Abstract
In Azotobacter vinelandii, a cyst-forming bacterium, the alternative sigma factor RpoS is essential to the formation of cysts resistant to desiccation and to synthesis of the cyst-specific lipids, alkylresorcinols. In this study, we carried out a proteome analysis of vegetative cells and cysts of A. vinelandii strain AEIV and its rpoS mutant derivative AErpoS. This analysis allowed us to identify a small heat-shock protein, Hsp20, as one of the most abundant proteins of cysts regulated by RpoS. Inactivation of hsp20 did not affect the synthesis of alkylresorcinols or the formation of cysts with WT morphology; however, the cysts formed by the hsp20 mutant strain were unable to resist desiccation. We also demonstrated that expression of hsp20 from an RpoS-independent promoter in the AErpoS mutant strain is not enough to restore the phenotype of resistance to desiccation. These results indicate that Hsp20 is essential for the resistance to desiccation of A. vinelandii cysts, probably by preventing the aggregation of proteins caused by the lack of water. To our knowledge, this is the first report of a small heat-shock protein that is essential for desiccation resistance in bacteria.
-
One supplementary figure and two supplementary tables are available with the online version of this paper.
-
Edited by: W. Achouak
Introduction
Azotobacter vinelandii is a Gram-negative bacterium of the family Pseudomonadaceae. Under adverse conditions, this bacterium undergoes differentiation to form cysts resistant to desiccation. A mature cyst consists of a contracted cell known as the central body, which is surrounded by a capsule consisting of a laminated outer layer called the exine and an inner layer called the intine, both lacking a lipid membrane. Both layers are composed of lipoproteins and alginate, an exopolysaccharide comprising two types of monomer: mannuronic acid and its epimer, guluronic acid. Guluronic acid residues originate from a polymer-level epimerization process catalysed by a family of seven alginate epimerases, AlgE1–AlgE7 (Høidal et al., 2000). Alginate is essential for the encystment process. Mutations that impair alginate synthesis also impair the formation of cysts (Campos et al., 1996; Mejía-Ruíz et al., 1997; Castañeda et al., 2001). Alginate monomer composition is also important for cyst formation, as the AlgE1–AlgE7 epimerases are, in combination, essential for the formation of mature cysts (Steigedal et al., 2008).
Upon induction of encystment, a family of phenolic lipids known as alkylresorcinols is synthesized, replacing the phospholipids of the cyst membrane and forming part of the exine layer (Reusch & Sadoff, 1983; Segura et al., 2009). Strains unable to produce alkylresorcinols, due to the inactivation of alkylresorcinol biosynthetic genes, produce cysts with an altered exine morphology that resists desiccation, implying that these phenolic lipids play a structural role in the exine layer, but are not essential for desiccation resistance (Segura et al., 2009).
Little is known about the regulation of gene expression during the encystment process. Mutations in the genes encoding the two-component system GacS/GacA impair alginate synthesis and, therefore, the formation of mature cysts (Castañeda et al., 2001). This system regulates alginate biosynthesis by post-transcriptional control of the alginate biosynthetic gene algD, through the Rsm system. GacA activates transcription of seven small RNAs that counteract the activity of RsmA, a protein, that binds the algD mRNA to inhibit its translation (Manzo et al., 2011).
In a previous work, we reported that inactivation of rpoS, which codes for the alternative sigma factor RpoS, resulted in the inability to form cysts resistant to desiccation and to produce the cyst-specific alkylresorcinols (Cocotl-Yañez et al., 2011). RpoS was shown to regulate the expression of ArpR, a LysR-type transcriptional regulator expressed only during encystment, which positively regulates transcription of the alkylresorcinol biosynthetic genes arsABCD. This activation is dependent on acetoacetyl-CoA, which could be a metabolic signal for encystment (Romero et al., 2013). Although inactivation of rpoS did not affect alginate synthesis, the cysts produced by an rpoS mutant are unable to form the intine and exine layers of the cyst capsule (Cocotl-Yañez et al., 2011). Therefore, RpoS controls unidentified genes that are essential for mature cyst formation. In the present study, we report that RpoS regulated the expression of a small heat-shock protein essential for the resistance to desiccation of A. vinelandii cysts.
Methods
Microbiological procedures.
