Transcription Factors CysB and SfnR Constitute the Hierarchical Regulatory System for the Sulfate Starvation Response in Pseudomonas putida

ABSTRACT Pseudomonas putida DS1 is able to utilize dimethyl sulfone as a sulfur source. Expression of the sfnFG operon responsible for dimethyl sulfone oxygenation is directly regulated by a σ54-dependent transcriptional activator, SfnR, which is encoded within the sfnECR operon. We investigated the transcription mechanism for the sulfate starvation-induced expression of these sfn operons. Using an in vivo transcription assay and in vitro DNA-binding experiments, we revealed that SfnR negatively regulates the expression of sfnECR by binding to the downstream region of the transcription start point. Additionally, we demonstrated that a LysR-type transcriptional regulator, CysB, directly activates the expression of sfnECR by binding to its upstream region. CysB is a master regulator that controls the sulfate starvation response of the sfn operons, as is the case for the sulfonate utilization genes of Escherichia coli, although CysBDS1 appeared to differ from that of E. coli CysB in terms of the effect of O-acetylserine on DNA-binding ability. Furthermore, we investigated what effector molecules repress the expression of sfnFG and sfnECR in vivo by using the disruptants of the sulfate assimilatory genes cysNC and cysI. The measurements of mRNA levels of the sfn operons in these gene disruptants suggested that the expression of sfnFG is repressed by sulfate itself while the expression of sfnECR is repressed by the downstream metabolites in the sulfate assimilatory pathway, such as sulfide and cysteine. These results indicate that SfnR plays a role independent of CysB in the sulfate starvation-induced expression of the sfn operons.

Sulfur, an essential element for bacterial growth, is assimilated from a range of sources. The preferred sulfur source for bacteria is inorganic sulfate, but in soil environments, sulfate may not be freely available because inorganic sulfate constitutes less than 5% of the total sulfur (1). The remainder is largely in the form of sulfate esters and carbon-bound sulfur such as peptides/amino acids and sulfonates (1). Volatile organic sulfur compounds, for example, dimethyl sulfide (DMS) and methanethiol, are also widely distributed in the environment because they are emitted from oceans (2), freshwater sediments (28,29), and soils (26,31). Thus, several soil bacterial species have developed a number of systems that allow the use of these organosulfur compounds.
Metabolic use of these organosulfur compounds, especially alkanesulfonates, taurine, and alkanesulfate ester, has been extensively investigated in Escherichia coli and Pseudomonas spp. (reviewed in reference 19). Bacterial organosulfur assimilation is mediated by the sulfate starvation response, which involves a set of proteins synthesized by several species of bacteria when the supply of their preferred sulfur sources, such as inorganic sulfate, cysteine, and thiosulfate, is limited. These proteins are known as sulfate starvation-induced (SSI) proteins and include enzymes and transport systems for scavenging sulfur from organic compounds (19).
Transcriptional regulation of the SSI genes has been investigated extensively in E. coli, but knowledge is limited to the two well-characterized operons involved in sulfonate and taurine utilization, i.e., ssuEADCB and tauABCD, respectively (reviewed in reference 38). In E. coli, a LysR-type transcriptional regulator, CysB, which is highly conserved among gramnegative bacteria, plays a key role not only in the expression of the cys genes responsible for sulfate assimilation but also in that of the SSI genes. Full expression of the CysB-regulated genes requires a signal for sulfur limitation provided by an internal inducer molecule, N-acetylserine or O-acetylserine (34). Additionally, sulfide and thiosulfate act as anti-inducers by inhibiting the binding of CysB protein to cys promoters (12,35). A LysR-type transcriptional regulatory gene, cbl, one such CysB-regulated gene, is also required for the expression of ssu and tau operons (15). Under sulfate starvation conditions, Cbl binds to upstream regions of the ssu and tau operons to activate their transcription. Under sulfate-rich conditions, however, adenosine-5Ј-phosphosulfate (APS), an intermediate in sulfate assimilation, functions as a core-pressor of Cbl and inhibits Cbl DNA binding, resulting in a lack of ssu and tau operon expression (4). In contrast, regulation of SSI gene expression in soil bacteria such as pseudomonads is believed to be quite different from that in E. coli because no equivalent to the cbl gene exists in the pseudomonad genome (Pseudomonas Genome Project [www.pseudomonas.com]). Some specific regulators have been reported to be involved in the regulation of the respective organosulfur assimilatory genes (5,18,40); however, it remains unknown how pseudomonads perceive sulfate starvation conditions and subsequently control the expression of a variety of organosulfur-assimilating pathways.
