Lactobacillus reuteri LTH2584, L. plantarum TMW1.460, L. hammesii DSM16381, L. spicheri Lp38 and Lactobacillus sanfranciscensis ATCC 27651 were anaerobically cultivated at 37°C (L. reuteri) or 30°C (all other strains) in modified De Man Rogosa Sharpe (mMRS) medium containing 0.1% Tween 80 (mMRS-Tween 80; Zhang et al., 2010). mMRS-Tween 20 was prepared by replacing Tween 80- with an equal weight of Tween 20. E. coli DH5α (New England Biolabs) served as a host for plasmids in the cloning procedures, and E. coli BL21 Star (DE3) (ThermoFisher Scientific) was used for protein overexpression. E. coli strains were cultivated in Luria-Bertani (LB) medium (BD, Mississauga, CA, USA) with agitation at 200 rpm and 37°C. Antibiotic-resistant E. coli carrying plasmid pET-28a(+), pUC19 or pJRS233 were cultured in media containing 50 mg/L kanamycin, 50 mg/L ampicillin, or 500 mg/L erythromycin, respectively. Erythromycin-resistant L. plantarum was grown in presence of 5 mg/L erythromycin. Aspergillus niger FUA5001 and Penicillium roqueforti FUA5005 were grown at 25°C for 72 h on malt extract agar.

Linoleic acid metabolism of lactobacilli was analyzed after incubation in mMRS-Tween80 broth supplied with 5% inoculum and 4 g/L linoleic acid anaerobically for 4 days. Lipids were isolated by addition of one volume of 85:15 (vol/vol) chloroform-methanol to cultures prior to incubation 4°C overnight; cultures were then extracted twice with additional two volumes of chloroform–methanol (85:15, vol/vol). The organic solvent was evaporated under reduced pressure and the dry residue stored at -20°C under nitrogen.

Genomic DNA was isolated using the Blood & Tissue Kit (Qiagen, Hilden, Germany). Plasmid DNA from E. coli was extracted with a QIAprep Spin Miniprep kit (Qiagen). PCR primers used in this study were synthesized by Integrated DNA Technologies (San Diego, CA, USA). The Taq DNA polymerase was purchased from TaKaRa Bio (Shiga, Japan). T4 DNA ligase and restriction enzymes were obtained from Thermo Scientific (Mississauga, CA, USA). PCR products were purified by using the DNA gel extraction kit (Qiagen). DNA sequencing was performed by Macrogen (Rockville, MD, USA).

Genes of putative LAHs in L. spicheri Lp38, L. reuteri LTH2584 and L. plantarum TMW1.460 were amplified with primers that are specific to LAHs in genome-sequenced strains of these species (Table 1). To identify the LAH in L. hammesii, lah sequences from the closely related L. spicheri and L. brevis were aligned and specific primers targeting conserved sequences upstream and downstream of lah coding sequences were designed (Table 1). The four lah genes were sequenced by service of Macrogen (Rockville, MD, USA).

The phylogenetic analysis of LAHs included the type strains of the 24 groups of the -genus Lactobacillus sensu lato (Zheng et al., 2015) and the genera Weissella, Leuconostoc, and Oenococcus. Sequences of biochemically characterized linoleate 10-hydratase and linoleate 13-hydratase from Lactobacillus spp., Bifidobacterium spp., and Streptococcus spp. were included in the phylogenetic analysis (Volkov et al., 2010; Rosberg-Cody et al., 2011; Yang et al., 2013; Kim et al., 2015). Protein sequences of LAH were retrieved from GenBank¹, using NCBI BLAST analysis with organism specific search. Protein phylogenetic tree was built in MEGA7.