Bacterial strains and plasmids used are listed in Table 1. Medium and growth conditions were as follows: A. vinelandii was grown at 30 °C in Burk's fixed nitrogen-free salts medium (Kennedy et al., 1986), supplemented with 2 % sucrose (BS) for vegetative growth, or 0.2 % n-butanol (BB) or 0.2 % β-hydroxybutyrate (BBHB) for encystment induction. Escherichia coli strain DH5α was grown on Luria–Bertani medium at 37 °C. Antibiotic concentrations used for A. vinelandii and E. coli were (μg ml−1): ampicillin, 0 and 100; nalidixic acid, 30 and 0; spectinomycin, 50 and 50; kanamycin, 1 and 30; and gentamicin, 1 and 10, respectively. Transformation and conjugation of A. vinelandii were carried out as described previously (Page & von Tigerstrom, 1978; Bali et al., 1992). The staining of alkylresorcinols was carried out as described previously (Segura et al., 2003). β-Glucuronidase activity was measured as reported by Romero et al. (2013).
Nucleic acid procedures.
DNA manipulations were performed according to standard protocols (Sambrook et al., 1989). Chromosomal DNA, used as the template for PCRs, was obtained from A. vinelandii AEIV WT strain. DNA sequencing was done with a Perkin Elmer/Applied Biosystems DNA Sequencer. The sequences of the primers used in this work are shown in Table S1 (available in the online Supplementary Material).
Proteomics methodology.
For the proteomic analysis, cells were collected at 30 h in BS medium (stationary phase) and 120 h in BBHB medium. Bacterial cell proteins were obtained by sonication at 24 kHz 1 min ON/1 min OFF for five cycles at 4 °C in a Vibra Cell (Sonics) in the presence of a protease inhibitor (Complete tablets; Roche Diagnostics). To further limit proteolysis, protein isolation was performed using phenol extraction (Hurkman & Tanaka, 1986). To solubilize and obtain completely denatured and reduced proteins, pellets were dried and resuspended as reported previously (Encarnación et al., 2003). Prior to electrophoresis, samples were mixed with 7 M urea, 2 M thiourea, 4 % CHAPS, 2 mM tributylphosphine, 2 % ampholytes and 60 mM DTT. Sample preparation, 2D PAGE and image analysis were performed as described previously (Encarnación et al., 2003). 2D PAGE was performed with the Investigator System (Genomic Solutions). Ampholines at pH 3–10 were used for the first dimension. The separation in the second dimension was performed in 12 % polyacrylamide gels. For the first dimension, ~500 mg total protein was loaded. Gels were stained with Coomassie Blue G-250. Protein spots on all gels were detected at 127×127 mm resolution using a PDI image analysis system and PD-Quest software (Protein Databases). We were interested in spots that showed a twofold change at least and met the conditions of a statistical Student's test (significance level 95 %). The experiments were performed three times. Selected spots from Coomassie-stained 2D gels were excised, reduced, alkylated, digested and transferred automatically to a MALDI-TOF Bruker Daltonics Autoflex (Bruker Daltonics Bellerica) analysis target by a Proteineer SP II and SP robot using the SPcontrol 3.1.48.0 v software (Bruker Daltonics). Mass spectra were obtained using a Bruker Daltonics Autoflex operated in the delayed extraction and reflectron mode. Spectra were calibrated externally using a peptide calibration standard (Bruker Daltonics 206095). Peak lists of the tryptic peptide masses were generated using FlexAnalysis1.2 v. SD1Patch2 (Bruker Daltonics). The search engine Mascot Server 2.0 was used to compare fingerprints against a A. vinelandii National Center for Biotechnology Information non-redundant data set, with the following parameters: one missed cleavage allowed, carbamidomethyl cysteine as the fixed modification and oxidation of methionine as the variable modification. We accepted those proteins with scores >50 and P<0.05.
Construction of hsp20 mutant strain.
