Genetic and Physiological Responses of Bifidobacterium animalis subsp. lactis to Hydrogen Peroxide Stress

Abstract

Consumer interest in probiotic bifidobacteria is increasing, but industry efforts to secure high cell viability in foods is undermined by these anaerobes' sensitivity to oxidative stress. To address this limitation, we investigated genetic and physiological responses of two fully sequenced Bifidobacterium animalis subsp. lactis strains, BL-04 and DSM 10140, to hydrogen peroxide (H₂O₂) stress. Although the genome sequences for these strains are highly clonal, prior work showed that they differ in both intrinsic and inducible H₂O₂ resistance. Transcriptome analysis of early-stationary-phase cells exposed to a sublethal H₂O₂ concentration detected significant (P < 0.05) changes in expression of 138 genes in strain BL-04 after 5 min and 27 genes after 20 min. Surprisingly, no significant changes in gene expression were detected in DSM 10140 at either time. Genomic data suggested that differences in H₂O₂ stress resistance might be due to a mutation in a BL-04 gene encoding long-chain fatty acid coenzyme A (CoA) ligase. To explore this possibility, membrane fatty acids were isolated and analyzed by gas chromatography-mass spectrometry (GC-MS). Results confirmed that the strains had significantly different lipid profiles: the BL-04 membrane contained higher percentages of C_14:0 and C_16:0 and lower percentages of C_18:1n9. Alteration of the DSM 10140 membrane lipid composition using modified growth medium to more closely mimic that of BL-04 yielded cells that showed increased intrinsic resistance to lethal H₂O₂ challenge but did not display an inducible H₂O₂ stress response. The results show that deliberate stress induction or membrane lipid modification can be employed to significantly improve H₂O₂ resistance in B. animalis subsp. lactis strains.

INTRODUCTION

Bifidobacteria are Gram-positive rods of irregular shape with a G+C content of 55 to 67% and are part of the normal gastrointestinal flora in human infants and adults (1, 2). Bifidobacteria have been associated with several health-related benefits, including a decrease in severity of the side effects associated with use of antibiotics, reduced incidence of infection in patients receiving irradiation therapy, decrease in the duration of diarrhea due to various etiologies, reduced frequency of allergic reactions, and alleviation of constipation (3–8). Although no conclusive data regarding a minimal effective dose of probiotics in humans are available, results from clinical trials suggest a direct dose-effect correlation with probiotic efficacy (9, 10). This means that bifidobacteria likely need to be consumed at very high levels (>10⁷ CFU) in bioactive foods to effect a probiotic outcome. At present, yogurt and fermented milks are the most common foods for delivery of probiotic bifidobacteria, but their incorporation into other foods is increasing. A major obstacle to production and storage of bioactive foods containing bifidobacteria is the susceptibility of these cells to oxidative stress. Bifidobacteria are anaerobic and therefore lack common enzymes for detoxification of oxidative free radicals produced in the cell, such as catalase and superoxide dismutase (11, 12, 13). However, previous research in our laboratory has demonstrated variability in the intrinsic and inducible resistance of bifidobacteria strains to hydrogen peroxide (H₂O₂) (14). The purpose of this study was to investigate the transcriptional and physiological responses of 2 closely related strains of Bifidobacterium animalis subsp. lactis, BL-04, a human fecal isolate, and DSM 10140, a strain originally isolated from French yogurt (12), to sublethal H₂O₂ exposure in an industrial growth medium. These strains were chosen based on their current use in industry as probiotics in bioactive foods, the availability of complete genome sequence information for both, and marked differences in their inducible and intrinsic H₂O₂ resistance (12, 14). Specifically, a 20-min exposure to a sublethal concentration (1.25 mM) of H₂O₂ was shown to significantly improve survival (P < 0.05) of B. animalis subsp. lactis BL-04, while survival of strain DSM 10140 was significantly decreased by this treatment (14). Additionally, B. animalis subsp. lactis BL-04 showed 2-fold-higher intrinsic H₂O₂ resistance than DSM 10140 (14).