Coding regions of the four lah were amplified from genomic DNA of the respective strains with primers listed in Table 1. Amplicons were purified and cloned into pGEM-T Easy vector (Promega, Madison, WI, USA). The lah fragments in recombinant pGEM-T vectors were digested with restriction endonucleases (Table 1), purified and ligated into expression vector pET-28a(+) (Novagen, Toronto, ON, Canada), yielding the respective constructs pET28a/LAH for each strain. Recombinant plasmids were introduced into chemically competent E. coli BL21 Star (DE3) (Thermo Fisher Scientific) and transformants were plated on LB agar containing 50 mg/L kanamycin. The gene cloning was verified by PCR amplification and sequencing.

Four recombinant E. coli BL21 (DE3) strains were grown to an optical density (OD) at 600 nm of 0.6. Protein expression was induced by the addition of isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 1.0 mM, followed by incubation for 4 h and harvesting of cells by centrifugation at 4°C.

Overexpressed LAHs were present mainly as inclusion bodies. Solubilization and refolding of inclusion bodies were carried out with the protein refolding kit (Novagen) according to the manufacturer’s instructions. The refolded proteins were concentrated by using a 10 KDa Amicon Ultra-15 centrifugal filter Unit (Millipore, Germany).

After concentration, the His-tagged LAHs were purified by Ni-NTA spin columns (Qiagen). The purified LAHs were finally dialyzed against 50 mM 4-morpholineethanesulfonic acid (MES) buffer (pH 6.1) (Sigma-Aldrich) overnight at 4°C. The purified enzymes were assessed by SDS-PAGE and staining with Coomassie blue. FAD was added to the purified enzymes at a final concentration of 0.2 mM and incubated at 4°C for 24 h (Joo et al., 2012).

To determine the enzymatic activity, 4.5 mg linoleic acid and 25 μg purified LAH were incubated in 1 ml of 50 mM MES buffer (pH 6.1) containing 50 mM NaCl, 2% ethanol and 10% glycerol at 25°C for 3 h. Fatty acids were extracted following the procedure described above. The organic phase of the extracts was collected and analyzed with LC-APPI-MS as described (Black et al., 2013a,b).

Gene disruption of lah in L. plantarum TMW1.460 was achieved by an in-frame, unmarked deletion (Su et al., 2011). The approximately 900 bp 5′-flanking regions (fragment A) and 1000 bp 3′-flanking regions (fragment B) of lah were amplified by PCR with primers listed in Table 1. The fragment A was digested with PstI, SphI, and fragment B was digested with SphI, HindIII. The resulting fragments were purified and sequentially ligated into vector pUC19 to generate pUC19/AB. The AB fragment in pUC19/AB was confirmed by sequencing, and sub-cloned into the PstI and HindIII restriction sites of pJRS233 to create pJRS233/Δlah. Recombinant plasmids were transformed into electrocompetent L. plantarum TMW1.460 at 12.5 kV/cm, 25 μF, and 200 Ω. The cells were grown in mMRS-Tween80 broth containing 5 mg/l erythromycin at 42–44°C for 80 generations to select for single-crossover mutants. Several colonies were isolated, cultured in mMRS-Tween80 broth for approximately 100 generations, and plated on mMRS-Tween80 agar at 30°C. The colonies were replica plated on mMRS-Tween80-erythromycin agar to identify erythromycin-sensitive double-crossover mutants. Gene replacement was confirmed by PCR amplification and sequencing. The phenotype was determined by LC-APPI-MS analysis of culture supernatant of L. plantarum TMW1.460Δlah grown in mMRS-Tween80 supplemented with 4 g/L linoleic acid.

Ethanol tolerance of L. plantarum TMW1.460 and TMW1.460Δlah was carried out with strains grown in mMRS-Tween 80, or mMRS-Tween 20. Ethanol was added as indicated and media were sterilized by filtration after addition of ethanol. Stationary phase cells were harvested by centrifugation, and resuspended in the same volume of mMRS-Tween 80 or -Tween 20 containing 20% ethanol. A 100 μl aliquot of cell suspension was analyzed as untreated control. The samples were incubated in 30°C and aliquots were removed in 1.5 h intervals and serially diluted in 0.85% NaCl. The appropriate dilutions were surface plated in duplicate on respective mMRS-Tween80 or -Tween20 agar and incubated at 30°C for 24 h. Ethanol resistance was determined in three independent experiments.

