To investigate cell determinants of resistance to gastric acidity in Lactobacillus plantarum using comparative proteomics.
To investigate cell determinants of resistance to gastric acidity in Lactobacillus plantarum using comparative proteomics.
Among ten Lact. plantarum strains that were tested for their acid resistance in vitro, three strains with different phenotypes were selected for comparative proteomic analysis: LC 804 (resistant), CIP A159 (intermediate) and CECT 4185 (sensitive). Constitutive differences between whole-cell proteomes of the three strains were studied. Among the differentially expressed proteins between strains, 17 have previously been reported to be involved in acid resistance processes. The effect of a low-pH exposure on these proteomic patterns was investigated, which led to the identification of five putative determinants of acid resistance (heat-shock protein GrpE, methionine synthase and 30S ribosomal protein S2) or sensitivity (phosphotransacetylase and adenylosuccinate synthase). Analysis also revealed three additional proteins involved in cell envelope biogenesis (3-oxoacyl-synthase II, dTDP-glucose 4,6-dehydratase and dTDP-4-dehydrorhamnose 3,5-epimerase) likely to be key factors of intrinsic acid tolerance in Lact. plantarum.
The approach used in this study enabled the identification of potential markers of acid tolerance in Lact. plantarum, which may serve for phenotyping and screening purposes.
The present work represents a further step towards the identification of bacterial biomarkers for each particular probiotic feature.
Lactobacilli constitute a heterogeneous group of lactic acid bacteria with a prominent role in food fermentation (Giraffa et al. 2010). Based on their proven health-promoting capacities (Weichselbaum 2009), some lactobacilli have been recognized as probiotics (Araya et al. 2002). Among the major species, Lactobacillus plantarum is part of a number of ethnic and commercial probiotic products where it has a long history of safe use (de Vries et al. 2006). It is also a flexible and ubiquitous species with a high level of phenotypic and genotypic diversity (Siezen et al. 2010).
One of the key requirements for a strain to comply with the probiotic concept is that it should remain viable during gastrointestinal (GI) transit (Araya et al. 2002). Before they can adhere to the intestinal mucosa, grow and exert beneficial effects, probiotic bacteria must in fact survive the passage through the stomach, where they are faced with the high acidity of the human gastric juice. The latter is mainly composed of pepsin and chlorhydric acid, generating a fasting pH of 1·5, which increases to between pH 3·0 and pH 5·0 during feeding (Shulkesa et al. 2006). Because these harsh conditions provide a natural barrier against the entry of living bacteria into the intestinal tract, acid resistance is considered as a key criterion in the selection of potential probiotics (Parvez et al. 2006).
To resist acid stress, lactobacilli employ several complementary strategies, including maintenance of intracellular pH homeostasis, repair of damaged proteins and cell envelope modifications (Lebeer et al. 2008; reference therein). In addition, acid resistance of several Lactobacillus strains is positively affected by prior exposure of the cells to moderately acidic conditions (Lorca and Valdez 2001; Rollan et al. 2003; Guillouard et al. 2004; Zhang et al. 2012), a mechanism known as the acid tolerance response (Foster and Hall 1991). Recent investigations have shed some light on the adaptative aspect of lactobacilli's tolerance to acidic conditions (Lee et al. 2008; Huang et al. 2011; Wu et al. 2011, 2012). Still, the molecular bases and mechanisms of acid resistance remain largely unclear.
In recent years, proteomics has significantly contributed to unravel the molecular basis of probiotic functionality (Siciliano and Mazzeo 2012). Notably in the Lactobacillus casei phylogenetic group, comparative proteomic studies enabled the identification of potential markers of the adhesion capacity (Izquierdo et al. 2009) and the bile tolerance (Hamon et al. 2011, 2012), two of the commonly used criteria in probiotic strain selection. Developing such an interstrain analytical approach is critical when it comes to selecting the strains that stand the best chance of success in clinical trials. Other probiotic features including acid resistance will have to be integrated in this approach with the aim of acquiring for each individual species a set of biomarkers covering various probiotic characters that would serve as a fingerprint for strain selection.