In contrast to alkanesulfonates and taurine, little is known about how bacteria scavenge sulfur from volatile organosulfur compounds such as organic sulfides and thiols. Pseudomonas putida DS1 is capable of utilizing DMS as a sulfur source (10), and the DMS metabolic pathway was investigated at the molecular level. DS1 desulfurizes DMS to sulfite via dimethyl sulfoxide, dimethyl sulfone (DMSO 2 ), methanesulfonate (MSA), and sulfite (Fig. 1A). The desulfonation of MSA requires an FMNH 2 -dependent monooxygenase, SsuD, which is conserved among many gram-negative and gram-positive bacteria (10). However, we found that the conversion of DMSO 2 to MSA requires a novel FMNH 2 -dependent monooxygenase, SfnG, which is encoded within the sfnFG operon (8). Expression of the sfnFG operon is directly regulated by a novel 54 -dependent transcriptional regulator, SfnR, which is encoded within the sfnECR operon ( Fig. 1B) (8,9). SfnR also activates the expression of the sfnAB operon, and sfnA is suggested to be involved in the use of methanethiol as a sulfur source (11). Expression of the sfnECR, sfnFG, and sfnAB operons is activated under sulfate starvation conditions.
The finding of SfnR implied that sulfone utilization of DS1 is controlled by a novel mechanism for the sulfate starvation response because all of the other known transcriptional regulators of gram-negative bacteria involved in organosulfur metabolism belong to the LysR family, e.g., Cbl and SdsB (5), AsfR (40), SftR (18), and SsuR (16). More interestingly, SfnR lacks an N-terminal regulatory domain, unlike many other 54 -dependent transcriptional regulators (8,9). Therefore, determining the role that SfnR plays in the sulfate starvation response would be of great interest. However, the transcription mechanism of the sfnR gene itself remains unclear. We previously demonstrated that CysB is necessary for the expression of the sfnECR operon (9), but it remains to be elucidated whether CysB directly activates the expression of the sfnECR operon by binding at its upstream region. Hence, in this study, we first investigated the detailed mechanism of the transcriptional regulation of the sfnECR operon.  (9). The expression of sfnECR is repressed by SfnR itself and activated by a LysR-type transcriptional regulator, CysB (see Results). ϩ, stimulation of transcription; Ϫ, inhibition of transcription. tase (Invitrogen) and an IRD800-labeled primer (Aloka), sfnE-PE1 (see Table  S1 in the supplemental material). The primer extension products were purified using phenol-chloroform extraction and ethanol precipitation. The products were subjected to electrophoresis together with a sequence reaction using the same primer and a Li-Cor model 4200I-2 Auto-DNA sequencer running Base ImaglR data collection software version 4.0 (Li-Cor) according to the manufacturer's instructions.
A plasmid, pET-C-ht-cysB, for the production of the C-terminally histidinetagged CysB (C-ht-CysB) protein was constructed as follows. We obtained the complete nucleotide sequence of the cysB gene of DS1 from the mini-Tn5 flanking DNA regions of cysH::mini-Tn5 mutants isolated previously (21) because cysB of DS1 lies adjacent to the cysH gene. The DNA fragment containing C-ht-cysB was amplified from total DNA of DS1 with the set of primers cysB-FW and C-ht-cysB (see Table S1 in the supplemental material). The PCR product was cloned into a pT7Blue (R) vector (Novagen), and the nucleotide sequence was confirmed. Clones were digested with both NdeI and EcoRI (sites incorporated in the primers) (see Table S1 in the supplemental material), and the To construct plasmid pBBR-C-ht-cysB, the XbaI-SacI fragment containing C-ht-cysB from pET-C-ht-cysB was cloned between the corresponding sites of pBBR1MCS-5 (Table 1).
To construct plasmids pG18cysNC::Km and pG18cysI::Km, a DNA fragment was first generated consisting of the Km r cassette flanked by two homology regions with the surrounding sequence of cysNC or the internal sequence of cysI by using a two-step PCR according to a method described previously (21). The nucleotide sequence of the surrounding region of cysNC was obtained from several mini-Tn5 flanking DNA regions of cysNC::mini-Tn5 mutants isolated previously (21). The internal nucleotide sequence of cysI was obtained by sequencing the PCR fragment, which was amplified from total DNA of DS1 with primers KT-cysI-F and KT-cysI-R (see Table S1 in the supplemental material) and designed from the cysI sequence of P. putida KT2440 whose whole genome sequence is available. In the first PCR, about 500 bp of the 5Ј and 3Ј regions were amplified, which partially overlapped with the Km r cassette, using total DNA from DS1 and the primers described below: the 5Ј region was amplified using primers cysNC-UpF/cysNC-UpR-Km or cysI-UpF/cysI-UpR-Km, and the 3Ј region was amplified using primers cysNC-DnF-Km/cysNC-DnR or cysI-DnF-Km/ cysI-DnR (see Table S1 in the supplemental material). Simultaneously, the Km r cassette was amplified as described previously (21). The resultant three PCR products were then assembled during the second PCR. The assembled DNA fragments were ligated into the SmaI site of pG18mobsacB (Table 1) to yield plasmids pG18cysNC::Km and pG18cysI::Km.
