The LysR-type transcriptional regulator CysB controls the repression of hslJ transcription in Escherichia coli

The LysR-type transcriptional regulator (LTTR) CysB controls the biosynthesis of cysteine and sulphur utilization in Escherichia coli. CysB is an activator of the cys regulon, tau operon, cbl gene and spt gene, and in association with the Cbl (CysB-like) protein it downregulates the expression of the ssu operon (Kredich, 1996; van der Ploeg et al., 2001; Parry & Clark, 2002). Like many other members of the LTTR family, CysB is a repressor of its own transcription (Kredich, 1996). The small metabolite N-acetylserine (NAS) is an inducer that facilitates binding of the CysB tetramers to the regulatory regions of target genes, while preventing binding to the cysB control region and the consequent negative autogenous control (Kredich, 1996). In the absence of cysteine, NAS is produced by nonenzymic conversion of O-acetylserine. This conversion and the synthesis of NAS are blocked in the presence of cysteine (Kredich, 1996).

The CysB protein consists of several functional domains (Lochowska et al., 2001). The N-terminal region comprises the helixturnhelix (HTH) DNA-binding motif flanked by other amino acids responsible for specificity of binding to DNA sequences. The central region is important for ligand (e.g. NAS) recognition and binding inside the cavity formed by the homodimer (Tyrell et al., 1997; Verschueren et al., 2001). The C-terminal domain is proposed to be involved in homo-oligomerization of this protein as well as in DNA binding of the HTH motif (Lochowska et al., 2001). Transcription factors that contain the HTH motif bind DNA as dimers with the binding motifs positioned in parallel (Pabo & Sauer, 1992; Perez-Rueda & Collado-Vides, 2000). The CysB crystal structure predicts HTH motifs being positioned perpendicularly in the dimer, suggesting an unusual mechanism of binding to DNA (Tyrell et al., 1997). Although there is no strict binding consensus sequence on DNA according to data collected to date, the CysB tetramer recognizes AT-rich sequences in the context of the features of the LysR motif with the characteristic sequence T-N₁₁-A as the core of an inverted repeat (IR) found to be the binding site for most of the LTTR proteins (Goethals et al., 1992; Lochowska et al., 2001). The specific structures of the CysB binding sites have been recognized as wide DNA regions covered with several divergently or convergently oriented half-sites separated either by a few nucleotides (activator sites) or by one or two turns of DNA helix (repressor sites) (Hryniewicz & Kredich, 1994, 1995; Kredich, 1996). Finally, DNA-bending sensitive sites were defined in some regulatory regions characteristic for the cys regulon (Kredich, 1996).

Besides being implicated in sulphur utilization and the synthesis of cysteine, CysB activates the expression of the adi and lysU genes (Shi & Bennett, 1994; Rowbury, 1997). Recently, we found a novel CysB-regulated gene, hslJ, involved in displaying a novobiocin resistance (Nov^R) phenotype in E. coli shown to be independent of previously characterized genes (e.g. gyrB, cls, nov) implicated in this phenomenon (Lilic et al., 2003). The hslJ gene encodes a putative outer-membrane protein (OMP) but the role of HslJ in the mechanism that enables cells to display novobiocin resistance is unknown. Originally, it was shown that in E. coli C600 cysB (SY380), AB1157 cysB(Ts) (SY381) and MC4100 cysB (SY602) mutants the novobiocin resistance was increased 615-fold in comparison to cysB wild-type (WT) isogenic strains (Rakonjac et al., 1991; Lilic et al., 2003). Using the random insertion of the lacZ reporter gene we discovered the hslJ gene, whose expression is negatively regulated by CysB (Lilic et al., 2003). Expression of the hslJ : : lacZ gene fusion is elevated in cysB mutants and the five- to sixfold overproduction of HslJ in either cysB⁺ or cysB mutant strains increases the novobiocin resistance. A cysB hslJ double mutant does not exhibit the resistance to novobiocin. The hslJ gene expression is found to be negatively autogenously controlled in trans. However, according to analysis of the deduced amino acid sequence of HslJ this protein resides in the outer membrane. This prediction is strengthened by the fact that the hslJ : : ΩKan mutation induces the psp operon (Lilic et al., 2003). The psp operon is specifically induced by overexpression of WT and mutant OMPs (e.g. pIV, and a number of other secretins) as well as by expression of mutant envelope proteins (Model et al., 1997). Hence, it is very likely that HslJ is localized in the outer membrane, implying the existence of an additional factor involved in negative regulation of hslJ expression.

The results presented by Lilic et al. (2003) did not answer the question whether CysB plays the role of a repressor acting directly and regulating the transcription of hslJ. In this study, we defined the hslJ promoter region and determined the mechanism of negative regulation imposed by CysB. Also, we addressed the question whether transcription of the hslJ is subject to double negative control by both CysB repression and the HslJ feedback autoregulation. We characterized the MC4100 cysB mutated allele in strain SY602 and tested it and other cysB alleles in the context of either hslJ regulation or control of the cys regulon.