MATERIALS AND METHODS

Culture conditions.

Bacterial strains were maintained as glycerol freezer stocks at −80°C, and working cultures were prepared by two successive transfers (1% inoculum [vol/vol]) into peptonized milk medium (MP5) (14) with anaerobic incubation at 37°C for 18 h. Batch cultures of each strain were prepared by dilution of the working culture to an absorbance at 600 nm (A₆₀₀) of 1.0 in MP5 medium, inoculated at 1% (vol/vol) into 1 liter of MP5 in a New Brunswick BioFlo III fermentor (New Brunswick Scientific, Edison, NJ), and then incubated at 37°C with an agitation rate of 100 rpm to prevent sedimentation. A gas mixture of 5% CO₂ and 95% N₂ was continuously passed over the headspace of the fermenter to achieve anaerobic conditions, and the pH was maintained at 6.5 by automatic addition of 15% (vol/vol) NH₄OH. The cultures were incubated until the cells reached early stationary-phase (approximately 12 h; ∼log 5.8) (14).

RNA isolation.

Cells from 5-ml samples grown under the conditions indicated above were harvested by centrifugation at 7,500 × g for 10 min. The cell pellets were suspended in 50 ml of prewarmed MP5 medium containing a sublethal H₂O₂ concentration of 1.25 mM and held at 37°C for 5 (T1) or 20 (T2) min (14). Immediately after treatment, 100 ml of RNAprotect bacterial reagent (Qiagen, Inc., Valencia, CA) was added to the cell suspensions to stop transcription and prevent mRNA degradation. A control sample was also prepared, which was not exposed to H₂O₂. Cells in RNAprotect were held at room temperature for 10 min, then collected by centrifugation at 9,500 × g for 10 min, and stored at −20°C until RNA isolation.

Cell pellets were thawed at room temperature and suspended in 900 μl of lysozyme solution (20 mg/ml in Tris-EDTA buffer) that also contained 20 U of mutanolysin (Sigma-Aldrich). Samples were incubated for 30 min at 37°C on a shaker incubator at 240 rpm, after which 20 μl of proteinase K (Omega Bio-Tek Inc., Norcross, GA) (20 mg/ml) was added and the samples were returned to the shaker/incubator for 30 min. The RNA was then isolated using an Aurum total RNA minikit (Bio-Rad, Hercules, CA) following the vendor's recommended procedures. The quantity of recovered RNA was measured with a NanoDrop 8000 spectrophotometer (ThermoFisher Scientific, Waltham, MA), and the quality of the RNA was assayed using an Agilent 2100 bioanalyzer (Agilent Technologies, Inc., Waldbronn, Germany). Samples that had sufficient quantities (>10 μg) of quality RNA were stored at −80°C until needed.

Synthesis and labeling of cDNA.

cDNA was synthesized and labeled as recommended by the Affymetrix (Santa Clara, CA) protocol for prokaryotic target preparation in the GeneChip expression analysis technical manual. The cDNA was fragmented into segments of approximately 50 to 100 bp using DNase I and labeled with GeneChip DNA labeling reagent (Affymetrix, Santa Clara, CA) and terminal deoxynucleotidyl transferase (Promega, Madison, WI). Fragmentation labeling efficiency was measured by gel shift assay.

DNA microarrays.

Sample hybridization was performed at the Center for Integrated Bio-systems at Utah State University against a custom Affymetrix bifidobacterial DNA microarray designed to include 1,761 shared plus unique chromosomal genes predicted to occur in B. animalis subsp. lactis BL-04 and DSM 10140 (12). The only predicted coding sequences not included in the microarray design were redundant transposases and rRNA genes. Hybridization was performed according to the Affymetrix protocol for prokaryotic target hybridization in the GeneChip expression analysis technical manual using a hybridization temperature of 50°C. The DNA microarrays were scanned using the HP GeneArray scanner (Affymetrix, Santa Clara, CA) to generate raw intensity values for each probe.

