Total protein extracts from L. plantarum WCFS1 wild type and ΔctsR mutant strains from unstressed or heat-treated cultures were analyzed by 2D-gel fractionation, using a linear pH gradient (pH 4.0–7.0) during electrofocusing. The second dimension exhibited a separation range from 10 to 250 kDa. Representative gels of the four experimental conditions are depicted in Figure 1.

Image analysis revealed a total of 305 ± 3 and 328 ± 4 well-resolved peptide spots for the wild type and the mutant grown at 30 °C, as well as 311 ± 10 and 333 ± 7 for the same strains after temperature upshift to 42 °C.

Quantification of the spot intensity, performed by PDQuest software analysis, pointed out statistically significant changes in the levels of 23 spots (Figure 1 and Table 1).

Their identification, after in situ digestion with trypsin, by MALDI-TOF-TOF peptide mass fingerprinting, revealed that in several cases, different spots corresponded to essentially the same polypeptide and thus constituted different isoforms, presumably arising from in vivo post-translational modifications or due to artifacts occurring during sample handling. The characteristics of the investigated peptides are summarized in Table 1 and detailed maps of selected spots are reported in Figure 2.

In summary, the 23 analyzed spots corresponded to 10 proteins: the cold shock protein CspC (spot 1); the heat shock proteins Hsp1 (spot 2) and Hsp3 (spots 3–5); typical chaperonines GroES, GroEl and DnaK (spots 6, 7 and 8), the seryl-tRNA synthetase SerRS2 (spots 9–10) and three proteins ClpP, ClpE, ClpB belonging to Clp family (spots 11–23).

The heat shock response in both strains resulted in an increase in intensity of seven spots (2–8 in Figure 1) corresponding to five different peptides. In addition, spot quantification always indicated a greater heat-induction factor of these peptides in the ΔctsR mutant than in the wild type strain, due to a lower basal level in the mutant grown at 30 °C (Table 1). Overall, there was a moderate increase of the general stress-induced chaperones GroES, GroEL and DnaK (spots 6, 7 and 8) ranging from 1.25-fold to 5.7-fold upon temperature upshift in both strains analyzed (Figure 1). The levels of other two peptides Hsp1 (spot 2), Hsp3 (spots 3–5), drastically increased when bacteria were submitted to heat-stress at 42 °C. In particular, the small heat shock protein Hsp1 (spot 2) was never detected in extracts from cultures grown under optimal temperature conditions. Similarly, spot 4 was absent in the mutant and only weakly detectable in the L. plantarum WCFS1 wild type untreated cultures (Figures 1 and 2). Spots 3, 4 and 5 were identified as the same heat shock protein, Hsp3, thus revealing the existence of three potential isoforms, exhibiting two different molecular mass (Mr) (about 16.9 kDa for spot 4 and 15.4 kDa for spots 3 and 5) and pI values (approximately pH 5.0 for spot 5 and pH 4.8 for spots 3 and 4) (Figure 2). Spot 5 showed the highest heat-induction, increasing up to approximately 30-fold in both strains, and exhibited a Mr 1.2 kDa lower than the predicted theoretical mass, thus suggesting a potential post-translational modification connected to heat shock response. Proteomics should serve as a highly reproducible tool to allow the identification of spots by comparing maps obtained from different studies. Indeed, we could recognize the characteristic three spots of Hsp3 even in previously reported 2D-gel analyses of total extracts from different L. plantarum strains [18,20]. Izquierdo and coauthors [20] detected only one spot; while, in contrast to our results, Hamon et al. [18], identified two of these three spots as the alkaline shock proteins Asp1 and Asp2. Therefore, to our knowledge, this would be the first study to indicate the existence of different isoforms of small heat shock proteins in L. plantarum WCFS1 as well as in other LAB species. Future studies should address this aspect especially in relation to possible mechanisms of regulation of sHsp activity.

Two spots (9 and 10) clearly differentiated between the mutant and the wild type proteomic map under all experimental conditions. Spot 9 was only present in the L. plantarum WCFS1 wild type protein extracts (Figure 1A,C), whereas spot 10 was only observed in the mutant strain samples (Figure 1B,D). These spots, slightly differing only in the isoelectric point value (5.13 and 5.17, respectively), were both identified as the seryl-tRNA synthetase SerRS2 (Figure 2). Intriguingly, the serRS2 gene of L. plantarum WCFS1 is located exactly upstream of ctsR. Even if the correct ctsR deletion had been previously verified [13], we cannot rule out the occurrence of polar effects that may have altered the expression of proximal genomic regions, including serRS2. Maybe such alteration could be associated to the detected post-translational modification. Aminoacyl-tRNA synthetases are essential components in the translation machinery in all forms of life. They are responsible for charging their cognate tRNAs with the correct amino acid [21]. We do not know the biological significance of this observed difference. To our knowledge, no previous proteomic study has reported similar results. We can also hypothesize that the form found in the ΔctsR extracts might arise from pleiotropic effects of the mutation.