␤-Galactosidase reporter assay. The P. putida strain harboring a reporter plasmid was grown on SFMM containing 1 mM sulfate or 1 mM MSA until the stationary phase at 30°C. A LacZ assay was performed with the bacterial cultures according to the method of Miller (30).
Purification of proteins. Escherichia coli BL21(DE3)(pETsfnR) or BL21(DE3) (pET-C-ht-cysB) cells producing N-ht-SfnR or C-ht-CysB were collected from 4 ml of culture and suspended into 0.5 ml of IMAC wash buffer (Bio-Nobile). The cell suspension was subjected to ultrasonication with a Branson Sonifier 250D. The cell extract was centrifuged at 10,000 ϫ g for 5 min to remove cell debris. The supernatant containing N-ht-SfnR or C-ht-CysB was subjected to metal chelation chromatography with a QuickPick IMAC (Bio-Nobile) metal affinity kit according to the manufacturer's instructions. The protein samples eluted by the QuickPick IMAC system were analyzed by sodium dodecyl sulfate (SDS)polyacrylamide gel electrophoresis and stored at 4°C for 48 h. Protein concentrations were determined with the Bio-Rad protein assay (Bio-Rad).
EMSA. EMSA was performed with a DIG Gel Shift kit, second generation (Roche Applied Science), according to the manufacturer's instructions. The EMSA with N-ht-SfnR was performed as reported previously (8). Probes PbED, PbEDM1, PbEDM2, and PbEDM3 were generated by PCR with the sets of primers listed in Table S1 in the supplemental material and labeled with digoxigenin-11-ddUTP at the 3Ј termini by terminal transferase. For EMSA with C-ht-CysB, the DNA fragment containing the region from Ϫ137 to ϩ20 relative to the sfnE transcription start point was generated by PCR with a set of primers, sfnE-137-FW and sfnE ϩ 20-RV (see Table S1 in the supplemental material) and labeled as described above. DNA-binding reactions were performed as described previously (8). The reaction mixtures were incubated for 15 min at 15°C to 25°C. After addition of 5 l of loading buffer to the mixture, 10-l aliquots were subjected to electrophoresis (150 V, 45 min, 15 to 25°C) on 6% polyacrylamide gels and Tris-borate-EDTA buffer. Probes in the polyacrylamide gels were transferred to Hybond-N ϩ (Amersham Biosciences) membranes and then detected with a CSPD system (Roche Applied Science).
DNase I footprinting. DNase I footprinting was performed according to a method described previously (8). Briefly, the 245-bp DNA fragments between Ϫ189 and ϩ59 relative to the transcriptional start site of sfnE were amplified by PCR with the appropriate pairs of primers (5Ј-biotin-labeled sfnE-186-FW/ IRD800-labeled sfnE ϩ 59-RV and IRD800-labeled sfnE-189-FW/5Ј-biotin-labeled sfnE ϩ 59-RV) (see Table S1 in the supplemental material) and used as probes. The probes binding to C-ht-CysB were digested with 3 units of DNase I (Takara Bio) for 30 s at 15°C to 25°C.
Gene disruption. The cysNC and cysI genes of P. putida DS1 were disrupted by homologous recombination (double crossover) according to a method described previously (21). Briefly, the plasmids pG18cysNC::Km and pG18cysI::Km were introduced into DS1 by filter mating with E. coli S17-1pir, and the doublecrossover strains containing the Km r cassette within the target genes were screened. The cysNC and cysI disruptants were designated ⌬cysNC and ⌬cysI, respectively.