The present study investigates the natural protein diversity within the Lact. plantarum species with relation to acid resistance and the subsequent capacity to survive GI tract conditions. It is based on the study of the proteomic patterns of three Lact. plantarum strains showing different levels of acid resistance in vitro.
Ten Lact. plantarum subsp. plantarum strains were used in this study: four collection strains (Lact. plantarum CECT 748T, CECT 749, CECT 4185 and CIP A159), one probiotic strain commonly used in commercial products (Lact. plantarum 299V, Probi, Lund, Sweden) and five laboratory isolates (Lact. Lact. plantarum LC 56, LC 660, LC 800, LC 804 and WHE 92). Strains were identified at the species level using recA PCR (Bringel et al. 2005) (data not shown). All cultures were maintained as frozen stocks held at −80°C in Cryobank cryogenic beads (Bio-Rad, Hercules, CA, USA). For experimental use, strains were cultured anaerobically (Anaerocult A system; Merck, Darmstadt, Germany) at 37°C in de Man–Rogosa–Sharpe broth (Biokar, Beauvais, France) supplemented with 0·05% (w/v) l-cysteine hydrochloride monohydrate (MRSC; Merck) to early-stationary phase, using three successive subcultures (1% v/v inoculation). In these conditions, early-stationary phase was reached after a culture time ranging from 12 to 15 h depending on strains, as determined from their respective growth curves at 600 nm (data not shown), with pH values at harvesting time ranging from 4·0 to 4·3.
Acid resistance was assessed by investigating the ability of strains to survive in a simulated gastric juice (SGJ), as previously described (Gibson et al. 2005). Early-stationary phase cells from a 10-ml broth culture were harvested, washed twice with phosphate-buffered saline (Biokar; pH 7·2), and cell pellets were resuspended in 10 ml of half-strength peptone water, made of 5 g l−1 of casein peptone (Biokar) and 2·5 g l−1 of sodium chloride (Merck) in distilled water. Bacterial suspensions, with a cell density of around 109 cells ml−1, were diluted 10-fold in SGJ medium, composed of half-strength peptone water and 0·3 g l−1 pepsin from porcine stomach mucus (Sigma-Aldrich, St Louis, MO, USA) and adjusted to pH 1·6, 1·8 and 2·0 using 1 mol l−1 hydrochloric acid (Merck). Controls consisted of bacterial suspensions in half-strength peptone water (pH 6·5). After a 20min incubation at 37°C, aliquots of these suspensions were taken and diluted serially in half-strength peptone water to determine cell viability. Cell numbers were estimated by plate counting using 10 μl of serially diluted samples that were spread onto MRSC agar plates and incubated anaerobically at 37°C for 48 h. Three independent experiments were carried out, and each assay was performed in triplicate. Counts of viable bacteria in SGJ, expressed as log CFU ml−1, were compared to those of controls. Using Statgraphics plus 5·1 software (Manugistics, Rockville, MD, USA), data were subjected to a two-way analysis of variance (anova) with strain and pH as variables. A multiple comparison test using least significant difference procedure was carried out to compare means for which the anova test indicated significant differences (P < 0·05).
Preparation of total protein extracts and two-dimensional electrophoresis (2-DE) were performed as previously reported (Hamon et al. 2011). Experiments were carried out for bacterial cells cultured in two different broths: MRSC (initial pH of 6·5) and MRSC initially adjusted to pH 4·0 with HCl. Significant differences in protein expression were investigated with three independent gels per strain and per condition using a GS-800 Calibrated Densitometer (Bio-Rad) and the PD Quest 8.0.1 software (Bio-Rad). Spot intensities were normalized to the sum of intensities of all valid spots in one gel. For analysis of changes in protein expression during acid exposure, a protein was considered to be under- or overproduced when changes in normalized spot intensities were at least 1·5-fold at a significance level of P < 0·05 (Student's t-test for unpaired samples), as previously described (Sanchez et al. 2007b). Regarding proteome comparison between strains, proteins were considered differentially produced when spot intensities passed the threshold of a twofold difference at a significance level of P < 0·05 (Student's t-test for unpaired samples), as previously defined (Hamon et al. 2011). Both thresholds were arbitrarily chosen.
