Susceptibility of L. lactis NCDO 2118 to acid stress and bile salts.

Concerning the acid stress, lowering the intracellular pH reduces the transmembrane pH difference and the activity of acid-sensitive enzymes and damages proteins and DNA [67].The first mechanism used by L. lactis species to cope with acid stress is to maintain a low intracellular pH (pHi) by using membrane ATPase FoF1 [68; 69] and the generation of alkaline substances through the catabolism of amino acids (deamination, for example) [70; 71]. Bile salts, on the other hand, are surface-active, amphipathic molecules with a potent antimicrobial activity, and they act as detergents that disrupt biological membranes [67]. The percentage of resistance to bile salts also tends to vary among LAB and even between strains of the same species [72].

Here, we have identified 25 and 16 genes previously shown to be involved in acid stress and bile resistance in other species, respectively. In an in vitro assay, however, only 48% of L. lactis NCDO 2118 was able to grow after pH challenge, and 95% of bacteria was inhibited by bile salts. Other authors have already found that bacteria with an intestinal origin tend to be more resistant to stomach acids [73]. Therefore, this finding corroborates our results because L. lactis NCDO 2118 was isolated from frozen peas. Most of the genes found in L. cremoris MG1363 were also identified in L. lactis NCDO 2118. Additionally, a work using proteomics analyses identified some genes related to acid response and they are present in L. lactis NCDO 2118 genome (clpEP, ahpC, tig, hpr and luxS) [74] showing that other approaches may better elucidate the mechanism of survival to acid stress on this strain.

The high susceptibility of L. lactis NCDO 2118 to bile salts, on the other hand, must be further explored in vitro and in vivo using transcriptomics analyses to determine the expression rates of the described genes.

Competitive exclusion mechanisms of L. lactis NCDO 2118.

There are several mechanisms used by bacteria to competitively exclude other species, such as bacteriocin production, space competition through the use of adhesins or receptors that bind to specific surface features, predation and even rapid growth [75].

Adhesins are responsible for the recognition and colonization of host tissues through specific binding. This process may activate the innate host cells or the expression of new genes. Adhesins may be characterized as hair-like attachments named pili or fimbriae or in other cases, named non-pilus adhesin, related to the microbial cell surface [76].

In L. lactis NCDO 2118, we have identified the gene chiA (NCDO2118_2053) and the genes coding for the Chitin binding protein (CBP–NCDO2118_2054) and the laminin-binding protein (NCDO2118_1446), which are normally related to adhesion in other bacteria. Chitin is degraded by chitinases that belong to members of the glycoside hydrolase of family 18 [77]. One example of bacteria that produces chitinase is Serratia marcescens, one of the most efficient organisms in chitin degradation [78]. When E. coli was cloned with a chitin-binding protein of Serratia marcescens, there was a significant increase in its ability to adhere to human colon cells [77].

Chitin-binding encoding genes are broadly distributed in many microorganisms. The L. lactis IL1403 genome, for example, harbors chitinolytic machinery represented by one family 33 CBP (yucG; referred as LlCBP33A), one family 18 chitinase (chiA, referred as LlChi18A) and one family 20 N-acetylhexosaminidase [3; 79]. Another example of bacteria that present a high adhesion degree is Borrelia burgdorferi, which is able to bind to mammalian laminin, an important extracellular matrix (ECM) component [80]. A laminin-binding protein has also been identified in L. lactis NCDO 2118.

Additionally, we have found using MATS experiments that L. lactis NCDO 2118 presents a 52% of association to xylene, which supports the presence of genes coding for adhesion-related proteins in this strain. The hydrophobicity is directly related to the capacity of strains to adhere to surfaces. This capacity is determined by hydrophobic components present in the outer membrane of microorganisms, and it is known that hydrophobic interactions have an important role in the adhesion of bacteria to the epithelium. The application of MATS experiments facilitates a qualitative assessment of the polarity or non-polarity of the bacterial surface, which is important because it indicates the potential for probiotic adhesion to apolar surfaces in the intestinal and vaginal epithelia. However, this test is only a primary indicator of the adherence of microorganisms [81; 82].