Quantitative RT-PCR. Quantitative RT-PCR was performed according to a method described previously (21). To investigate the mRNA levels of cysB and sfnECR in the DS1 derivatives, total RNA from cells grown on SFMM with 1 mM sulfate or MSA until the log phase (optical density at 550 nm of 1.0 to 1.4) was extracted and subjected to an RT reaction with SuperScript II reverse transcriptase (Invitrogen) and Random Primers (Invitrogen). Real-time PCR was carried out with a set of primers, cysB-qRT-F/cysB-qRT-R (see Table S1 in the supplemental material) or sfnR-qRT-F2/sfnR-qRT-R2 (21). To investigate the mRNA levels of sfnFG and sfnECR in the wild-type strain DS1 and the ⌬cysNC and ⌬cysI strains, each strain was grown on SFMM with 1 mM cysteine until the log phase (optical density at 550 nm of 1.0 to 1.4). The cells were collected, washed twice with SFMM, and resuspended in the same medium with 10 M cysteine added or without 1 mM sulfate. After further cultivation for 1 h, total RNA was extracted and subjected to quantitative RT-PCR as described above using the set of primers sfnG-qRT-F/sfnG-qRT-R (21) and sfnR-qRT-F2/sfnR-qRT-R2. The mRNA levels of the target genes (cysB, sfnFG, and sfnECR) were normalized to that of the 16S rRNA gene, which was amplified with the primers 16S-qRT-F and 16S-qRT-R (21).
Nucleotide sequence accession number. The nucleotide sequence data for the cysB gene reported in this study have been registered in the DDBJ, EMBL, and GenBank nucleotide sequence databases with accession number AB374359.

Determination of the transcription start point of the sfnECR operon.
To characterize the sfnECR promoter, we first investigated the transcription start point of the sfnECR operon. Because we had found a 54 -dependent promoter-like sequence upstream of the sfnE translation start point, we first believed that the expression of sfnECR was autoregulated by SfnR. However, in our previous study, the expression of sfnECR was observed in an sfnR-defective mutant (Dfi74J) (9). Therefore, we postulated that the expression of sfnECR is controlled by another 54 -dependent transcriptional regulator or that additional promoters exist in the upstream region of sfnECR. To test these hypotheses, we carried out primer extension analysis using total RNA from the sfnR disruptant (SRK1) (8) as well as from the wild-type DS1. As a result, one transcription start point was found at 132 bp upstream of this 54 -dependent promoter-like sequence (Fig. 2), indicating that this sequence does not function as a promoter. Instead, a 70 -promoter-like sequence was located upstream of the transcription start point. The Ϫ10 region corresponding to this start point is 5Ј-TATAAA-3Ј (position Ϫ13 to Ϫ8); the Ϫ35 region would lie somewhere within the sequence 5Ј-TTTTTC AT-3Ј (position Ϫ36 to Ϫ29). Interestingly, the amount of the primer extension product from the sfnR disruptant was much larger than that from the wild type, suggesting that SfnR represses the expression of sfnECR containing its own gene. An SfnR-binding motif, which is conserved upstream of sfnFG and sfnAB (5Ј-CTGTN 10 ACAG-3Ј; consensus sequence), was also found downstream of the sfnECR transcriptional start point (5Ј-CTGTN 10 ACAC-3Ј; position ϩ28 to ϩ45). These results suggest that SfnR represses the expression of sfnECR by binding to this motif.
Involvement of the SfnR-binding motif in repression of sfnECR. To investigate involvement of the SfnR-binding motif located downstream of the transcription start point in the negative regulation of sfnECR, we performed deletion analysis of the DNA region containing this motif using lacZ as a reporter gene. A 3Ј-deletion series of the DNA region downstream of the sfnECR transcription start point was constructed (Table 1) and was introduced into P. putida DS1 or SRK1, and the LacZ activity of the resultant transformants was measured under sulfate starvation conditions. In the wild type, LacZ activity was increased by deletion of the region containing the SfnRbinding motif (ϩ60 to ϩ21 with respect to the sfnECR transcription start point) (Fig. 3). However, in SRK1, deletion of the same region did not have any effect on the induction of LacZ activity. These results strongly suggest that SfnR binds to the region from ϩ60 to ϩ21.
To confirm the binding of SfnR to this DNA region, we carried out an EMSA using purified N-terminally histidinetagged SfnR (N-ht-SfnR) and a 40-bp probe from ϩ60 to ϩ21 relative to the sfnE transcription start point (PbED) (Fig. 4A).
An sfnF promoter probe, Pb10 (8) (Fig. 4A), was used as a positive control. As reported previously (8), Pb10 exhibited two shifted bands (Fig. 4B) because it contains two partially overlapping imperfect inverted repeat sequences (IIRSs; IR1 and IR2; Fig. 4A) for the binding of SfnR. Probe PbED also exhibited two shifted bands (Fig. 4B), suggesting that it contains two SfnR-binding sites in this region. However, the amounts of complex II of PbED were smaller than those of Pb10. In the sequences between Pb10 and PbED, IR1 was highly conserved, but IR2 was not clearly conserved (Fig. 4A). Therefore, we thought that binding of SfnR to IR2 was weaker in PbED than in Pb10. To investigate the involvement of these IIRSs in the binding of SfnR, we performed EMSA with probes mutated in IR1 and/or IR2 (PbEDM1, PbED2, and PbEDM3) (Fig. 4A). PbEDM2 with a mutation in IR2 gave rise to only one shifted band and did not form complex II (Fig. 4B). This result demonstrated that SfnR binds to IR2 in PbED despite the low degree of conservation of IIRS. PbEDM1 with a mutation in IR1 did not produce any shifted bands, even at an N-ht-SfnR concentration of 480 g/ml. Therefore, we concluded that IR2 alone is insufficient to bind to SfnR in PbED.