Statistical analysis of microarray data was performed using Bioconductor (www.bioconductor.org) in the open-source statistical platform R (www.r-project.org). The raw probe data were preprocessed using the RMA-MS method (15) and filtered to include only genes that had a high signal intensity and a low coefficient of variation. To test for differential expression, the preprocessed, filtered data were analyzed using the limma/eBayes method (16). Genes were determined to be significantly differentially expressed if they had a false discovery rate-corrected P value less than 0.05. The significantly differentially expressed genes were grouped according to function and by treatment times and strain.

Microarray validation.

To validate the microarray data, quantitative real-time PCR (RT-PCR) was performed for 6 different genes (Table 1) using cDNA produced after each treatment as described by Smeianov et al. (17). A log-fold change (LFC) was calculated between control and treatment samples and plotted against the LFC calculated from the microarray data. A positive LFC represents upregulation of a particular gene in treated cells versus the control, while a negative LFC reflects gene downregulation.

Table 1.

Target gene oligonucleotide primers for RT-PCR

Protein function (gene ID)	Primer sequence		Amplicon size (bp)	Annealing temp (°C)
	Forward	Reverse
Peroxiredoxin (Balac_0865)	CCGTGTGAAGGCGTCGCAGT	GCTCGGCTCGAGCGTTTCGT	91	61.5
Ribonucleotide reductase (Balac_0326)	CACCACGCTCGCCGAGATCC	TGCTCATCGTGATGCGCCCG	104	61.5
Long-chain acyl-CoA synthetase (Balac_1406)	TCCAGGGCTACGGCCTGACC	CGCCGGTGGGTGAGATACGC	123	61.5
DnaK (Balac_1557)	ACGCCGCTGTCCCTCGGTAT	ACGGCTGGTTGTCTTCGGCG	121	61.5
3-Oxoacyl-ACP reductase (Balac_0317)	AAGCTCGTGCGTGACCTGGC	TGGGGTCGTTCGCGTTCGTG	94	61.5
Multidrug resistance efflux pump (Balac_1405)	TGCGTGGAAACCGGCGACTC	CCGCCCACTTCGTTCTGCGT	149	61.5

Membrane fatty acid analysis.

To determine whether H₂O₂ exposure altered cytoplasmic membrane fatty acid (CMFA) composition, cells were grown in batch culture as described before and treated with a sublethal H₂O₂ concentration of 1.25 mM for 5 (T1) or 20 (T2) min. Cells in 20-ml samples were collected by centrifugation at 5,000 × g for 5 min and then washed twice with phosphate-buffered saline. Membrane fatty acids were then isolated from the pelleted cells according to the protocol of Sasser (18) and identified using gas chromatography as described previously (19). An untreated control sample was also prepared.

To determine the effect of exogenous fatty acids in the growth medium on CMFA composition, cells were grown to early stationary phase in MP5 broth containing 1% Tween 80 (C_18:1n9), 1% Tween 20 (C_12:0), or no exogenous fatty acid substrate. Bacteria in 20-ml samples were collected and analyzed as described above.

Inducible and intrinsic H₂O₂ resistance.

Cells were grown to early stationary phase in MP5 medium with 1% Tween 80 or no added exogenous fatty acids, exposed to a sublethal H₂O₂ concentration of 1.25 mM for 20 or 60 min, and then challenged with a 30-min exposure to a lethal H₂O₂ concentration of 2.55 or 5.25 mM (14). Control cells that received no H₂O₂ treatment were also prepared. Samples were plated on MRS agar containing 0.05% filter-sterilized cysteine after 0 and 30 min and then incubated anaerobically at 37°C for 48 h before enumeration. Results are expressed as percent survival, which is calculated by dividing the log₁₀ CFU/ml of surviving cells after 30 min by the log₁₀ CFU/ml of cells after 0 min (14). The Student t test was used to identify significant differences (P < 0.05) between treatment means (20).

