region for their phenotype of growth on MTR as the sole sulfur source. Table 1 displays the results obtained. As we can see, mutants in mtnK (previously identified as coding for MTR kinase [6], strain BFS1850), mtnS (strain BSHP7010 [6]), ykrU (renamed mtnU, strain BFS1851), ykrW (renamed mtnW, strain BSHP7014 allowing the expression of downstream genes) and ykrY (renamed mtnY, strain BSHP7016) failed to grow on the substrate. In the absence of IPTG ykrX (renamed mtnX, strain BFS1852) also failed to grow, but it recovered its growth properties when IPTG was added to the medium, suggesting some polar effect of the transposon insertion. The mutant of ykrZ (renamed mtnZ, strain BFS1853) presented only a very weak (residual) growth on MTR, suggesting the presence in the cell of some other enzymatic activity able to partially complement the lack of mtnZ gene product. Disruption of ykrV (renamed mtnV, strain BSHP7020) had no visible effect on growth on MTR as the sulfur source.

Identification of promoters
Several genes in the region have been shown by Henkin and co-workers to be expressed from promoters regulated by the S-box attenuation system [7]. This is the case of mtnKS and mtnWXYZ transcription units. Some of the genes, however, are not regulated in this way. Expression of the mtnU and mtnV genes is not subject to that regulation since no S-box is present in their leader transcript. As shown in Figure 2A the promoter of mtnU is located 35 nt from the translation start point. Its start was found to lie 5 nt downstream from a putative -10 box identified in the sequence (TTAAAT). Upstream from this box separated by 18 nt is a -35 box (ATGATA) with sequence similar to the consensus sequence TTGACA that is typical of B. subtilis sigmaA-dependent promoters [11].
The promoter of mtnV is located 42 nt upstream from the translation start point. Its start lies 8 nt downstream from a putative -10 box identified in the sequence (TATGAT) separated by 17 nt from -35 box (TTTACT) (see Fig. 2.B). The mtnU and mtnV genes share the same promoter region (94 nt) but are transcribed in divergent orientation from overlapping promoters. Thus, the -10 box of the mtnV promoter is situated between the -10 and -35 boxes of the mtnU promoter and the -10 box of the mtnU promoter is situated between the -10 and -35 boxes of the mtnV promoter.
The mtnKS promoter region is 326 nt long. Its start was found to lie 7 nt downstream from a putative -10 box identified in the sequence (TACCAT) (see Fig. 2.C). Upstream from this box and separated by 18 nt is a -35 box (TTGACA), a typical B. subtilis sigmaA-dependent promoter. Downstream of this promoter lies an S-box regulatory sequence.
Genes mntWXYZ are expressed from two overlapping promoters that are situated in a 195 nt long region. The upstream P1 promoter's start was found to lie 7 nt downstream from a putative -10 box identified in the sequence (GATAAT) separated by 17 nt from a consensus -35 box (TTGACA). The P2 promoter's start was found to lie 7 nt downstream from a putative -10 box identified in the sequence (TAAAAT) upstream from which is a -35 box (ATGGGA) (see Fig. 2.D). This promoter region is also reminiscent of regions recognized by the developmental sigmaH transcriptional regulator. The -10 box of the P1's promoter and the -35 box of P2 are partly overlapping. The relative intensity of the signals indicates that transcription from the P1 promoter is more abundant than from P2 (see Fig. 2.D).

Transcription organisation of the mtn locus
To further investigate transcription of the mtnVWXYZ genes, RNA synthesis was analyzed by Northern blotting. RNA was extracted from exponentially growing cells, in minimal medium containing either sulfate or methionine as sulfur source. As shown in Figure 3.A, a band of about 1200 nt, corresponding to the expected length of a transcript initiated at the mtnV promoter and terminating near its stop codon, was observed for the mtnV gene probe. An equal intensity of the signal was observed for mtnV transcripts prepared from cells either grown with sulfate or with methionine as sulfur source (lanes 1 and 2, Fig. 3.A).
When mtnW and mtnZ gene specific probes were used, two bands were revealed: one of about 2.5 kb and second of about 3.2 kb (Fig. 3.B and 3.C)). The larger band corresponds to the expected length of a transcript initiated at the mtnW promoter and terminating in a stem and loop structure at the end of the mtnWXYZ transcriptional unit. The smaller band can possibly be the result of RNA processing at the end of the S-box regulatory sequence of the 5' extremity of the transcript. The intensity of bands when hybridizing RNA from cells grown with sulfate as sulfur source was higher than when using RNA from cells grown in the presence of methionine (lane 1 and 2 in Fig.  3.B and 3.C).
As shown previously, mtnK and mtnS are expressed as an operon, while mtnU is expressed independently [6].