A set of 13 spots (11–23, Figures 1 and 2 and Table 1) showed a differential pattern in the two strains analyzed. Under optimal growth temperature, the levels of these peptides were higher (from 1.75- to 12-fold) in the mutant. After heat shock, the spots increased in intensity in both strains, the resulting peptide levels being on average 1.8-fold higher in the ΔctsR mutant than in its parental strain. MASCOT analysis revealed that these spots were isoforms of just three proteins, namely ClpP, ClpE and ClpB. Summing together the spots recognized as the same protein showed that the levels of all of them were enhanced as a consequence of the ΔctsR mutation without the need of heat-stress. Thus, ClpB, ClpP and ClpE levels were, respectively, approximately 11-, 6- and 5-fold higher in the ΔctsR mutant than in the parental strain (Table 1). This finding is consistent with previous transcriptional analysis [14], and further confirms that clp genes belong to the CtsR regulon of L. plantarum WCFS1. Indeed, potential CtsR boxes were previously found in their 5′ regions. Moreover, in line with the proteomic data presented here, their basal mRNA level was found to be higher in the ΔctsR deficient strain, indicating lack of repression by CtsR [14]. Analyzing L. plantarum cultures exposed to alkaline conditions, Lee et al. [19] observed that an increase in ClpB level corresponded to a reduction in the DnaK concentration. Similarly, in this study, we found that typical stress-response proteins, such as DnaK, GroEL, GroES, the small heat shock proteins Hsp1 and Hsp3 and the cold shock protein CspC were always downregulated in the unstressed mutant strain. Our interpretation is that the notably higher levels of proteins belonging to the Clp family observed in the ΔctsR mutant could provide the correct refolding or proteolytic degradation of the damaged proteins at physiological conditions, thus limiting the activation of other chaperone systems.

We detected two putative isoforms (spots 12 and 13) of the 21.5 kDa proteolytic subunit ClpP. Spot 11 was also identified as ClpP, although its molecular mass (33 kDa) and isoelectric point did not correspond to the theoretically predicted ones. This particular form of ClpP was previously observed by Cohen et al. [15] in L. plantarum WCFS1 at stationary-phase, suggesting that this might be due to a complex with an unidentified small molecule. Three isoforms (spots 21–23) with a Mr of 95.6 kDa were identified as the ATP-binding subunit ClpB, varying slightly in their isoelectric points (Figure 2). Seven well-defined spots (14–20) were identified as being isoforms of the ATP-binding subunit ClpE. These peptides were resolved in the gel into two main groups (four, spots 17–20, and three, spots 14–16) with a Mr difference of 5.8 kDa (Figure 2). Recently, Elsholz et al. [7] demonstrated that the heat stress regulation of CtsR was regarded to be dependent on an intrinsic heat-sensing property of CtsR in all low GC, Gram-positive bacteria. Therefore, de-repression is considered as a physiological process, which is independent from CtsR degradation. It was reported that ClpE plays a role in the control of CtsR stability, but the mechanisms of how this occurs has not yet been clarified [7,22–24]. Two ClpE forms, differing by 6 kDa in their Mr, as revealed by western blot analysis, were previously reported in extracts from Lactococcus lactis strains subjected to either high temperature [22] or oxidative stress [24]. Based on these previous studies and on our data, we can therefore suggest that a variation of the relative concentration of these forms may be associated to a cellular heat stress condition. However, in contrast to what was observed by Elsholz et al. [24], and taking into account our proteomic analyses, it is the small subunit that seems to characterize physiological conditions in the wild type, whereas the heavier Mr isoforms appear only in response to heat shock. Accordingly, in the mutant strain, which constitutively exhibits a proteomic pattern that is typical of a stress condition, all ClpE forms are evident in extracts from both unstressed and heat stressed cells. The peptide mass fingerprint of the four heavier isoforms revealed the occurrence of a 15 amino acid fragment corresponding to the N-terminal zinc finger region (Figure 3). However, this peptide was never detected in the digestion pattern of the three small ClpE isoforms; instead, a different 13 amino acids N-terminus proximal fragment, starting from position 66, was found. Such a fragment was never observed after trypsin digestion of ClpE spots with heavier Mr (Figure 3).