To investigate the membrane fluidity, LAURDAN (6-dodecanoyl-2-dimethylaminonaphthalene) (Thermo Fisher Scientific) was employed to measure generalized polarization (GP) value. L. plantarum TMW1.460 and TMW1.460Δlah were cultivated at 30°C for 20 h in mMRS-Tween80 or mMRS-Tween 20. The membrane fluidity of the cells influenced by ethanol at the concentration ranging from 0 to 16% was assessed as described (Molina-Höppner et al., 2004). The effect was determined in three independent experiments.

Cell surface hydrophobicity was assessed by quantification of microbial adhesion to solvents (MATS; Kankaanpää et al., 2004). Solvents used were: chloroform (polar and electron acceptor) and tetradecane (non-polar), ethyl acetate (polar and electron donor) and octane (non-polar). The MATS is based on the comparison between microbial cell affinity to a polar and non-polar solvent within a couple that pose similar van der Waals surface tension components.

Lactobacillus plantarum TMW1.460 and TMW1.460Δlah were grown in mMRS-Tween 20, or mMRS without Tween but supplemented with 1 g/L oleic acid or linoleic acid. Cells were harvested by centrifugation, washed twice and resuspended in 150 mM NaCl to a final cell concentration of 10⁸ CFU/ml. A 1 ml aliquot was removed as untreated control (A₀). A 2.4 ml aliquot of cell suspension was mixed with 0.4 ml of solvent by vortexing for 60 s, respectively. The emulsified mixture was allowed to stand for 20 min to ensure the complete separation of the two phases. A sample (1 ml) was carefully taken from the aqueous phase (A). The optical cell density of sample A₀ and A was measured at 600 nm. The microbial adhesion percentage to each solvent was calculated with the equation: percent affinity = 100 × [1-(A/A₀)]. Each measurement was determined in three independent experiments.

Extraction and purification of 13-HOE and 10-HOE was based on a protocol developed for 10-HOE (Black et al., 2013b). Fermentation of L. hammesii or L. plantarum TMW1.460Δlah in mMRS-Tween 80 with 4 g/L linoleic acid was conducted for production of 10-HOE or 13-HOE, respectively. The crude extraction of cellular lipids was performed as described above. Subsequently, the purification of 10-HOE or 13-HOE was based on a protocol described previously (Black et al., 2013b). The fermentation of L. hammesii or L. plantarum TMW1.460Δlah in mMRS-Tween 80 with 4 g/L linoleic acid and the crude extraction of fatty acid mixtures were performed as described above. For further purification, 25 mg of sample dissolved in chloroform was applied to a 500-mg Sep-Pak silica cartridge (Waters, Ltd, Mississauga, ON, Canada) previously equilibrated with 6 ml chloroform. The cartridge was successively washed with the following gradient of isopropanol in chloroform: 35 ml of chloroform, followed by 18 ml of 1, 5, 10, and 50% (vol/vol) isopropanol in chloroform. Eluates were collected and concentrated to dryness under nitrogen. The dry residues were dissolved in chloroform for analysis by LC-APPI-MS to identity of the product, and the removal of contaminating lipids.

The minimum inhibitory concentration (MIC) was determined to assess the toxicity of linoleic acid to lactobacilli, and to determine the antifungal activity of 10-HOE, 13-HOE, coriolic acid, ricinoleic acid, and linoleic acid. Lipids dissolved in ethanol were serially diluted twofold in mMRS-Tween80 using 96-well microtiter plates. Ethanol in the samples was evaporated under a sterile laminar flow hood before inoculation with indicator strains. Stationary phase cells of lactobacilli were harvested by centrifugation, washed twice in mMRS-Tween80 and diluted to approximately 10⁷ CFU/ml in the same medium. The plates were incubated at 30°C.

