The transcriptional response to cadmium, organic hydroperoxide, singlet oxygen and UV-A mediated by the σE–ChrR system in Caulobacter crescentus

Authors


*E-mail sulgomes@iq.usp.br; Tel. (+55) 11 3091 3826; Fax (+55) 11 3091 2186.

Summary

Caulobacter crescentusσE belongs to the ECF (extracytoplasmic function) subfamily of RNA polymerase sigma factors, whose members regulate gene expression in response to distinct environmental stresses. During physiological growth conditions, data indicate that σE is maintained in reduced levels due to the action of ChrR, a negative regulator of rpoE gene expression and function. However, once bacterial cells are exposed to cadmium, organic hydroperoxide, singlet oxygen or UV-A irradiation, transcription of rpoE is induced in a σE-dependent manner. Site-directed mutagenesis indicated that residue C188 in ChrR is critical for the cadmium response while residues H140 and H142 are required for the bacterial response to organic hydroperoxide, singlet oxygen and UV-A. Global transcriptional analysis showed that σE regulates genes involved in protecting cells against oxidative damages. A combination of transcriptional start site identification and promoter prediction revealed that some of these genes contain a putative σE-dependent motif in their upstream regions. Furthermore, deletion of rpoE and two σE-dependent genes (cfaS and hsp20) impairs Caulobacter survival when singlet oxygen is constantly generated in the cells.

Introduction

Solar radiation reaching the surface of the Earth is one of the most common stresses encountered by bacteria in their environment due to the generation of reactive oxygen species (ROS). After absorbing this irradiation, endogenous cellular compounds (sensitizers) are excited to highly reactive triplet states, and these excited compounds can either decay to the ground state by energy dissipation or undergo photochemical reactions. The transfer of an electron from the excited sensitizer to molecular oxygen is responsible for the increased intracellular concentrations of superoxide radical (O2-), hydrogen peroxide (H2O2) and hydroxyl radical (OH). Alternatively, the excited compounds can directly transfer their energy to molecular oxygen, generating singlet oxygen (1O2) (Wondrak et al., 2006).

Exposure of bacterial cells to heavy metals is another environmental condition known to mediate its biological effect through the action of ROS. Some heavy metals, including iron, copper and chromium, are redox-active and can directly produce ROS. On the other hand, the non-redox-reactive heavy metals, such as cadmium, lead and mercury, can bind thiol groups, leading to glutathione depletion and oxidation of sulphydryl groups in proteins. Thus, the increase in the intracellular concentration of ROS during cell exposure to these non-redox-reactive metals occurs indirectly, as a consequence of the decreased antioxidant defence of the cell. In addition, some heavy metals can also mediate their toxicity in a ROS-independent manner. Zinc or iron displacement from metalloproteins, resulting in the inactivation of these proteins, is one example (Stohs and Bagchi, 1995).

Reactive oxygen species generated under these environmental conditions can lead to cell death by damaging important cellular compounds (Pryor et al., 2006). To protect themselves against the oxidative injuries, cells increase the expression of genes involved in ROS elimination prior to damage occurrence, repair of damaged molecules and metal detoxification. Several of these genes in bacteria are under the direct control of activators, repressors or ECF (extracytoplasmic function) sigma factors (Storz and Imlay, 1999; Fuangthong et al., 2001; Paget et al., 2001; Anthony et al., 2005). Whereas the activity of the activators and repressors is directly modulated by the redox state of the cells, ECF sigma factors involved in such response are indirectly controlled by ROS via their corresponding anti-sigma factors (Storz and Imlay, 1999; Helmann, 2002). Hence, during the reducing condition normally found in the cytoplasm of bacterial cells, these sigma factors are maintained in an inactive state by protein–protein interaction with their cognate anti-sigma factors. However, once cells are exposed to environmental conditions that induce ROS generation, the sigma/anti-sigma complex is disrupted, releasing the sigma factor to direct transcription of the target genes.

Caulobacter crescentus is a free-living bacterium found mostly in nutrient-poor aquatic environments (Poindexter, 1981). Due to its lifestyle, C. crescentus cells might be exposed to solar light and heavy metals. A whole-genome transcriptional analysis of cadmium stress in this bacterium showed that important components of the response to such condition is the maintenance of reduced intracellular concentration of the metal and the protection against oxidative stress (Hu et al., 2005). Some genes predicted to be involved in the control of gene expression are also part of this response, suggesting a key role for these genes in sensing and responding to the heavy metal. Among these regulatory genes, one finds CC0648, which encodes one of the 13 putative ECF sigma factors of C. crescentus (Nierman et al., 2001).

This work shows that CC0648, here named rpoE, is responsible for triggering part of the transcriptional response to cadmium, controlling genes involved in protecting cells against oxidative damages. Besides this heavy metal, data revealed that rpoE is also involved in the bacterial response to tert-butyl hydroperoxide, 1O2 and UV-A irradiation. In addition, evidences are presented indicating that chrR (CC0647), which encodes a putative anti-sigma factor, negatively regulates rpoE expression. Furthermore, a conserved cysteine residue in ChrR was shown to be critical for cadmium response, whereas two also conserved histidine residues are required for the bacterial response to tert-butyl hydroperoxide, 1O2 and UV-A. This is the first characterization of amino acid residues in ChrR directly involved in the response to specific stresses.