Spots of interest were subjected to tryptic in-gel digestion and analysed by chip-liquid chromatography–quadrupole-time-of-flight (chip-LC-QTOF) using an Agilent G6510A QTOF mass spectrometer equipped with an Agilent 1200 Nano LC system and an Agilent HPLC Chip Cube, G4240A (Agilent Technologies, Santa Clara, CA, USA), as described previously (Hamon et al. 2011). Protein identification was performed against the four genomes of Lact. plantarum (ATCC 14917T, JDM1, ST-III and WCFS1) available at the NCBI Website (http://www.ncbi.nlm.nih.gov; accessed 4 March 2013), based on consensus scoring between in-house installed versions of Phenyx (Geneva Bioinformatics, Geneva, Switzerland) and OMSSA (Geer et al. 2004) MS/MS search algorithms to confirm protein identities and limit the risk of false positives (Kapp et al. 2005). Annotations were made according to the cluster of orthologous genes (COG) functional groups.
Cluster analysis was performed as described by (Dumas et al. 2008) on the protein spots that were differentially expressed in standard growth conditions between three Lact. plantarum strains showing different acid resistance phenotypes in vitro (CECT 4185, sensitive; CIP A159, intermediate; and LC 804, resistant), focusing exclusively on proteins (a total of 17 found in 28 spots) reportedly involved in acid resistance. When a protein spot was reliably absent in one strain, the absent values were replaced by the lowest value for the gel. Each protein spot volume was divided by the average value of all existing values for this protein spot in all gels and then subjected to a base-2 logarithmic transformation (Meunier et al. 2007). The hierarchical clustering analysis was carried out with PermutMatrix, which allows a representation of clustering results as dendograms of the gels and of the protein spots (Caraux and Pinloche 2005). Clustering results were calculated using an algorithm resulting from the combination of Ward's aggregation method and Pearson-based distance metrics (Meunier et al. 2007).
Strains of Lact. plantarum were exposed to a 20-min acid shock in SGJ (pH 1·6, 1·8 and 2·0), and their ability to maintain viability was investigated (Table 1). Two-way anova revealed significant effects of the strain, the pH value and their interaction (P < 0·05). A stepwise lowering of the pH resulted in a gradual decrease in the viability of bacteria (P < 0·05). Although strains showed similar behaviours at pH 2·0 (P < 0·05) with none of them exceeding 0·5 log reduction in viability as compared to controls, the impact of pH 1·6 and 1·8 was unequal as it depended on the considered strain. Bacteria could be assigned to three different groups according to their acid resistance phenotype: a resistant phenotype including Lact. plantarum LC 804 with limited viability losses following acid challenge at pH 1·6 and 1·8, reaching respectively 2·57 ± 0·61 and 0·26 ± 0·15 logarithmic units, as compared to the control; a sensitive phenotype with Lact. plantarum CECT 4185, which showed no viable cells after exposure to pH 1·6 and the highest viability decrease at pH 1·8 (4·53 ± 0·14 logarithmic units); an intermediate phenotype with the eight other strains, which were moderately acid resistant and had viability reductions in the range from 2·80 ± 0·38 to 6·03 ± 0·85 logarithmic units. Lact. plantarum CECT 4185 (highest sensitivity), Lact. plantarum CIP A159 (intermediate sensitivity) and Lact. plantarum LC 804 (lowest sensitivity) were used for comparative proteomic analysis in standard conditions and following low-pH exposure.
