![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Laboratoire des Interactions Plantes-Microorganismes (LIPM), UMR 2594-441 CNRS-INRA, BP52627, Castanet-Tolosan F-31320, France
Received 2 February 2007/ Accepted 19 March 2007
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Much of our knowledge regarding the mechanisms of induction of general stress responses derives from studies of the model bacterium Escherichia coli and its close relatives. In these bacteria, entry into stationary phase, as well as a number of different stress conditions, leads to the activation of alternative sigma factors, which redirect the RNA polymerase holoenzyme to hundreds of genes, some of which confer multiple-stress resistance on the cells. These alternative sigma factors are of two different classes. The first class is represented by E. coli 
S (encoded by rpoS), which is considered the master regulator of the general stress response in this bacterium. The S content of the cell is regulated at multiple levels, including transcription, translation, and proteolysis, with different stress conditions affecting different levels of control (for reviews, see references 29 and 52). Members of the S regulon encode proteins with diverse functions, including stress response (for instance, katE, encoding a catalase) (72). rpoS mutants are thus sensitive to multiple stresses and have a decreased capacity for survival in stationary phase (28). A second class of alternate sigma factors is represented by E (encoded by rpoE), a sigma factor of the ECF family (for extracytoplasmic function) (27). The activity of E is negatively controlled by a membrane-bound anti-sigma factor, which binds E and renders it inactive. E is activated by perturbations in the cell envelope, like the accumulation of misfolded proteins in the periplasm. This signal triggers proteolysis of the anti-sigma factor, thus releasing into the cytoplasm an active form of E, which can then interact with the core RNA polymerase (for recent reviews, see references 1, 2, and 57). It now appears that E can also be activated upon entry into stationary phase through a different, incompletely characterized mechanism (14). In addition to its own operon, targets of E include genes encoding proteins with functions in stress response, such as the heat shock sigma factor H, as well as proteins involved in the synthesis, folding, or degradation of outer membrane proteins (for reviews, see references 20 and 56). Inactivation of E in E. coli is lethal, and existing rpoE mutants contain suppressive mutations (10, 16). In other bacteria, like Salmonella enterica serovar Typhimurium, the resistance to multiple stresses, as well as the survival in stationary phase, of rpoE mutants is affected (32, 37, 66).
Two classes of sigma factors with similar functions in stress adaptation and/or stationary-phase survival were also identified in gram-positive bacteria (26, 27, 42). Interestingly, in addition to the regulation of stress responses, both classes of sigma factors can also control virulence genes in certain gram-negative and gram-positive pathogenic bacteria (28, 36, 42, 56).
Rhizobia are soil bacteria existing either in free-living forms or in symbiosis within nodules of legume plants, where they differentiate into nitrogen-fixing bacteroids (23, 41, 48, 65). Both in the soil and in planta, rhizobia are subject to stress. For instance, rhizobia experience oxidative stress during root hair infection and nodule invasion (33, 59, 60, 63), as well as oxygen limitation in the fixation zone of the nodule (64). In the soil, variations of temperature, osmolarity, or pH, as well as nutrient starvation, are the stress conditions most frequently faced by rhizobia (70, 71, 75). Therefore, general stress response must be important for the survival of rhizobia, both in the soil and in planta. However, little is known about this response in rhizobia. In Rhizobium leguminosarum, stationary-phase bacteria were found to be more resistant than exponentially growing bacteria to heat, salt, and oxidative and acidic stresses, suggesting that this kind of response also exists in rhizobia (67). Nevertheless, little is known about the mechanisms of regulation of this response, although the involvement of a quorum-sensing system has been reported in R. leguminosarum (68). Interestingly, most rhizobia belong to the alpha subdivision of proteobacteria, and no S homologue could be found in the complete genomes of alphaproteobacteria, which include those of six rhizobia. This suggests that other regulators could be involved in the general stress response of these bacteria. In contrast, many ECF sigma factors are encoded by the genomes of rhizobia, but little is known about their functions (74).
As a first step toward understanding the rhizobial general stress response, we performed transcriptomic studies of the responses of Sinorhizobium meliloti to heat shock and starvation. Among the genes up-regulated under both stress conditions (referred to as "stress response genes"), we found rpoE2, which encodes a putative ECF sigma factor that is well conserved among alphaproteobacteria. We show that this sigma factor controls its own transcription and is activated by various stress conditions, and we present evidence that the product of the gene cotranscribed with rpoE2 could play the function of anti-sigma factor. We identified 44 genes under the control of RpoE2, including katC and rpoH2, which encode proteins with predictable functions in stress responses (a catalase and a heat shock sigma factor, respectively), as well as rpoE5, encoding another putative ECF sigma factor. Strikingly, more than one-third of the 60 stress response genes identified here are under RpoE2 control, which suggests that RpoE2 is a major global regulator of the general stress response in S. meliloti.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
Strain and plasmid constructions.