Results

ChrR is a negative regulator of rpoE expression

Caulobacter rpoE is located upstream of two other genes (chrR and CC0646), all of them being transcribed in the same orientation. Proteins encoded by rpoE and chrR are significantly similar (58.5% and 65.7% of similarity in global alignments respectively) to σE and the anti-sigma factor ChrR, respectively, of Rhodobacter sphaeroides (Newman et al., 1999). The third gene of the putative Caulobacter operon, CC0646, encodes a protein from the photolyase/cryptochrome family. The possible organization of rpoE, chrR and CC0646 genes in an operon was investigated by RT-PCR experiments and results suggested the presence of an mRNA encompassing all three genes in Caulobacter (data not shown).

To investigate the possible role of ChrR in the expression of rpoE, the construction of a null mutant strain in the chrR gene was attempted. However, deletion of chrR was obtained in only two instances: in the presence of a plasmid-encoded functional copy of chrR under the control of the xylose-dependent promoter PxylX (pRL390), generating strain SG250, or after extended incubation periods during selection for the second recombination event, producing strain SG251. As inferred by β-galactosidase assays carried out with cells harbouring a transcription fusion between the upstream region of rpoE and the lacZ reporter gene (construct pRSR), rpoE promoter activity in strain SG250 growing in the presence of xylose is increased relative to that observed in the parental strain NA1000 (Fig. 1). Moreover, replacement of xylose for glucose, which should cause a reduction in chrR expression, leads to an even larger increase in rpoE promoter activity (Fig. 1).

Figure 1.

Expression of rpoE is negatively controlled by ChrR. Activity of rpoE promoter was determined by β-galactosidase assays carried out with cells from parental strain NA1000, conditional chrR mutant SG250 and chrR null mutants SG251-1 and SG251-2. Strain SG250 was grown in the presence of either 0.2% d-xylose (xyl) or 0.2% d-glucose (glu) for 4 h after previous removal of xylose, whereas strains SG251-1 and SG251-2 were grown in the absence of any sugar added (−). NA1000 cells growing under the same conditions of strains SG250 and SG251 were used as control. Results are mean values of three independent experiments; bars represent the standard error.

Analysis of bacterial colonies of chrR null mutant strain SG251 obtained after prolonged incubation to isolate a double cross-over revealed two types of null mutants: (i) SG251-1, in which rpoE promoter activity was found to be increased twofold relative to the parental strain NA1000, and (ii) SG251-2, with no change in rpoE promoter activity relative to the parental strain (Fig. 1). Nucleotide sequence analyses revealed that both types of null mutants present secondary mutations in the rpoE gene: a single-nucleotide substitution (G181A) in strain SG251-1 (resulting in the non-conservative substitution G61S in σE), and an insertion after nucleotide 383 in the null mutant strain SG251-2, producing an entirely different protein sequence from there on (data not shown). Together, these results suggest that ChrR is necessary for bacterial growth under normal conditions and negatively regulates rpoE expression.

rpoE is involved in the response to cadmium, organic hydroperoxide, 1O2 and UV-A

Data from β-galactosidase assays revealed that rpoE promoter activity increases when bacterial cells are exposed to cadmium (Fig. 2), as previously indicated (Hu et al., 2005). Besides this heavy metal, it was observed that rpoE promoter activity is induced following Caulobacter treatment with tert-butyl hydroperoxide (tBOOH, which represents stress by alkyl hydroperoxides) or methylene blue (MB) in the presence of white light (source of 1O2) and during UV-A irradiation even in the absence of an exogenous sensitizer (Fig. 2). Both MB and white light are required for the increase in rpoE promoter activity (data not shown). On the other hand, no induction occurred during exposure of Caulobacter cells to paraquat (source of O2-), diamide (thiol-specific oxidant), H2O2 and other heavy metals, including cooper, iron, cobalt, nickel and zinc (data not shown).

Figure 2.

rpoE promoter activity is induced by cadmium, tBOOH, 1O2 and UV-A irradiation in an rpoE-dependent manner. β-Galactosidase assays carried out with cells from rpoE null mutant containing the empty vector pJS14 and parental strain NA1000 bearing the construct pRL380 (overexpressing chrR) or the empty vector pJS14 exposed or not for 1 h to 25 μM CdSO4, 100 μM tBOOH, 2.5 μM MB in the presence of white light (WL – 6 W·m−2) or UV-A irradiation (emission peak at 365 nm) at irradiance of 10 W·m−2 (total dose of 36 kJ·m−2). Results are mean values of three independent experiments; bars represent the standard error.

To verify whether Caulobacter rpoE controls its own expression, an rpoE null mutant was constructed. Even though rpoE promoter activity is not affected by deletion of the rpoE gene during normal bacterial growth, its induction by cadmium, tBOOH, 1O2 and UV-A is lost in the absence of this gene (Fig. 2). Similarly, rpoE promoter activity is not affected by the inducers in parental cells overexpressing chrR. Together, these results indicate that rpoE is positively autoregulated under these stress conditions.

To investigate if induction of rpoE promoter activity leads to an increase in σE levels and to determine the kinetics of its accumulation, a polyclonal antibody against a recombinant σE protein expressed in Escherichia coli was raised in rabbits. Immunoblot assays using this antibody revealed that σE levels are transiently increased after treatment of Caulobacter cells with cadmium (Fig. 3A) or tBOOH (Fig. 3B). The amount of σE also increases following 1O2 exposure and high levels of the protein are maintained as long as bacterial cells containing MB are kept under white light irradiation, as 1O2 is constantly generated in such condition (Fig. 3C). However, the amount of σE decreases when the white light source is removed from NA1000 cultures previously exposed to MB and light (Fig. 3C). Similarly, it was verified that σE levels increase during UV-A irradiation and continued exposure under this condition is needed to sustain the high levels of the protein (Fig. 3D).

Figure 3.