Bacterial strains, media and growth conditions.
The strains of E. coli and plasmids used in this study are listed in Table 1. Strains SY511, SY512, SY513 and SY514 were constructed by transducing strains MC4100, SY602, EC2275 and EC2549, respectively, with P1_vir phages grown on RH90 and selecting for Tet^R transductants. Strain SY515 was constructed by transducing strain SY602 with P1_vir phage grown on SY605 and selecting for Kan^R transductants. Media used for bacterial growth were rich medium (LB) or minimal medium (A) solidified when necessary with 1·5 % (w/v) agar (LA) as described by Miller (1992). All strains were grown at 37 °C as described previously (Lilic et al., 2003). When required, the antibiotics were added to the media at the following concentrations: 100 µg ampicillin (Amp) ml^-1; 25 µg kanamycin (Kan) ml^-1; 50 µg spectinomycin (Spc) ml^-1; 15 µg tetracycline (Tet) ml^-1; 10400 µg novobiocin (Nov) ml^-1. To detect LacZ⁺ cells, 0·1 ml of a 10 mg ml^-1 solution of X-Gal in N,N-dimethylformamide was added to LB agar plates or LB soft agar. Transformation of bacteria and transduction by P1_vir were performed as described by Miller (1992).

Table 1. Bacterial strains and plasmids used in this study

DNA manipulations.
Plasmids pVGM3, pVGM4 and pVGM5, containing the 666 nt (region A), 150 nt (region B) and 93 nt (region C) upstream of the hslJ start codon, respectively, fused to promoterless lacZYA genes on plasmid pRS415 were constructed and used for determination of the hslJ promoter region. We used PCR to amplify different DNA fragments, regions A, B and C. As a template we used the MC4100 chromosomal DNA isolated as described by Hopwood et al. (1985). The PCR was performed using the forward primers (Operon Tech): region A, HSLJ-rr1 (5'-GGAATTCGGTACCGGTAAAATC-3'); region B, HSLJ-fw1 (5'-GGAATTCTGGAAAAAGGCG-3'); and region C, HSLJ-fw2 (5'-GGAATTCCAGGGGAGCTGATTC-3') (the EcoRI site is underlined). The reverse primer used for all fragments was HSLJ-rev1 (5'-GCGGATCCTTGTGGTTCTCAAT-3') (the BamHI site is underlined). The purified PCR products were digested with EcoRI and BamHI and ligated into the high-copy-number vector pRS415 digested with EcoRI and BamHI, creating plasmids pVGM3, pVGM4 and pVGM5. To transfer regions A, B and C onto a low-copy-number (56) vector pMS421 we subcloned the constructed hslJlacZ operon fusions from pVGM3, pVGM4 and pVGM5 by digesting these plasmids with EcoRI and SalI and ligating the corresponding DNA fragments (Table 1) with pMS421 digested with the same enzymes, creating the plasmids pVGM6, pVGM7 and pVGM8, respectively. All constructs were checked by DNA sequencing performed by the dideoxy chain-termination method (Sanger et al., 1977).

To characterize the cysB mutation in strain SY602 (MC4100 cysB) we isolated the chromosomal DNA of SY602 and performed the PCR using the primers (IMGGE, Belgrade) CYSB-fw1 (5'-TTTAGCATGCAATTACAACAAC-3'), carrying the SphI restriction site (underlined), and CYSB-rev1 (5'-GAAGATCTTTTTTCCGGCAGTTT-3'), carrying the BglII restriction site (underlined). A 1 kb DNA fragment from PCR carrying the mutated cysB allele was then digested by SphI and BglII and cloned into vector pQE70, creating plasmid pVGM9. The entire cysB allele was sequenced.

RNA manipulations.
Total RNA was isolated according to Aiba et al. (1981) and Gerendasy & Ito (1990). Briefly, strains MC4100 and SY602 (MC4100 cysB) were cultured in 10 ml minimal medium A supplemented with cysteine overnight at 37 °C. The cells were harvested and resuspended in 10 ml protoplast buffer (15 mM Tris/HCl pH 8, 0·45 M sucrose, 8 mM EDTA) and then 80 µl lysozyme (1 mg ml^-1) was added. Protoplasts were centrifuged and the pellet was resuspended in 0·5 ml lysis buffer (20 mM Tris/HCl pH 8, 10 mM NaCl, 1 mM NaCl, 1 mM sodium citrate, 1·5 % SDS), then 15 µl diethyl pyrocarbonate (DEPC) was added and the suspension was incubated for 5 min at 37 °C and then put on ice. After treatment with 250 µl NaCl (40 %, w/v) and incubation on ice for 10 min, the suspension was centrifuged for 1 h. The RNA was extracted with phenol/chloroform (1: :1, v/v) at room temperature and precipitated by adding 1 ml ethanol (at -70 °C). The RNA pellet was washed with 70 % ethanol and dissolved in water. We added 3 vols 4 M sodium acetate to concentrate the RNA solution and then reprecipitated. The RNA quality and concentrations were determined by measurement of A₂₆₀ and A₂₈₀. RNA with a A₂₆₀/A₂₈₀ ratio of 1·52·0 was used for further experiments.