Identification of the DNA region required for expression of the sfnECR operon. We then carried out a 5Ј-deletion analysis of the sequence upstream of the sfnE transcription start point, using lacZ as a reporter gene, to determine the minimal region necessary for SSI expression of the sfnECR operon. Strain DS1 was transformed by each reporter plasmid (pMEsfnE-137 through pMEsfnE-45) (Fig. 5), and the LacZ activity of the resultant transformant was measured in the presence and absence of sulfate. Deletion from Ϫ500 to Ϫ138 did not affect the LacZ activities (data not shown). The LacZ activity of pMEsfnE-104, pMEsfnE-80, or pMEsfnE-63 increased in the absence of sulfate, although these LacZ activities appeared to be partially decreased in comparison to that of pMEsfnE-137. However, the LacZ activity of pMEsfnE-45 did not increase in the absence of sulfate and was almost the same as that of pMElacZ (control vector). These results indicate that the DNA region up to 63 bp from the transcription start point of sfnE is necessary for SSI expression of the sfnECR operon. In this region, we found an inverted repeat sequence (5Ј-AAGAATATAAA ATCTTATTCTT-3Ј, at position Ϫ57 to Ϫ36) (Fig. 2B), which could be involved in transcription activation under sulfate starvation conditions.
Binding of CysB protein to the sfnECR promoter region. We previously demonstrated that CysB of strain DS1 (CysB DS1 ) is necessary for expression of the sfnECR operon by in vivo transcription analysis using the cysB disruptant (9); however, it has remained unknown whether CysB DS1 directly activates the expression of the sfnECR operon by binding at the region upstream of its promoter. The DNA binding of CysB protein of Salmonella enterica serovar Typhimurium (CysB ST ) has been well characterized with the cysJIH, cysK, and cysP promoters (12,13,32); these studies demonstrated that CysB ST binds to the AT-rich 40-bp sites just upstream of the Ϫ35 regions of the three cys promoters (designated CBS-J, CBS-K1, and CBS-P1, respectively) in the presence of an inducer, O-acetylserine or N-acetylserine, although these three binding sites showed only a low level of similarity to one another and lacked a strong consensus sequence. As well as these three binding sites of CysB ST , the 40-bp region just upstream of the Ϫ35 region of the sfnE promoter is quite AT rich (the AϩT content of the region from Ϫ74 and Ϫ35 is 83%); therefore, we postulated that CysB DS1 could bind to this promoter region.
To test this hypothesis, we performed EMSA with the purified CysB DS1 protein and the sfnE promoter probe. Because only a partial cysB DS1 sequence was obtained in our previous study, we first determined the complete cysB DS1 sequence (GenBank accession no. AB374359) with the method described in Materials and Methods. The cysB DS1 gene showed 94% nucleotide identity and 100% amino acid identity with cysB of P. putida KT2440 (Pseudomonas Genome Project). For the production of CysB DS1 , we constructed a plasmid, pET-C-ht-cysB, in which the histidine tag to facilitate the purification was added to the C-terminal region of CysB DS1 because LysR-type transcriptional regulators contain a DNA-binding domain in their N terminus. C-ht-CysB DS1 was overproduced with E. coli BL21(DE3)(pET-C-ht-cysB) and was purified in a single step by metal ion affinity chromatography (see Fig. S1 in the sup-plemental material). The molecular mass of the purified protein was approximately 36 kDa, which is consistent with the deduced molecular mass of C-ht-CysB DS1 . We previously reported that the cysB disruptant (⌬CYSB) cannot utilize sulfate as a sulfur source and that the expression level of sfnECR is greatly diminished in ⌬CYSB (9). Therefore, we constructed a broad-host-range plasmid that produced Cht-CysB DS1 , pBBR-C-ht-cysB (Table 1), and introduced it into ⌬CYSB to confirm that C-ht-CysB DS1 maintains regulatory function in vivo. As a result, ⌬CYSB harboring pBBR-C-ht-cysB restored the ability to utilize sulfate (data not shown). We then used quantitative RT-PCR analysis to investigate the expression levels of sfnECR in ⌬CYSB (pBBR-C-ht-cysB). As a result, it was shown that ⌬CYSB (pBBR-C-ht-cysB) restored the ability to express sfnECR under sulfate starvation conditions (data not shown), indicating that C-ht-CysB DS1 maintains regulatory function in vivo.