σE levels increase following Caulobacter exposure to cadmium, tBOOH, 1O2 and UV-A irradiation. Results shown are from immunoblot assays performed with total protein extract of NA1000 cells exposed to each stress condition. Aliquots for preparation of extracts were obtained at the indicated time points after the stress started or after shift to the dark. (A) 25 μM CdSO4; (B) 100 μM tBOOH; (C) 2.5 μM MB and white light (WL – 6 W·m−2); (D) UV-A irradiation (emission peak at 365 nm) at irradiance of 10 W·m−2. The blots were developed using anti-σE antibody and anti-rabbit IgG conjugated with alkaline phosphatase. The band corresponding to σE is indicated by an arrowhead. In addition, the blots also show a lower band (used as an internal control) corresponding to a cross-reactivity of the antibody as revealed by protein extracts of rpoE null mutant (not shown).

Immunoblot assays using anti-σE antibody were also employed to monitor rpoE expression in the presence of paraquat, diamide, H2O2 and the heavy metals cooper, iron, cobalt, nickel and zinc. However, no increase in σE levels was observed upon treatment with these stress agents, suggesting that σE is not involved in the bacterial response to such stress conditions, as also revealed by β-galactosidase assays (not shown).

Conserved histidine and cysteine residues of ChrR are required for the response to specific inducers

As ChrR orthologues in distinct bacteria have conserved histidine and cysteine residues in their C-terminal domain (see Fig. S1 in Supporting information and Campbell et al., 2007), it was asked whether the conserved residues in Caulobacter ChrR (H140, H142, H176, C186 and C188) could be involved in the response to cadmium, tBOOH, 1O2 and UV-A irradiation. For that, site-direct mutagenesis was used to individually replace these histidine and cysteine residues for alanine and serine, respectively, as described in Experimental procedures. As revealed by quantitative RT-PCR analyses, mutant ChrR proteins containing one of the substitutions H140A, H142A, C186S and C188S are still functional in vivo, as rpoE expression is close to its basal level in the absence of stress (Fig. 4). On the other hand, it was not possible to obtain a mutant strain expressing ChrR/H176A employing the same approach, suggesting that this substitution could cause a complete inactivation of ChrR, which is not tolerated by the cell, similarly to what was observed during the construction of chrR null mutant strain.

Figure 4.

Substitution of ChrR residues H140, H142, C186 and C188 affects differently the bacterial response to cadmium, tBOOH, 1O2 and UV-A. Results shown are from quantitative RT-PCR performed with total RNA extracted from cells expressing wild-type ChrR or a mutant version of the protein (ChrR/H140A, ChrR/H142A, ChrR/C186S, ChrR/C188S and ChrR/C186-188S) before and during the following stress conditions: 25 μM CdSO4 for 60 min, 100 μM tBOOH for 15 min, 2.5 μM MB and white light (6 W·m−2) for 30 min or UV-A irradiation (emission peak at 365 nm) at irradiance of 10 W·m−2 for 30 min (total dose of 18 kJ·m−2). Values represent the fold increase of rpoE expression in the corresponding strain exposed or not to the stress conditions compared with that of the parental strain NA1000 growing under normal conditions. Results were normalized using gene CC0088 as an endogenous control, which was shown to be constitutively expressed in the samples analysed. Data are mean values of two independent experiments; bars represent the standard error.

Even though ChrR/H140A and ChrR/H142A are able to negatively control rpoE expression, quantitative RT-PCR experiments showed that cells expressing one of these mutant proteins completely lost the ability of inducing rpoE expression following exposure to tBOOH, 1O2 and UV-A irradiation, whereas rpoE induction during cadmium stress is not affected in these strains (Fig. 4). These results suggest a role for residues H140 and H142 in Caulobacter response to tBOOH, 1O2 and UV-A irradiation. On the other hand, rpoE induction during cadmium exposure was partially affected in cells expressing ChrR/C186S and completely abolished in cells expressing ChrR/C188S, with no change in the response to tBOOH, 1O2 or UV-A (Fig. 4). Thus, whereas the requirement of residue C188 for the cadmium response is evident from data presented here, residue C186, however, does not appear to be crucial for this response, as ChrR/C186S is still apparently inactivated in the presence of this heavy metal, even though to a lesser extent.

In addition, residues C186 and C188 of ChrR were also simultaneously replaced for serine. As revealed by RT-PCR analyses, rpoE expression levels in this double mutant are increased relative to those observed in the parental strain NA1000 during normal growth conditions (Fig. 4). However, the mutant protein ChrR/C186-188S is probably not completely inactive, as rpoE expression is further increased following tBOOH, 1O2 and UV-A exposure (Fig. 4). On the other hand, the absence of both cysteines C186 and C188 in ChrR completely prevented induction of rpoE expression in the presence of cadmium. Therefore, the behaviour of Caulobacter cells expressing ChrR mutant proteins in responding to different stresses indicates that cadmium leads to ChrR inactivation by a mechanism distinct from the one during tBOOH, 1O2 and UV-A exposure.

Identification of σE-dependent genes

Using transcriptome analysis, a total of 56 genes showed increased expression (more than 1.5-fold) in NA1000 cells overexpressing rpoE from a plasmid-encoded copy of the gene under the control of PxylX promoter (pRL290) in comparison with cells carrying the empty vector pUJ142 (Table S1 in Supporting information). Among these putative σE-dependent genes, 14 were also previously identified as members of the cadmium stimulon in C. crescentus (Hu et al., 2005) (Table 1), suggesting that σE might be responsible for induction of most of these genes, if not all. Interestingly, some of these genes are upregulated specifically by cadmium among the metals tested, while others were also assumed as part of either chromium or uranium stimulons (Hu et al., 2005) (Table 1).