Primer extension analysis was performed according to Sambrook et al. (1989) using primer HSLJ-rev3 (5'-GCCATCAGCAGGCTTAGC-3') (Genosys), 4326 nt downstream from the translational start, and 100 µg of total RNA isolated from strain MC4100 or SY602 (MC4100 cysB). RNA samples were resuspended in RNase-free water and incubated at 37 °C for 1 h in the presence of RNase-free DNase (Promega). Primer was labelled at its 5' end by using [γ-³²P]dATP and T4 polynucleotide kinase (New England Biolabs). The cDNA was extended at 37 °C for 40 min with MMLv Reverse Transcriptase (Pharmacia). The products were loaded on a 5 % polyacrylamide gel together with a nucleotide sequence (Sanger et al., 1977) generated with the same (unlabelled) primer and plasmid pHV3002 as a template.

β-Galactosidase assay.
E. coli strains carrying the plasmid hslJlacZ operon fusions were grown overnight at 37 °C in LB broth containing the appropriate antibiotic and diluted 100-fold into the same medium. Following growth to mid-exponential phase (OD₆₀₀ 0·4), cultures were assayed for β-galactosidase activity by the method of Miller (1992).

Overproduction and purification of CysB-His₆ protein.
Plasmid pVGM1 was used for overexpression and purification of the CysB-His₆ fusion protein following the protocol in the Qiagen manual. The E. coli cysB gene was amplified by PCR using plasmid pJOH1 as a template and the pair of primers CYSB-fw1 and CYSB-rev1. The PCR product was digested with SphI and BglII and cloned into the SphI/BglII sites of vector pQE70 to generate the plasmid pVGM1. This construct was checked by sequencing (Sanger et al., 1977) using primers Type III/IV (Qiagen) and Reverse sequencing (Qiagen). Plasmid pVGM1 was used to transform E. coli strains SY602 and SY380 previously transformed with plasmid pREP4. The recombinant strains were grown in LB medium at 37 °C to mid-exponential phase (OD₆₀₀ 0·4). IPTG (1 mM) was then added and incubation continued for 3 h. Purification of the overproduced protein CysB-His₆ (strain SY380/pREP4/pVGM1) was carried out at room temperature using the Ni-NTA agarose matrix of the QIAexpressionizt kit (Qiagen). Briefly, the cells were centrifuged, resuspended in 1/25 of the culture volume of sonication buffer, disrupted by freezing (in dry ice/ethanol, with thawing in cold water), and sonication, and the cell debris was removed by centrifugation. E. coli crude extracts were loaded on a 1 ml Ni-NTA agarose column previously equilibrated with sonication buffer and the CysB-His₆-tagged protein was eluted with an imidazole gradient (100500 mM). The eluted fractions were subjected to SDS-PAGE as described by Laemmli (1970). The protein eluate was later dialysed overnight against a storage buffer (50 mM Tris/HCl pH 8·0, 1 mM EDTA and 20 % glycerol). The concentration of CysB-His₆ was determined using the Bradford protein assay (Bio-Rad) (Bradford, 1976). The purified protein was stored at -80 °C.

Nondenatured protein molecular mass determination.
Native molecular masses of the fusion protein CysB-His₆ were determined by the method of Bryan (1977). Purified CysB-His₆ (10 µg), and molecular size protein markers (20 µg) (nondenatured protein molecular mass marker kit, Sigma) in 20 mM Tris/HCl (pH 7·4), 1 mM EDTA and 50 mM NaCl buffer were subjected to electrophoresis on a set of native protein gels that contained various concentrations of acrylamide (6, 7, 8·5 and 10 %), and stained with Coomassie blue. Considering the pI (around 7) of CysB-His₆, besides using gels of pH 8·8, we used native gels of pH 9·5. Both sets of gels worked equally well. The relative mobilities (R_F), construction of the Ferguson plots and the coefficients of retardations (K_r; negative slope) of each protein species deduced from the slopes were determined. The logarithm of the K_r of the markers was then plotted versus the logarithm of their molecular masses. The plots obtained were used to determine the molecular sizes [molecular mass (Da)x1000] of CysB-His₆ by extrapolation of their respective K_r.

Electrophoretic mobility-shift experiments.
DNA mobility-shift experiments were performed essentially as described by Prentki et al. (1987). DNA probes carrying different fragments of the hslJ promoter region were obtained by PCR amplification using pHV3002 as a template and different sets of primers: probe 301, HSLJ-fw3 (5'-GCAAAACTTAAGCAATCTGGAAAAAGGCG-3') (Genosys) plus HSLJ-rev2 (5'-GTCACGGGCTTACCG-3') (Genosys); probe 135, HSLJ-fw4 (5'-TGAAGAAAGTAGCCGCG-3') (Genosys) plus HSLJ-rev2; and probe C, HSLJ-fw2 plus HSLJ-rev1. DNA probes 301 and 135 were labelled at the 5' end by using [γ-³²P]dATP and T4 polynucleotide kinase (New England Biolabs). Probe C was used as an unlabelled specific competitor in 50-fold molar excess. Briefly, CysB protein was pre-incubated in binding buffer (10 mM Tris/HCl, pH 7·5; 70 mM KCl; 5 mM MgCl₂; 1 mM DTT; 1 mM EDTA; 12·5 %, v/v, glycerol; 0·1 % Triton X-100; 200 µg ml^-1 BSA) for 5 min at 4 °C before the labelled DNA (12 ng) and in some cases NAS (1, 3 or 5 mM) and/or unlabelled specific probe were added. All reaction mixtures contained 1000-fold weight excess of poly(dI-dC) (Pharmacia). Binding reaction mixtures (total volume 20 µl) were incubated for 20 min at 37 °C. After pre-electrophoresis (2 h), the samples were loaded onto 5 % native polyacrylamide gels in 0·5x TBE and run at a voltage of 12 V cm^-1. The gels were then dried and exposed to X-ray film.