In EMSA with the purified C-ht-CysB DS1 and the labeled DNA fragment containing the region from Ϫ137 to ϩ20 rela- tive to the sfnE transcription start point as a probe, a single complex (complex I) was observed at 0.078 g/ml of C-ht-CysB DS1 , and additional slower complexes (complex II and complex III) appeared at higher concentrations (Fig. 6A). The reason for the formation of these slower complexes was unclear, but the same number of complexes has been observed with C-ht-CysB of E. coli (C-ht-CysB EC ) and the cysP promoter (27). The shifted band was removed by the addition of a 400fold excess of unlabeled sfnE promoter probe, showing that CysB DS1 specifically bound to this region. In contrast to the cases of CysB ST and CysB EC , O-acetylserine did not have any effect on either the abundance or the relative mobilities of complexes formed by CysB DS1 (Fig. 6).
To determine the CysB DS1 -binding site in the sfnE promoter region, DNase I footprinting was performed with a 245-bp probe containing the region from Ϫ189 to ϩ56 relative to the sfnE transcriptional start point. As shown in Fig. 6B, the 40-bp region from Ϫ71 to Ϫ32 was protected from DNase I digestion in the presence of C-ht-CysB DS1 . The position of this protected region relative to the transcriptional start point was almost consistent with those of the binding sites of CysB ST (CBS-J, CBS-K1, and CBS-P1).
Constitutive expression of the cysB gene under sulfate starvation conditions. The results presented above clearly indicate that CysB directly regulates the expression of sfnECR. Next, we used quantitative RT-PCR analysis to investigate whether expression of the cysB gene is activated under sulfate starvation conditions. The mRNA level of the cysB gene (1.24 Ϯ 0.19) in DS1 cells grown on MSA relative to that of cells grown on sulfate did not increase under the sulfate starvation conditions, in contrast to that of sfnECR (33.9 Ϯ 5.5) (values were calculated from four assays). This indicates that CysB is the master regulator located at the top of the hierarchy in the regulation of sulfate starvation responses of the sfn operons.
The role of cysteine biosynthetic intermediates for repression of the sfn genes. Expression of the sfn operons is activated under sulfate starvation conditions, but it remains unknown whether sulfate starvation is sensed only by CysB or whether SfnR plays a role independent of CysB in the SSI expression of the sfn operons. To estimate individual functions of CysB and SfnR, we investigated whether the expression of their target genes, sfnECR and sfnFG, respectively, is repressed in mutants that were blocked at different steps in cysteine biosynthesis from sulfate.
In P. putida, cysteine has been reported to be synthesized via the following sequence of reactions (3,20,39) (Fig. 1A): conversion of sulfate into APS (catalyzed by the cysD and cysNC gene products), reduction of the sulfate moiety to sulfite (catalyzed by the cysH gene product) and then to sulfide (catalyzed by the cysI and cysJ gene products), and finally, incorporation of the sulfide into O-acetylserine to yield cysteine (catalyzed by the cysM or cysK gene product).
We constructed the cysNC or cysI disruptant (⌬cysNC or ⌬cysI, respectively), which was unable to utilize sulfate or sulfite, respectively, as a sulfur source (data not shown). We subsequently measured the expression levels of sfnECR and sfnFG in both disruptants in the presence or absence of sulfate by using quantitative RT-PCR analysis (Fig. 7). As expected, in the wild-type strain DS1, mRNA levels of both sfnECR and sfnFG decreased in the presence of sulfate. In the ⌬cysNC strain, the mRNA level of sfnFG decreased even in the absence of sulfate and further decreased by half in the presence of sulfate. This result suggests that the expression of sfnFG is repressed by sulfate itself. The reason for the low level of expression of sfnFG in the absence of sulfate can be explained if it is assumed that a "sulfur-free" mineral medium contains residual amounts of sulfate, as reported by Bykowski et al. (4); in the wild-type strain, residual amounts of sulfate are probably immediately metabolized, while in the ⌬cysNC strain, sulfate would accumulate to a level sufficient to prevent the expression of sfnFG.
