The B. abortus 104-M strain was obtained from the Lanzhou Institute of Biological Products in China (Yu et al., 2015). The cells were subjected to a control treatment [grown on tryptic soy broth (TSB), condition #1] and eight different stress treatments as previously described, with some modifications (Teixeira-Gomes et al., 2000; Al-Dahouk et al., 2008, 2009, 2013; Lamontagne et al., 2010). These included: (i) seven single-stress conditions: #2 serum stress [addition of 10% serum (obtained from healthy volunteers after informed consent; stored at −20°C)]; #3 nutrient starvation stress (grown in Sauton's glycerol medium); #4 physical/chemical stress [grown in an acidic, high-temperature, hyperhaline, and high osmotic pressure condition (pH 4.5, 45°C, NaCl 1 mol/L final concentration, sorbitol 1 mol/L final concentration)]; #5 peroxide/nitric oxide stress (addition of 50 mM H₂O₂ and 5 mM DETA-NO, final concentrations); #6 oxygen deficiency stress (incubation sealed); #7 iron-limited stress [addition of 50 μM final concentration of the Fe²⁺-chelator 2,2′-dipyridyl (DIP; Sigma-Aldrich, Shanghai, China)]; #8 antibacterial stress (addition of polymyxin B to 100 μg/mL final concentration); and (ii) a multi-stress condition, consisting of a combination of conditions #2 to #8 (#9, see Table 1 for details).

The cells were first cultured in TSB with continuous shaking (200 rpm) at 37°C for approximately 24 h until mid-log phase (OD_600nm = 1.0). The cells were harvested by concentrating the solution, and were then resuspended in the eight different growth conditions listed above and incubated for 3 h in a shaking incubator. Cultures were then serially diluted and plated on tryptic soy agar to determine their viability post-challenge. The survival percentage of the 3 h post-stress challenge was calculated by dividing the number of colony-forming units obtained from each stress treatment to that obtained from the control treatment, multiplied by 100. The survival experiments were performed at least three times for each treatment. All experiments involving live B. abortus 104-M were conducted in BSL-2 labs in line with health and safety guidelines.

The protein samples were prepared as described previously (Zai et al., 2017). Briefly, cells cultured in each condition were harvested by centrifugation (7,000 × g for 15 min), and then washed three times with phosphate-buffered saline (PBS). The bacterial cells were resuspended in lysis buffer and disrupted by ultrasonication (25% amplitude, 15 min at 0°C). The resultant suspension was centrifuged (40,000 × g for 30 min) and the protein concentrations in the collected supernatants were measured using a BCA (bicinchoninic acid) protein assay kit (Thermo Fisher Scientific, Waltham, USA). The cell protein extracts were reduced in 1 mM dithiothreitol (25°C for 1 h) and then alkylated in 5.5 mM iodoacetamide (25°C for 1 h, in the dark). Sequencing-grade trypsin (Promega) was added to a final ratio of 1:50 (V:V) and the proteins were digested in solution overnight at 37°C.

All experiments were performed on an LTQ Q-Exactive HF mass spectrometer (Thermo Scientific, USA) coupled online with a nano-HPLC (Ultimate 3000, Thermo Scientific) (Scheltema et al., 2014). The peptides were loaded onto a trap column (C18, 3 μm particles, 100 μm × 2 cm) and separated on EASY-Spray columns (C18, 1.9 μm particles, 15 μm × 12 cm) with trapping at a flow rate of 600 nL/min (Kentache et al., 2017). The mobile phase A was 0.1% formic acid in water and the mobile phase B was 0.1% formic acid in acetonitrile. The peptides were eluted using a gradient (6–95% mobile phase B) during a 195-min LC run and then sprayed directly into the MS instrument.

The mass spectrometer was operated using the data-dependent top-15 method with automatic switching between MS and MS/MS scans (Kalayou et al., 2016). Full MS scans were acquired at a resolution of 120,000, with an automatic gain control target value of 3 × 10⁶ ions or maximum injection time of 80 ms within the scan range 300–1,400 m/z. Peptide fragmentation was performed by higher energy collision dissociation (HCD) with the normalized collision energy set to 27 (Tutturen et al., 2014). The 15 highest-intensity ions were then selected for the collision-induced fragmentation at a resolution R = 15,000, an automatic gain control target value of 5 × 10⁴ ions, or maximum fragment accumulation time of 45 ms. After the fragmentation event, dynamic exclusion of precursor ion masses for 12 s was used to avoid the repeated fragmentation of peaks. We excluded precursor ions with single, unassigned, or ≥ seven charge states from the fragmentation selection.