|Strains||Viable counts of the control (log CFU ml−1)||Viability reduction vs. control (Δlog CFU ml−1)|
|pH 1·6||pH 1·8||pH 2·0|
|299V||8·82 ± 0·14||4·4 ± 0·49‡,§||3·14 ± 0·47‡,§,¶||0·38 ± 0·12|
|CECT 748T||8·58 ± 0·13||4·27 ± 0·85†,‡||2·87 ± 0·34‡,§||0·14 ± 0·08|
|CECT 749||8·53 ± 0·14||4·62 ± 0·63‡,§||3·19 ± 0·17‡,§,¶||0·21 ± 0·13|
|CECT 4185||8·53 ± 0·09||> 7·5||4·53 ± 0·14**||0·24 ± 0·17|
|CIP A159||8·50 ± 0·02||4·91 ± 1·30‡,§||3·25 ± 0·21§,¶||0·31 ± 0·11|
|LC 56||8·22 ± 0·19||4·89 ± 1·80‡,§||3·45 ± 0·25¶||0·33 ± 0·16|
|LC 660||8·27 ± 0·03||3·58 ± 0·47†,‡||2·97 ± 0·15‡,§||0·14 ± 0·11|
|LC 800||8·76 ± 0·07||4·61 ± 1·18‡,§||2·80 ± 0·38‡||0·16 ± 0·09|
|LC 804||8·77 ± 0·06||2·57 ± 0·61†||0·26 ± 0·15†||0·10 ± 0·12|
|WHE 92||8·70 ± 0·08||6·03 ± 0·85§||4·10 ± 0·30**||0·38 ± 0·12|
Lact. plantarum CECT 4185, CIP A159 and LC 804 were cultured in MRSC under nonstressing conditions to early-stationary phase (pH at harvesting time of 4·2, 4·0 and 4·0, respectively), and whole-cell proteins were extracted and analysed by 2-DE in a pI range of 4·0–7·0 and a mass range of 10–250 kDa. In these conditions, the estimated coverage of the theoretical proteome of Lact. plantarum was 52·2%, as inferred from the in silico proteomic data obtained from the four Lact. plantarum genomes available at NCBI (data not shown). Proteomic patterns of the three strains were compared to establish a link between a strain's constitutive proteome and its level of acid resistance. Figure 1 shows representative 2-D patterns for these strains when cultured in standard conditions. Although the constitutive proteomic patterns were overall similar, 102 out of an average of 461 detected protein spots displayed different expression levels (Table S1). These spots were excised, and the corresponding proteins were subjected to tryptic digestion followed by liquid chromatography–mass spectrometry (LC-MS) analysis and proteomic database search using Phenyx and OMSSA search engines. Of the 102 spots, 94 representing 67 different proteins were identified, some of these proteins having been found in more than one spot, suggesting the presence of protein isoforms. Sequence alignment analysis was based on the four sequenced Lact. plantarum genomes (strains ATCC 14917T, JDM1, ST-III and WCFS1) and revealed a systematic occurrence of the corresponding genes with high level of similarity (>98%, results not shown). The identified proteins fell into 14 functional categories covering most of the biochemical functions encountered in bacterial cells (Table S1).
Among the identified proteins showing different expression levels between strains in standard conditions, 17 (found in 28 spots) have previously been reported to be involved in acid resistance (Table 2). Protein relative abundances were scrutinized using the hierarchical clustering analysis developed by Meunier et al. (2007) (Table S2). This made it possible, on the one hand, to sort gels according to the global patterns of the 28 spots of interest and, on the other hand, to gather spots with similar expression profiles across the gels (Fig. 2). Good data reproducibility could be observed as gels from one strain were clustered together with no significant intergel discrepancies. Protein spots could be divided into two main groups. The first group includes proteins that are overexpressed in the resistant phenotype (LC 804): heat-shock protein GrpE (GrpE), spot 4; inosine-5′-monophosphate dehydrogenase (GuaB), spot 66; 3-oxoacyl-synthase II (FabF), spot 69; mannose-6-phosphate isomerase (Pmi), spot 70; multiple sugar ABC transporter (MsmK1), spot 73; l-lactate dehydrogenase (LdhL), spots 74 and 80; lysyl-tRNA synthase (LysS), spots 79 and 94; phosphoketolase (Xpk2), spots 85, 86 and 87; methionine synthase (MetE), spot 88; ATP-dependent Clp protease ClpL (ClpL), spot 96; and 30S ribosomal protein S2 (RpsB), spot 102. The second group includes proteins that are either underexpressed (GuaB, spot 33; and glyceraldehydes 3-phosphate dehydrogenase (Gap), spots 61 and 65) or undetected (LdhL, spot 5; dTDP-glucose 4,6-dehydratase (RfbB), spots 14 and 51; dTDP-4-dehydrorhamnose 3,5-epimerase (RfbC), spot 42; phosphotransacetylase (Pta), spot 43; Gap, spot 46; Pmi, spot 49; and 6-phospho-β-glucosidase (Pbg5), spot 55; ClpL, spot 62; and adenylosuccinate synthase (PurA), spot 63) in the resistant phenotype (LC 804) (Fig. 2). In all, four proteins (ClpL, GuaB, LdhL and Pmi) showed some isoforms overexpressed in the resistant phenotype (ClpL: spot 96; GuaB: spot 66; LdhL: spots 74 and 80; and Pmi: spot 70) while other isoforms were underexpressed (GuaB: spot 33) or not detected (ClpL: spot 62; LdhL: spot 5; and Pmi: spot 49).