This finding may suggest a putative cleavage site after the N-terminal zinc finger region that justifies the observed difference of 5.8 kDa between the two ClpE groups, although, to confirm it, N-terminus of the different spots should be experimentally determined. The zinc finger motif in the N-terminal domain is highly conserved in the ClpE family [25]. Varmanen et al. [22] reported that in heat stressed L. lactis, the small ClpE isoform was lacking in a ΔclpP mutant, while the same isoform prevailed in bacteria expressing a mutation in the ClpE zinc finger domain, suggesting that this region is implicated in ClpP-dependent processing. Recently, Miethke and coauthors [23] suggested that ClpEP is involved in CtsR degradation after heat shock and demonstrated that the ClpE N-terminal zinc finger domain is essential for basal in vitro ATPase activity. Elsholz et al. [7] found that, during heat exposure, CtsR concentration diminished in L. lactis wild type but it was not affected in a ΔclpE mutant, confirming that a ClpEP complex is responsible for CtsR quality control in low GC, Gram-positive bacteria lacking McsA and McsB modulators. Strikingly, we found that, in unstressed wild type strain, the small ClpE variant was 10-fold more abundant than the higher molecular mass ClpE. Conversely, in the mutant strain and when wild type cells were exposed to heat, the ratio between both forms was unchanged (almost 1:1 ratio). According to the reported studies, our results suggest that when L. plantarum WCFS1 cells are exposed to heat stress, a ClpEP complex may bind CtsR via a putative ClpE zinc finger domain interaction, allowing CtsR degradation and, thus, the expression of the CtsR regulon.

Changes in the expression of the genes encoding the proteins under investigation and/or under putative CtsR control were monitored by qRT-PCR transcriptional analysis. The mRNA levels in both wild type and mutant strains were analyzed for clpE, clpB, clpP and hsp1, identified in the protein map, as well as for the clpC gene, whose promoter, like that of hsp1, was previously shown to be specifically recognized and bound by the recombinant L. plantarum WCFS1 CtsR [14]. In order to evaluate heat stress induction at the transcriptional level, the mRNA pattern of these L. plantarum WCFS1 genes, presenting CtsR operators, was monitored prior to and after short heat exposure, in both wild type and mutant strains. Relative to control temperature condition, increased mRNA levels were detected after heat shock for all the analyzed genes in both strains (Figure 4). Overall, the transcriptional profile matched the proteomic data (stress induction factors of Table 1), except for clpC, whose protein could not be identified by the 2D-gel analysis, and clpP level in the wild type, whose mRNA was poorly heat-induced (1.3 fold), in contrast to the increased protein concentration (4.9 fold).

In the parental strain, hsp1, clpC, clpB, and clpE transcripts were strongly induced by heat shock, confirming proteomic analysis and their involvement in heat-stress response mechanisms. In contrast, clpP expression was poorly heat induced (1.7-fold) in the mutant strain and only weakly triggered in the wild type (1.3-fold); such a discrepancy might reflect a different kinetics/temporal pattern of transcriptional activation [26] and/or a particularly short half-life of the clpP mRNA. Indeed, the transcriptional induction of some CtsR-controlled genes, including clpP, was found to be rather transient in response to heat shock [7,22,27].

The significant heat induction of putative CtsR regulon genes, at both transcriptional and protein level, in the ΔctsR mutant, suggests that additional, CtsR-independent, regulatory mechanisms contribute to control their expression. A heat shock control mediated by additional regulators was previously hypothesized in L. plantarum WCFS1 for the CtsR-controlled ftsH, as well as in other bacterial species for other CtsR-repressed genes [5,13,28–30]. The heat shock induction was particularly marked for hsp1 (28-fold increase in the mutant strain), which is likely to be under a dual repression control, mediated by both CtsR and HrcA, as suggested by previous studies revealing the presence of a CIRCE element in hsp1 promoter [9,31]. This hypothesis could also explain the failure in detecting Hsp1 in the proteomic map of both wild type and CtsR-deficient strains at optimal growth temperature. Possibly, under physiologic conditions, the strong repression exerted by CtsR and HrcA keeps the Hsp1 level below the detection limit for 2D-gel analysis.