All plasmid constructions were performed in E. coli DH5
. DNA sequences of oligonucleotide primers used for PCR amplifications are available in Table S1 (http://bioinfo.genopole-toulouse.prd.fr/annotation/iANT/bacteria/rhime/DOC/Bruand2007/index.html).
For constructing plasmid pCBT104, a DNA fragment containing the smc01505 and smc01506 open reading frames (ORFs) was generated by PCR using OCB529 and OCB530 as primers and genomic DNA of strain Rm1021 as a template. This fragment was then digested with BamHI-XbaI and ligated into BamHI-XbaI-cut pJQ200-mp19. For constructing plasmid pCBT113, the smc01506 ORF in plasmid pCBT104 was disrupted at the BssHII site by insertion of the hph gene, generated by PCR using OCB461 and OCB462 as primers and plasmid pVO205 as a template.
Plasmid pCBT105 was constructed by subcloning the BamHI-XbaI DNA fragment of pCBT104 into the similarly cut pBBR1MCS-5. For constructing plasmid pCBT109, a DNA fragment carrying the smc01505 ORF was generated by PCR using primers OCB529 and OCB531, digested with BamHI and XbaI, and inserted into the similarly cut pBBR1MCS-5.
For constructing pLS4.1, a DNA fragment containing the smc01505 and smc01506 ORFs harboring a 129-bp deletion in smc01505 (bp 19 to 147 from the start of the ORF) was generated in a two-step PCR. In the first step, the region upstream from the deletion was amplified using OCB529 and OCB533 as primers, Rm1021 genomic DNA as a template, and Pfu Ultra (Stratagene) as the DNA polymerase. OCB533 carries sequences –20 nucleotides upstream and +20 nucleotides downstream from the deletion endpoints (i.e., bp –2 to +18 and 148 to 167 of smc01505). In a second step, the PCR fragment generated in the first step was used as a primer, together with OCB530, to amplify by PCR a DNA fragment using the OCB539-OCB534 PCR fragment as a template and GoTaq (Promega) as the DNA polymerase (the OCB539-OCB534 PCR fragment was used as a template because only the downstream part of OCB533 anneals in this region). The resulting PCR product was then ligated into pGEM-T, generating plasmid pLS2.7. Finally, the smc01505
-smc01506 region was subcloned from pLS2.7 into pBBR1MCS-5 as an ApaI-XbaI fragment to give pLS4.1.
Similarly, for constructing the plasmid pLS5.1, a first PCR fragment was generated using OCB535 and OCB533 as primers, Rm1021 genomic DNA as a template, and Pfu Ultra (Stratagene) as the DNA polymerase. This PCR fragment was used as a primer, together with OCB536, to amplify a second PCR product using the OCB539-OCB534 PCR fragment as a template and the GoTaq (Promega) DNA polymerase. The resulting PCR product was then ligated into pGEM-T, generating plasmid pLS3.4. Finally, the entire region was subcloned in pJQ200-mp19 as a BamHI-XbaI fragment, giving plasmid pLS5.1.
For the construction of pLS6.20 and pLS6.32, a DNA fragment carrying the smc01504-smc01505 intergenic region was generated by PCR using OCB529 and OCB533 as primers, Rm1021 genomic DNA as a template, and Pfu Ultra (Promega) as the polymerase and ligated into pCZ750 digested with XbaI and blunted with Pfu Ultra. The orientation of the insert in the resulting clones was screened by PCR and checked by DNA sequencing. pLS6.20 and pLS6.32 correspond to plasmids in which the lacZ gene is transcribed from the promoters of smc01504 and smc01505-smc01506, respectively.
The S. meliloti DNA regions cloned into pCBT104, pCBT109, pLS2.7, pLS3.4, pLS6.20, and pLS6.32 were verified by DNA sequencing.
The plasmids were introduced in S. meliloti by triparental mating using pRK2013 as a helper. The construction of S. meliloti strains is described in Results.
Preparation of samples for microarrays, qRT-PCR, and ß-galactosidase assays.
For microarray and quantitative reverse transcription-PCR (qRT-PCR) analyses of the entry into stationary phase after carbon starvation, Rm1021 cells were grown exponentially at 28°C in 500 ml of VMM containing 2 mM sodium succinate. At an optical density at 600 nm (OD600) of 
0.12 (i.e., 1 h 30 min before the first signs of slow down of the culture), cells (25 ml) were harvested by filtration and immediately frozen in liquid nitrogen (exponential-phase sample). Three hours later (the beginning of the plateau; OD600, 0.25), cells were harvested again (stationary-phase sample). For qRT-PCR analyses of the entry into stationary phase after nitrogen starvation, Rm1021 cells were grown exponentially at 28°C in 500 ml of VMM containing 0.62 mM NH4Cl. At an OD600 of 0.1, cells (25 ml) were harvested by filtration as described above (exponential-phase sample); 6 h later (OD600, 0.22), cells were harvested again (stationary-phase sample).