Determination of the hslJ transcription start point
The hslJ : : lacZ gene fusion activity was increased in a cysB mutant (Lilic et al., 2003). Hence, we isolated total RNA from the cysB mutant strain, and by using the oligonucleotide HSLJ-rev3 and the primer extension reaction we determined the hslJ transcription start point (TSP). The TSP (+1) was found to be an A nucleotide positioned 36 bp upstream of the hslJ ATG translation start codon (Fig. 1a, b). We also performed the primer extension experiment with RNA isolated from the cysB⁺ strain MC4100 (data not shown). The analysis of the bands obtained in the primer extension experiments showed about the same moderate signal and a very similar steady-state level of hslJ transcription in both the WT and the cysB mutant strains. This result implies that an additional negative regulator, other than CysB, controls the transcription of hslJ.

(46K): Fig. 1. Determination of the hslJ TSP. (a) Primer extension analysis. Lane 1 contains primer with no RNA; lane 2 contains the primer extension product from RNA isolated from strain SY602. Sequence ladders generated with the same primer are shown. The TSP (A, +1) is indicated. (b) Sequence of the promoter region of hslJ. Locations of the TSP (A, +1), σ70 (underlined -35 and -10 sequences) and σS (open boxes) putative promoters of hslJ, the LysR motifs' (two head-to-head arrows; arrows indicate the inverted repeat), ribosome-binding site (RBS, bold nucleotides), and the hslJ (italic) start codon (bold and underlined) are shown on the nucleotide sequence depicted from the hslJ regulatory region (region C).

Regarding the hslJ TSP, we localized a putative σ⁷⁰ hslJ promoter with the sequence TAATCC-N₁₇-TCTATT at position 7143 nucleotides upstream of the hslJ ORF (Fig. 1b). However, the major parts of the consensus sequence characteristic for the σ^S promoter were also found to reside in this DNA region (Fig. 1b). In order to discriminate between these two putative hslJ promoters, we analysed hslJ transcription in SY511 (rpoS^- cysB⁺) and in SY512, SY513 and SY514 cysB mutants carrying the inactive σ^S factor and the constructs pVGM5 or pVGM8 with an hslJlacZ operon fusion (see below) respectively, under different growth conditions. We showed that hslJ transcription does not depend on either the σ^S factor or the stationary growth phase (data not shown).

We searched for the sequence proposed to be the consensus for binding of LTTRs such as the CysB protein. Nucleotide sequence analysis for direct and inverted repeats was done using the PCgene program. Inspection of the hslJ regulatory region revealed the IR sequence CTATTcttaaAATAG positioned 32 nucleotides upstream of the hslJ ORF (Fig. 1b). This AT-rich sequence overlaps the -10 hexamer of the putative σ⁷⁰ hslJ promoter as well the hslJ TSP and resembles the conserved base pairs of a generic T-N₁₁-A LysR motif present in the binding sites for CysB and most LTTRs (Goethals et al., 1992; Lochowska et al., 2001). In addition, a second putative CysB binding site aTaATccccaATgAc was found 10 bp upstream of the first one (Fig. 1b). This site is not AT-rich but it contains the core of the LysR motif and the imperfect IR, overlaps the -35 hexamer of the putative hslJ promoter, and in concert with the first site resembles the organization of the CysB repressor sites CBS-B and CBS-K2 found in the cys regulon promoters (Hryniewicz & Kredich, 1995).

Definition of the hslJ promoter region
Different hslJlacZ operon fusions were constructed to dissect the regulatory region and to determine a DNA fragment that carries an active hslJ promoter. The constructs carried either the largest portion of the hslJ regulatory region A (pVGM3) or 5' deletion derivatives resulting in regions B (pVGM4) and C (pVGM5) (Fig. 2). All constructs contained the putative σ⁷⁰ hslJ promoter. Since hslJ was proposed to be a CysB-regulated gene, the activities of the hslJlacZ operon fusions were measured in the WT MC4100 and in a cysB mutant strains. As shown in Fig. 2, the deletion analysis showed that all constructs, including one with region C carrying the putative σ⁷⁰ hslJ promoter with the rest of 5' region deleted, were active. The β-galactosidase activities were at the same level in all constructs and slightly higher in the cysB mutant strain. This result defined region C to be the hslJ promoter region containing both the determined putative σ⁷⁰ hslJ promoter and the potential CysB repressor site (two LysR motifs).