Regulatory features
To substantiate the results obtained with RNA analysis and further investigate the expression of genes from the

Figure 2
Identification of the mtn region promoters by primer extension. A. Identification of the transcription start site of the mtnU operon. The size of the extended product is compared to a DNA-sequencing ladder of the mtnU promoter region. Primer extension and sequencing reaction were performed with the same primer. The +1 site is marked by an arrow. B. Identification of the transcription start site of the mtnV gene. The size of the extended product is compared to a DNA-sequencing ladder of the mtnV promoter region. Primer extension and sequencing reaction were performed with the same primer. The +1 site is marked by an arrow. C. Identification of the transcription start site of the mtnKS gene. The size of the extended product is compared to a DNA-sequencing ladder of the mtnKS promoter region. Primer extension and sequencing reaction were performed with the same primer. The +1 site is marked by an arrow. D. Identification of the transcription start site of the mtnWXYZ operon. The size of the extended product is compared to a DNA-sequencing ladder of the mtnWXYZ promoter region. Primer extension and sequencing reaction were performed with the same primer. Two +1 sites are marked by arrows. E. Sequences of the corresponding promoter regions. Promoter sites are in capital letters and underlined (-35 and -10 boxes), and the transcription start sites are indicated by broken arrows (+1).
mtn region, we constructed mutants carring lacZ transcriptional fusions as well as used some mutants constructed during the functional analysis program (also corresponding to lacZ transcription fusions). Table 2 shows the results obtained with these strains when using sulfate or methionine as sulfur source. The mtnU gene (strain BFS1851, mtnU::lacZ) is expressed constitutively at a fairly low level and its expression is independent of the sulfur source used (62 U/mg of protein in the exponential growth phase in presence of sulfate and 53 U/mg of protein in the exponential growth phase in presence of methionine). In contrast, the mtnV gene (strain BSHP7020, mtnV::lacZ) although expressed in the similar way (constitutive and sulfur source independent expression) is expressed at a significantly higher level (217 U/mg of protein in the exponential growth phase in presence of sulfate and 181 U/mg of protein in the exponential growth phase in presence of methionine).
The genes from the mtnWXYZ transcriptional unit (strains BSHP7014, BFS1852, BSHP7016 and BFS1853 for mt-nW::lacZ, mtnX::lacZ, mtnY::lacZ and mtnZ::lacZ, respectively) are expressed in a coordinated and sulfur sourcedependent way. The expression of the first gene in the operon (mtnW) is higher than that of the last one (mtnZ) with intermediary values for intermediary genes mtnX and mtnY. This suggests the effect of some transcription polarity during the process of transcription (see Table 2). A 5fold difference is observed between the expression of the mtnWXYZ genes in the presence of sulfate and that in the presence of methionine (579 U/mg of protein in the exponential growth phase in the presence of sulfate and 113 U/ mg of protein in the exponential growth phase in the presence of methionine for the mtnW::lacZ transcriptional fusion and 280 U/mg of protein in the exponential growth phase in presence of sulfate and 57 U/mg of protein in the exponential growth phase in presence of methionine for mtnZ::lacZ transcriptional fusion). This observation is in accordance with the presence of S-box regulatory element in the promoter region of mtnWXYZ operon which modulates gene expression as a function of methionine availability [7].