To generate the hsp20 mutant strain (Fig. S1), primers upHsp20/lwHsp20 were used to amplify a PCR fragment of 1 kb, containing the complete hsp20 gene. This fragment was ligated into the pJET1.2/blunt vector (ThermoScientific) to produce pSMhsp20. A SphI digest of pSMhsp20, made blunt with Klenow enzyme (ThermoScientific), was ligated to a SmaI fragment containing the gentamicin cassette excised from vector pBSL98 (Alexeyev et al., 1995), resulting in pSMhsp20 : : Gm, that is unable to replicate in A. vinelandii. The AEIV strain was transformed with this plasmid and a gentamicin-resistant transformant (AEhsp20) was isolated, and confirmed to carry the hsp20 : : Gm mutation and the absence of the WT hsp20 gene by PCR using upHsp20/lwHsp20 primers.
Complementation of hsp20 and rpoS mutants with hsp20.
A 591 bp fragment, corresponding to the hsp20 gene lacking the first and last codons, was amplified by PCR using Phusion High-Fidelity DNA Polymerase (ThermoScientific) and Hsp20Bam/Hsp20Hind primers. The resultant PCR product was digested with BamHI and HindIII, and ligated into pET-21a (Novagen). E. coli DH5α was transformed with this plasmid and a transformant carrying pMHsp20His was isolated. This plasmid was digested with ScaI and BglII. The 1.8 kb fragment containing the hsp20 gene with a tag of six histidines at the C terminal was cloned into pJET1.2/blunt, resulting in pJetHsp20. A ScaI digest of pJetHsp20, made blunt with Klenow enzyme, was ligated to a SmaI fragment containing the kanamycin cassette excised from vector pBSL97, resulting in pMC625, which is unable to replicate in A. vinelandii. The AEhsp20 strain was transformed with pMC625 for co-integration into the chromosome, producing strain AEhsp625 (Fig. S1).
A promoterless hsp20 gene, flanked by HindIII and BamHI restriction sites, and with a tag of six histidines at the N terminus, was amplified by PCR using AEIV chromosomal DNA and primer HisAmhindhsp20, whose sequence includes the His-tag, and primer HisAmbamhsp20. The product was digested and cloned into pBBR1MCS-2, and digested with the HindIII and BamHI enzymes. This produced pBMCA6, with hsp20 under the control of the kanamycin promoter (Fig. S2). pBMCA6, which is able to replicate in A. vinelandii, was transferred by conjugation into strains AEhsp20 and AErpoS to produce strains AEhsp20/pBMCA6 and AErpoS/pBMCA6, respectively.
Quantitative reverse transcription-PCR (qRT-PCR).
Expression of hsp20 was measured by qRT-PCR as reported previously (Noguez et al., 2008). For RNA extraction, the cultures were grown in BS or BB medium. Cells were collected at 30 h for BS and 120 h for BB. The primers used for the qRT-PCR assays were: upRT-hsp20/dwRT-hsp20 for hsp20 expression and fw-gyrA/rev-gyrA for gyrA expression. Amplification conditions were 10 min at 95 °C, a two-step cycle at 95 °C for 15 s and 60 °C for 60 s, for a total of 45 cycles. The size of all amplimers was 100 bp. gyrA expression was used as internal control to normalize the results. All assays were performed in triplicate. The data are presented as fold change (mean±sd) of mRNA levels of the mutant strain relative to those of the WT.
Primer extension analysis assays.
A 572 bp fragment, corresponding to the promoter region of hsp20 (nt −326 to +246), was amplified by PCR using primers upHsp20rr/dwHsp20rr. The product was cloned into pJET1.2/blunt, resulting in pMChsp20. Total RNA was isolated from AEIV cultures grown for 30 h in BS or 120 h in BB liquid medium. Primer extension experiments were carried out at 42 °C using 50 µg RNA and avian myeloblastosis virus reverse transcriptase (Roche), with primer primhsp20. The cDNAs were end-labelled with [γ-32P]dATP using polynucleotide kinase (Roche). The sequencing ladders were generated with the same primers using a Thermo Sequenase Cycle Sequencing kit (USB) and pMChsp20 as template.
Construction of hsp20 : : gusA translational fusion.