(15K): Fig. 2. Construction of different hslJlacZ operon fusions and determination of the hslJ promoter region. Schematic representation of plasmids pVGM35, carrying the operon fusions (hslJAlacZ), (hslJBlacZ) and (hslJClacZ), respectively, constructed by ligating regions A (666 bp), B (150 bp) and C (93 bp) to the lacZ gene in EcoRI- and BamHI-digested promoterless vector pRS415 (bla; AmpR). Shaded box (PhslJ), σ70 putative promoter of the hslJ gene; Atg, hslJ start codon; black box (+1), hslJ TSP. DNA fragments are not to scale. The β-galactosidase activities (Miller units, MU) of the MC4100 cysB+ (WT) and SY602 cysB- (Mutant) strains carrying plasmids pVGM3, pVGM4 or pVGM5, or vector pRS415, are shown. The β-galactosidase activities represent the mean values of six independent assays with standard deviations in parentheses. The β-galactosidase activity in LB medium was assayed after growing cells at 37 °C in the presence of 0·1 mM cysteine.

Negative regulation of hslJ transcription
Considering the work of Lilic et al. (2003) and the results presented above, the moderate increase of the activities of the hslJ : : lacZ gene fusion and the hslJlacZ operon fusions, respectively, in the cysB mutant, might be due to action of the additional negative regulator that works in concert with CysB. This idea is substantiated by the finding that the WT hslJ presented on a plasmid, in trans, strongly negatively regulates the expression of the chromosomal hslJ gene while increasing the resistance to novobiocin (Lilic et al., 2003). In this previous work the hslJ : : lacZ gene fusion activities were always measured in conditions where the hslJ gene was inactivated either by the lacZ gene or by the Kan cassette (Lilic et al., 2003) and did not reflect the regulation of hslJ transcription. Therefore, in order to solve this question, the constructs with different hslJlacZ operon fusions, pVGM6, pVGM7 and pVGM8, described above (Fig. 2), were introduced into MC4100 isogenic strains carrying different combinations of the cysB and hslJ mutated genes (MC4100, cysB⁺ hslJ⁺; SY602, cysB^- hslJ⁺; SY605, cysB⁺ hslJ^-; SY515, cysB^- hslJ^-). In parallel with β-galactosidase activities we determined the Nov^R status of the transformants. The results we obtained using different fusions (pVGM6, pVGM7 and pVGM8) and different cysB mutated alleles (EC2549 cysB^null, EC2275 cysB and SY602 cysB) were essentially the same (data not shown). Hence we present the values obtained with the construct pVGM8 carrying the hslJ promoter region C, hslJ_ClacZ, introduced into the strains with combined mutations, SY602 cysB allele and hslJ : : ΩKan (Table 2). These results revealed that negative regulation of hslJ transcription is both CysB- and HslJ-dependent. This control is stronger when one of the proteins is not functional (cysB⁺ hslJ^- or cysB^- hslJ⁺) in comparison with the WT situation (cysB⁺ hslJ⁺) (Table 2), suggesting that negative regulations imposed by these functionally active proteins are in competition. The similar steady-state level of hslJ transcription in either WT or cysB mutant strains obtained in the primer extension experiments is in accordance with this result. The strongest repression was obtained by CysB alone while the maximal hslJ transcription is achieved in the absence of both functional CysB and HslJ proteins (cysB^-hslJ^-) (Table 2). This was the exact genetic background when we found that the hslJ : : lacZ gene fusion expression is elevated in a cysB mutant (Lilic et al., 2003). These results imply that hslJ transcription is under a double negative control effected by the expression of CysB regulator and the HslJ protein. These regulations are in competition but do not exclude each other.

Table 2. Activity of the hslJClacZ operon fusion from the pVGM8 construct in cysB, hslJ or double cysB hslJ mutant backgrounds

When the hslJ : : ΩKan allele encoding the functionally inactive HslJ is present, besides making the strain Nov^S, the HslJ-dependent negative regulation is absent as well (Table 2). The hslJ transcription regulated by both LTTR CysB and the expression of HslJ protein is insufficient to produce the critical concentration of the HslJ product and to increase the novobiocin resistance of the WT strain (Table 2). The strong negative regulation of hslJ transcription and the parallel increase in novobiocin resistance observed in the strain where the hslJ gene is intact and the CysB-imposed negative regulation is abolished (cysB^- hslJ⁺) (Table 2) could be due to transient elevation of HslJ production and the immediate downregulation of hslJ expression. This negative feedback hslJ autoregulation could be important because of the toxicity of HslJ observed upon overexpression of this protein (Lilic et al., 2003). Considering the structure of the HslJ protein and the fact that the hslJ : : ΩKan allele induces the psp operon, HslJ has been predicted to be located in the outer membrane of E. coli (Lilic et al., 2003). As mentioned above, among other stimuli, the psp operon is induced by expression of mutant OMPs (Model et al., 1997). Here we found the hslJ : : ΩKan allele to be inactive in autogenous control. Therefore, the HslJ-dependent negative regulation of the hslJ transcription is likely to be indirect, involving an additional negative regulator. The action of this regulator might be dependent on the proper HslJ cell localization.

However, a CysB-dependent negative regulation of the hslJ transcription should be direct. Since all results were obtained with the construct carrying the hslJ promoter region C, and since this region contains a putative CysB repressor site composed of two LysR motifs' found to be the consensus sequences for binding of the LTTRs, CysB could bind the hslJ regulatory region and act as a direct repressor of the hslJ transcription.