Reconstruction of the metabolic pathway
In order to identify the methionine salvage pathway we made constructs allowing us to decipher the order of the gene products in the pathway, together with in silico, physiologic and genetic analysis of the effect of metabolites of the pathway. This is reminiscent of the way advocated by Koonin et al. for the use of in silico approaches as complement to in vivo experiments [12].
As a first goal we showed that the end product of the pathway is indeed methionine. This was demonstrated by showing that MTR, which is a good sulfur source, can be used as the methionine source in methionine auxotrophs (Fig. 4, Fig. 5 and data non shown).
Two genes in the pathway are dispensable, mtnV and mt-nX. The first one encodes a transaminase of which there are nine putative paralogs in the genome of B. subtilis (YwfG, AlaT, AspB, PatA, YhdR, YdfD, PatB, YisV, and HisC). In the same way, MtnX (YkrX) is a member of the phosphatase family pfam00702 ( [13], Fig. 6), and therefore of a ubiquitous class of hydrolases (several phosphatase genes in particular are present in the genome of B. subtilis). This is likely to account for the lack of phenotype under our growth conditions. Inactivation of mtnZ provides only a very weak, residual growth on MTR. Inactivation of mtnK, mtnS, mtnY and mtnW result in resistance to 3F-MTR and lack of growth on MTR. Inactivation of mtnW with a polar effect on the distal genes (by insertion of a disrupting plasmid) has a phenotype similar to that of mtnY (i.e. lack of growth on MTR, and lack of influence of MTR on sulfate supplemented plates). In contrast, we discovered that MTR is toxic when the distal genes are present (when used as sole sulfur source or even in the presence of sulfate, see Fig. 7). Because of the weak phenotype of a mtnZ mutant and the absence of phenotype of a mtnX mutant, we can be confident that MtnY acts before MtnW (this is a regular feature in operons, where it is often observed that the more distal genes code for proteins acting in the more proximate steps of the pathway).
The methionine salvage pathway has been deciphered in K. pneumoniae. It is possible, combining this knowledge to the genetic and physiologic results just described, to use it at the basis for reconstructing in silico the corresponding metabolic pathway in B. subtilis. The first steps are similar in both organisms: methylthioadenosine is converted into MTR by a nucleosidase (MtnA, [5]). Subsequently, MTR is phosphorylated into MTR-1-phosphate by MtnK [6]. On the other end of the pathway, methionine is syn-thesized directly from its keto acid precursor, 2-keto-4methylthiobutyrate, by a transaminase. MtnV is the likely preferred enzyme for this activity. In K. pneumoniae a dioxygenase is converting 2,3-diketo-5-methylthio-1-phosphopentane into 2-keto-4-methylthiobutyrate [2]. Using dynamic programming (FASTA) we compared the sequence of the corresponding protein to the complete proteome of B. subtilis. YkrZ comes out as the first hit, as the most similar enzyme present in the proteome. Furthermore, it displays a strong consensus similarity with the dioxygenases of the family pfam03079 (Fig. 8) [13]. In order to check whether dioxygen was indeed involved in the case of B. subtilis we grew the cells anaerobically, with nitrate as an electron acceptor, and tested for growth on MTR: while the wild type strain grew well when sulfate was the carbon source, it failed to grow with MTR (Table  1).
Since this dioxygenase is coded in the mtn operon we can infer that it indeed displays the corresponding activity [12], and we therefore renamed it MtnZ. In K. pneumoniae, the immediate precursor activity is that of a coupled phosphatase. The presence of MtnX, which belongs to family pfam00702 comprising phosphatases is strongly suggestive of its involvement at this step [13]. We are thus left with two enzymes, and two steps. We also know, from the genetic data, that the steps are catalyzed in the order Mt-nY, MtnW. Finally, the reaction needed upstream of MtnZ is active on a molecule phosphorylated in position 1.
MtnW is very similar to ribulosephosphate carboxylase oxygenase (Rubisco). It is therefore likely to be active on a ribulose-1-phosphate derivative. Hence MtnY, which is similar to the araD gene product of E. coli (ribulose-5- At this stage it is difficult to explicitly identify the activity of MtnW. Even in the case of the paradigmatic Rubisco, with many crystal structures known, the exact mechanism of catalysis is still a matter of controversy. However we can note (as did [8], at a time when they could not propose a function for the protein) that all the residues involved in catalysis have been conserved, the only residues modified being those involved in the binding of the phosphate at position 5 of ribulose diphosphate. The reaction is that of a dehydratase, but the pathway of the reaction is not yet known. Further work will establish the details of the reaction.
Finally MtnU is also defective for MTR recycling. However, this protein is synthesized at a level much lower that that of the other components of the pathway. We can therefore surmise that it is involved in a regulatory step in the pathway.