For microarray and qRT-PCR analyses of heat shock, wild-type (Rm1021 or Rm2011) or rpoE2 mutant (CBT208 or 2011mTn5STM.3.12.B10) cells, carrying or not carrying plasmids, as indicated, were grown exponentially at 28°C in 100 ml VMM. At an OD600 of 0.4, two 25-ml aliquots of the culture were transferred into preheated flasks and incubated at 40°C, while the other half of the culture was kept at 28°C. After 30 min, cells were harvested as described above from each culture. For qRT-PCR analyses of salt stress, Rm1021 cells were grown exponentially at 28°C in 100 ml of VMM. At an OD600 of 0.15, NaCl was added to half of the culture at a final concentration of 250 mM, while the rest of culture was kept untreated. After 30 min at 28°C, cells (25 ml) were harvested from treated and untreated cultures as described above.
For microarrays or qRT-PCR, RNA was prepared from the collected samples as previously described (11), followed by DNase I treatment (QIAGEN clean-up procedure).
For ß-galactosidase assays of heat shock, Rm1021 (wild-type) or CBT208 (rpoE2) cells harboring the plasmid pCZ750, pLS6.20, or pLS6.32 were grown exponentially at 28°C in 80 ml of VMM. At an OD600 of 0.2, two 25-ml aliquots of the culture were transferred into preheated flasks and incubated at 40°C, while the other half of the culture was kept at 28°C. After 1 h, a 50-µl sample of each culture was collected. For ß-galactosidase assays of the exponential- to stationary-phase transition, cells were grown exponentially in the same medium at 28°C, and 50 µl of cells was collected at an OD600 of 0.2 (exponential-phase sample). The culture was then incubated for an additional 24 h at 28°C, and 50 µl of cells was collected at an OD600 of 1.2 (stationary-phase sample). ß-Galactosidase activity was assayed in the collected samples as described previously (44).
Labeling of hybridization probes, microarray hybridizations, and analyses. Cy3- and Cy5-labeled cDNAs were prepared according to the method of DeRisi et al. (18) from 10 to 25 µg of RNA. For each experiment, RNA preparations from three independent cultures were used. Sm6koligo microarrays were purchased from A. Becker (University of Bielefeld, Bielefeld, Germany). They consisted of glass slides carrying mainly 70-mer oligonucleotides representative of the 6,208 predicted ORFs of S. meliloti spotted in triplicate, as well as a number of control spots (39). Hybridizations were performed as described previously (6). Data were acquired on a GenePix 4000 scanner (Axon Instruments), and quantifications of mean signal intensities for each spot were performed using GenePix Pro 3.0 (Axon Instruments). Data analyses were carried out using EMMA 2.2. software (CeBiTec; Bielefeld University, Bielefeld, Germany). Heat shock experiment data were normalized using the median of the signals. For experiments with entry into stationary phase, since many genes were down-regulated in this condition (>1,700) (12) (see Table S2 at http://bioinfo.genopole-toulouse.prd.fr/annotation/iANT/bacteria/rhime/DOC/Bruand2007/index.html), data could not be normalized using a global method. Instead, we first identified by microarrays and qRT-PCR analysis three reference genes (of unknown function), which were expressed at similar levels during exponential and stationary phases of growth (sma2239, smb21413, and smc00817) (data not shown). The raw data were then normalized (using Microsoft Excel) with the mean signal of these three reference genes. For all experiments, M values (log2 experiment/control ratio) and P values (t test) were calculated with EMMA. All data are available in Table S1 (http://bioinfo.genopole-toulouse.prd.fr/annotation/iANT/bacteria/rhime/DOC/Bruand2007/index.html).
Quantitative RT-PCR analyses. Reverse transcription was performed using Superscript II reverse transcriptase (Invitrogen) with random hexamers as primers. RNA samples isolated from at least three independent experiments were tested for each condition. Real-time PCRs were run on a LightCycler system (Roche) using the FastStart DNA MasterPLUS SYBRGreen I kit (Roche) according to the manufacturer's instructions. 16S rRNA was used as a reference for normalization. The sequences of the primers used are available in Table S1 (http://bioinfo.genopole-toulouse.prd.fr/annotation/iANT/bacteria/rhime/DOC/Bruand2007/index.html).
Stress resistance assays. To test resistance to salt stress, bacterial cultures in VMM containing 2 or 10 mM Na2 succinate were grown to exponential or stationary phase and treated with NaCl at a final concentration of 2.5 M, and cell viability was measured by plating cells at 0, 30, 60, and 90 min after the salt shock.