Microarray data accession number.

Microarray hybridization data have been deposited in Gene Expression Omnibus under accession number GSE44382.

RESULTS AND DISCUSSION

Influence of H₂O₂ stress on global gene expression.

To explore the cellular responses of B. animalis subsp. lactis strains to oxidative stress, we analyzed the transcriptional response of strains BL-04 and DSM 10140 after 5 or 20 min sublethal H₂O₂ exposure. B. animalis subsp. lactis BL-04 showed a total of 138 significant (P < 0.05) differentially expressed (DE) genes after a 5-min exposure and 27 DE genes after 20 min (Fig. 1). Among the DE genes detected after 5 or 20 min, 112 (81%) and 22 (82%), respectively, have an assigned function (Fig. 1; also, see Table S1 in the supplemental material). In contrast, strain DSM 10140 showed no statistically significant (P < 0.05) DE genes at either treatment time compared to control cells.

Fig 1.

Numbers of B. animalis subsp. lactis BL-04 genes, grouped according to functional category, that were significantly upregulated (black bars) or downregulated (white bars) after 1.25 mM H₂O₂ exposure for 5 min (A) or 20 min (B).

RT-PCR analysis of 6 selected genes was used to validate microarray data obtained from strain BL-04. As shown in Fig. 2, RT-PCR did not detect any contradictions between the two platforms, and there was a strong positive correlation (r² = 0.83) between the fold change for gene induction or repression predicted from the microarray and the respective values determined by RT-PCR.

Fig 2.

Correlation of fold change values from DNA microarray and real-time quantitative PCR results. Fold change values were obtained for the 6 genes listed in Table 1. Symbols denote expression values from B. animalis subsp. lactis BL-04 cells after 5 min (open diamonds) or 20 min (filled diamonds) exposure to 1.25 mM H₂O₂. The best-fit curve is shown along with the calculated equation and r² value.

Bifidobacterium spp. lack the most common genes associated with oxidative stress defense, such as superoxide dismutase and catalase. However, grouping of DE genes into predicted functional categories showed that exposure of BL-04 to an oxidative stress triggered upregulation of genes involved in the thioredoxin reductase system (Table 2). Under favorable conditions, this system functions with ribonucleoside reductase to use NADPH to reduce the 2′ OH group of ribose for deoxynucleotide production, as well as to maintain cytoplasmic redox for disulfide bond production in proteins (21, 22). During oxidative stress, however, cells can use thioredoxin reductase and peroxiredoxin to direct NADPH toward the removal of oxidative free radicals via the reduction of H₂O₂ and toxic lipid hydroperoxides (23–25). Schell et al. (26) suggested that these enzymes might be one of the primary defense mechanisms against oxidative stress in bifidobacteria, and other research has shown upregulation of thioredoxin, thioredoxin reductase, and peroxiredoxin genes in response to oxygen stress (27, 28). Interestingly, exposure to bile can also produce an oxidative stress response via generation of oxygen free radicals (29), and Sanchez et al. (30) found that bile stress induced a thioredoxin-dependent thiol peroxidase in B. animalis subsp. lactis. Collectively, our results and these prior data confirm that thioredoxin reductase (Balac_0866) and peroxiredoxin (Balac_0865) provide a primary defense mechanism against oxidative stress in B. animalis subsp. lactis.

Table 2.