However, the expression of sfnECR was not repressed in the presence of sulfate in either the ⌬cysNC or the ⌬cysI strain, suggesting that the expression of sfnECR is repressed by metabolites downstream of sulfite, such as sulfide or cysteine. These results indicate that SfnR plays a role independent of CysB in the SSI expression of the sfn operons. Interestingly, in The labeled probe containing the region from Ϫ137 to ϩ20 relative to the sfnE transcription start point was incubated with purified C-ht-CysB DS1 (0.078, 0.63, and 5.0 g/ml) in the presence (ϩ) or absence (Ϫ) of 10 mM O-acetylserine. A 400-fold excess of unlabeled probe was added as a specific competitor where indicated by the plus sign. (B) DNase I footprinting of C-ht-CysB DS1 binding to the sfnECR promoter region. The 245-bp fragment was labeled with IRD800-labeled primer at the sfnECR sense or antisense strand. The following amounts of C-ht-CysB DS1 were used in lanes 1 to 5: 0, 1, 2, 4, and 8 g, respectively. Lanes G, A, T, and C correspond to the sequence ladders generated with the same primer. Solid bars and arrowheads indicate the protection region of C-ht-CysB DS1 and the hypersensitive nucleotides, respectively.

DISCUSSION
In this study, we identified the sfnECR promoter region and demonstrated that CysB directly binds to this region. Furthermore, we identified the CysB-binding sequence using DNase I footprinting. The binding sites of CysB of enteric bacteria have been reported to lack a strong consensus sequence (12,13,32), and we also did not find any consensus sequences among the binding sequences of CysB DS1 , CysB EC , and CysB ST . However, the location of the CysB DS1 -binding site relative to the transcriptional start point was almost consistent with those of binding sites of CysB ST (CBS-J, CBS-K1, and CBS-P1). CysB ST binds to these three sites only in the presence of N-acetylserine or O-acetylserine, which functions as an inducer to stimulate transcription initiation (12,32,34). In the absence of the inducer, the ability of CysB ST to bind to the cysJIH promoter region (CBS-J) is greatly diminished, and at the cysK and cysP promoter regions, CysB ST binds to different sites from CBS-K1 and CBS-P1, called CBS-K2 (position Ϫ115 to Ϫ79) and CBS-P3 (position Ϫ63 to Ϫ11), respectively. Moreover, in EMSA with CysB ST and the cysP and cysK promoter probes, O-acetylserine has been reported to stimulate the formation of the protein-DNA complex and increase its mobility (12,32). In contrast, CysB DS1 bound to the sfnE promoter region without O-acetylserine at a position corresponding to CBS-J, CBS-K1, and CBS-P1 (Fig. 6B). In addition, in EMSA with CysB DS1 and the sfnE promoter probe, O-acetylserine did not have any effect on either the abundance or the relative mobilities of complexes formed by CysB DS1 (Fig. 6A). These findings imply that O-acetylserine does not affect the nature of DNA binding of CysB DS1 . In previous studies, the function of CysB of Burkholderia cenocepacia appeared to be independent of the presence of O-acetylserine (16); therefore, CysB may vary with respect to the recognition of an inducer.
Measurements of mRNA levels in the disruptants of sulfate assimilatory genes cysNC and cysI (Fig. 7) suggest that the expression of sfnECR is repressed by metabolites downstream of sulfite, such as sulfide or cysteine. This result will provide an important clue to identify an inducer or corepressor molecule for the repression of sfnECR under sulfate-rich conditions. In S. enterica serovar Typhimurium, sulfide and thiosulfate were reported to act as anti-inducers for CysB ST and inhibit both the transcription initiation and binding of CysB ST to the target promoter region (12,35). In addition, cysteine was reported to have a repressive effect on the expression of cys genes because this compound inhibits activity of the serine acetyltransferase enzyme required for synthesis of an inducer, O-acetylserine (25). We postulated that sulfide, thiosulfate, or cysteine acts as a corepressor of CysB DS1 , and we therefore performed EMSA with C-ht-CysB and the sfnE promoter probe in the presence of up to 10 mM of each compound; however, these compounds affected neither the abundance nor the relative mobilities of the protein-DNA complexes (data not shown). Further study is therefore necessary to identify the inducer or corepressor molecules of CysB DS1 .