Primers MelAF/MelAR and Pfu DNA polymerase (ThermoScientific) were used to amplify a PCR fragment of 1.1 kb containing the melA gene. This fragment was cloned into pUC19 vector, which had been restricted previously with EcoRI and HindIII, and blunted in order to remove its multiple cloning site, resulting in pUMA. A XhoI digest of pUMA was ligated to a XhoI fragment containing the tetracycline cassette obtained from vector pBSL190 (Alexeyev et al., 1995). The resulting plasmid was called pUMATc. pUMATc was digested with PstI and HindIII to clone the gusA reporter taken from pAHFUTd-Tc (Hernandez-Eligio et al., 2012), resulting in pUMATcgusAPT.
To generate an hsp20 : : gusA translational fusion, primers uphspXbafus/dwhspPtsI were used to amplify a 247 bp DNA fragment containing the promoter, Shine–Dalgarno sequence and three codons of hsp20. This PCR fragment was digested with XbaI and PstI, and cloned into pUMAgusAPT, digested with the same enzymes, to obtain pUMAhspPT.
AEIV WT strain was transformed with pUMAhsp20PT. A tetracycline-resistant transformant was selected, resulting in strain AEhsp20PT. The integration of the hsp20 : : gusA translational fusion by gene replacement into the chromosome of AEIV was confirmed by PCR analysis using uphspXbafus/gusArev primers. The absence of the WT melA gene was confirmed by PCR analysis using melAFw/melARv primers.
Resistance to desiccation, and electron microscopy assays.
Transfer of washed vegetative cells from 30 h liquid cultures to BS to BB medium plates induced encystment. Desiccation resistance assays were carried out as described previously (Campos et al., 1996; Segura et al., 2009). Approximately 106 c.f.u. of each strain were collected after 5 days in the induction medium and were desiccated at 30 °C on 0.2 mm membranes for 5 days. Surviving cells, quantified by viable cell count, were considered mature cysts resistant to desiccation. Electron microscopy was carried out as reported previously (Mejía-Ruíz et al., 1997).
Results
Proteome analysis of the WT strain AEIV and its rpoS mutant derivative AErpoS
In order to identify genes under RpoS control involved in encystment, we carried out a search for proteins of the WT strain that were upregulated during encystment, as compared to vegetative cells. We also identified which of these proteins were not upregulated in the rpoS mutant during encystment. Proteome expression profiles were carried out (as described in Methods) in the WT strain AEIV and its rpoS mutant derivative AErpoS, under vegetative and encystment conditions (Fig. 1). Visually, we detected changes in the expression of over 30 proteins in the WT encysted cells, as compared with vegetative cells. These protein spots were excised from the gel and analysed by MALDI-TOF. This analysis resulted in the identification of the 20 proteins listed in Table S2. Nine of these proteins were upregulated during encystment in the WT, but not in the AErpoS strain (Fig. 1).
Results of 2D PAGE from the WT AEIV and rpoS mutant AErpoS vegetative and encysted cells.
These proteins, as annotated in the genome of A. vinelandii DJ (Setubal et al., 2009), are (Table 2): a small heat-shock protein Hsp20, which was among the most abundant proteins in cyst cells of the WT strain; the alginate epimerases AlgE1 and AlgE6, which have been shown previously to be involved in encystment (Steigedal et al., 2008); a calcium-binding protein encoded within an alginate epimerases gene cluster; two proteins involved in polysaccharide export and biosynthesis; an haemerythrin HHE cation-binding protein, which was detected with two different molecular masses and pIs, suggesting that it undergoes a post-translational modification; and a hypothetical protein. Due to its abundance in cysts, Hsp20 was chosen for further analysis, to investigate its regulation by RpoS and its role in the encystment process. The gene encoding Hsp20 is located downstream of and in the opposite direction to Avin25760, which codes for a putative dihydroorotate oxidase, and upstream of and in the opposite direction to Avin25780 (GroL). The Hsp20 amino acid sequence contains the typical α-crystalline domain present in small heat-shock proteins (Arrigo & Landry, 1994) and shares 29 % identity with Hsp16.5 from Methanococcus jannaschii, a protein whose crystal structure has been determined (Kim et al., 1998). This identity is shared between the α-crystalline region, but not in the amino or C-terminal regions.