Resistance to heat shock was tested in several ways. Exponential- or stationary-phase cultures were grown at 28°C in rich (TY or LBMC) medium or minimal medium (VMM) and exposed to high temperature (40°C, 45°C, 47°C, or 49°C), and the cell viability was measured by plating cells at 0, 30, 60, and 90 min after the temperature upshift. Alternatively, cultures were grown in VMM at various temperatures, and the growth rates were compared.
Resistance to H2O2 was measured either by a disk inhibition assay (5 µl of 3% H2O2) or by adding up to 4 mM H2O2 to exponential- or stationary-phase cultures grown at 28°C in VMM and measuring cell viability by plating them at 0, 15, 30, and 60 min after exposure to stress.
Resistance to low pH was measured by diluting 100 µl of exponential- or stationary-phase cultures grown at 28°C in VMM in 900 µl of citrate-phosphate buffer, pH 3.5 (0.035 M citric acid, 0.03 M Na2HPO4), and measuring cell viability by plating cells at 0, 30, 60, and 90 min after low-pH exposure.
Plant assays.
For plant assays of symbiotic phenotypes, seeds of Medicago sativa cv. Europe or Medicago truncatula Gaertn. cv. Jemalong A17 were surface sterilized, germinated on agar plates, and allowed to grow on nitrogen-free Fahreus medium in test tubes for 3 days. Ten plants were inoculated with 5 x 104 bacteria/plant of the mutants or the corresponding wild-type strains. The nodulation kinetics and aspect of the plant were followed for 30 days. The whole test was performed at least twice independently on M. sativa.
Microarray data accession numbers. The entire set of microarray data has been deposited in the ArrayLims database (https://www.cebitec.uni-bielefeld.de/groups/brf/software/arraylims/).
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
To identify genes induced in logarithmic phase under stress conditions, we exposed exponentially growing bacteria to high temperature (40°C) for 30 min and compared, using microarrays, their transcription profiles to that of control, unstressed bacteria (28°C). A total of 169 genes were found to be up-regulated (M > 1; P < 0.05) at 40°C in comparison to 28°C, including expected genes, like the chaperone-encoding groESL operons, as well as genes encoding putative heat shock proteins (smb21294, smb21295, smc02577, and smc04040) (see Table S2 at http://bioinfo.genopole-toulouse.prd.fr/annotation/iANT/bacteria/rhime/DOC/Bruand2007/index.html).
To determine the nature of genes induced upon entry into stationary phase, we compared the transcription profiles of bacteria growing exponentially to that of bacteria taken 1.5 h after entry into stationary phase. This kind of experiment had been described previously (12) but was repeated here using a minimal medium containing one-fifth of the original concentration of the carbon source (sodium succinate), allowing the bacteria to enter stationary phase at a much lower density (OD600, 0.25, in comparison to 1.2 in the previous study [12]). This was done to ensure that the only limiting element in the medium was the carbon source and to rule out possible effects linked to high-density cultures. A total of 374 genes were up-regulated (M > 1; P < 0.05) in bacteria entering stationary phase in comparison to exponentially growing bacteria (see Table S2 at http://bioinfo.genopole-toulouse.prd.fr/annotation/iANT/bacteria/rhime/DOC/Bruand2007/index.html).
Sixty genes were induced, both in exponential phase after a heat shock and upon entry into stationary phase (Table 2). The induction of some of these genes under these conditions was confirmed by qRT-PCR (Table 3). The seven genes tested were also induced under two additional conditions, i.e., upon entry into stationary phase following nitrogen starvation, as well as in exponential phase after a treatment with 0.25 M NaCl (Table 3; note that fixN3 was hardly induced under the latter conditions). This confirmed that these genes are part of the general stress response of S. meliloti.