Discussion
Several genes in the vicinity of mtnK have been shown to have significant relationships with sulfur metabolism. In particular, it has been known for some time that genes ykr-WXYZ were preceded by an S-box, typical of sulfur mediated regulation [7]. In addition, while analyzing the function of ribulose-1,5-diphosphate carboxylase (Rubisco), Hanson and Tabita discovered a class of highly related enzymes that were involved in sulfur metabolism [8].
Allowing cells to grow in the presence of the toxic MTR analog 3F-MTR, resistant mutants were found in genes mtnK, mtnW and mtnY. The trifluoromethyl group of 3F-MTR is the toxic group of the molecule and it was expected that

Figure 4
The MTR recycling pathway in B. subtilis.

Figure 5
Growth of mutants from the mtn region with MTR as sole sulfur source. Panel A: ED1 minimal medium plate with 1 mM IPTG containing 0.2 mM MTR as sole sulfur source WT, metI (BSIP1143), mtnS (BSHP7010), mtnK (BFS1850), mtnU (BFS1851), mtnV (BSHP7020), mtnW (BSHP7014), mtnX (BFS1852), mtnY (BSHP7016) and mtnZ (BFS1853) were inoculated for over-night growth at 37°C. No growth of mtnS, mtnK, mtnW and mtnY is represented by an example of absence of growth around a disc with MTR of mtnY mutant in panelB. Normal growth of mtnV and mtnX is represented by an example of normal growth around a disc with MTR of mtnV mutant in panel C. The partial growth of the mtnZ mutant is illustrated by its growth around a disc with MTR in panel C. Panel B: The mtnY strain (BSHP7016) was inoculated on ED1 minimal medium plate with no added sulfur source. 10 µl of methionine (met) or MTR was put on paper discs and the plate was incubated over-night at 37°C. Panel C: The mtnV strain (BSHP7020) was inoculated on ED1 minimal medium plate with no added sulfur source. 10 µl of methionine (met) or MTR was put on paper discs and the plate was incubated over-night at 37°C. Panel D: The mtnZ strain (BFS1853) was inoculated on ED1 minimal medium plate with no added sulfur source. 10 µl of methionine (met) or MTR was adsorbed on paper discs and the plate was incubated over-night at 37°C. Methionine was used as a control.
inactivation of any gene coding for the steps, including permeation, that led from MTR to methionine (the ultimate methyl donor, since AdoMet is an essential metabolite for the cell) would result in a resistant phenotype.
Remarkably, no permease gene was found, suggesting that MTR enter the cells via several entries. In addition, apart from the mtnKS and mtnWXYZ operons no other genes was found, suggesting that all essential steps for recycling are coded for by these genes (or that other steps are coded for by redundant genes). The first step of the metabolic pathway is phosphorylation of MTR. The last step, presumably, is transamination, with mtnV being the preferred transaminase.
Interestingly, the pathway described in this work, although similar to that found by the pioneering work of Abeles and co-workers, uses an enzyme, MtnW, which is extremely similar to Rubisco but not present in K. pneumoniae orK. oxytoca [14,15]. While most of the genome sequence of this bacterium is known [http:// wit.integratedgenomics.com/GOLD/] no counterpart of MtnW could be found (data not shown). The corresponding activity exists in K. pneumoniae, but no corresponding gene has yet been isolated. As discovered by Hanson and Tabita, MtnW counterparts constitute a special class (class IV) of Rubisco-like enzymes, which are involved in sulfur metabolism: we can presume that they are all part of the methionine salvage pathway in these organisms [8]. Interestingly, the expected reaction required to metabolize 5thiomethyl-ribulose-1-phosphate is that of a dehydratase that may use a co-factor as a substrate for the reaction [16]. Rubisco, in the presence of carbon dioxide (resp. dioxygen), acts as a carboxylase (resp. dioxygenase) which cleaves the substrate. In the present case we expect that, instead of cleavage, we have maintenance of a five carbon molecule that is dephosphorylated (by MtnX) and subsequently cleaved by dioxygen in the reaction mediated by MtnZ. The counterpart of this activity in other genomes, including those with a methionine salvage pathway is not known.
As a strong support of this schema, we found counterparts of MtnK and of MtnZ in K. pneumoniae, substantiating the proposed pathway. In this latter organism the counterpart of MtnY is not known, and the corresponding step (opening of the MTR-1-P ring with epimerisation) is not known in any organism yet. MtnY is part of a very wide family of aldolases-epimerases-transketolases and in silico prediction of function alone, at this stage is highly problematic (wrong assignment is frequent for similar functions [17]), but combination with genetic data make the prediction highly probable [12]. We therefore propose that MtnY be used as a basis for annotation of similar gene products. For example in Xylella fastidiosa, gene XF2209 and in Pseudomonas aeruginosa gene PA1683 probably encodes the cognate activity. Noticeably, a counterpart exists in the Human Genome, where a similar pathway operates.
Two gene products are not directly accounted for in the present schema, MtnU and MtnS. MtnU is expressed at a very low level (ten times lower) as compared to MtnW, and this would hardly fit with the expected stoichiometric enzyme concentration usually found in multistep metabolic pathways. In addition, we found in this work that its synthesis is not submitted to any regulation by the sulfur source. Similarly, MtnS, which is highly similar to an eukaryotic translation initiation factor eIF-2B involved in GTP/GDP exchange is a member typical of a class of GTPdependent regulators. The presence of two regulator molecules in this pathway indicates that it must have an important role in the cell. B. subtilis is likely to strive on the phylloplane. It is therefore regularly submitted to very high local concentrations of oxygen, and we speculate that this pathway, in addition to providing an excellent means to recycle the energy costly methionine, is used as a means to protect the cell against oxygen.