Table 1.  Overlapping between the σE-regulon and metal stress stimulons (Hu et al., 2005).
Gene numberGene nameCadmiumChromiumUranium
CC0351 +  
CC0419 +  
CC0459folE+  
CC0646 +  
CC0647chrR+  
CC0648rpoE+  
CC1039msrA++ 
CC1179 ++ 
CC1427cfaS+  
CC1428phr+  
CC2258hsp20+ +
CC2600 +  
CC2602 +  
CC2748 ++ 

To confirm the results obtained by transcriptome analysis, five selected genes putatively dependent on σE were tested in quantitative RT-PCR experiments. Transcript levels of these genes were found to be higher in cells overexpressing rpoE relative to those observed in the parental strain (Fig. 5A). All five genes are also highly induced in NA1000 cells exposed to MB and white light, without any increase being observed in the rpoE null strain under the same condition (Fig. 5B). In addition, absence of rpoE does not affect the expression levels of these genes during normal growth condition (Fig. 5B). Thus, these results validate the analysis for identification of σE-dependent genes and suggest that this ECF sigma factor is in an inactive state in Caulobacter cells under normal growth conditions.

Figure 5.

Expression of σE-dependent genes increases in cells overexpressing rpoE and during 1O2 exposure in an rpoE-dependent manner. Data are from quantitative RT-PCR.
A. Total RNA was extracted from the parental strain NA1000 harbouring construct pRL290 (to overexpress rpoE) or the empty vector, after growing in the presence of 0.2% xylose for 30 min. Values represent the fold increase in gene expression in the strain overexpressing rpoE compared with that of the parental strain.
B. Strain NA1000 and the rpoE null mutant were exposed to 2.5 μM MB and white light (6 W·m−2). Total RNA was obtained before and 30 min after stress started. Grey bars correspond to strain NA1000 under stress condition. White and black bars refer to rpoE null mutant exposed to 1O2 and growing under no stress condition respectively. Values represent the fold increase in gene expression compared with that of the parental strain NA1000 growing under normal conditions.
Data are mean values of two independent experiments; bars represent the standard error.

Identification of σE-dependent promoters

As σE controls its own expression, a 5′RACE experiment was performed from cells exposed to 1O2 with the aim to identify a putative σE-dependent promoter upstream of rpoE. A single transcription start site (TSS) was observed at position +7 relative to the translation start site (+1) predicted by the TIGR annotation (data not shown), which is in good agreement with the position previously determined by a high-throughput analysis of transcription start sites of Caulobacter genes in cells exposed to cadmium (position +6) (McGrath et al., 2007). Analysis of the region upstream of rpoE TSS revealed the putative promoter elements −35 and −10 corresponding to the sequences TCATC and CGTAT respectively (Fig. 6A). These motifs are highly similar to those found upstream of the rpoE gene from distinct bacteria (Dufour et al., 2008).

Figure 6.

Identification of the putative σE-dependent promoter motif in the upstream region of genes controlled by σE.
A. Table showing the putative −35 and −10 promoter elements of the genes predicted to be directly controlled by σE. The σE-binding motif (TCATC-N17−18-CGTAW, where W is A or T) derived from the rpoE 5′RACE experiment was used to perform a search in region encompassing nucleotides −300 to +5 relative to the predicted translation start site (+1) of all putative σE-dependent genes, allowing for two substitutions. The ‘dna pattern’ tool of RSA website (http://rsat.ulb.ac.be/rsat) was used in this search. The co-ordinate represents the position of the 3′ end nucleotide of the putative σE-binding motif relative to the translation start site (+1). Genes in parenthesis are proposed to be co-transcribed with the gene containing the putative σE-binding motif.
B. Sequence logos corresponding to the putative consensus for σE-dependent promoters in C. crescentus and R. sphaeroides. In both panels, the logos were generated with WebLogo (http://weblogo.berkeley.edu/) using the putative promoters identified by an in locus sequence analysis: 17 promoter sequences in C. crescentus and 15 in R. sphaeroides (Dufour et al., 2008).

An in locus search in the region upstream of all putative σE-dependent genes for the sequence TCATC-N17−18-CGTAT revealed promoter elements −35 and −10 similar to those observed for rpoE in the region upstream of 10 other transcription units (Fig. 6A). Mutational analysis previously carried out with the R. sphaeroides system showed that σE apparently does not discriminate between TAT and TAA in the −10 motif of promoters recognized by this sigma factor (Newman et al., 1999). If this feature is also true for the C. crescentus system, genes CC0358, CC0462, CC1160, CC1625, CC2600 and CC2806 would be probably under the direct control of σE, as a sequence similar to TCATC-N17−18-CGTAA is found in the region upstream of these genes (Fig. 6A). To obtain further evidence that σE controls these transcription units by recognizing conserved motifs in their upstream region, the TSS of genes CC0459 and CC1428 were determined by 5′RACE carried out from cells exposed to 1O2 (data not shown). In agreement, the TSS of these genes is in a perfect match to the one identified by the promoter motif search. Even though the technique used for identification of TSS could also identify processing sites, the 5′ end in the correct distance to a predicted promoter sequence is very likely a TSS.