Protein identification was performed by submitting raw data files to Proteome Discoverer software (Thermo Scientific, USA, v. 1.2). The MS/MS searches were performed using the SEQUEST (v. 28) algorithm against a database constructed from the UniProt entries for B. abortus 104-M (taxonomy: 1210454), which contained 3,072 protein sequences. The search parameters included specific digestion with trypsin with up to two missed cleavages allowed; carbamidomethylation (C) on cysteine was set as a fixed modification; and oxidation (M) on methylene and acetyl (protein-N term) on asparagine & glutamine were applied as variable modifications. The initial allowed mass deviation of the precursor ion was set to 15 ppm, and the allowed value for the fragment mass was set to 0.02 Da. Protein identifications that contained at least two identified peptides were accepted with a false discovery rate (FDR) less than 1.0% (Zai et al., 2017).

For protein quantification, a label-free experiment was performed as previously described (Pettersen et al., 2016; Schmidt et al., 2016). Briefly, raw data were imported into Proteome Discover 1.4 following the MS analysis. Protein abundance of was calculated on the basis of label-free quantitation intensity [LFQ]. For comparison, the protein abundance in the control group was set as a reference and the protein abundance in the other eight stress groups were aligned. Proteins identified in at least two out of nine groups were considered for label-free quantification. Those proteins exhibiting fold change >1.5 (P-value ≤ 0.05) between the treatment and the control were deemed up-regulated or down-regulated respectively. The MS proteomics data have been deposited in the ProteomeXchange Consortium via the jPOSTrepo (Japan ProteOme STandard Repository) with the data set identifier PXD007548 (Vizcaino et al., 2014; Okuda et al., 2017).

The calculation of protein molecular masses, pI, and peptide grand average of hydropathicity (GRAVY) values were carried out using the ProtParam tool from the ExPASy toolbox (Gasteiger et al., 2003). Protein transmembrane helices were predicted using TMHMM 2.0 (Krogh et al., 2001). The protein functions were assigned by the Clusters of Orthologous Groups (COG) database (Galperin et al., 2015). The pathways of proteins were analyzed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Kanehisa et al., 2017). Possible interactions between identified proteins were tested using the STRING tool (Search Tool for the Retrieval of Interacting Genes/Proteins) (Szklarczyk et al., 2017). Heatmaps of the proteins were generated using the versatile matrix visualization and analysis software Morpheus, available from the Broad Institute.

The proteomes of bacteria grown under the nine groups were investigated. For each condition, two biological replicates and two technical replicates were sampled, resulting in a total of four samples per condition for LC-MS/MS. Protein identifications should contain at least two identified peptides with a FDR less than 1.0%. Significant label-free changes in proteins were determined using the statistical analysis-based variance (ANOVA) test, which was performed on the protein LFQ values. Additionally, principal coordinate analysis (PCoA) was used to visualize the correlations among the nine groups. The Pearson's r correlation coefficient between the nine groups was also visualized by means of a correlation matrix. The expression patterns of the proteins grown under the nine conditions were presented in a heatmap with hierarchical clustering performed using a Euclidean distance metric and the average linkage method (Pettersen et al., 2016).

Brucella spp. are facultative intracellular bacteria. During the invasion of host tissue, Brucella are subjected to various harsh environmental conditions including nutrient limitation, low pH, antimicrobial defenses, and extreme levels of ROS (Roop et al., 2009; Barbier et al., 2011; Olsen and Palmer, 2014). Correspondingly, Brucella is well-equipped from both a physiological and metabolic perspective to adopt to environmental stresses (Roop et al., 2009). Analysis of the differential proteome expression patterns of Brucella under stress should improve our understanding regarding its adaptation and pathogenesis. Thus, we chose a multiple-environmental-stress strategy to reveal the global metabolic adaptations of B. abortus to intravacuolar environmental conditions. These conditions included: (i) a control condition (growth on TSB, condition #1); (ii) seven single-stress conditions: (conditions #2–8) and (iii) a multi-stress condition (#9, see Table 1 for details). The multi-stress condition constituted a combination of each single-stress condition, and may more realistically simulate the conditions experienced by Brucella during infection of the host. We investigated the survival rate and differential protein expression of B. abortus in the single-stress and multi-stress conditions compared with the control condition (Figure 1). The survival rates of B. abortus under the different stress conditions ranged from 3.17 to 73.17%. Nutrient starvation, physical/chemical starvation, peroxide/NO starvation, and anaerobic starvation resulted in low survival rates (Table 1). The multi-stress condition resulted in the lowest survival rates, and may constitute a more accurate reflection of the in vivo conditions of B. abortus under intracellular infection.