Conclusions
This work demonstrates that a complete methionine salvage pathway exists in B. subtilis. This pathway is chemically similar to that in K. pneumoniae, but recruited different proteins to this purpose. In particular a paralogue or Rubisco, MtnW, is used at one of the steps in the pathway. A major observation stemming from the present experiments is that in the absence of MtnW MTR becomes extremely toxic to the cell. This sensitivity opens an unexpected target for new antimicrobial drugs, since analogs of 5-methylthio-ribulose1-phosphate might have a strong inhibitory effect on growth on bacteria containing this methionine salvage pathway, including Bacillus anthracis.

Materials and methods
Bacterial strains and plasmids, and growth media E. coli and B. subtilis strains as well as plasmids used in this work are listed in Table 3 When methionine was used as sulfur source (1 mM), MgSO 4 was replaced by MgCl 2 at the same magnesium concentration (2 mM). For assaying growth on plates, either the MgSO 4 containing medium or the sulfur-free basal medium was used (MgSO 4 was replaced by MgCl 2 as decribed above). In the latter case, 10 µl of the sulfur source under investigation was applicated onto paper discs (MTR, 200 mM stock solution and methionine, 100 mM stock solution) deposited at the center of the plate, after bacteria had been uniformly spread at the surface of the plate, and growth was measured around the disk. In some cases MTR was used directly in the plate as sulfur source (0.2 mM). When necessary IPTG was included at 1 mM concentration. When xylose was added to the medium (0.5%) in order to trigger the expression of genes under the control of Pxyl inducible promoter, fructose was used as carbon source instead of glucose. LB and ED plates were prepared by addition of 17 g/liter Bacto agar or Agar Noble (Difco), respectively, to the medium. When included, antibiotics were added to the following concentrations: ampicillin, 100 mg/liter; chloramphenicol, 50 mg/liter; spectinomycin, 100 mg/liter; erythromycin plus lincomycin, 1 mg/liter and 25 mg/liter. Bacteria were grown at 37°C. The optical density (OD) of bacterial cultures was measured at 600 nm. MTR was prepared from MTA (Sigma, D5011) by acid hydrolysis as described by Schlenk [19]. 3-fluoromethythiorybose (3F-MTR, 5-thio-5-S-trifluoromethyl-D-ribose) was synthesised accordingly to [6,20]. For mutant selection, 3F-MTR was used at 100 mg/liter concentration in the ED minimal medium plates containing magnesium sulfate as sulfur source and glucose as carbon source. When applicated onto paper discs 10 µl of 100 mM stock solution of 3F-MTR was used. For anaerobic growth on plates, the Anaerocult A (Merck) within an anaerobiosis jar for CO2 production with simultanious O2 absorbtion was used. Sulfur-free ED minimal medium plates were supplemented to 1% glucose final concentrtion and with 0.5% sodium pyruvate and 20 mM sodium nitrate as electron acceptor. Plates were incubated at 37°C for 4 days with the sulfur source under investigation.