The putative consensus for the CaulobacterσE-dependent promoter derived here from an in locus search is highly similar to that proposed to be recognized by RhodobacterσE (Fig. 6B), which is in agreement with the high degree of similarity between these two ECF sigma factors.

rpoE, cfaS and hsp20 are necessary for Caulobacter survival during 1O2 stress

The capacity of the rpoE null mutant and the parental strain NA1000 in surviving to cadmium, tBOOH or 1O2 stresses was analysed, but no increased sensitivity was observed for the mutant strain following exposure to cadmium or tBOOH, even under conditions more severe than those used to monitor rpoE expression (data not shown). On the other hand, the rpoE null strain displayed increased sensitivity relative to the parental strain NA1000 when both strains were constantly exposed to 5 μM MB and white light (Fig. 7).

Figure 7.

rpoE, cfaS and hsp20 genes affect C. crescentus survival to 1O2 exposure. Survival curves of bacterial cells from parental strain and single mutants in rpoE, cfaS or hsp20 genes during exposure to 5.0 μM MB in the presence of white light (6 W·m−2) are shown. Aliquots were taken for cell viability determination at the indicated time points after stress started. Results are mean values of three independent experiments; bars represent the standard error.

To test the involvement of some σE-dependent genes in the C. crescentus survival to 1O2 exposure, null mutants in genes msrA, cfaS and hsp20 were constructed. However, no increase in sensitivity to 1O2 was observed in cells lacking msrA, and only a small increase was detected in cells deleted for the hsp20 gene (Fig. 7). On the other hand, cells lacking cfaS were found to be much more sensitive than the parental strain to the presence of 1O2 (Fig. 7). Cell survival during 1O2 stress was also investigated in a double mutant in msrA and hsp20, but this strain was as little impaired in bacterial survival under such condition as cells lacking only hsp20 (data not shown). As expected, no difference in viability was observed in the parental and null mutant strains when cells were exposed to white light in the absence of MB (data not shown).

Discussion

This report describes a molecular system involved in Caulobacter response to cadmium, organic hydroperoxide, 1O2 and UV-A irradiation. It is apparent from the data presented here that the C. crescentus system is closely related to the well-characterized R. sphaeroidesσE–ChrR system in several aspects: (i) rpoE gene is located in a probable transcriptional unit with chrR, (ii) chrR negatively regulates the expression levels of the σE-dependent genes, (iii) rpoE gene expression is induced following 1O2 exposure, (iv) σE directly controls its own expression and genes involved in photoreactivation of pyrimidine dimmers and unsaturated fatty acid modification, and (v) σE recognizes similar −10 and −35 sequences in the promoter regions of target genes.

Despite these similarities between the σE–ChrR systems from C. crescentus and R. sphaeroides, an important difference is revealed by the alignment of amino acid sequences of ChrR orthologues: the lack of the zinc-biding motif HxxxCxxC in the N-terminal portion of the Caulobacter protein, ruling out the involvement of this metal in maintaining the structure of this domain, in contrast to what was shown for the Rhodobacter protein (Newman et al., 2001). In spite of this difference, an overall conservation is evident in the N-terminal sequences of both ChrR proteins, suggesting that this portion of the Caulobacter protein is still the domain responsible for the anti-sigma activity. Zinc-binding ligands are also missing in E. coli RseA, which displays an N-terminal domain structurally similar to Rhodobacter ChrR, despite the low sequence similarity (Campbell et al., 2007). The lack of the zinc-binding motif in Caulobacter ChrR also suggests that zinc binding to the N-terminal domain is not required for the bacterial response to the inducers. Curiously, only five other bacteria presenting ChrR orthologues also lack this metal-binding motif (see Fig. S1), three of them (Maricaulis maris, Hyphomonas neptunium and Oceanicaulis alexandrii) were recently proposed to belong to the group of Caulobacterales (Badger et al., 2005), indicating that this might be a characteristic of this order, differentiating them from Rhodobacterales. Notably, the histidine residues H140 and H142 of Caulobacter ChrR required for tBOOH, 1O2 and UV-A responses along with H176 and E146 are part of a conserved motif recently shown in the Rhodobacter system to be involved in zinc binding in the cupin-like domain (Campbell et al., 2007), suggesting that the Caulobacter protein might associate with the metal in its C-terminus. However, the zinc ion at this position is easily dissociated from Rhodobacter ChrR and it is not required for maintaining the protein structure (Campbell et al., 2007), remaining to be investigated whether this metal has a physiological role during inactivation of ChrR under tBOOH, 1O2 and UV-A stresses.

Results shown here indicated that environmental conditions experienced by C. crescentus in which the intracellular levels of organic hydroperoxide and/or 1O2 increase lead to induction of rpoE expression. However, it is not possible to conclude if organic hydroperoxide and 1O2 independently cause ChrR inactivation or whether only one of these ROS is able to act upon ChrR, since organic hydroperoxide can stimulate 1O2 generation and vice versa. As UV-A irradiation is known to mediate its biological effects through ROS generation (Wondrak et al., 2006), it is reasonable to suppose that organic hydroperoxide and/or 1O2 are responsible for the increase in rpoE expression under such environmental condition. On the other hand, cadmium exposure, another environmental condition possibly encountered by C. crescentus in its natural habitat, appears to lead to rpoE induction independently of the action of ROS. The slower response to cadmium exposure compared with other inducers may imply that the effect of this heavy metal could be more indirect, possibly depleting thiols and hence causing a thiol-oxidative environment. However, the possibility of cadmium binding to ChrR cannot be excluded, since the sulphydryl-binding property of this metal is known (Stohs and Bagchi, 1995). Therefore, although the stress conditions that affect the activity of σE and ChrR are known, the inducers directly sensed by this system remain to be determined.