Hsp20 was transcribed from an RpoS-dependent promoter
As shown in Fig. 1, the level of the Hsp20 protein is very high under encystment conditions and is reduced significantly in vegetative cells or in the rpoS mutant. To confirm the role of RpoS in the regulation of hsp20 expression, qPCR was used to determine the effect of the rpoS : : Sp mutation on the expression of hsp20, measured under vegetative (BS) and encystment conditions (BB). The qPCR assays showed that inactivation of rpoS diminishes the expression of hsp20 under both conditions (0.0084±0.0035 in BS and 0.02±0.02 in BB), as compared with the WT (1.0).
We also determined the transcription start site for the hsp20 gene. Primer extension analysis was carried out with total RNA isolated from the WT and the rpoS mutant grown on encysting and vegetative media in stationary-phase cultures. A single transcriptional start site was identified, located 120 nt upstream of the hsp20 ATG start codon (Fig. 2). No transcript was detected in the rpoS mutant, suggesting that this promoter is recognized by RpoS. In fact, the −10 region of the Phsp20 promoter has the CTATCCT sequence, which corresponds to the RpoS-dependent promoter sequence reported for phbB and algD genes in A. vinelandii (Castañeda et al., 2001; Peralta-Gil et al., 2002). The primer extension analysis (Fig. 2) suggested that the level of the hsp20 transcript is higher under vegetative than encystment conditions. This result was unexpected, as the proteome analysis showed that the level of the Hsp20 protein was higher in cysts than in vegetative cells. We therefore measured and compared the level of hsp20 transcripts by qPCR, under encystment versus vegetative conditions. In agreement with the results observed in the primer extension analysis, the relative expression level of the hsp20 mRNA was lower under encysting (0.42±0.23) than under vegetative conditions (1.0).
(a) Identification of the transcription initiation start site of hsp20 by primer extension analysis using total RNA isolated from strain AEIV under encystment (lane 1) and vegetative conditions (lane 3), and from strain AErpoS under encystment (lane 2) and vegetative conditions (lane 4). (b) The hsp20 promoter (Phsp20) is indicated, and the transcriptional start site (+1), the −10 box and the hsp20 ATG translation start codon are shown in bold and underlined. The predicted RsmA-binding site is shown in bold italic.
Post-transcriptional regulation of Hsp20
The proteome analysis, the primer extension analysis of hsp20, and the qPCR performed under vegetative and encystment conditions suggested that hsp20 is transcribed mainly in vegetative cells, but its translation is favoured under encystment conditions. To test this hypothesis, strain AEhsp20PT, carrying a translational hsp20 : : gusA gene fusion integrated into the chromosome, was constructed. The induction kinetics of hsp20 translation were determined in vivo by measuring the β-glucuronidase activity in strain AEhsp20PT, under vegetative and encysting conditions. As shown in Fig. 3, although Hsp20 translation occurred under both conditions, the maximum translation was observed in cells undergoing encystment. These data indicated a post-transcriptional regulation of hsp20.
Translation of hsp20 during growth in liquid BS or BB media, measured as β-glucuronidase activity of an hsp20 : : gusA translational fusion in the WT strain AEIV. One unit β-glucuronidase activity corresponds to 1 nmol substrate (X-Gluc) hydrolysed min−1 (mg protein)−1. Means±sd are shown.
Inactivation of hsp20 impaired cyst resistance to desiccation
In order to determine the role of Hsp20 protein in the encystment process, strain AEhsp20, an AEIV derivative carrying a hsp20 : : Gm mutation, was constructed. The growth and alginate production of this strain in BS medium was similar to the WT AEIV strain (data not shown). The AEhsp20, AErpoS and WT AEIV strains were induced for encystment by transferring cells grown in liquid BS medium to BB plates. Similar to the rpoS mutant, strain AEhsp20 was unable to produce cysts resistant to desiccation (Table 3).
Electron microscopy of the cysts made by the hsp20 mutant was carried out and compared to cysts made by the WT and the rpoS mutant. As shown in Fig. 4(a), the cysts formed by the hsp20 mutant were indistinguishable from those formed by the WT. Similar to the WT strain, cysts of the hsp20 mutant were composed of a compacted cell surrounded by the intine and exine layers. To confirm that the inability to resist desiccation of the cysts formed by the hsp20 mutant strain was caused by the absence of the Hsp20 protein, we carried out the complementation of the AEhsp20 strain. pMC625, carrying a WT hsp20 copy tagged with a tail of six histidines in the C terminus, was co-integrated into the chromosome of AEhsp20, producing strain AEhsp625 (Fig. S1). Integration of the plasmid was confirmed by PCR and Western blot analysis (data not shown). The AEhsp625 strain produced cysts resistant to desiccation (Table 3). These results confirmed that hsp20 is essential for the survival of mature cysts under desiccation.