For clpC, clpB and clpE, a lower transcriptional heat-induction was detected in the ΔctsR genetic background, compared to wild type (Figure 4). These results correlate with the proteomic data on ClpB and ClpE: under physiological conditions, their basal level in the ΔctsR strain was found to be higher than in the wild type; accordingly, giving the lack of repression by CtsR, a lower heat induction was expected and indeed observed in response to heat stress (Table 1). A similar pattern was observed for clpC, whose changes in expression could only be appreciated at the mRNA level, since the corresponding protein was not detected in the proteomic map of either wild type or ΔctsR L. plantarum. We suppose that ClpC level may be below the detection limit in both strains, either in normal or heat stress condition. As a support to our hypothesis, other proteomic studies, using similar 2-D gel techniques, did not detect ClpC in extracts from L. plantarum cultures either under control conditions or subject to different kinds of stress [15–20].

Under optimal temperature conditions, the concentration of CspC (spot 1) was 37% lower in the mutant relative to wild type strain (Table 1). A lower expression of cspC gene in the ΔctsR strain was also confirmed by qRT-PCR analysis, indicating a 50% decrease in its mRNA level with respect to wild type, as detected under physiologic conditions (data not shown). Sequence analysis of the cspC five prime untranslated region (5′-UTR) revealed the presence of a putative CtsR-binding site (5′-AGGTAAT TTT GGTCAGA-3′), which partially matches the consensus CtsR box (5′-RGTCADN NAN RGTCADN-3′), defined so far for Gram-positive bacteria [5]. However, based on our data, a direct CtsR repression on transcription of this gene appears unlikely.

The reference strain L. plantarum WCFS1 [32] and the previously described L. plantarum ΔctsR deletion mutant [13] were used in this study. Strains were routinely grown in MRS broth (Pronadisa, Madrid, Spain) at 30 °C.

Total protein extracts of L. plantarum strains were obtained from exponentially growing cultures (OD₆₀₀ = 0.6) before (control) or after transfer from 30 °C to 42 °C and further incubation for 30 min (heat stress). Growth curve analyses were performed to assure that both strains were stressed at corresponding growth phases. Colony forming units (CFU) counting indicated similar survival rates at heat challenge for both strains. Cells were sedimented by centrifugation (11,000× g, 20 min, 4 °C), washed with PBS at pH 7.1, and pellets frozen at −70 °C. Total protein extracts were obtained by mechanical disruption with a French Press as previously described by Mohedano et al. [33]. The protein concentration in the extracts was quantified by using the Quant-iT™ Protein Assay Kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. All experiments were performed in triplicate.

Samples of the bacterial lysates, containing 80 μg of proteins, were analyzed by 2D electrophoresis as previously described [34] with the following modifications. The first dimension was run on IPG strips (pH 4.0–7.0, 17 cm) (Bio-Rad, Hercules, CA, USA) with a 4-step program (50 V for 16 h, 300 V for 45 min, 3500 V for 22 h 45 min, and 5000 V for 30 min). A total of >80,000 V h were reached in all cases. The second dimension was run on 12% SDS-PAGE at 5 watts/gel for 30 min and then at 17 watts/gel until the die-front reached the bottom edge (approximately 5 h). Reference markers consisted of 10 μL of Precision Plus ProteinTM Standards Dual Color (Bio-Rad).

The gels were stained with SYPRO Ruby protein gel stain (Bio-Rad) according to the manufacturer’s instructions, and images of the gels were taken with a EXQuest Spot Cutter (Bio-Rad).

The spots were detected and quantified with the PDQuest 2D analysis 7.4.1 software (Bio-Rad). For this quantification, 305 stained spots were matched in the 12 gels analyzed (three biological replicates for each experimental condition), and used to normalize the average intensity. In order to define the appropriate spot detection parameters, the faintest, the smallest and the largest spots were marked on the gel scan, followed by manual matching of landmarks and significant spots. Intensities were measured via both absolute and normalized spot volumes. Differentially expressed proteins were highlighted by statistical analysis. Spots showing statistically significant differences (Student’s t test, p-value ≤ 0.05) and a minimum of a 1.8-fold change of intensity were selected for identification. When different spots were identified as being the same protein (see 2.1), the total volume of the various isoforms was calculated.