The goal of this study was to achieve substantial coverage of the B. abortus proteome under in vitro stress conditions. Accordingly, we applied a proteomics approach on whole cell lysates prepared from B. abortus grown in the nine groups described above. LC-MS/MS analysis of the resulting peptide mixtures generated 1.1 million spectra. The acquired raw MS data files were then analyzed and the spectral files matched to 27,076 unique peptides with an FDR of 1.0%. We mapped the unique peptides to the B. abortus 104-M UniProt database (3,072 protein sequences), and only proteins that were identified by at least two unique peptides were confirmed (Zai et al., 2017). A total of 2,289 proteins were identified by two repeats, which represents approximately 74.5% coverage of the predicted proteome (Figure 2A). Proteins identified in at least two out of nine groups (2,272) were considered for label-free quantification (Figure 2B). The distribution of the identified proteins with respect to pI, molecular weight, hydrophobicity, and transmembrane regions was consistent with the annotated proteins (Figure S1). Among the 2,289 identified proteins, 1,570 were annotated in the COG database. Almost all of the pivotal categories for Brucella were identified, suggesting good coverage and representation of the genomic content of B. abortus by the proteome. Table S1 lists all identified proteins with their accession numbers and calculated score.

We used high-resolution MS to quantitatively describe the protein profiles of B. abortus under stress conditions, and compared them with the expressed proteome in the control. Protein quantification allowed the characterization of proteins that were differentially expressed across the nine groups. Here, we obtained LFQ intensities for 2,272 quantified proteins. Their quantitative levels covered a 5-log₁₀ dynamic range. The density plot of the log2 ratio between the stress group and control group closely matched a normal distribution, which indicated that the experimental procedure was performed without systematic bias toward the different samples. PCoA analysis indicated good homogeneity of the biological replicates and was able to discriminate nine distinct protein populations (Figure 3A). The Pearson's correlation coefficient r varied between 0.95 and 0.99, indicating that there were differences in the protein levels between the different stress conditions (Figure 3B).

We next explored differences between the protein profiles of B. abortus under the various stress conditions. Hierarchical clustering was performed based on the LFQ intensities of the 2,272 proteins quantified in this study, and indicated distinguishable protein expression profiles between the nine different groups (Figure 3C). For each growth condition, a range of 1,967–2,109 proteins per group were identified (Table 2). About 87.6% of the detected proteins were common to all groups. Several of the identified proteins demonstrated a significant difference in abundance in stress compared to the control condition. The study identified DEPs in each stress condition whose quantitative levels varied by more than 1.5 fold from the respective LFQ intensity means in the control, i.e., 213, 413, 311, 451, 314, 357, 401, and 306 proteins in the eight stress conditions, respectively (Figure S2). Table S2 lists all the DEPs along with their accession numbers and LFQ intensities for each condition.

To understand the functional classification and metabolic pathways that were involved in the environmental stress response, the DEPs in each stress condition were functionally analyzed. All of the enriched KEGG pathways of B. abortus in response to each stress treatment are listed in Table 3. The results suggested that the metabolic adaptations of B. abortus to each environmental stress condition varied.

Brucella can persist for several weeks in the blood of an intraperitoneally-infected host (Vitry et al., 2014). Considering the role that serum stimulation has in intraperitoneally-infected hosts of Brucella, serum stress was chosen as single-stress condition #2. In this treatment, we discovered 125 and 88 up-regulated and down-regulated proteins, respectively, compared with the control. As shown in Figure S3A, the main enriched KEGG pathways for regulated proteins included biosynthesis of siderophore group non-ribosomal peptides, ABC transporters, synthesis and degradation of ketone bodies, thiamine metabolism, and two-component system. This suggests that Brucella may enhance its iron acquisition through the regulation of ABC transporters and biosynthesis of siderophores in order to adapt to the serum stress.