Differentially regulated genes associated with oxidative stress response of B. animalis subsp. lactis BL-04

Gene ID	Predicted function	Log fold change vs. controla
		T1	T2
Balac_0326	Ribonucleoside diphosphate reductase beta chain (EC 1.17.4.1)	5.39	2.77
Balac_0327	Ribonucleoside diphosphate reductase alpha chain (EC 1.17.4.1)	4.23	NS
Balac_0328	NrdI protein/ribonucleotide reductase stimulatory protein	1.71	NS
Balac_0865	Peroxiredoxin (EC 1.11.1.15)	1.10	NS
Balac_0866	Thioredoxin reductase (EC 1.8.1.9)	1.84	1.19
Balac_0118	Oxidoreductase (EC 1.1.1.-)	1.40	NS
Balac_0120	Vanillate O-demethylase oxidoreductase/ferric reductase	1.59	NS
Balac_0121	Flavodoxin	2.14	NS
Balac_0123	Flavodoxin	1.54	NS
Balac_1314	Anaerobic ribonucleoside-triphosphate reductase-activating protein (EC 1.97.1.4)	1.05	NS
Balac_1315	Anaerobic ribonucleoside-triphosphate reductase (EC 1.17.4.2)	0.72	NS
Balac_0573	NTP pyrophosphohydrolases, including oxidative damage repair enzymes	NS	1.34
Balac_0025	Oxidoreductase	0.56	NS
Balac_1337	MoxR-like ATPase (EC 3.6.3.-)	−0.88	NS
Balac_0086	Penicillin-binding protein	−0.74	NS
Balac_1247	DNA repair protein RecO	−0.97	NS
Balac_1114	RecA protein	0.58	NS
Balac_1212	LexA repressor (EC 3.4.21.88)	0.60	NS
Balac_1437	Multidrug resistance protein B	0.52	−1.22
Balac_1555	DnaJ-class molecular chaperone	NS	−1.34
Balac_1556	GrpE protein	0.47	NS
Balac_0440	Acyl-coenzyme A:6-aminopenicillanic-acid-acyltransferase precursor (EC 2.3.1.-)	NS	−1.63
Balac_0441	Aminopeptidase C (EC 3.4.22.40)	0.76	NS
Balac_0442	Glutamate/gamma-aminobutyrate antiporter	0.74	−1.37
Balac_0443	Carboxypeptidase S1 (EC 3.4.16.6)	1.28	−1.27
Balac_0444	Amino acid permease	0.77	NS
Balac_1501	Sugar kinases, ribokinase family	3.03	2.41
Balac_1502	Tetracycline resistance permease/tetracycline efflux pump/MFS transporter	2.98	2.53
Balac_1503	Inosine-uridine preferring nucleoside hydrolase (EC 3.2.2.1)	2.79	2.54
Balat_0464	5′-Nucleotidase (EC 3.1.3.5)	1.05	NS
Balac_1081	RNase D (EC 3.1.26.3)	1.25	NS
Balac_1597	Raffinose transport system permease protein	3.46	NS
Balac_1598	Raffinose transport system permease protein	3.20	NS
Balac_1599	Raffinose-binding protein	3.12	NS
Balac_1601	α-Galactosidase (EC 3.2.1.22)	2.10	NS
Balac_1567	4-α-Glucanotransferase (EC 2.4.1.25)	3.13	NS
Balac_1568	α-Glucosidase (EC 3.2.1.20)	1.43	NS
Balac_1569	Multiple sugar transport system permease protein MsmG	2.58	NS
Balac_1570	Sugar transport system permease protein	2.55	NS
Balac_1572	Maltose/maltodextrin-binding protein	1.90	NS
Balac_1573	Trehalose-6-phosphate hydrolase (EC 3.2.1.93)	0.99	NS

^aNS, no significant change relative to control cells.

Transcriptome data also showed that ribonucleoside-diphosphate reductase alpha and beta chains (Balac_0326 and Balac_0327) were differentially expressed with a high LFC, as were several other genes (Balac_1501, Balac_1503, Balac_1081, and Balat_0464 [which corresponds to BL-04 gene Balac_0464]) involved in nucleotide turnover (Table 2). These genes encode proteins used for deoxynucleoside triphosphate (dNTP) production and to hydrolyze nucleic acids for DNA/RNA turnover and scavenging (31). Under H₂O₂ stress conditions, where peroxiredoxin consumes NADPH for detoxification (and therefore makes it less available for deoxyribonucleotide synthesis), the observed high-level induction of genes for nucleotide turnover could be a reflection of the need to maintain a constant pool of dNTPs to support excision and repair of oxidatively damaged DNA (21, 32, 33).