|Functional category||Protein identity||Geneb||Referencesc||Spot number||Variation factor: acid vs. standard conditionsd|
|CECT 4185||CIP A159||LC 804|
|Translation, ribosomal structure and biogenesis||Lysyl-tRNA synthase||lp_0550 (lysS)||Budin-Verneuil et al. (2007)||79||n.a.||n.a.||1,7|
|30S ribosomal protein S2||lp_2055 (rpsB)||Budin-Verneuil et al. (2007)||102||11,6||4,3||3,4|
|Cell wall biogenesis||dTDP-glucose 4,6-dehydratase||lp_1189 (rfbB)||Budin-Verneuil et al. (2007)||14||n.a.||1,2e||n.a.|
|dTDP-4-dehydrorhamnose 3,5-epimerase||lp_1188 (rfbC)||Budin-Verneuil et al. (2007)||42||1,4||n.a.||n.a.|
|Stress proteins, chaperones||Heat-shock protein GrpE||lp_2028 (grpE)||De Angelis et al. (2001)||4||2||2||2,3|
|ATP-dependent Clp protease ClpL||lp_3583 (clpL)||Wall et al. (2007)||62||1,4||−11||−4,4|
|Energy production and conversion||L-lactate dehydrogenase||lp_0537 (ldhL)||Budin-Verneuil et al. (2007)||5||1,2e||−1,1e||n.a.|
|Phosphotransacetylase||lp_0807 (pta)||Wolfe (2005)||43||–||n.a.||n.a.|
|Carbohydrate transport and metabolism||Glyceraldehyde 3-phosphate dehydrogenase||lp_0789 (gap)||Budin-Verneuil et al. (2007); Sanchez et al. (2007a)||46||–||n.a.||n.a.|
|Mannose-6-phosphate isomerase||lp_2384 (pmi)||Budin-Verneuil et al. (2007)||49||−1,0e||n.a.||n.a.|
|6-phospho-beta-glucosidase||lp_2778 (pbg5)||Budin-Verneuil et al. (2007)||55||3,9||−1,7||n.a.|
|Multiple sugar ABC transporter||lp_0180 (msmK1)||Budin-Verneuil et al. (2007)||73||n.a.||–||–|
|Phosphoketolase||lp_3551 (xpk2)||Pieterse et al. (2005); Sanchez et al. (2007a)||85||n.a.||n.a.||1,2e|
|Amino acid transport and metabolism||Methionine synthase||lp_1375 (metE)||Sanchez et al. (2007a)||88||++||n.a.||2,1|
|Nucleotide transport and metabolism||Inosine-5′-monophosphate dehydrogenase||lp_3194 (guaB)||Budin-Verneuil et al. (2007)||33||2,2||2,0||−1,7|
|Adenylosuccinate synthase||lp_3270 (purA)||Budin-Verneuil et al. (2007)||63||–||–||–|
|Lipid transport and metabolism||3-oxoacyl-synthase II||lp_1673 (fabF)||Budin-Verneuil et al. (2007)||69||n.a.||1,0e||1,1e|
Cells cultured in MRSC to early exponential phase (optical density at 600 nm of 0·5; 5·5 ≤ pH ≤ 5·6) were harvested and resuspended in stressing conditions using MRSC initially adjusted to pH 4·0. This sublethal pH, which represents a 1·5 pH drop on average compared with harvesting conditions, was chosen because it allowed strains to resume growth after the initial acid shock, which was not the case with lower pH values as strains could, in some instances, maintain viability, but were unable to grow (results not shown). The shocked cells were cultured in stressing conditions for 4·5 h (strain LC 804), 5·5 h (strain CIP A159) and 7 h (strain CECT 4185), which enabled the harvesting of all cells at the early-stationary phase, as in control conditions (data not shown). As protein expression is growth-phase dependent, having cells in a comparable physiological state was in fact key in this investigation. During this low-pH challenge, a growth rate decrease was observed for the three strains (LC 804, 19%; CIP A159, 37%; and CECT 4185, 56%), as compared to standard conditions, with broth pH at harvesting time of 3·0, 3·2 and 3·4, respectively, confirming the differential acid resistance of the strains. Analysis of changes in protein expression during acid stress exposure was based on the 17 proteins previously reported to play a role in acid resistance. Figure 1 