|
|
|
|
1-kb region containing the rpoE2 gene disrupted with the hph gene was cloned into the plasmid pJQ200mp19, which is nonreplicative in S. meliloti and carries the Bacillus subtilis sacB gene, which is toxic in gram-negative bacteria grown in the presence of saccharose (54). The resulting construct (pCBT113) was introduced into S. meliloti by triparental mating, and Hygr colonies were selected. In these cells, the plasmid had been inserted into the chromosome by single-crossover recombination. These cells therefore contained both a wild-type and a mutated copy of rpoE2. Several independent purified colonies were grown in LBMC medium in the absence of antibiotic and then plated on the same medium supplemented with 5% saccharose in order to select bacteria in which the plasmid had been lost by homologous recombination; among the resulting Sacr Gms colonies, two kinds of recombinants were theoretically expected, one Hygs, in which the recombination had regenerated the wild- type situation, and one Hygr, in which the rpoE2 gene was disrupted by the hph gene. If the rpoE2 gene was nonessential, we expected 50% of each type, since the Hygr gene was inserted near the middle of the DNA fragment cloned in pCBT113. If rpoE2 was essential, we expected only Hygs colonies. Of a total of 95 Sacr Gms colonies isolated from three independent initial integration events, 45 (47%) were Hygr, and the presence of only one disrupted copy of rpoE2 was verified by PCR in 11 of these colonies. One of these (CBT208) was kept for further analyses. These observations suggested that the rpoE2 gene is not essential. The presence of two rpoE2::Tn5 mutants in a library of 5,000 transposon mutants constructed by Pobigaylo and collaborators (53) is in agreement with this observation. As rpoE2 was disrupted at the 61st nucleotide of the ORF in our rpoE2::hph mutant, rpoE2 transcription could not be easily measured by qRT-PCR in this strain. We therefore measured only the transcription of the upstream ORF, smc01505, which is likely in an operon with rpoE2. To gain insight into other possible genes controlled by RpoE2, we also quantified the transcripts of the adjacent divergent gene smc01504. Whereas the basal levels of expression of these genes at 28°C were similar in the wild type and the rpoE2 mutant (not shown), the transcription of both smc01505 and smc01504 was no longer further elevated in the rpoE2 mutant upon temperature upshift (Table 5) (the groEL5 gene, used as a control, was induced by heat shock in both wild-type and mutant strains). To check that this result was not specific to the genetic background (Rm1021), we performed a similar experiment in a rpoE2::Tn5 mutant isolated in the Rm2011 strain (53). In this strain, the transposon was inserted further downstream in rpoE2, allowing us to quantify rpoE2 transcripts as well. We could confirm the data obtained as described above for smc01505 and smc01504 in the Rm1021 background, and we additionally verified that rpoE2 was no longer inducible in the mutant (n = 1; data not shown). We further confirmed these observations using the plasmid-borne lacZ transcriptional fusions described above (Table 4), and additionally showed that both smc01504 and smc01505-rpoE2 promoters were no longer induced in the rpoE2 mutant in stationary phase (Table 4). Note, however, that given the high background levels of ß-galactosidase expression in nonstressed bacteria and in the rpoE2 mutant (Table 4), we cannot exclude the possibility that these genes may have two promoters, one recognized by RpoE2 and the other recognized by an unknown sigma factor. We conclude from these data that (i) RpoE2 is required for proper stress induction of the transcription of its own operon, as well as that of the neighboring gene smc01504, and (ii) RpoE2 is therefore activated by these stress conditions.
|
|
|
Interestingly, among the 60 general stress genes identified in the present study, 20 were identified as RpoE2-dependent genes by microarrays, and 2 could be predicted as RpoE2 targets because they are in operons with actual RpoE2 targets (smb21442 and smb20934). Moreover, we found by qRT-PCR analysis that smb20094, located close to an RpoE2 target (smb20092), is also dependent on RpoE2 (Table 5). In contrast, five additional randomly chosen general stress genes showed RpoE2-independent induction (Table 5).
Altogether, these data show that RpoE2 controls a large regulon and is an important regulator of the general stress response, as more than one-third (23/60) of the stress response genes identified in this study are controlled by RpoE2.
Phenotypic analysis of the rpoE2 mutants. RpoE2 appears to respond to several stress conditions by inducing a large regulon, including genes involved in stress response. To exclude any strain-specific effect, the two types of rpoE2 mutants available to us (see above and Table 1) were tested. We could not detect any apparent difference between the rpoE2 mutants and their wild-type counterparts. Both strains exhibited the same doubling time in rich and in minimal media and showed comparable viabilities, both in exponential and stationary phases of growth. We determined the capacities of rpoE2 mutant cells to resist various stresses and found that rpoE2 mutants did not differ from the wild types in their resistance to heat, salt, acidic pH, or H2O2, in both exponential and stationary phases of growth (see Materials and Methods) (data not shown). The capacities of the rpoE2 mutants to establish symbiotic interactions with M. sativa and M. truncatula plants were also tested. No difference could be observed between the mutants and the wild-type strains in the kinetics of nodule formation or in the number and aspect of nodules, or in the final aspect of the plants (data not shown).
The smc01505 gene encodes a negative regulator of the RpoE2 regulon.