During the invasion of host tissue, a major hurdle in the infection of a host cell by Brucella is the lack of nutrients within the phagosome (Essenberg et al., 2002; Hanna et al., 2013; Barbier et al., 2017). Considering the role that nutrient starvation has in the intracellular replication of Brucella, nutrient stress was set as single-stress condition #3. Of the DEPs detected under this condition, 139 and 274 proteins were respectively up-regulated or down-regulated compared with the control. The primary enriched KEGG pathways for regulated proteins included two-component system, glycerophospholipid metabolism, ABC transporters, nitrogen metabolism, and porphyrin and chlorophyll metabolism (Figure S3B). This implies that Brucella may decrease its energy usage and secondary metabolite biosynthesis through the regulation of glycerophospholipid metabolism, nitrogen metabolism, and porphyrin and chlorophyll metabolism in response to nutrient starvation stress.

One major mechanism of Brucella pathogenesis is the capacity to survive in the acidic, high-temperature, hyperhaline, and high osmotic pressure environment inside macrophages (Detilleux et al., 1991; Liu et al., 2015). Given that physical/chemical stimulation influences the intracellular replication of Brucella, physical/chemical stress was set as single-stress condition #4. We discovered 120 and 191 up-regulated and down-regulated proteins present in the physical/chemical stress condition compared with the control. ABC transporters, thiamine metabolism, biosynthesis of siderophore group nonribosomal peptides, glycerophospholipid metabolism, and two-component system constituted the main enriched KEGG pathways (Figure S3C). Brucella may therefore decrease its energy usage and secondary metabolite biosynthesis through the regulation of glycerophospholipid metabolism and thiamine metabolism in order to the physical/chemical stress.

The exogenous production of ROS such as O²⁻ and H₂O₂ by the host immune system has also been shown to be important for the survival of Brucella (Jimenez De Bagues et al., 2007). Nitric oxide produced by the macrophages were also crucial for phagocytes to control the intracellular replication of Brucella (Roop et al., 2009; Ronneau et al., 2016). Peroxide/nitric oxide stress was therefore set as single-stress condition #5. Of the DEPs detected in this treatment, 235 and 216 proteins were up-regulated or down-regulated respectively compared with the control. The main enriched KEGG pathways for regulated proteins included nitrogen metabolism, ABC transporters, biosynthesis of siderophore group nonribosomal peptides, thiamine metabolism, and two-component system (Figure S3D). The result suggests that Brucella may decrease its amino acid usage and secondary metabolite biosynthesis through the regulation of nitrogen metabolism and thiamine metabolism, meanwhile enhancing its iron acquisition through regulation of the two-component system and biosynthesis of siderophores in response to peroxide/NO stress.

The presence of oxygen has positive effects on the levels of proteins that are functional in aerobic respiration and purine metabolism, while low access to oxygen results in the induction of enzymes of mixed-acid fermentation and gluconate metabolism (James et al., 1995). Considering the role that oxygen deficiency plays in the intracellular replication of Brucella, oxygen deficiency stress was set as single-stress condition #6. In this treatment we discovered that 165 and 149 proteins were up-regulated or down-regulated respectively compared with the control. The main enriched KEGG pathways for the regulated proteins were two-component system, flagellar assembly, glycerophospholipid metabolism, quorum sensing, and ABC transporters (Figure S3E). Thus, Brucella may decrease its energy usage and virulence through the regulation of glycerophospholipid metabolism, two-component system, flagellar assembly and quorum sensing to manage oxygen deficiency stress.

Iron is an essential element for Brucella and is pivotal in host-pathogen interactions (Eskra et al., 2012; Roop, 2012). In the host, free iron levels are extremely low, resulting in iron limitation being a crucial stress (Braun, 2001). Iron-limitation stress was thus set as single-stress condition #7. We found that 169 and 188 proteins were respectively up-regulated or down-regulated in the iron-limited stress condition compared with the control. As indicated in Figure S3F, the main enriched KEGG pathways included oxidative phosphorylation, two-component system, biosynthesis of siderophore group nonribosomal peptides, nitrogen metabolism, and citrate cycle (TCA cycle). This result suggests that Brucella may decrease its energy usage through the regulation of oxidative phosphorylation, nitrogen metabolism and the TCA cycle, while enhancing its iron acquisition through the regulation of siderophore biosynthesis to adapt to the stress resulting from limited iron.