shows representative 2-D patterns of the three selected strains with different acid resistance phenotypes when cultured in stressing conditions. Overall, these patterns were relatively different compared to those obtained in standard conditions and suggested quantitative change for most of the protein spots observed. Table 2 reports changes in spot intensities following exposure to acid stress conditions. Thirteen of the 17 proteins linked to acid resistance in previous studies appeared to respond to acid shock (absolute value of fold-change factor r > 1·5): three proteins (GrpE, spot 4; MetE, spot 88; and RspB, spot 102) were induced, three others (Pta, spot 43; PurA, spot 63; and MsmK1, spot 73) were repressed and could not be detected after acid challenge, while the seven remaining proteins (LdhL, spots 5, 74 and 80; GuaB, spots 33 and 66; Gap, spots 46, 61 and 65; Pbg5, spot 55; ClpL, spots 62 and 96; LysS, spots 79 and 94; and Xpk2, spots 85, 86 and 87) displayed modifications in expression levels that depended on the considered strain or isoform. The expression levels of four proteins (RfbB, spots 14 and 51; RfbC, spot 42; Pmi, spots 49 and 70; and FabF, spot 69) were however not impacted by the acid challenge (|r| ≤ 1·5).
Two-DE and MS analyses were applied to the investigation of key proteins in the resistance of lactobacilli to gastric acidity, a major factor when it comes to the survival of probiotics to the GI tract. Despite known limitations, 2-DE has proven to be a valuable tool in the proteomic phenotyping of bacterial strains (Enroth et al. 2000; Dumas et al. 2008; Wang et al. 2010). With regard to probiotic research, this technique has been used to explore adhesion capacities of Lact. plantarum (Izquierdo et al. 2009) and Bifidobacterium longum (Aires et al. 2010), as well as bile tolerance properties of Lact. plantarum (Hamon et al. 2011) and Lact. casei (Hamon et al. 2012). By extending the use of this approach to study another probiotic trait, the present work represents a further step towards the identification of bacterial biomarkers for each particular probiotic feature.
An in vitro test that mimics the physiological gastric environment was used to assess acid resistance of ten Lact. plantarum strains. This allowed the direct comparison of bacterial strains, which is difficult to achieve in in vivo experiments. Yet, this model also presents some limitations. On the one hand, plate counting only determines the cells' ability to be cultured and underestimates the real number of viable bacteria (Lahtinen et al. 2006). On the other hand, this approach does not take into account the fact that bacteria are usually ingested along with a food matrix whose buffering capacity may ensure a substantial protective effect, particularly in dairy products (Marteau et al. 1997). Despite these limitations, comparative studies like the one presented here give a preliminary estimate of the strains' relative resistance to gastric conditions (Izquierdo et al. 2008). In our case, significant variations in acid resistance were observed between strains, which is in accordance with previous reports showing a strain-specific behaviour of lactobacilli with regard to pH (Jacobsen et al. 1999; Succi et al. 2005). Interestingly, the probiotic Lact. plantarum 299V displayed an intermediate level of acid resistance as compared to the other strains with no reported probiotic activity.