To learn the possible function of smc01505, located immediately upstream from and in the same operon as rpoE2, we decided to mutagenize it. In order not to perturb the transcription-translation of rpoE2, we intended to introduce an in-frame deletion into smc01505. For this, we cloned in pJQ200mp19 an 1-kb region surrounding smc01505-rpoE2 and containing a 129-bp in-frame deletion in smc01505 (see Materials and Methods). This plasmid (pLS5.1) was introduced into S. meliloti Rm1021 by triparental mating, and Gmr cells were selected in which the plasmid was integrated in the genome through single-crossover recombination. These cells therefore contained both a wild-type and a deleted copy of smc01505. In a second step, we selected for excision of the plasmid by growing cells in the presence of 5% saccharose. If smc01505 was not essential, we expected to get wild-type and mutant cells at equivalent frequencies, as described above for the disruption of rpoE2, since the deletion is located in the middle of the DNA fragment cloned in pLS5.1. However, PCR analyses performed on 66 Sacr Gms colonies, obtained from two independent events of plasmid integration, showed that all of them contained the wild-type structure, except one in which smc01505 contained the desired deletion. Nevertheless, further analysis of the latter clone revealed that the chromosomal region normally located downstream from smc01505, including the rpoE2 coding sequence, was not present in this strain (not shown). Altogether, these observations indicated that the smc01505 ORF could not be disrupted unless rpoE2 was absent from the strain.
The activities of ECF sigma factors are generally regulated by anti-sigma factors encoded in the same operon as the sigma factor itself (27). We therefore suspected that smc01505 could be the RpoE2 anti-sigma factor and that disruption of smc01505 might lead to permanent activation of RpoE2, which might be toxic for the cells, as previously observed in other instances for ECF sigma factors (31, 61). To test this hypothesis, we first constructed a strain expressing RpoE2 from a multicopy plasmid. For this, we inserted into pBBR1MCS-5 a DNA fragment containing the smc01505-smc01506 region preceded by its own promoter but carrying an in-frame deletion in smc01505. Whereas we could easily introduce in S. meliloti the vector alone or derivatives carrying either smc01505 (pCBT109) or the wild-type smc01505-smc01506 region (pCBT105), the rpoE2-expressing plasmid (pLS4.1) gave rise to tiny, slow-growing colonies. These observations suggested that indeed, RpoE2 could be toxic if overexpressed without simultaneous overexpression of SMc01505.
To test the possibility that smc01505 encodes a protein with anti-RpoE2 activity, we introduced in S. meliloti a plasmid expressing smc01505 under the control of its own promoter sequences (pCBT109) and tested the expression of various genes in this strain using qRT-PCR. smc01505 was strongly expressed at 28°C in this strain, i.e., 80-fold as much as in the wild-type strain containing the empty vector at 28°C (not shown). This high basal level of transcription was not dependent on RpoE2, as it was similar in an rpoE2 mutant strain (not shown). This shows that smc01505 is overexpressed in this strain, and we assume that the multiple copy number of the plasmid associated with the basal promoter activity of smc01505 and/or the presence of active promoter sequences in the plasmid (for instance, the lacZ promoter located next to the cloning site), is responsible for this constitutive high expression of smc01505.
We then looked at the expression of six RpoE2-dependent genes (i.e., rpoE2 itself, smc01504, katC, rpoH2, rpoE5, and smc00885) in the strain carrying pCBT109 in comparison to the control strain carrying the empty vector. Whereas the expression levels of these genes at 28°C were similar in both strains (not shown), all of them were induced at 40°C at a much lower level (and were even sometimes no longer induced) in the smc01505-overexpressing strain in comparison to the control strain (Table 7) (the groEL5 gene used as a control was still fully inducible in both strains). Therefore, this confirmed our hypothesis that the smc01505 product is a negative regulator of RpoE2 activity and suggested that this polypeptide could have an anti-sigma factor function.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Among the 60 stress-induced genes, we characterized in more detail rpoE2 (smc01506), encoding a putative ECF sigma factor. We showed that stress induction of the rpoE2 gene is dependent on RpoE2 itself, which indicates that this sigma factor is activated by stress. We observed that several different stress conditions could activate RpoE2: heat shock and salt shock, as well as entry into stationary phase after carbon or nitrogen starvation. Using a combination of transcriptomics, bioinformatics, and qRT-PCR, we identified 44 genes under the control of RpoE2 and defined putative promoter sequences recognized by this sigma factor. Strikingly, all these genes were also induced by salt and/or osmotic stress in a recent transcriptomic study by Domínguez-Ferreras et al. (19), which confirms that RpoE2 is activated by salt stress. Recently, Bobik et al. (9) showed that 74% (98/132) of the S. meliloti genes induced by oxygen limitation are under the control of the FixLJ two-component regulator (see Table S2 in reference 9). None of these genes was found to be dependent on RpoE2. In contrast, 16 out of the remaining 34 genes induced by oxygen limitation independently of FixJ were induced by heat shock in an RpoE2-dependent manner in the present study. We therefore conclude that RpoE2 likely controls the expression of these genes under micro-oxic conditions. RpoE2 would thus be the second most important regulator of gene expression under microaerobic conditions in S. meliloti. Finally, more than one-third of the general stress genes identified in the present study proved to be RpoE2 targets. Altogether, these data indicate that RpoE2 is a major regulator of the general stress response in S. meliloti.