Purification of the active CysB regulator
We wanted to analyse the CysB binding properties of the hslJ regulatory region in vitro. In order to approach this issue, by adding His tag at the C-terminal end of CysB (36·3 kDa), we constructed the fusion protein CysB-His₆ (37·3 kDa) encoded from the plasmid pVGM1. The in vivo activity of this protein is proved by complementation of the cysteine auxotrophy. The cysB cysteine auxotrophs SY602/pREP4 and SY380/pREP4 were transformed with pVGM1. The transformants grew on minimal medium plates in the absence of cysteine (data not shown). Hence, we used strain SY380/pREP4 harbouring plasmid pVGM1 to overexpress and purify the fusion protein CysB-His₆ placed under the IPTG inducible promoter (see Methods). After 3 h induction in the presence of 1 mM IPTG, we went through the protein purification procedure and eluted >90 % pure CysB-His₆ fusion protein (37·3 kDa) in the fraction obtained by 0·250·3 M imidazole (Fig. 3). Only 10 % of the induced protein was soluble (90 % left in the pellet after lysis) (data not shown).

(39K): Fig. 3. Purification of the active CysB-His6 fusion protein. Lanes 1 and 2, purified CysB-His6, 37·3 kDa (arrow), after elution with 0·25 M (lane 1) and 0·3 M imidazole (lane 2). Lane 3, Protein standards.

It has been shown that in order to be active, the CysB regulator binds DNA of the target promoters as a homotetramer and positively or negatively regulates transcription (Kredich, 1996). Hence, the activity of the LTTR CysB depends on its potential to oligomerize and form homotetramers. We analysed this property of the purified CysB fusion protein by employing PAGE under nondenaturing conditions and at specific pH, screening the profiles of the CysB mobilities in different gel concentrations, and comparing the profiles with the mobilities of protein standards (see Methods). Knowing the molecular mass of the CysB fusion protein (37·3 kDa) we estimated the molecular masses of the oligomers obtained: CysB-His₆ tetramer, 150 kDa; CysB-His₆ hexamer, 225 kDa. CysB fusion protein oligomerizes up to tetramers and hexamers (data not shown); higher oligomers were also observed but we did not estimate their molecular masses.

CysB binding in the hslJ regulatory region
We used the purified CysB fusion protein to analyse binding of this transcription factor in the hslJ regulatory region. It has been shown for the cys regulon target genes that CysB binds DNA in both promoter and coding regions (-110 to +30) (Kredich, 1996). The protein in different concentrations, probe 301 comprising the 166 nt of the hslJ regulatory region (region B, 150 nt, is inside this portion of probe 301) and the 135 nt of the 5' region of the hslJ gene, and probe 135 covering only the 5' region of the hslJ gene were used (Fig. 4a). Gel retardation assay revealed that the CysB fusion protein binds only a DNA fragment containing the hslJ regulatory region (166 nt), making distinct DNAprotein complexes Cpx1 and Cpx2 (Fig. 4b, lanes 13). The appearance of these complexes was dependent on the protein concentration. Using region C as an unlabelled probe (Fig. 4c, lane 5) diminished the binding of CysB to probe 301. Formation of two complexes, such as Cpx1 and Cpx2, usually indicates two binding sites. As shown above, region C is the hslJ promoter region that carries the putative hslJ promoter and two LysR motifs' (the putative CysB repressor site composed of two binding sites) (Fig. 1b, Fig. 2). Therefore, CysB most likely binds the sequence predicted to be the CysB repressor site in an hslJ regulatory region, and upon binding acts directly as a repressor of hslJ transcription.

(40K): Fig. 4. Binding of CysB-His6 to the hslJ regulatory region. (a) Schematic representation of probes 301, 135 and C in relation to the hslJ gene (arrow) and the hslJ promoter region [black box (PhslJ), hslJ promoter; small black box (+1), hslJ TSP; ATG, hslJ start codon]. (b) CysB-His6 binds the hslJ regulatory region in a concentration-dependent manner. Lanes 13, CysBDNA 301 complexes, Cpx1 and Cpx2; lanes 46, CysBDNA 135 binding. (c) NAS facilitates binding of the CysB-His6 to the hslJ regulatory region. Lane 1, free DNA 301; lanes 24, binding of the CysB fusion protein in increasing concentrations to probe 301 (Cpx1); lane 5, CysB-His6 preincubated with unlabelled specific competitor (probe C) at 50-fold molar excess; lane 6, binding of CysB-His6 to probe 301 in the presence of 3 mM NAS (Cpx13); lane 7, binding of CysB-His6 to probe 301 in the presence of 3 mM NAS when the protein was preincubated with unlabelled specific competitor (probe C) at 50-fold molar excess.