Transformation
Standard procedures were used to transform E. coli [21] and transformants were selected on LB plates containing ampicillin, spectinomycin or ampicillin plus spectinomycin. B. subtilis cells were transformed with plasmid DNA following the two-step protocol described previously [22]. Transformants were selected on LB plates containing erythromycin plus lincomycin or spectinomycin or chloramphenicol.

Molecular genetics procedures
Plasmid DNA was prepared from E. coli by standard procedures [21]. B. subtilis chromosomal DNA was purified as described by Saunders [23]. Restriction enzymes and T4 DNA ligase were used as specified by manufacturers.
DNA fragments used for cloning experiments were prepared by PCR using PfuTurbo DNA polymerase (Stratagene). Amplified fragments were purified by QIAquick PCR Purification Kit (Qiagen). DNA fragments were purified from a gel using Spin-X columns from Corning Costar by subsequent centrifugation and precipitation.
The mtnXYZ region (nucleotides -31 relative to the mtnX translation start point and ending 3 bp after the stop codon of mtnZ) was amplified by PCR using primers introducing a SpeI cloning site at the 5' end and a BamHI cloning site at the 3' end of the fragment. This fragment was then inserted into the SpeI and BamHI sites of xyloseinductible pX plasmid [24] producing plasmid pHPP7015. Prior to transformation, this plasmid was linearised at its unique ScaI site. Complete integration of the plasmid was obtained by a double cross-over event at the amyE locus, giving strain BSHP7015.
The DNA downstream from the mtnW gene (nucleotides +41 to +257 relative to the translation start point) was amplified by PCR using primers introducing an EcoRI cloning site at the 5' end and a BamHI cloning site at the 3' end of the fragment, then inserted into the EcoRI and BamHI sites of plasmid pJM783 [25] producing plasmid pHPP7014.
To introduce an additional antibiotic resistance gene into plasmid pHPP7014, a SmaI restricted spectinomycin resistance cassette [26] was inserted into the ScaI restriction site of the bla gene producing plasmid pHPP7014bis. The plasmid in which the mtnW gene was disrupted as well as fused (transcriptional fusion) with the lacZ gene was introduced into the chromosome of BSHP7015 strain by a single cross-over event, giving strain BSHP7014.
For transcriptional fusion of mtnY with the lacZ gene, a DNA segment downstream from the mtnY gene (nucleotides + 57 to + 264 relative to the translation start point) was amplified by PCR using primers introducing an EcoRI