In an attempt to explain the observed differences in acid resistance capacity between three Lact. plantarum strains (LC 804, resistant; CIP A159, intermediate; and CECT 4185, sensitive), potential proteomic markers were investigated among their constitutive proteins. Such proteomic approach allows discriminating strains with similar gene content on the contrary to gene-based methods, which focus on flexible gene pool, as applied in a recent study screening the ability of lactic acid bacteria to survive passage trough GI tract (Turpin et al. 2011). With this regard, the use of sequenced Lact. plantarum strains for protein sequence alignment and identification was not compulsory in this study.
The differentially expressed proteins between strains all appeared to be encoded by highly conserved genes in the Lact. plantarum species. As their relative abundance is likely to be assessed for any Lact. plantarum strain, these core-genome proteins represent good marker candidates. These proteins covered a wide range of biochemical functions, most of which did not seem to be linked to acid resistance. Particular interest was in differentially expressed proteins with a reported putative involvement in acid resistance of other lactic acid bacteria. This led to the identification of 17 proteins likely to be implicated in acid resistance of the selected strains. Eleven of these proteins were more abundant in the constitutive proteome of the LC 804 strain and may account for its high acid resistance. They could therefore constitute an inherent and characteristic proteomic profile that is indicative of intrinsic acid resistance. Proteins of interest included molecular chaperones GrpE (spot 4) and ClpL (spot 96), as well as FabF (spot 69), respectively, involved in cell protection activities (De Angelis et al. 2001; Frees et al. 2007) and in the modulation of membrane composition (de Mendoza et al. 1983), two of the commonly used lactobacillial strategies to resist acid stress (Lebeer et al. 2008). The eight other proteins of this group are key components of central metabolism, which suggests a major role of the cell's general physiology in acid resistance.
In contrast, six additional proteins seemed to play a minor or even a counterproductive role in the acid resistance of Lact. plantarum, as they were underexpressed in the resistant phenotype. Interestingly, two of them (ClpL, spot 62; and GuaB, spot 33) are isoforms of proteins that are overexpressed in the resistant phenotype, suggesting an unequal contribution of these isoforms to protein activity and to the subsequent Lact. plantarum resistance to acidity.
To confirm the putative involvement of the 17 proteins of interest in the acid resistance process and get an overview on how acidity affects their levels of expression, proteomic analysis of strains response to acid stress was carried out. Thirteen proteins appeared to be directly implicated in acid stress adaptation, as their expression was significantly affected by exposure to a low pH (P < 0·05). Among these, GrpE (spot 4), MetE (spot 88) and RpsB (spot 102) were induced following acid challenge, which is consistent with previous studies (De Angelis et al. 2001; Budin-Verneuil et al. 2007; Sanchez et al. 2007a). As they also were more abundant in the constitutive proteome of the resistant phenotype, these proteins may serve as potential markers of acid resistance in Lact. plantarum.
Three other proteins (MsmK1, spot 73; Pta, spot 43; and PurA, spot 63) were repressed in response to low-pH exposure. In addition, Pta and PurA were found in lower amount in the constitutive proteome of the resistant phenotype. This suggests that both proteins could impart bacterial sensitivity to acid stress, as shown for Pta in Escherichia coli (Wolfe 2005). Accordingly, Pta and PurA may represent putative makers of acid sensitivity in Lact. plantarum. This is however not the case for Msk1 whose relative abundance was higher in the inherent proteome of the resistant phenotype.
The seven remaining proteins involved in acid stress adaptation (ClpL, Gap, GuaB, LysS, LdhL, Pbg5 and Xpk2) showed modifications in expression levels that varied according to the considered strain or isoform. This unequal regulation suggests that different strategies may exist in the acid resistance process of Lact. plantarum species, some strains favouring certain specific pathways while others downplaying them, as previously observed in the bile tolerance response of Lact. casei and Lact. plantarum (Hamon et al. 2011, 2012).