Antimicrobial peptides can limit the colonization of bacterial during infection in the innate defense, and therefore it is likely that Brucella encounters antimicrobial peptides within host microenvironments during infection (Martinez De Tejada et al., 1995). These peptides may be involved in environmental signaling that triggers changes in bacterial gene expression. Considering the role that antimicrobial peptide stimulation has in the intracellular replication of Brucella, antibacterial stress was set as single-stress condition #8. Of the DEPs in this treatment, 146 and 255 proteins were up-regulated or down-regulated respectively in comparison with the control. The main enriched KEGG pathways included two-component system, nitrogen metabolism, oxidative phosphorylation, biosynthesis of siderophore group nonribosomal peptides, and base excision repair (Figure S3G). This suggests that Brucella may decrease its energy usage through the regulation of oxidative phosphorylation and nitrogen metabolism, while enhancing its antimicrobial peptide resistance through regulation of the two-component system and base excision repair in response to antibacterial stress.

During the invasion of host tissue, Brucella experiences several stresses simultaneously, including nutrient limitation, low pH, antimicrobial defenses, and extreme ROS levels from the host immune response. Thus, the combination of stress conditions 2 to 8 was set as the multi-stress condition #9 that may better simulate the environments that Brucella may occur during host infection. We discovered that 116 and 190 proteins were respectively up-regulated or down-regulated in the multi-stress condition compared with the control. The primary enriched KEGG pathways included two-component system, nitrogen metabolism, thiamine metabolism, biosynthesis of siderophore group nonribosomal peptides, and ABC transporters (Figure S3H). Brucella may decrease its energy usage and secondary metabolite biosynthesis via the regulation of nitrogen metabolism and thiamine metabolism, while enhancing its iron acquisition and antimicrobial peptide resistance by regulating the two-component system and siderophore biosynthesis in order to cope with multiple stresses.

Compared with the single-stress condition, the enriched pathways of B. abortus in response to multi-stress conditions were more extensive and covered the primary enriched pathways of each single-stress condition. For example, the two-component system pathway enriched in the multi-stress condition was also enriched in all eight of the single-stress conditions, indicating its significance in the metabolic adaptation to multiple environmental stresses. The nitrogen metabolism pathway was enriched in the multi-stress condition and also in numerous single-stress conditions, including nutrient starvation stress, physical/chemical stress, peroxide/nitric oxide stress, iron-limited stress, and antibacterial stress. The DEPs included in this enriched pathway were similar in their regulatory patterns in both the single-stress and multi-stress treatments. The results suggested that the metabolic adaptations of B. abortus to multiple stresses constitute the synthesis of each environmental stress condition, and that the multiple stress treatment may better simulate the metabolic adaptations of Brucella during host infection.

The functional analysis above indicated that B. abortus exhibited different metabolic adaptations to each environmental stress condition. To better understand the primary metabolic adaptations of B. abortus to various stress conditions, we summed all the DEPs in each stress condition, resulting in a total of 1,221 proteins that were differentially expressed in at least one stress condition compared with the control. The protein functions of all 1,221 DEPs in the eight stress conditions were assigned by the COG database. We identified 20 different COG phylogenetic protein groups (Figure 4), with the highest represented subsets in category E (amino acid transport and metabolism, 104 proteins), category K (transcription, 93 proteins), category C (energy production and conversion, 90 proteins), category P (inorganic ion transport and metabolism, 87 proteins), and category G (carbohydrate transport and metabolism, 63 proteins). Furthermore, 114 proteins were assigned only a putative function (category R), while 127 proteins remained without an allocated biological role (category S). The COG functional analysis suggested that energy usage-related categories (E, C, and G) and the iron acquisition-related category (P) were involved in the main metabolic adaptations of B. abortus to various stress conditions.

A putative protein-protein interaction network was then constructed from all high-confidence B. abortus protein interaction pairs that were matched with the 1,221 DEPs using the STRING tool (Szklarczyk et al., 2015). The KEGG pathway analysis showed a significant enrichment of DEPs (P-value ≤ 0.05). All 1,221 DEPs in the eight stress conditions were primarily enriched in oxidative phosphorylation, ABC transporters, two-component systems, biosynthesis of secondary metabolites, porphyrin and chlorophyll metabolism, glycerol phospholipid metabolism, the TCA cycle, thiamine metabolism, nitrogen metabolism, or were associated with carbon metabolism (Figure 5). The KEGG pathway analysis suggested that the energy usage-related pathways and regulatory mechanism pathways constituted the primary pathways involved in the metabolic adaptation of B. abortus to various stress conditions. Table S3 lists the DEPs that were related to the main metabolic changes of B. abortus in response to stress treatments.