The observation that residues required for Caulobacter response to stress conditions are located in the C-terminal domain of the protein confirms the hypothesis of modularity of ChrR function, in which the N-terminus of the protein is responsible for binding and sequestering the cognate sigma factor whereas the C-terminal domain is involved in responding to the stress conditions (Campbell et al., 2007). Interestingly, although the histidine residues shown here to be required for the tBOOH, 1O2 and UV-A response are widely conserved in ChrR from distinct bacteria, the corresponding cysteine residue necessary for the Caulobacter response to cadmium is lacking in some bacteria (see Fig. S1), suggesting that the σE–ChrR system from such bacteria is possibly involved in the cellular response to tBOOH, 1O2 and UV-A but not to cadmium.

Despite the overall conservation between the putative consensus sequences for σE-dependent promoters in C. crescentus and R. sphaeroides, there is a notable difference regarding their −35 regions: while the sequence TCA is present in the −35 region of most of the σE-dependent promoters in C. crescentus, sequences proposed to be recognized by RhodobacterσE contain mainly a TGA motif in their −35 regions (Dufour et al., 2008). In accordance, the alanine residue of CaulobacterσE inferred by amino acid sequence alignment with E. coliσE to contact the cytosine in the TCA motif is not conserved in the Rhodobacter protein, in which a serine was proposed to interact with the guanine in the corresponding sequence TGA (Lane and Darst, 2006; Dufour et al., 2008). Interestingly, in the majority of the σE–ChrR systems, the interaction between the sigma factor and the second nucleotide of the −35 region appears to be similar to that inferred for R. sphaeroides (Dufour et al., 2008). However, this particularity of the Caulobacter system apparently does not impact the set of genes that probably compose the core σE–ChrR regulon in this bacterium, as a combination of transcriptome analysis and promoter prediction reported here suggests that genes rpoE, chrR, phr, cfaS and probably CC2600 are also directly regulated by σE in C. crescentus. In addition, CaulobacterσE apparently controls the expression of eight genes (or 14 genes considering that CaulobacterσE does not discriminate between TAT and TAA in the −10 motifs) that comprise an extended σE–ChrR regulon. Among the functions under the direct control of CaulobacterσE, photoreactivation of pyrimidine dimers, unsaturated fatty acid modification and tetrahydrofolate (THF) synthesis appear to play important roles mainly during the photo-oxidative stress. On the other hand, C. crescentus lacks an RpoHII-like sigma factor that directly controls genes whose upstream regions do not contain a putative σE-dependent promoter, in contrast to R. sphaeroides and some other bacteria (Dufour et al., 2008; Nuss et al., 2008). Thus, it is not clear how these genes are regulated once σE is activated in the cells.

In conclusion, this report provides important new biological insights into the role and regulation of the conserved σE–ChrR system: (i) the involvement of this system in stress responses other than 1O2, as shown here for Caulobacter, precluding the previous implication of this system as a pathway specifically activated by 1O2, (ii) the demonstration that a ChrR orthologue can be functional and responsive to stresses even lacking the N-terminal zinc-binding motif HxxxCxxC, (iii) the characterization of conserved histidine and cysteine residues of ChrR required for the response to specific inducers, suggesting distinct mechanisms for ChrR inactivation, and (iv) the essential role of cfaS, a widely conserved member of the core σE–ChrR regulon, for bacterial viability during 1O2 stress.

Experimental procedures

Bacterial strains and growth conditions

Caulobacter crescentus strains were grown at 30°C in peptone-yeast extract (PYE) medium (Poindexter, 1964). When appropriate, the growth medium was supplemented with 0.2% d-glucose, 0.2% d-xylose, 3% sucrose, nalidixic acid (20 μg ml−1), kanamycin (10 μg ml−1), tetracycline (2 μg ml−1) or chloramphenicol (1 μg ml−1). Plasmids were propagated in E. coli strain DH5α (Invitrogen) and mobilized into C. crescentus by bacterial conjugation using E. coli strain S17-1 (Simon et al., 1983). Both E. coli strains were grown at 37°C in Luria–Bertani (LB) medium (Miller, 1972), with the appropriate antibiotics: ampicillin (100 μg ml−1), kanamycin (50 μg ml−1), tetracycline (12.5 μg ml−1) or chloramphenicol (30 μg ml−1).

Construction of mutant strains

Single mutant strains for genes rpoE (CC0648), chrR (CC0647), msrA (CC1039), cfaS (CC1427) and hsp20 (CC2258) were obtained by an in-frame deletion in the corresponding gene. In these mutant strains, most of the gene was deleted as follows: rpoE (182 amino acids or 89% of σE protein), chrR (146 amino acids or 69% of ChrR protein), cfaS (370 amino acids or 90% of CfaS protein), msrA (180 amino acids or 92% of MsrA protein) and hsp20 (148 amino acids or 97% of the protein). To create truncated copies of the genes, two fragments flanking the region to be deleted were amplified by PCR in each case and directly cloned into pGEM-T (Promega). A complete list of primers used in this study is in Table S2. The 5′ and 3′ fragments relative to the deletions were isolated from the vectors and simultaneously subcloned into pNPTS138 (Tsai and Alley, 2000). The resulting constructs were transferred to C. crescentus strain NA1000 (Evinger and Agabian, 1977) by conjugation with E. coli S17-1.

Codons for the conserved histidine and cysteine residues of ChrR were replaced for a codon corresponding to alanine and serine, respectively, by overlapping PCR with a pair of complementary primers designed for each substitution. Each part of chrR was amplified separately by PCR using one of each complementary primer set and a primer hybridizing upstream or downstream of chrR. The partially complementary PCR products were used together as templates in a second amplification reaction with the primers hybridizing upstream and downstream of chrR. The amplicons obtained were cloned into pGEM-T and sequenced. The insert was excised from vector and subcloned into pNPTS138. The resulting plasmid was transferred to C. crescentus strain NA1000 by conjugation with E. coli S17-1.