Effect of hsp20 mutation on cyst formation and alkylresorcinol synthesis. (a) Electronic micrographs of the cysts formed by strains AEIV, AEhsp20 and AErpoS. Bar, 0.5 µm. E, Exine; I, intine. (b) Alkylresorcinol staining of the WT strain AEIV, the hsp20 mutant AEhsp20 and the rpoS mutant AErpoS. The cells were induced to encyst on BB medium for 5 days before staining.
Synthesis of cyst-specific lipids alkylresorcinols was not affected in the hsp20 mutant strain
Inactivation of alkylresorcinol biosynthetic genes results in strains that produce cysts with an altered exine morphology, but that are able to resist desiccation (Segura et al., 2009). As shown in Fig. 4(a), the AEhsp20 strain produced cysts with WT exine morphology, suggesting that the synthesis of alkylresorcinols is not affected by the hsp20 mutation. Indeed, when cells of AEhsp20 induced for encystment were stained with Fast Blue to visualize alkylresorcinol production, we found that the hsp20 mutant strain developed a red colour similar to that of the WT strain, indicating the presence of alkylresorcinols, while the rpoS mutant strain (used as a negative control) remained white (Fig. 4b).
Expression of hsp20 from an RpoS-independent promoter did not restore resistance to desiccation to the rpoS mutant
The question was raised of whether, in addition to ArpR (the activator of alkylresorcinol synthesis) and Hsp20, other proteins under the control of RpoS are required for the formation of the cyst capsule (mature cysts) and desiccation resistance. We therefore tested whether expression of the hsp20 gene, from an RpoS-independent promoter, restored resistance to desiccation in the rpoS mutant. pBBR1MCS and its derivative pBMCA6, which carry a WT copy of hsp20 (tagged with six histidines in the N terminus) under the control of the kanamycin promoter, were conjugated into the rpoS mutant strain AErpoS and the hsp20 mutant strain AEhsp20. The expression of Hsp20 in the strains harbouring pBMCA6 was confirmed by Western blot analysis (data not shown). As expected, pBMCA6 restored cyst desiccation resistance to strain AEhsp20 (Table 3). The cysts formed by AErpoS harbouring pBMCA6 neither resisted desiccation nor produced alkylresorcinols. This indicated that, in addition to ArpR and Hsp20, other proteins under the control of RpoS are necessary for the formation of mature cysts resistant to desiccation.
Discussion
Cysts of A. vinelandii are metabolically dormant cells that are resistant to desiccation. In the laboratory, they are viable for >10 years upon storage in dry soil (Vela, 1974). The RpoS sigma factor is essential to the formation of functional cysts resistant to desiccation and to synthesize alkylresorcinols (Cocotl-Yañez et al., 2011; Romero et al., 2013). In order to find genes specifically induced upon encystment and regulated by RpoS, we carried out a proteome analysis that revealed the presence of several proteins putatively under the control of RpoS that are specific to cysts. Among these, we found a haemerythrin-like protein and the alginate epimerases AlgE1 and AlgE6. Haemerythrins are non-haem oxygen-binding proteins in which oxygen binds to a di-iron centre (Stenkamp, 1994). In prokaryotes, these proteins have putative functions such as binding oxygen either as a storage (Karlsen et al., 2005), sensory or detoxification mechanism (Xiong et al., 2000; Isaza et al., 2006). Whether the haemerythrin-like protein present in cysts of A. vinelandii fulfils these roles remains to be investigated. The presence of AlgE1 and AlgE6 epimerases in cysts is not surprising, since the AlgE1–AlgE7 epimerases in combination are essential to the formation of cysts resistant to desiccation (Steigedal et al., 2008). Moreover, the inactivation of rpoS did not prevent alginate synthesis. The novel finding in this case is the regulation of AlgE epimerases by RpoS.