Oxidative phosphorylation is the process in which ATP is formed through using enzymes to oxidize nutrients (Dimroth et al., 2000). TCA cycle is a series of chemical reactions that produces ATP through the oxidation of acetyl-CoA. In this study, several proteins involved in oxidative phosphorylation and the TCA cycle, such as those associated with the NADH dehydrogenase (NDH) family, succinate dehydrogenase (SDH), and cytochrome c oxidase CcO, were abundantly down-regulated in response to both the single-stress and multiple-stress treatments (Figures 6A,C). A major hurdle in the infection of a host cell by Brucella is the lack of nutrients within the phagosome (Barbier et al., 2017). These results suggest that immediately after phagocytosis, Brucella reduces energy metabolism via the TCA cycle as the available extracellular nutrients decrease.

Two-component systems are adapted to respond to a wide variety of stress, including nutrients, quorum signals, antibiotics, temperature, pH, and so forth (Mascher et al., 2006). In this study, several proteins involved in two component system-controlled metabolic pathways, such as chromosomal replication initiator protein dnaA and nitrate reductase Nar, were decreased in abundance. Two-component system sensor histidine kinase PhoR/GlnL, hybrid sensor histidine kinase/response regulator ChvG, and transcriptional regulator RegB were up-regulated in response to several single-stress and multiple environmental stresses (Figure 6B). During the invasion of host tissue, Brucella must survive under several severe stresses through generating a suitable adaptive response to various signals (Viadas et al., 2010). Here, we found that the two-component system proteins were regulated in response to stress, which is essential for the persistence of Brucella within stressed environments in the host organism.

Cationic antimicrobial peptides (CAMPs) are crucial for the host defense against invasive bacterial infection (Alegado and Tan, 2008). The resistance of pathogenic bacteria toward antimicrobial peptides may also account for their virulence. In this study, several proteins involved in CAMPs resistance, such as the serine peptidase DegP and Hemolysin D (NL70_15120), were increased in abundance in our study. Transporter TolC and N-acetylmuramoyl-L-alanine amidase (NL70_04450) were down-regulated in response to several single environmental stresses, especially antibacterial stress (Figure 6E). These results indicated that the adopted mechanism of Brucella to resist host antimicrobials is important for persistent infection.

Siderophores are an important group of structurally diverse natural products that chelate iron and are important in the acquisition of the essential trace element iron by most microorganisms (Miethke and Marahiel, 2007). In the host, free iron levels are extremely low as the metal is largely bound to proteins. To overcome iron limitation, some bacteria and fungi produce siderophores. In this study, several proteins involved in the biosynthesis of siderophore group nonribosomal peptides, such as enterobactin biosynthetic enzymes EntA/B/E/F, were found to be up-regulated particularly in response to serum stress and iron-limitation stress (Figure 6F). These results indicated that Brucella has evolved strategies to overcome iron limitation and to compete with the iron sequestration immune mechanisms of the host.

We also observed specific variations in the amounts of multiple proteins involved in nitrogen metabolism, thiamine metabolism, and purine metabolism. In this study, nitrate reductase (Nar), nitrite reductase (NirK), and thiamine pyrophosphokinase (ThiN/E/G), which play essential roles in amino acid metabolism, and several enzymes participating in glycine, serine, and threonine metabolism, were relatively significantly differentially expressed in the different conditions (Figure 6D). Serine peptidase HtrA and heat shock protein Hsp20 were observed to increase considerably under both the single-stress and multiple-stress treatments. The Cu-Zu superoxide dismutase SodC was found to increase during oxidative stress and multiple-stress treatments, but remained unchanged under the physical/chemical stress conditions of heat shock or acidic pH. Other proteins such as exopolyphosphatase (NL70_13665), ribonuclease H (RnhA), single-stranded DNA exonuclease (RecJ), and urea amidohydrolase (UreA/B/C) were also reduced in at least one of the stress conditions in our study.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.