Although most (>65%) of the RpoE2-regulated genes encode proteins of unknown function, it is striking that eight of them encode putative transcription regulators: three sigma factors (rpoE2 itself, rpoE5, and rpoH2) and five members of two-component systems (exsF-exsG, sma0113-sma0114, and smc01504). Whether some of these regulators are responsible for indirect control of RpoE2-dependent genes is presently not known. Interestingly, several RpoE2 targets are known or supposed to be related to stress. Some of them had previously been functionally shown to be up-regulated under various stress conditions and in stationary phase: katC, encoding a catalase homologous (48% identity) to the RpoS-regulated E. coli catalase KatE (63), and rpoH2, encoding a sigma factor of the heat shock family (49, 50). smc01504 encodes a putative protein displaying significant similarity (52% identity) to the recently described PhyR regulator of Methylobacterium extorquens, involved in stress response and phyllosphere colonization (25). An additional RpoE2-dependent gene encodes another ECF sigma factor (rpoE5). In addition to katC, some other genes share homologies with genes that are controlled by the master regulator of the general stress response, RpoS, in other gram-negative bacteria: smc00371, whose product is 40% identical to the E. coli YciF protein of unknown function (30), as well as the glgA2 and glgX2 genes, possibly involved in glycogen synthesis. However, the functions of these proteins in stress response are not known. Interestingly, 55% of the RpoE2 targets map on pSymB. Recently, Domínguez-Ferreras et al. have highlighted the important number of pSymB genes that are upregulated in response to an increase in external osmolarity (19). Our data confirm the importance of this plasmid in S. meliloti stress response.
Despite the fact that the RpoE2 regulon is induced by stress conditions and contains several stress response genes, no deficiency could be associated with rpoE2 mutations, either in vitro or in planta. Although we cannot exclude the possibility that the RpoE2 regulon is involved in resistance to a limited number of unknown stresses, as observed for some other ECF sigma factors in other bacteria (for instance, sigF in Caulobacter crescentus [3]), the large number of genes regulated by RpoE2, as well as the presence of genes clearly associated with stress responses, makes us favor two other hypotheses. First, we cannot formally exclude the existence of secondary mutations compensating for the absence of rpoE2 in the mutant strains tested. Second, the lack of phenotype of rpoE2 mutants could be due to the redundancy of the S. meliloti genome (24). Thus, for instance, RpoE2 controls the transcription of katC and rpoH2, but the S. meliloti genome encodes two additional enzymes with catalase activities (KatA and KatB) and another heat shock sigma factor (RpoH1). Although no clear symbiotic phenotype could be associated with the lack of katC (63) or rpoH2 (49, 50), the katA katC, katB katC, and rpoH1 rpoH2 double mutants were clearly affected in their symbiotic phenotypes (7, 33, 50, 63), showing that other genes can compensate for the absence of some RpoE2 targets. Similarly, we can hypothesize that other regulators, controlling the same genes as RpoE2, can compensate for its absence. Although the heat induction of most genes tested in the present study was completely dependent on RpoE2, one gene (smb21456) was still significantly induced by heat in rpoE2 mutant cells, supporting the possibility that it is controlled by another regulator in response to heat. If this gene, or other still unknown similarly regulated genes, is involved in stress resistance, it could explain the absence of phenotype of rpoE2 mutants. Interestingly, the genome of S. meliloti contains nine genes encoding putative ECF sigma factors, in addition to rpoE2, and we cannot exclude the possibility that some of them complement the absence of RpoE2, as observed in other systems (for instance, the large overlaps between the regulons of B. subtilis ECF sigma factors) (27). One of them (rpoE5) was found to be under the control of RpoE2 and is therefore unlikely to do so. We are presently testing whether other S. meliloti ECF sigma factors are able to complement the lack of RpoE2.
The heat shock sigma factor RpoH1 has been previously implicated in regulation of gene transcription after heat shock in S. meliloti and could possibly be active in stationary phase (49). Nevertheless, we could not find its consensus binding sequence (CTTNAAN17CCANNT [46]) upstream from any of the 60 general stress response genes identified here, although we found it upstream from two genes specifically up-regulated by heat shock (clpB and ibpA). This suggests that RpoH1 does not play a major regulatory role in the S. meliloti general stress response.