Additionally, there is an apparent operon in BL-04 (Balac_0440 to Balac_0444) that contains genes involved in protein degradation, which showed significant upregulation after 5 min (T1) and a significant downregulation after 20 min (T2) (Table 2). Previous research has shown that some bacteria utilize proteolytic enzymes to detoxify proteins that have been irreparably damaged by oxidative stress (34, 35). Our data suggest this operon might be used by BL-04 to perform a similar function.

Finally, 28 genes were associated with energy production or sugar transport in strain BL-04, with 19 (68%) of those genes being upregulated in response to H₂O₂ stress (Fig. 1). In this study, the DE genes involved in sugar metabolism with the highest LFC included those involved in raffinose (Balac_1597 to Balac_1601) and maltose transport and metabolism (Balac_1567 to Balac_1573) (see Table S1 in the supplemental material). The influence of these sugars on H₂O₂ resistance was not explored here, but other studies have suggested that complex carbohydrates can enhance bile salt resistance in bifidobacteria (36, 37).

Membrane fatty acid analysis.

Several studies have shown that cell envelope lipid composition plays a crucial role in bacterial response to environmental stresses (38–42). We therefore investigated the membrane fatty acid composition of BL-04 and DSM 10140 after H₂O₂ exposure. Analysis showed no significant change in membrane composition in BL-04 cells grown in medium containing 2.55 mM H₂O₂ for 5 min (T1) or 60 min (T2) compared to cells grown in control medium (Fig. 3). In contrast, strain DSM 10140 showed a significant decrease (P < 0.05) in C_16:0 and a significant increase (P < 0.05) in C_18:1n9 after exposure to 1.25 mM H₂O₂ (Fig. 3). Surprisingly, direct comparison between the bifidobacterium strains showed dramatic differences even in control cells, with BL-04 having 20% more C_16:0 and 15 to 20% less C_18:1n9. Both strains had similar total amounts of C_16:1, with BL-04 having predominantly C_16:1n7 and DSM 10140 having more C_16:1n9 (Fig. 3). These differences are similar to those seen in other bacteria (43, 44, 45) and likely factor into the increased survival of BL-04 under oxidative stress compared to DSM 10140 (14).

Fig 3.

Membrane fatty acid composition for B. animalis subsp. lactis BL-04 (A) and DSM 10140 (B). The graphs show data from cells grown in MP5 medium with no H₂O₂ (control) and cells exposed to 1.25 mM H₂O₂ in MP5 broth for 5 or 20 min. Error bars correspond to the standard error of the mean (SEM). Means with the same letters within each strain are not significantly different (P < 0.05).

In an effort to identify the basis for the dramatic differences we observed in the gene expression profiles and membrane lipid composition of BL-04 versus DSM 10140, we reviewed the comparative genome analysis of these bacteria (12). The two strains are highly clonal, with only 39 coding single-nucleotide polymorphisms and 4 insertion/deletions totaling 443 bp. However, we found that one of these lesions produced a 45-bp deletion in a BL-04 gene (Balac_0771) predicted to encode a long-chain fatty acid coenzyme A (CoA) ligase. In other bacteria, this gene has been shown to activate exogenous long-chain fatty acids for incorporation into the cellular membrane (46) and therefore might result in a different membrane lipid profile for MP5-grown BL-04 compared to DSM 10140. To test whether this lesion affected the ability of BL-04 to incorporate exogenous fatty acids into its membrane, both strains were grown in MP5 medium modified to contain 1% Tween 80 (C_18:1), 1% Tween 20 (C_12:0), or no fatty acids; then lipids were extracted for membrane fatty acid analysis.