Spots of interest were automatically excised using the EXQuest Spot Cutter, deposited in 96-well plates and analyzed at the Centro de Investigaciones Biológicas Proteomic Unit as follows. The spots were digested in gel automatically using a DigestPro MS apparatus (Intavis AG, Koeln, Germany). The digestion protocol used was based on Schevchenko et al. [35] with minor variations: gel pieces were washed first with 50 mM ammonium bicarbonate (Sigma-Aldrich, St. Louis, MO, USA) and second with acetonitrile (ACN) (Scharlau, Barcelona, Spain). Trypsin (Promega, Madison, WI, USA), at a final concentration of 12.5 ng/μL in 50 mM ammonium bicarbonate solution, was added to the gel pieces for 8 h at 37 °C. Finally, 70% ACN containing 0.5% trifluoroacetic acid (TFA) (Sigma-Aldrich) was added for peptide extraction. The extracted peptides were dried by speed-vacuum centrifugation and resuspended in 4 μL of 30% ACN with 0.1% TFA.

One microliter of each peptide mixture was deposited onto an 800 μm AnchorChip (Bruker- Daltonics, Bremen, Germany) and dried at RT. One microliter of matrix solution (3 mg/mL α-cyano-4-hydroxycinnamic acid) in 33% ACN with 0.1% TFA was then deposited onto the digest and allowed to dry at room temperature (RT).

Samples were analyzed with an Autoflex III TOF/TOF mass spectrometer (Bruker-Daltonics). Typically, 1000 scans for peptide mass fingerprinting (PMF) and 2000 scans for MS/MS were collected. Automated analysis of mass data was performed using FlexAnalysis software (Bruker-Daltonics). Internal calibration of MALDI-TOF mass spectra was performed using two trypsin autolysis ions with m/z 842.510 and m/z 2211.105; for MALDI-MS/MS, calibrations were performed with fragment ion spectra obtained for the proton adducts of a peptide mixture covering the m/z 800–3200 region. The typical error observed in mass accuracy for calibration was usually below 20 ppm. MALDI-MS and MS/MS data were combined through the BioTools 3.0 program (Bruker-Daltonics) to interrogate the NCBI non-redundant protein database (NCBI: 20100306) using MASCOT software 2.3 (Matrix Science, London, UK). Relevant search parameters were set as follows: enzyme, trypsin; fixed modifications, carbamidomethyl (C); oxidation (M); 1 missed cleavage allowed; peptide tolerance, 50 ppm; MS/MS tolerance, 0.5 Da. Protein scores greater than 75 were considered significant (p-value ≤ 0.05).

Total RNAs were extracted from exponentially growing L. plantarum cultures (OD₆₀₀ = 0.6), before and after heat stress (incubation at 42 °C for 10 min) using the UltraClean microbial isolation kit (MO BIO, Carlsbad, CA, USA), according to the manufacturer’s instructions. The quality of the RNA samples, estimated by integrity of rRNA, was verified by electrophoresis on 1.2% agarose gels and analysis with Quantity One software (Bio-Rad). Total RNA preparations were quantified by using the Quant-iT™ RNA Assay Kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. One microgram of total RNA was reverse transcribed to cDNA using the Quantitect reverse transcription kit (Qiagen, Valencia, CA, USA), which includes a DNase I treatment step. The absence of chromosomal DNA contamination was confirmed by real-time PCR on corresponding DNase I-treated RNAs.

Quantitative real-time PCR (qRT-PCR) was performed on an Applied Biosystems 7300 real-time PCR instrument using SYBR green I detection. The ldhD gene of L. plantarum WCFS1 was used as internal control [14]. Each reaction mixture contained 5 μL of 20-fold-diluted cDNA, 10 μL of Power SYBR green PCR master mix (Applied Biosystems, Foster City, CA) and 100 nM of each sense and antisense primer [14]. The cspC forward and reverse primer sequences used were: 5′ ATC ACT CGC GAA AAC GGT AG 3′; and 5′ CGA TCG CTT TCT TCA ACG TC 3′, respectively. The amplification profile comprised 10 min at 95 °C, followed by 35 cycles consisting of 20 s at 95 °C, 30 s at 58 °C, and 30 s at 72 °C. Fluorescence was monitored during each extension phase, and melting-curve analyses were performed to confirm the amplification of specific transcripts. Data were analyzed using the AB 7300 software, by applying the two standard-curves quantification method. Each assay included triplicates of PCR of the samples, negative no-template controls, and standard curves for both the internal-control and target genes. Using one-tailed, two-tailed Student’s t-test, any p-value ≤ 0.05 was considered significant.