Figure 7
Toxicity of MTR for BSHP7014 strain. Strain BSHP7014 (mtnW::lacZ amyE::pxyl mtnXYZ) was grown on ED1 minimal medium plates in the presence of sulfate as sulfur source (panel A) or in the absence of any added sulfur source (agar as sole sulfur source, panel B). Xylose was added to the medium in order to trigger the expression of mtnXYZ from the pxyl promoter. 10 µl of methionine (met) or MTR was adsorbed on paper discs and plates were incubated overnight at 37°C. Methionine was used as a control for growth and/or toxicity of the sulfur source.
cloning site at the 5' end and a BamHI cloning site at the 3' end of the fragment, then inserted into the EcoRI and BamHI sites of plasmid pJM783 producing plasmid pHPP7016. The plasmid in which the mtnY gene was disrupted as well as fused (transcriptional fusion) with the lacZ gene was introduced into the chromosome by a single cross-over event, giving strain BSHP7016.
To construct a mtnV transcriptional fusion with the lacZ gene, a DNA fragment downstream from the mtnV gene (nucleotides + 44 to + 259 relative to the translation start point) was amplified by PCR using primers introducing an EcoRI cloning site at the 5' end and a BamHI cloning site at the 3' end of the fragment, then inserted into the EcoRI and BamHI sites of plasmid pJM783 producing plasmid pHPP7011. The plasmid in which the mtnV gene was disrupted as well as fused (transcriptional fusion) with the lacZ gene was introduced into the chromosome by a single cross-over event, giving strain BSHP7020.  Table 3.

Transposon mutagenesis
A transposon bank was constructed by introduction of the mini-Tn10 delivery vector pIC333 [27] into the B. subtilis 168 strain as described previously [28]. Several thousand independent clones were pooled together and 5 samples of chromosomal DNA were prepared for further use.

RNA isolation and manipulation
Total RNA was obtained from cells growing on ED1 minimal medium with sulfate or methionine as sulfur source to an OD 600 of 0.5 using "High Pure RNA Isolation Kit" from Roche. The RNA concentration was determined by light absorption at 260 nm and 280 nm. 2 µg of RNA were loaded onto 1.2% agarose gel to check the RNA purity and integrity.
RNA molecules were separated on 1% agarose gels and transfered to nylon membranes (Hybond-N, Amersham). Efficiency of transfer was monitored by analysis of ethidium bromide-stained material. Membranes were prehybridized at 50°C for 1 hr in DIG Easy Hyb buffer from Roche. Hybridization was performed under the same conditions with mtnV, mtnW or mtnZ specific probes using a non-radioactive DNA labeling and detection kit "Dig-UTP labeling" from Roche.
Primer extension analysis using reverse transcriptase AMV (Roche) was performed as described by [29] with two oligonucleotides for each promoter identification. For mtnKS promoter the followings primers were used: 5'ACCAGCGTCTCGGCGCGAAAAAAATGCGCCCC3'  and 5'TCACAATGGAATTACGGTCGGTTGCTTTTGG3' (+137 to +169 and +172 to +203 with respect to the translation start point, respectively); for the mtnU promoter the followings primers were used: 5'AGTTCATCAAGATTGGCCAGATCATATCCG3' and 5'CAGGCAGAACAAGAACATCAGCATGTTTGC' (-133 to -103 and -90 to -60 with respect to translation start point, respectively); for the mtnV promoter the followings primers were used: 5'GTTTCATCTCCTCAACAATATGCTCAG-GAG' and 5'TCCCAGATTGATAACGTCATGTCCTTCTGC' (-166 to -146 and -114 to -84 with respect to the translation start point, respectively); for the mtnWXYZ promoter the followings primers were used: 5'CGTTTCTCGTC-CGAATCTTATCTCTCAGCC' and 5'AGCTGCAAGAATT-AGCACCGTGCTTTATAAG' (+43 to +73 and +76 to +107 with respect to the translation start point, respectively). The same primers were used for the generation of sequence ladders. Reaction products were separated on 7% denaturing polyacrylamide gel containing 8 M urea. DNA sequences were determined using Sanger's dideoxy chaintermination method with "Thermo Sequenase radiolabeled terminator cycle sequencing kit" from Amersham Pharmacia Biotech.

Enzyme assays
B. subtilis cells containing lacZ fusions were assayed for βgalactosidase activity as described previously [30]. Specific activity was expressed in Units per mg protein. The Unit used is equivalent to 0.28 nmols min-1 at 28°C. Protein concentration was determined by Bradford's method using a protein assay Kit (Bio-Rad Laboratories). At least two independent cultures were monitored.
Amylase activity was detected after growth of B. subtilis strains on Tryptose Blood Agar Base (TBAB, Difco) supplemented with 10 g/liter hydrolyzed starch (Sigma). Starch degradation was detected by sublimating iodine onto the plates.