The small molecule NAS has been considered to be the CysB ligand (inducer) that binds the cavity formed by two CysB monomers and, depending on the specific binding site in the target promoter, facilitates binding of the perpendicularly positioned HTH motifs of the CysB tetramer(s) to the activator sites while preventing binding to the repressor site in the cysB regulatory region (Kredich, 1996). The binding of CysB protein to probe 301 was analysed in the presence of NAS. CysB alone at a concentration of either 0·15 or 0·5 µg ml^-1 weakly bound probe 301 (Fig. 4c, lanes 3 and 4). However, in the presence of increasing concentrations of NAS (15 mM; result shown for 3 mM), CysB used at 0·15 µg ml^-1 bound probe 301 forming the more pronounced DNAprotein complexes Cpx1 and Cpx2, and the additional Cpx3 complex (Fig. 4c, lane 6). The efficacy of binding depended on NAS concentration as shown for other CysB-NAS-dependent DNA bindings (Kredich, 1996). Using region C as an unlabelled probe (Fig. 4c, lane 7) diminished this binding of CysB in the presence of NAS. Hence, once again, the portion of the hslJ regulatory region carrying the putative hslJ promoter and the putative CysB repressor site successfully competes for binding of CysB.

In all CysB-dependent regulations described to date, NAS plays the role of an inducer when used in concentrations like those used in this study (Kredich, 1996). We showed that CysB negatively regulates hslJ transcription, where it is proposed to bind the LysR motifs' that resemble the topology of the CysB repressor site that overlaps the hslJ promoter and the hslJ TSP. However, in the control of hslJ transcription where CysB acts as a repressor, NAS enhances the binding of this regulator to DNA. Therefore, it seems that the term inducer cannot be applied to NAS as regards CysB-mediated regulation of hslJ transcription.

cysB mutations
The CysB regulator is a multi-domain protein that contains the functional domains responsible for DNA binding (N-terminus), interaction with ligands (e.g. NAS) (central domain), and oligomerization and DNA binding (C-terminus) (Schell, 1993; Lochowska et al., 2001). The different MC4100 cysB mutants used in this work carry cysB alleles with specific mutations affecting the function of the entire protein or the specific functional domain. ΔcysB is a null mutation with the major portion of the cysB gene deleted and hence strain EC2549 entirely lacks the CysB product (Lochowska et al., 2001). However, the CysB I33N mutant protein produced by strain EC2275 carries a missense mutation in the recognition helix of the HTH motif in the CysB DNA binding domain (Lochowska et al., 2001) (Fig. 5b). This mutation has been shown to diminish the binding of CysB to both the activator and the repressor sites in the regulatory regions of the cys regulon genes and consequently the active regulation failed (Colyer & Kredich, 1994). The interallelic complementation of this allele with the other specific cysB allele encoding CysB T149Ter restored the functionality of the CysB regulator (Colyer & Kredich, 1994). In the previous and in this work we used the spontaneous MC4100 cysB mutant SY602 obtained by selection of Nov^R colonies on plates supplemented with 400 µg novobiocin ml^-1 (Lilic et al., 2003). This mutant, besides exhibiting resistance to novobiocin, was a cysteine auxotroph. We further characterized this mutant and found that the phenotypes obtained were due to mutation in the cysB gene (Lilic et al., 2003). The previous work by Rakonjac et al. (1991) and Lilic et al. (2003) and this study established that the Nov^R phenotype related to the cysB mutations is the consequence of elevated HslJ expression upon lack of hslJ transcription repression otherwise imposed by CysB. Hence, the repression of hslJ transcription, the Nov^S phenotype and the biosynthesis of the cysteine are the outcomes of the WT CysB activities.

(26K): Fig. 5. Characterization of the cysB allele from mutant SY602. (a) Schematic representation of the IS1 element insertion into the 3' region of the cysB allele from strain SY602. IS1 insertion generated the cysB831 : : IS1 mutation. IS1, insertion sequence 1 with left (IRL) and right (IRR) inverted repeats presented; Δ, deletion of the GGT nucleotides at the 5' end of IRL presented below as IRLΔGGT. The position of IS1 insertion in the 3' region of the cysB gene cysB831 : : IS1 and the corresponding amino acid sequence (CysB S277Ter) are presented below. Capital letters, cysB831 : : IS1 sequence; bold nucleotides show the 11 nt duplication upon IS1 insertion; small letters, IS1 sequence; 832/278, point of IS1 insertion. Underlined amino acids are those encoded by IS1. (b) Schematic representation of the CysB protein and its mutant forms, CysB I33N and CysB S277Ter. CysB functional domains are presented as described by Lochowska et al. (2001). HTH, DNA-binding motif. Positions of the I33N and S277Ter mutations in relation to the CysB domains are shown.

In this study we have characterized the MC4100 cysB allele in detail by cloning and sequencing the cysB gene from the mutant SY602 (see Methods). The sequencing revealed in-frame insertion of the 0·75 kb transposable element IS1 after nucleotide 831 in the 3' region of the cysB gene (Fig. 5a). We named this cysB allele as cysB831 : : IS1. This result is in agreement with the restriction pattern and the 1·7 kb length of the PCR fragment obtained using SY602 chromosomal DNA and a pair of primers framing the cysB gene assuming the insertion of the 0·75 kb DNA in the 3' region of the 1 kb long WT cysB gene. The IS1 element with 5'-cysB831 : : IR_L-(IS1)-IR_R-3' orientation is flanked by duplication of the 11 nt from the cysB coding region (Fig. 5a). The first three nucleotides GGT of the IR_L (left inverted repeat) are deleted, most likely leaving the IS1 element fixed and inactive for further transposition (Fig. 5a). All these characteristics of the IS elements (e.g. IS1) and their transpositions have been seen previously and have been reviewed by Galas & Chandler (1991).