Measurement of mRNA levels in the disruptants of sulfate assimilatory genes (Fig. 7) also provided genetic evidence that sulfate itself serves as a signaling molecule for sulfate excess in the transcription regulation of sfnFG. The expression of sfnFG requires a 54 -dependent transcriptional activator, SfnR. Many 54 -dependent transcriptional regulators share a common domain architecture that includes three domains: a C-terminal DNA-binding domain, a central module for ATPase activity and interaction with 54 -RNA polymerase, and an N-terminal regulatory domain (37). However, as reported previously, SfnR lacks an N-terminal regulatory domain (8,9); therefore, it is of interest in this context to examine how sulfate is sensed and represses the expression of sfnFG. Several studies have reported on 54 -dependent transcriptional regulators lacking an N-terminal regulatory domain like SfnR. The best characterized of these is the phage shock protein of E. coli, PspF, which controls expression of the pspABCDE operon required for survival in stationary phase in an alkaline environment (17). PspF has been reported to activate the psp operon without phosphorylation or an activation ligand. Instead, its regulatory function is controlled by formation of a repressive complex with the PspA protein (7). In a similar manner, SfnR may be negatively regulated by its partner protein in the presence of sulfate. As reported in our previous study, the phosphoenolpyruvate-dependent phosphotransferase system (PTS) family protein is also involved in the expression of sfnFG (21). Further studies are being considered to examine the relationships among PTS, SfnR, and the recognition of sulfate.
SfnR is also involved in the transcription regulation of sfnECR containing its own gene by binding to the region downstream of the sfnE transcription start point and interrupting transcription. This type of autoregulation has been described for several genes encoding a prokaryote transcriptional activator. Interestingly, the mRNA level of sfnECR appeared to increase following the accumulation of sulfate in ⌬cysNC and ⌬cysI strains (Fig. 7). Therefore, sulfate likely derepresses the expression of sfnECR, probably by inhibiting the DNA binding of SfnR, although the mechanism for the recognition of sulfate remains to be elucidated.
The results obtained in the previous and present studies allow us to propose a mechanism for transcription of the sfn The wild-type strain (WT) and the ⌬cysNC and ⌬cysI strains were grown on SFMM with 1 mM cysteine added as a sulfur source until the log phase was reached (turbidity at 550 nm of 1.0 to 1.5). The cells were collected, washed with SFMM, and resuspended in the same medium with 10 M cysteine added with or without 1 mM sulfate. After further cultivation for 1 h, total RNA was extracted and subjected to quantitative RT-PCR. The values of the mRNA levels in the DS1 derivatives are relative to those of wild-type cells in the presence of sulfate (taken as 1.0). The error bars represent the standard deviations calculated from four assays. VOL. 190, 2008 REGULATION OF THE sfn OPERONS OF PSEUDOMONAS PUTIDA 4529 genes as described below. The regulatory system controlled by CysB and SfnR is responsible for the hierarchical utilization of sulfur sources. In the absence of downstream metabolites in the sulfate assimilatory pathway, such as sulfide and cysteine, CysB activates the expression of sfnECR, probably in addition to that of the cys genes responsible for sulfate assimilation. Sulfate itself had no direct effect on the expression, but its conversion to the downstream metabolites resulted in about a 30-fold reduction in the expression levels of sfnECR (see Results). This repressive effect provided by sulfate in strain DS1 appears to be greater than that in E. coli and S. enterica serovar Typhimurium because the expression of cys genes controlled by CysB in these enteric bacteria is only partially repressed in the presence of sulfate (23,24). When sulfate itself is lacking, in addition to the downstream metabolites, SfnR activates the expression of sfnFG. SfnR also represses sfnECR transcription. This negative autoregulation may contribute to maintaining appropriate levels of sfnR expression. The regulation by SfnR would ensure rapid switching of utilized sulfur sources from DMSO 2 to sulfate when sulfate, one of the most preferred sulfur sources, is supplied.
The system for regulation of sfn genes by CysB and SfnR is similar to that for regulation of the ssu genes by CysB and Cbl in E. coli, although in E. coli, APS, which is synthesized in one step from sulfate in the sulfate assimilatory pathway, acts as a signaling molecule for sulfate excess (4). A similar hierarchical regulatory system may also exist for arylsulfatase gene (atsA) expression in Pseudomonas aeruginosa, because at least two independent corepressor molecules are suggested to be involved in the sulfate starvation-inducible expression of atsA, although sulfate itself is not the active corepressor (14). However, as described in the introduction, no equivalent to the cbl gene exists in the pseudomonad genome. Therefore, pseudomonads appear to have developed a different mechanism for a sulfate starvation response from that of E. coli, especially in terms of the recognition of sulfate. In the pseudomonad genome, some orthologs of the sfnR gene are present, for example, PP2771 in P. putida KT2440 (82% amino acid identity), PFL_4158 and PFL_2926 in Pseudomonas fluorescens Pf-5 (79% and 60% amino acid identities, respectively), and PA2354 and PA2359 in P. aeruginosa (79% and 59% amino acid identities, respectively), although all of their functions remain unknown. Therefore, investigating how these SfnR homologs form transcriptional networks and are related to the sulfate starvation response in each of these bacteria would be of great interest.