Finally, four of the 17 proteins of interest took no part in the acid stress response of Lact. plantarum, because their expression was not affected by exposure to low pH. These include three proteins involved in membrane (FabF, spot 69) and cell wall (RfbB, spots 14 and 51; and RfbC, spot 42) biogenesis, which suggests a minor role of cell envelope modification pathways in the defence mechanisms of Lact. plantarum against acid shock. Yet, these proteins showed constitutive differences in their expression levels between the three phenotypes, which is in accordance with observations for Lactococcus lactis MG1363 and its acid-resistant derivatives (Budin-Verneuil et al. 2007). They could therefore represent potential markers of constitutive acid resistance in Lact. plantarum in relation to particular membrane and cell wall compositions.
With regard to acid resistance factors, proton translocating (H+)-ATPase is probably what first comes to mind, as it plays a major role in the maintenance of intracellular pH homeostasis (Hutkins and Nannen 1993; Cotter and Hill 2003). However, none of the putative determinants of acid resistance identified in this study refer to (H+)-ATPase activity. In fact, it appears that Lact. plantarum, like many other lactic acid bacteria, does not utilize these enzymes to maintain a near-neutral internal pH in response to acid stress. Instead, it allows the internal pH to decrease as the outer pH falls, maintaining a relatively constant pH gradient (McDonald et al. 1990). This bacterial feature is believed to reduce the energy demand for proton translocation through the (H+)-ATPase (Siegumfeldt et al. 2000).
This work used comparative proteomics to analyse cell-free protein extracts from three Lact. plantarum strains with different acid resistance levels. This approach revealed potential proteomic markers of acid resistance (GrpE, MetE, RpsB) and sensitivity (Pta and PurA). Several proteins involved in cell envelope biogenesis (FabF, RfbB and RfbC) could also be key determinants of constitutive acid tolerance.
Yet, when these results are compared with previous transcriptomic data, very few overlapping could be found. For instance, none of the proteins/genes putatively involved in acid resistance of Lact. plantarum as identified here are mentioned in the transcriptomic studies on Lact. plantarum WCFS1 with regard to GI tract conditions (Bron et al. 2004; van Bokhorst-van de Veen et al. 2012), while only three of them (3-oxoacyl-synthase II, lp_1673; adenylosuccinate synthase, lp_3270; and phosphoketolase, lp_3551) appear in a third study on lactic acid stress (Pieterse et al. 2005). The later genes are neither induced nor repressed, whether under osmotic stress, acidic pH or lactate conditions, except when lactic acid is used at low or high absolute growth rates of the culture (Pieterse et al. 2005). Actually, transcriptomic and proteomic data often differ when it comes to the identification of a specific pattern related to a particular bacterial feature. For this reason, a comparative assessment of our proteomic results with previous transcriptomic studies to confirm the true involvement of the putative markers identified has proved inconclusive. Such discrepancy has also been shown in a previous study focusing on the bile stress response of Lactobacillus rhamnosus GG (Koskenniemi et al. 2011). All conditions being equal, the authors found that, when the strain was grown in the presence of 0·2% oxgall, the transcript levels of 316 genes changed significantly and 42 proteins were differentially abundant, with only 14 of these proteins displaying abundance changes correlated with transcriptome level changes (Koskenniemi et al. 2011).
Finally, this paper completes two previous Lact. plantarum studies, which identified putative biomarkers of bile tolerance and adhesion capacity, using similar approaches. These markers, together with those identified here, may serve for the screening of strains with the best ability to survive the GI tract conditions and to adhere to the intestinal mucosa, a prerequisite for cells to exert beneficial effects in the host organism. Future efforts will aim at extending the use of this approach to study specific probiotic effects.
This work was supported by the ‘Ministère de l'Enseignement Supérieur et de la Recherche’ and by the “Ministère de l'Agriculture et de la Pêche” through the “Unité Mixte Technologique 06.03: Méthodes analytiques et nutrimarqueurs”.
The authors declare no conflict of interest.