The two-step recombination procedure was performed using the construct into pNPTS138 to replace the wild-type copy of the gene for the corresponding mutated copy in the NA1000 background. A double mutant in msrA and hsp20 was generated by replacing the wild-type copy of msrA for its equivalent truncated copy in the hsp20 single mutant. Mutant strains were isolated by screening colonies by PCR and DNA sequencing.

To obtain the conditional chrR mutant, a chrR gene copy under the control of the inducible xylose promoter PxylX was constructed. For that, the chrR coding region was amplified by PCR and cloned into pGEM-T, obtaining the construct pRL300. This construct was digested and the fragment containing chrR was subcloned into the medium copy number vector pUJ142. The resulting construct (pRL390), which consists in a translational fusion between the first 151 codons of the xylX gene and the complete chrR coding region, was transferred by conjugation to C. crescentus harbouring the truncated copy of chrR cloned into pNPTS138. In this background, the second recombination event was performed in the presence of either 0.2% d-xylose or 0.2% d-glucose. The correct deletion in the chromosomal copy of the chrR gene was verified by PCR using a pair of oligonucleotides hybridizing to the rpoE gene and CC0646 gene in order to avoid amplification from the plasmid-encoded copy of chrR.

To overexpress chrR in C. crescentus cells, the insert present in pRL300 was isolated by digestion and ligated into pJS14, generating pRL380. This construct was introduced into C. crescentus strain NA1000 by conjugation with E. coli S17-1.

Promoter activity assay

β-Galactosidase assays were carried out with cultures of different Caulobacter strains carrying the rpoE–lacZ transcription fusion pRSR (R.C.G. Simão, unpublished). For analysis of the conditional mutant strain SG250 in the presence of either d-glucose or d-xylose, overnight cultures in complex medium PYE containing 0.2% d-xylose were washed three times and then re-suspended in PYE. The cultures were then diluted to an OD600 of 0.1, and either d-glucose or d-xylose was added at a final concentration of 0.2%. To test the effect of ROS or heavy metals on rpoE promoter activity, Caulobacter cells at exponential phase (OD600 of 0.5) were treated for different periods of time (30 min−2 h) with H2O2 (0.05–1.0 mM), paraquat (0.05–0.5 mM), diamide (0.01–5.0 mM), tBOOH (0.01–1.0 mM), MB (0.1–10 μM) in the presence of white light (6 W·m−2, measured with radiometer model IL1700 and detector SED033 – International Light) or with FeSO4, CuSO4, ZnSO4, CoSO4, CdSO4 and NiCl2, all at concentrations ranging from 10 μM to 1 mM. Untreated cultures were used as controls in these analyses. Because MB absorbs light around the 600 nm, the density of cultures treated with this sensitizer was monitored at 500 nm and results were normalized using OD600 from untreated cultures. Caulobacter cells were also irradiated with UV-A at irradiance of 10 W·m−2 for 60 min (total dose of 36 kJ·m−2). For that, 20 ml of the culture was placed in Pyrex plates under a UV-A bulb (Spectroline model ENF-260C) with emission peak at 365 nm. The UV-A irradiation was passed through a filter that blocks the passage of UVB light (BG39, Schott). β-Galactosidase activity was measured as previously described (Miller, 1972). All experiments were performed in duplicates and repeated on three different occasions.

Recombinant protein purification for antiserum production

To obtain a recombinant σE protein, a DNA fragment corresponding to rpoE coding region was amplified by PCR, ligated in pGEM-T and subcloned into pPROEXHTa (Invitrogen). The resulting construct was introduced by transformation into E. coli DH5α, and recombinant σE protein was overproduced in this strain as an N-terminal His6-tagged fusion protein by addition of 1 mM isopropyl-β-d-thiogalactopyranoside to exponential phase cultures. After incubation for 3 h at 37°C, bacterial cells were harvested and lysed by sonication in 10 mM Tris-HCl buffer (pH 7.5) containing 100 mM NaCl, 10 mM imidazole and 1 mM phenylmethanesulphonyl fluoride. The cell lysate was centrifuged (10 000 g for 5 min), and the His6–σE fusion protein was detected in the insoluble pellet as inclusion bodies. His6–σE protein was purified from inclusion bodies using 0.3% Sarkosyl, as previously described (Marshak et al., 1996).

Immunoblot experiments

To analyse the effect of 1O2 on σE levels, cultures of the parental strain NA1000 in the exponential growth phase were treated with 2.5 μM MB in the presence of white light (6 W·m−2) and next kept under stress condition or shifted to the dark. Caulobacter cells were also exposed to UV-A irradiation as described above. After irradiation for 30 min (total dose of 18 kJ·m−2), cells were maintained under such condition or transferred to the dark. In both treatments, aliquots were taken before, during stress and after shifting to the dark. In addition, cultures of NA1000 cells were treated with 100 μM tBOOH or 25 μM CdSO4 for 2 h and aliquots were obtained before and throughout the treatments. Equal amounts of total protein were resolved by SDS-PAGE and transferred to nitrocelulose membrane, according to the method previously described (Towbin et al., 1992). Membranes were incubated overnight at 4°C with anti-σE (1:1000) in 10 mM Tris-HCl, pH 8.0, containing 150 mM NaCl, 0.02% Tween 20, and 0.03% Triton X-100. The blots were developed using anti-rabbit IgG conjugated to alkaline phosphatase (Sigma).