Heat-shock proteins are distributed widely in every domain of life, their function being to prevent the misfolding or aggregation of proteins exposed to stress. Heat-shock protein expression is activated by heat, oxidative stress, starvation, UV radiation, desiccation and other stresses (Vayssier & Polla, 1998.). Proteome analysis also revealed the presence of Hsp20 as one of the most abundant proteins in the cysts, although a low level of this protein was also detected in vegetative cells. Unexpectedly, the hsp20 mRNA level found in qPCR and by primer extension analysis was higher in vegetative cells than in cysts, suggesting that hsp20 translation is inhibited in vegetative cells and favoured under encystment conditions. Indeed, the gusA : : hsp20 fusion used here showed that translation of hsp20 is lower in vegetative cells and that translation of its mRNA begins following the encystment process. Thus, we propose that in vegetative cells, translation of the hsp20 mRNA is inhibited by the translational repressor RsmA, which was shown recently to be a repressor of algD (Manzo et al., 2011), encoding a key enzyme of the alginate biosynthetic pathway that is essential for the formation of mature cysts (Campos et al., 1996). In fact, the hsp20 Shine–Dalgarno region possesses a potential RsmA-binding site that matches in seven out of 12 positions with the SELEX (systematic evolution of ligands by exponential enrichment)-derived consensus binding sequence (Dubey et al., 2005) and overlaps the Shine–Dalgarno sequence (Fig. 2b). This sequence matched the RsmA-binding sequence present in the algD and phbR transcripts (Manzo et al., 2011; Hernandez-Eligio et al., 2012). Interestingly, in Artemia franciscana cysts, the levels of small heat-shock proteins have been suggested to be under transcriptional and translational regulation (King et al., 2013).
Transcription of hsp20 is dependent directly on the RpoS sigma factor, which is unusual because in bacteria, and particularly in E. coli, the master regulator of heat-shock proteins, including small heat-shock proteins, is RpoH (Tilly et al., 1986; Grossman et al., 1987). Additionally A. vinelandii has an RpoH sigma factor, and no promoter sequence recognized by RpoH is present in the hsp20 regulatory region.
We found that expression in the rpoS mutant of hsp20 from an RpoS-independent promoter is not enough to restore cyst resistance to desiccation, implying that in addition to Hsp20 and ArpR, other proteins under RpoS control are necessary for cyst formation and desiccation resistance. We propose that these proteins include some alginate epimerases, as AlgE1 and AlgE6 were identified here as putative targets of RpoS.
This study shows that the Hsp20 protein plays an essential role in tolerance to desiccation, as cysts of the hsp20 mutant strain were unable to resist desiccation. This inability is not caused by changes in the structure of the cyst, since they are similar to the cysts formed by the WT strain. Consistent with this phenotype, the hsp20 mutant strain is not affected in the synthesis of alkylresorcinols, as these phenolic lipids play a role in cyst structure, but are not essential for desiccation resistance (Segura et al., 2009). In the desiccation stage, the lack of water in the cell is one of the most damaging stresses. Protein–protein aggregation occurs as a consequence of a loss of water. Small heat-shock proteins are able to bind proteins to protect them by avoiding aggregation (Ehrnsperger et al., 1997; Lee et al., 1997; Haslbeck et al., 1999). As a small heat-shock protein, Hsp20 is likely to interact with proteins that are important in maintaining the viability of A. vinelandii cysts, in order to protect them against misfolding or aggregation under desiccation.
Finally, small heat-shock proteins have been found to be related to desiccation protection in eukaryotic organisms, as they are found at high levels in the encysted embryos of several branchiopod crustaceans (Clegg, 2001, 2005; Willsie & Clegg, 2001; King et al., 2013). Additionally, sHsp20 and P23 transcripts accumulated in the larvae of the chironomid Polypedilum vanderplanki during desiccation (Gusev et al., 2011). To our knowledge, the present paper is the first report in bacteria of a small heat-shock protein that is essential for cyst desiccation resistance.
Acknowledgements
This research was supported by CONACYT (grant no. 131856). M. C.-Y. wishes to thank CONACYT for financial support during his PhD studies. We wish to acknowledge A. G. Martinez-Batallar and M. Hernández for their technical support in the proteomics methodology, H. Sámano-Sánchez for critical reading and Guadalupe Zavala and Paul Gaytan for technical support.