We have identified smc01505, which is cotranscribed with rpoE2, as encoding a possible anti-RpoE2 sigma factor, although this small protein of 55 amino acids does not show any similarity to known anti-sigma factors. As frequently observed for anti-sigma factors, this protein is encoded in the same operon as the sigma factor. However, its gene is located upstream from rpoE2 instead of downstream, as is usually seen in other systems. The frequent coexpression of the sigma factor with its anti-sigma factor can be interpreted as a need for tight regulation of sigma factor activity. Accordingly, we found that unbalanced expression of either one of these two proteins leads to deregulation of the response: (i) inactivation of smc01505 or overproduction of smc01506 both had a toxic effect, presumably by excessive activation of RpoE2, and (ii) overexpression of the smc01505 gene product hindered activation of RpoE2 by stress conditions. Similar effects were previously described for other sigma-anti-sigma pairs: overexpression of E. coli RpoE or its Pseudomonas aeruginosa homologue AlgU was found to be toxic to E. coli cells (55, 61), and overexpression of the corresponding anti-sigma factors was found to inhibit RpoE or AlgU activity (45, 62). Interestingly, the SMc01505 peptide does not carry any signal sequence or transmembrane domain, which is unusual for anti-ECF sigma factors. If this protein is indeed the anti-RpoE2, this would indicate that this sigma factor is not activated in response to periplasmic signals, as generally believed for this type of regulator, but rather by cytoplasmic signals. The Streptomyces coelicolor ECF R is thus regulated by a soluble anti-sigma factor that responds to redox change (34, 51). Although the molecular mechanisms leading to RpoE2 activation are still unknown, we can anticipate that at least two signal transduction pathways exist, since RpoE2 is activated by both physical/chemical (heat and salt) and nutrient stresses, which are usually signaled in different ways (for example, the recent study of E. coli E [14]).
The best RpoE2 homologues were found in almost all free-living alphaproteobacteria, and a striking synteny of the region was observed (Table 8). Thus, in bacteria phylogenetically close to S. meliloti, a homologue of smc01505 was systematically found upstream of the sigma factor-encoding ORF, reinforcing our hypothesis that coexpression of these two proteins is important (Table 8) (note that in Brucella species, this ORF had not been previously detected/annotated in the published genome sequences). Also, an ORF encoding a response regulator-like protein similar to SMc01504 was systematically found next to the smc01505-like genes. Examination of the intergenic regions between smc01505- and smc01504-like genes revealed in every case the presence of the putative –35 and –10 promoter sequences defined in the present study, which further supports the hypothesis that they are binding sites for these sigma factors (not shown). Phylogenetic analysis revealed that the tree of RpoE2-like sigma factors is congruent to the tree of alphaproteobacteria species, which indicates that this class of sigma factors is of ancient origin (not shown). Although little is known about these sigma factors, the R. leguminosarum rpoE2 homologue (called rpoZ) had been inactivated in a previous study and was suggested to be autoregulated, although the inducing signal was not known (73). More recently, the homologue of rpoE2 in Brucella (called rpoE1 in that case) was disrupted, and the resulting mutant showed attenuated virulence in mice (17).
|
Interestingly, some members of the RpoE2 regulon have been observed to be up-regulated during the process of infection of M. sativa by S. meliloti. Thus, using a transcriptional fusion to lacZ, Jamet et al. observed preferential expression of the katC gene in infection threads, as well as in the infection zone of the nodule (33). With a transcriptional fusion to gus, Oke et al. showed expression of rpoH2 at the apex of the nodule, which could correspond to the infection zone (49). Using a transcriptomic approach, we previously showed that 37 of the 44 RpoE2-dependent genes identified here are preferentially expressed in bacteria isolated from infection threads (12). Although it has still to be proven that expression of these genes during infection depends on RpoE2, we can speculate that the many stress conditions potentially encountered by the bacteria in infection threads are signals for activation of RpoE2 and therefore up-regulation of its regulon. However, we were unable to detect any symbiotic defect of the rpoE2 mutant strain(s).
In summary, we have identified RpoE2 as an important regulator of the general stress response in S. meliloti. However, the lack of any phenotype of RpoE2 mutants suggests that other, unidentified stress responses are activated in these strains. These are presently being investigated by looking for regulators of the RpoE2-independent general stress genes identified here.
|
ACKNOWLEDGMENTS |
|---|
This work was supported in part by the Département Santé des Plantes et Environnement of INRA.
|
FOOTNOTES |
|---|
Published ahead of print on 30 March 2007.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
E in Escherichia coli. Curr. Opin. Microbiol. 7:157-162.[CrossRef][Medline]E downregulates extracytoplasmic stress responses and outer membrane vesicle production in Escherichia coli. J. Bacteriol. 189:1523-1530.E, by guanosine 3',5'-bispyrophosphate (ppGpp). J. Bacteriol. 188:4627-4634.E is an essential sigma factor in Escherichia coli. J. Bacteriol. 179:6862-6864.E subunit of Escherichia coli RNA polymerase: a second alternate sigma factor involved in high-temperature gene expression. Genes Dev. 3:1462-1471.