RNA extraction

Total RNA used in quantitative RT-PCR was isolated from strains before and after treatment with 25 μM CdSO4 for 60 min, 100 μM tBOOH for 15 min and 2.5 μM MB and white light (6 W·m−2) for 30 min. NA1000 cells were also exposure to UV-A irradiation as described above for 30 min (total dose of 18 kJ·m−2). For transcriptome analysis, total RNA was extracted from exponential phase NA1000 cells harbouring the empty vector pUJ142 and a strain overexpressing rpoE, both cultured during 30 min in the presence of 0.2% xylose. NA1000 cells exposed to 2.5 μM MB and white light (6 W·m−2) for 30 min were also used for 5′RACE. Cells were harvested by centrifugation at 16 000 g for 1 min and RNA was extracted by the Trizol method, as described by the manufacturer (Invitrogen). RNA samples were treated with 1 U μg−1 DNase I Amplification Grade (Invitrogen), according to manufacturer's instructions and tested for the absence of DNA contamination by PCR. The quality and yield of RNA were evaluated by agarose/formaldehyde gel electrophoresis and A260 analysis respectively.

Quantitative RT-PCR

Reverse transcription for RT-PCR was performed using 5 μg of RNA, 200 U of Superscript II reverse transcriptase (Invitrogen) and 500 ng of random primer, following manufacturer's instructions. Quantitative PCR amplification of the resulting cDNA was performed with Platinum SYBR Green (Applied Biosystems) and gene-specific primers (see Table S2). These primers were designed using the PrimerExpress software (Applied Biosystems). Results were normalized using CC0088 gene as the endogenous control, which was previously used (Alvarez-Martinez et al., 2007) and shown to be constant in the samples analysed. Relative expression levels were calculated using the 2−ΔΔCT method (Livak and Schmittgen, 2001).

Microarray analysis

Microarrays containing 50-mer oligonucleotides for all predicted C. crescentus genes (TIGR annotation) used in the experiments were a kind gift of Dr M.T. Laub (Massachusetts Institute of Technology) (Biondi et al., 2006). Four distinct biological RNA samples of each strain were reverse transcribed and labelled using the SuperScript Plus Indirect cDNA Labelling System (Invitrogen). Briefly, the cDNA was synthesized from total RNA (20 μg) using the enzyme SuperScript III in the presence of modified aminoallyldeoxynucleotide triphosphate and random primer. The resulting amino-modified cDNA was chemically labelled by incorporation of the dyes Alexa Fluor 555 (Cy3) or Alexa Fluor 647 (Cy5). Hybridization and scanning was performed as described previously (Laub et al., 2002; Biondi et al., 2006).

Data analysis and normalization was performed using SpotWhat tool (Koide et al., 2006). The fluorescence intensity of Cy3 and Cy5 for each spot was determined by the equation: mean (foreground) − median (background). Spots containing flags or displaying a mean foreground value lower than the 90% quantile of the background were excluded. Normalization was performed by the LOWESS method (Yang et al., 2002). Normalized results of all replicates for each gene were used for the estimation of a normal distribution using a robust approach (median and median absolute deviation) and the probability of a fold change of 1.5 or more (log2σE++/NA1000 > 0.58 or log2σE++/NA1000 < −0.58) was estimated for genes represented in at least three replicates. Genes showing a probability equal or greater than 80% and showing the decrease or increase in gene expression in the strain overexpressing rpoEE++) in at least three biological replicates were considered as differentially expressed between the two strains.

5′RACE

The RNA 5′ ends of genes of interest were determined using the 3′/5′RACE kit (Roche). For that, the RNA was reverse transcribed using a gene-specific primer (Table S2), purified and poly(dA) tailed at their 3′ ends. The resulting cDNA was amplified by PCR using the forward poly(dT)-anchor primer provided by the kit to anneal at the poly(dA) tail and a second gene-specific primer. The PCR products were used in a second PCR reaction using a primer complementary to the poly(dT)-anchor primer and a distinct gene-specific nested primer. PCR products were ligated into the pGEM-T vector and several distinct clones were sequenced.

Stress sensitivity tests and growth curves

For stress sensitivity tests, overnight cultures were diluted to an OD600 of 0.1 and cells were grown until exponential phase (OD600 ∼ 0.5) before beginning of treatment. To test the effect of 1O2 on the viability of the strains, MB (2.5–5.0 μM) was added and the cultures were irradiated with white light (6 W·m−2) throughout the duration of the experiment. As a control, cultures were exposed to white light in the absence of MB. In both experiments, aliquots were taken before and after 16, 20 and 24 h of MB addition. To test the effect of other stresses, either CdSO4 (25 μM) or tBOOH (0.1–1 mM) was added to the cultures and aliquots were taken before and 1, 2, 4 and 24 h after exposure to stress. To measure cell viability, aliquots were removed, diluted and plated.

Growth curves of strain SG250 in the presence of either d-glucose or d-xylose were obtained by growing the cultures in PYE in the presence of either 0.2% d-xylose or 0.2% d-glucose after removing xylose as described above.

Acknowledgements

We are grateful to Rita C.G. Simão and Michael T. Laub for providing the rpoE–lacZ transcription fusion (pRSR) and the chips used in transcriptome analysis respectively. We also thank Paolo Di Mascio for many helpful discussions and Carlos F. Menck and André P. Schuch for assistance with UV-A irradiation. This study was supported by a grant from Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP). During this work, R.F.L. was supported by a predoctoral fellowship from FAPESP and S.L.G. was partially supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq-Brazil).

Ancillary