The insertion of IS1 element interrupts the region that encodes the CysB domain responsible for the oligomerization and binding of this regulator to DNA (Fig. 5b). In order to answer whether the CysB mutant regulator encoded by the cysB831 : : IS1 allele is produced as a CysB S277Ter protein (Fig. 5a, b) and whether its functionality can be restored by complementation with the other, different, CysB mutant(s), we performed an interalellic complementation experiment. We followed the resistance to novobiocin as a measure of hslJ expression and the biosynthesis of cysteine as a functionality of the cys regulon control enabled by CysB. The results are presented in Table 3. Plasmid pVGM9 carrying the cysB831 : : IS1 allele complemented the chromosomal copy of the EC2275 cysB allele encoding the CysB I33N protein, decreasing the novobiocin resistance, but it failed to complement the cysteine auxotrophy of strain EC2275. As a control, neither the cysB831 : : IS1 allele nor the EC2275 cysB allele complemented a ΔcysB mutation in strain EC2549, while the WT cysB⁺ efficiently restored the Nov^S and the cysteine prototrophy phenotypes. Hence, we showed that the functional restoration of CysB regarding the repression of the hslJ transcription and the consequent Nov^S phenotype can be obtained to a certain extent by complementation of the alleles that encode the CysB I33N and CysB S277Ter proteins. However, the combination of these mutated CysB proteins did not complement the cysteine auxotrophy. This result suggests that the restored oligomer of the CysB regulator partially represses the hslJ transcription while not functioning in regulating the transcription of the cys regulon genes. This result is different from that obtained by Colyer & Kredich (1994), where the interallelic complementation of the CysB I33N and the CysB T149Ter mutant regulators restored the functional regulation of the cys regulon genes and complemented the cysteine auxotrophy. One may suspect that the DNA-binding property of the LTTR CysB slightly differs in the hslJ system in comparison to the cys regulon target promoters. It could be that binding is correct but the transcription activation (CysB as an activator) important for the cys regulon genes but not for the hslJ transcription regulation (CysB as a repressor) failed.

Table 3. Interallelic complementation of cysB mutant genes that encode the CysB I33N and CysB S277Ter regulators

Conclusions
In this work we determined the hslJ TSP, the putative σ⁷⁰ promoter and the hslJ promoter region. We found that hslJ transcription is subject to a complex double negative control by CysB and autogenously upon expression of HslJ. These two regulations compete in controlling hslJ transcription but do not exclude each other. CysB is a repressor of hslJ transcription that binds a DNA sequence (region C) which carries the putative hslJ σ⁷⁰ promoter and the potential CysB repressor site. The putative repressor site is composed of two binding sites, LysR motifs, that overlap the -10 and -35 hexamers of the hslJ σ⁷⁰ promoter, and the hslJ TSP. The CysB ligand, NAS, hitherto considered to be an inducer in CysB-mediated gene regulations, instead of preventing, enhances the CysB-binding in hslJ promoter region. The interallelic complementation of the mutated cysB genes that encode the CysB I33S and the CysB S277Ter regulators suggests that there may be differences in DNA binding of CysB to the hslJ regulatory region in comparison to the regulatory regions of the cys regulon genes. The negative autogenous control of hslJ transcription is observed but the mechanism of regulation remains elusive. This autoregulation might be indirect and enabled by an assumed factor X dependent on the proper HslJ localization. The toxicity of the overexpressed HslJ, which may be explained by the proposed localization in the outer membrane, could be the reason for the tight negative regulation of the hslJ transcription. Only the small concentration of the HslJ produced in WT cells or the limiting higher amount of the HslJ transiently produced in the cysB mutants is tolerated by the bacterium. However, the HslJ transiently produced in the higher concentration (e.g. in cysB mutants) is sufficient to provide the novobiocin resistance followed by the immediate negative feedback control and the repression of hslJ transcription. This model could be analysed by using different approaches. The mechanism of HslJ-expression-dependent autogenous control could be illuminated by using a chase experiment in a cysB mutant background, by determining the exact localization of the HslJ protein, and by screening for the HslJ-dependent regulation factor which, when mutated, in the cysB mutants leaves the hslJ transcription open. Finally, the mechanism of novobiocin resistance, specific or not, the biological significance of the HslJ protein and its role in the physiology of bacteria remain for future studies. This work was supported by the Ministry of Science and Technology of the Republic of Serbia (MSTRS, grant no. 03E12). We thank Monika Hryniewicz for kindly providing strains EC2275 and EC2549.

Footnotes

^†, Dragutin J. Savic †Present address: Laboratory of Structural Microbiology, RU Box 52, The Rockefeller University, 1230 York Avenue, New York, NY 10021, USA. ‡Present address: Department of Microbiology and Immunology, Health Sciences Center, University of Oklahoma, PO Box 26901, Oklahoma City, OK 73190, USA.