For evaluation of the antifungal activity, lipids diluted with serial twofold dilutions. Conidiospores of A. niger and P. roqueforti were prepared as reported (Black et al., 2013b). The plate was inoculated with A. niger as an indicator organism and incubated at 25°C for 2 days, while the plate inoculated with P. roqueforti was incubated for 3 days. Fungal growth without addition of lipids served as the positive control and media alone as the negative control. The MIC was defined as the lowest concentration of lipids that inhibited the growth of fungi when growth was visible in the positive control. MIC values were determined by six independent experiments.

Data analysis was conducted with R 3.1.2 (R Core Team, 2014). Significant differences in the assessment of cell survival and membrane fluidity under ethanol stress were determined by one-way analysis (ANOVA). Significance was assessed at a 5% probability of error (P < 0.05).

The sequences of linoleate 10-hydratase in L. reuteri, L. plantarum, L. hammesii, and L. spicheri were deposited in GenBank with accession numbers KX827285, KX827286, KX827287, KX827288, respectively.

The products of linoleic acid conversion by five lactobacilli were analyzed by negative ion LC/APPI-MS/MS (Table 2). MS/MS spectra of the products confirmed the position of hydroxyl groups (Figure 1; Black et al., 2013a). The strains of lactobacilli differed in with respect to their conversion of linoleic acid to hydroxy fatty acids (Table 2). L. reuteri, L. hammesii, and L. spicheri produced 10-HOE only, while L. plantarum produced 10-HOE, 13-HOE and 10,13-dihydroxy octadecanoic acid. L. plantarum TMW1.460Δlah produced 13-HOE but not 10-HOE or 10,13-dihydroxy octadecanoic acid, demonstrating that their formation by L. plantarum TMW1.460 is attributable to a dedicated linoleate 13-hydratase acting on linoleic acid and 10-HOE, respectively. L. sanfranciscensis did not convert linoleic acid.

Phylogenetic relationships of putative LAHs from lactobacilli were compared to the corresponding enzymes of the 24 type strains in the genus Lactobacillus, and type strains from the genera Pediococcus, Weissella, Leuconostoc, and Oenococcus (Figure 2). All four hydratases from lactobacilli that were investigated in this study belonged to myosin cross reactive antigen family. L. sanfranciscensis harbored no LAH. The topology of the protein tree did not conform to the evolutionary relationship of the organisms (Zheng et al., 2015). The tree displayed two clusters but the two clusters do not differentiate between linoleate 10-hydratases and linoleate 13-hydratases (Figure 2).

Sequence analysis did not distinguish 10-hydratases from 13-hydratases, therefore, LAHs from four strains of lactobacilli were characterized biochemically after heterologous expression in E. coli and purification by affinity chromatography. A single band was observed by SDS-PAGE analysis after purification of the four enzymes (Figure 3); this band was absent in crude cellular extracts of E. coli strains prior to induction (data not shown). The genes from L. reuteri, L. plantarum, L. hammesii, and L. spicheri encoded a protein of 590, 564, 564, 564 amino acids, respectively, matching the size of the major band observed by SDS-PAGE analysis (Figure 3). The addition of cofactor FAD to apoenzyme was essential for activity. LC-APPI-MS analysis revealed that all four recombinant proteins produced 10-HOE from linoleic acid, as demonstrated by the fragment ion at m/z 185.1 in the MS/MS spectra (Table 2). Hence, all four recombinant proteins were shown to be 10-hydratases. The substrate specificity of the linoleate 10-hydratase of L. plantarum TMW1.460 confirmed that formation of 13-HOE and 10,13-dihydroxy octadecanoic acid by this strain is likely attributable to a second and dedicated linoleate 13-hydratase.