As shown in Fig. 4, DSM 10140 cells grown in MP5 with Tween 20 showed dramatically and significantly higher (P < 0.05) amounts of C_12:0 and a significant decrease (P < 0.05) in the pooled total of C_18:1n9 and its derivatives (C_19:0 cyclic propanol [Cyc], C_19:0 Cyc plasmalogens [Plas], and C_18:1n9 Plas) (19) than cells grown in MP5 with no fatty acid supplementation (21.4% versus 35.7%, respectively). Conversely, when DSM 10140 cells were grown in MP5 containing Tween 80, their membranes showed significantly (P < 0.05) higher percentages of C_18:1n9, as well as an increase in C_18:1n9 Plas which was not significant, versus cells grown in MP5 without added fatty acid (Fig. 4). DSM 10140 cells grown in MP5 supplemented with Tween 20 or 80 also had significantly less (P < 0.05) C_16:0 than cells grown in MP5 without fatty acids. These results confirm that DSM 10140 is able to efficiently incorporate exogenous fatty acids into its lipid membrane.

Fig 4.

Membrane fatty acid composition as a function of growth medium composition for B. animalis subsp. lactis BL-04 (A) and DSM 10140 (B) cells. The graphs show data from cells grown in MP5 medium with no exogenous fatty acids and cells grown in MP5 with 1% Tween 20 (C_12:0) or 1% Tween 80 (C_18:1n9). Error bars correspond to the standard error of the mean (SEM). Means with the same letters within each strain are not significantly different (P < 0.05).

In contrast to DSM 10140, membrane lipid profiles of BL-04 cells grown under the same conditions showed far less change in response to the exogenous fatty acid type, which supports our hypothesis that the 45-bp lesion in Balac_0771 impairs the function of its cognate enzyme. Supplementation with Tween 20, for example, did produce a significant increase (P < 0.05) in the membrane level of C_12:0 relative to cells grown without added fatty acids, but the degree of change was substantially lower than that seen in DSM 10140 (from 5.2 to 6.8% in BL-04 versus 3.0 to 27.3% in DSM 10140). Additionally, levels of C_18:1n9 and its derivatives were not significantly different (P > 0.05) from those in BL-04 cells grown without fatty acid supplementation (totals of 26.5% versus 27.5%, respectively) (Fig. 4). Growth of BL-04 in MP5 with Tween 80 did produce a significant increase (P < 0.05) in the concentration of C_18:1n9 and its derivative fatty acids relative to cells grown in MP5 without fatty acids (totals of 38.1% versus 27.5%, respectively). These differences, which would have been present in stress-treated cells, could affect membrane fluidity and, potentially, transduction of environmental stress signals, either of which could explain the observed contrasts in intrinsic and inducible H₂O₂ stress resistance (14). As a whole, these data support our hypothesis that the mutation in the BL-04 long-chain fatty acid-CoA ligase limits the ability of this strain to incorporate certain exogenous fatty acids into its cytoplasmic membrane.

In our experiments with different exogenous fatty acids, we noted that DSM 10140 cells grown in MP5 containing no fatty acids had a gross lipid profile that most closely matched the lipid profile of BL-04 grown in medium containing Tween 80 (Table 3). Because membrane fluidity could influence the efficiency of environmental stress triggers, we wondered if an inducible stress response in DSM 10140 might be restored by modification of its membrane fatty acid composition to more closely match the profile of BL-04. To explore this possibility, DSM 10140 cells were grown in MP5 medium that contained no supplemented fatty acids, treated with a sublethal H₂O₂ concentration [1.25 mM], and subsequently exposed to a lethal H₂O₂ concentration [2.55 or 5.25 mM] (14). Results showed no significant change in survival after H₂O₂ challenge of induced versus control cells grown in MP5 with Tween 80 and a significant decrease in survival of induced cells compared to control cells grown in medium with no exogenous fatty acids (Fig. 5). As is also shown in Fig. 5, however, cells grown with no exogenous fatty acid had significantly greater survival (P < 0.05) after lethal challenge at 5.25 mM H₂O₂ than cells grown in MP5 with Tween 80 (Fig. 5B). The increase in intrinsic H₂O₂ resistance could be associated with the higher percentage of cyclic fatty acids in the membrane (47, 48). Previous studies have shown that in response to low pH and osmotic stress, cells modify their membranes through chain length, saturation, and cyclopropanation of fatty acids, which alters the transition temperature of the membrane and makes it less permeable to organic acids and salts (44, 45, 49). More importantly, cyclopropanation decreases the susceptibility of the cell membrane to lipid peroxidation by stabilizing the unsaturated bond through addition of a methyl group (50). These properties would make the membrane less permeable to oxidative free radicals and more resistant to lipid peroxidation.

Table 3.

Membrane fatty acid (FA) composition of B. animalis subsp. lactis strains grown with different exogenous FAs

Strain and treatment	Mean % of each FA species in total cytoplasmic membrane lipid pool				Saturated/unsaturated
	Cyclic	Plasmologenc	Saturatedd	Unsaturatede
BL-04
No Tween	18.81	22.03	62.35	37.65	1.66
Tween 20a	11.71	16.04	70.07	29.93	2.34
Tween 80b	16.61	30.52	49.87	50.13	0.99
DSM 10140
No Tween	13.87	25.08	54.94	45.06	1.22
Tween 20a	10.51	15.47	74.41	25.59	2.91
Tween 80b	7.96	26.60	43.39	56.61	0.77

^aPolyoxyethylene(20) sorbitan monolaurate (C_12:0).

^bPolyoxyethylene(20) sorbitan monooleate (C_18:1).

^cPlasmologens (ether-linked lipids).

^dPercentage of saturated FAs in membrane, including cyclic FAs and plasmologens.

^ePercentage of unsaturated FAs in the membrane, including plasmologens.

Fig 5.

Experimental stress induction in B. animalis subsp. lactis DSM 10140. Graph shows percent survival of DSM 10140 cells after 20 min (A) or 60 min (B) exposure to a sublethal (1.25 mM) H₂O₂ followed by 30-min challenge at lethal concentrations of 2.55 mM H₂O₂ or 5.25 mM H₂O₂. White bars, cells grown in MP5 with no exogenous fatty acid (FA) source and no induction (control); hatched bars, cells grown with no exogenous FA source and given induction treatment; filled bars, cells grown in MP5 with Tween 80 as FA source and no induction (control); crosshatched bars, cells grown in MP5 with Tween 80 and given induction treatment. Each value is the mean of four replicates. Error bars correspond to the standard error of the mean (SEM). Means with the same letters within each strain are not significantly different (P < 0.05).

In summary, B. animalis subsp. lactis BL-04 and DSM 10140 are highly clonal yet display significant differences in their intrinsic and inducible resistance to H₂O₂ (14). Transcriptome data demonstrate H₂O₂ exposure triggers induction of an oxidative stress response in BL-04, but this mechanism is somehow impaired in DSM 10140. Genetic and membrane lipid data suggest that some of the differences in H₂O₂ resistance between these cells may be associated with membrane lipid composition, which in turn is affected by the activity of a long-chain fatty acyl-CoA ligase which is functional in DSM 10140 but impaired in BL-04. However, confirmation of this relationship will require functional studies involving genetic manipulation of B. animalis subsp. lactis, where genetic tools are only poorly developed. While efforts to restore an inducible H₂O₂ stress response in DSM 10140 via modification of its CMFA composition were unsuccessful, modification did significantly increase intrinsic H₂O₂ resistance. These data show deliberate H₂O₂ stress induction or membrane lipid modification can be used to significantly improve H₂O₂ resistance in B. animalis subsp. lactis.