Tween 80 is a derivative of oleate and a component of mMRS (Johnsson et al., 1995; Gänzle et al., 2000). mMRS-Tween20 did not support the growth of L. reuteri, L. hammesii, and L. sanfranciscensis but did support the growth of L. plantarum and its Δlah mutant (data not shown). However, unsaturated fatty acids (UFAs) may also exhibit antibacterial activity (Desbois and Smith, 2010). Therefore, the MICs of oleic and linoleic acid to lactobacilli including L. plantarum TMW1.460Δlah were assessed. The MICs of oleic and linoleic acids were >8 g/L for all strains, which indicates a high tolerance toward oleic and linoleic acid. L. sanfranciscensis showed less tolerance and its growth was inhibited by 1 g/L oleic acid and 0.5 g/L linoleic acid.

To investigate the effect of hydroxy fatty acids or 10-LAH itself on ethanol resistance, the survival of L. plantarum TMW1.460 and TMW1.460Δlah was assessed in mMRS-Tween80 and mMRS-Tween20 containing 20% ethanol. The ethanol tolerance of L. plantarum TMW1.460 and TMW1.460Δlah did not differ (Figure 4). However, the presence of Tween 80 in the growth medium enhanced bacterial survival, especially for L. plantarum TMW1.460 (Figure 4).

The ethanol-dependent membrane phase behavior of strains grown in mMRS-Tween80 or -Tween 20 was analyzed (Figure 5). The GP values decreased with increasing ethanol concentration, indicating that ethanol increased the fluidity of the membrane. The response of the membrane fluidity of L. plantarum TMW1.460 and TMW1.460Δlah to ethanol was similar. Tween 80 supplemented medium induced a more fluid membrane in both strains.

The MATS of L. plantarum TMW1.460 and TMW1.460Δlah cultivated in mMRS-Tween20, or mMRS-Tween20 supple mented with oleic or linoleic acid are shown in Figure 6. The deletion of lah modified the properties of the cell surface. The adhesion of L. plantarum TMW1.460Δlah was more solvent-dependent when compared to the wild-type strain and the affinity of cells to the four solvents was generally higher for Δlah mutant than for wild-type strain. Similar trends were noted with strains grown in mMRS with different supplements, suggesting that the differences between wild-type and mutant strains are attributable to 10-LAH, not the product of this enzyme. Similar behavior was observed when strains cultivated in Tween 80 (Data not shown).

In this study, both wild-type and mutant strains exhibited low affinity for tetradecane and octane (non-polar solvents), indicating that the cell surface of both strains was hydrophilic rather than hydrophobic. To determine the effect of deletion of lah on the Lewis electron donor/electron acceptor property, the bacterial affinity to chloroform and ethyl acetate between different strains were also compared (Figure 6). L. plantarum wild-type strain showed similar adhesion ability to chloroform and ethyl acetate. However, regardless of the different supplement in medium, the adhesion of the mutant strain was always higher to chloroform (electron acceptor and acidic solvent) than to ethyl acetate (electron donor and basic solvent).

13-HOE and 10-HOE were purified from organic extracts of cultures of L. plantarum TMW1.460Δlah and L. hammesii, respectively, when 4 g/L linoleic acid was used as substrate. The major impurity detected in the organic phase was linoleic acid; other more oxidized forms of hydroxy C₁₈ fatty acids were also present. The compounds were purified by silica solid phase microextraction and analyzed by LC-APPI-MS/MS. 10-HOE or 13-HOE were eluted in the 1% isopropanol fraction and a single peak was observed in the LC-APPI-MS/MS chromatogram after purification. The antifungal activity of purified 13-HOE and 10-HOE against A. niger and P. roqueforti is shown in Table 3, and compared to reference lipids differing in the number and position of hydroxyl groups or double bonds. Both coriolic acid and ricinoleic acid were active against all fungi indicators with MICs between 0.26 and 0.29. Linoleic acid showed the lowest antifungal activity.