The other bacterial competitive exclusion mechanism assayed here was the production of exclusion antimicrobial peptides, named bacteriocins. Bacteriocins produced by a bacterium may be activated against others, even ones from the same species, while the producer is immune to its own peptides [43]. This exclusion mechanism is very important for probioses, as it renders probiotic organisms able to compete with and kill pathogenic ones, promoting a health benefit to the host [2]. We have predicted one bacteriocin for each of the three classes in L. lactis NCDO 2118 (class I-III), which may be important for exclusion mechanisms of this bacteria. However, the lack of nisT and the pseudogenization of nisP on the class I gene cluster, the lack of ABC-transporters in the class II cluster and, also, the lack of information regarding the product of the putative bacteriocin in the class III cluster have to be further studied using in vitro analyses to elucidate whether those bacteriocins are produced and present antimicrobial activity or not.

We have also performed a deferred agar spot assay for the initial determination of antagonistic activity produced by L. lactis NCDO 2118. This test indicates the activity against various Gram-positive and -negative bacteria. This inhibitory effect may be due to H₂O₂, lactic acid, bacteriocins, antibiotic-like substances, or a combination of these compounds [83]. However, L. lactis NCDO 2118 showed no effect on the growth of the pathogenic strains assayed here.

Secreted proteins and immunomodulatory effects.

According to Luerce et al., (2014), the secreted proteins of L. lactis NCDO 2118 are possibly responsible for the immunomodulatory effects of this transient bacterium inside the host. In a comparison of the anti-inflammatory effects between L. lactis NCDO 2118, L. lactis IL1403 and L. cremoris MG1363 strains, only the L. lactis NCDO 2118 supernatant was able to decrease the IL-8 production (45%), showing its immunomodulatory ability against inflammation [18].

Here, we predicted 5 proteins that are present in the 26 secreted proteins exclusive from L. lactis NCDO 2118 and in the 867 expressed proteins from proteomic analyses and may thus be related to the probiotic effect of this strain (Table 7). From those 5 exclusive, secreted and expressed genes of L. lactis NCDO 2118, epsK and epsL are part of the operon epsABCDEFGHIJKLX, whereas there is an epsR gene located in another genomic region.

The EPSs are a type of biopolymer able to facilitate intense interactions of biofilm cells through adhesion, aggregation of bacterial cells, cohesion of biofilms, protective barriers, and cell component export [84]. Through microarray and electron microscopy analyses, Denou et al., 2008 found an eps cluster of genes exclusive from a probiotic Lactobacillus strain compared to a type strain and they have shown that deletion of this cluster from the probiotic strain results in lack of the fuzzy layer on the outside of the cell wall [85].

Altogether, the lack of further knowledge of the eps cluster of genes and the presence of three other genes coding hypothetical exclusive/secreted/expressed proteins highlight the need for additional studies to better elucidate the underlying mechanisms involved in the anti-inflammatory and immunomodulatory activities of this strain.

Heatmap of genome similarities and 16S phylogenetic tree.

The heatmap analyses of the 17 strains were performed with Gegenees [86]. The input files consisted of complete genomes in.fna format. Streptococcus thermophilus LMD-9, a closely related species, was used as an outgroup to root the tree. The analyses were performed with default parameters for comparative analyses using the alignment method BLASTn. Gegenees performs an all-versus-all alignment process of the fragments generated from the 17 genomes. The result was exported from Gegenees as a heatplot image. Additionally, a phylogenetic tree was made using the 16S sequences from all genomes as identified by RNAmmer [87]. After that, they were aligned in MUSCLE [88], and the phylogenetic tree was inferred using the Neighbor-Joining method with 1000 bootstrap replicates.

Genome synteny.

The genome synteny analyses were performed using Mauve, with the "progressiveMauve" option and all genome sequences in the.fna format. Mauve predicts gene synteny by merging locally collinear blocks of conserved genome orthologous regions and ordering them according to a reference genome [89].

Genome plasticity.

The genome plasticity analyses were performed by searching for horizontally acquired regions such as genomic islands and phage sequences. The genomic islands were searched using the software GIPSy: Genomic Island Prediction Software [90], which updates the methodology of the software PIPS: Pathogenicity Island Prediction Software. Briefly, GIPSy performs the prediction of four different classes of genomic islands: Pathogenicity Islands, Resistance Islands, Metabolic Islands and Symbiotic Islands. In this work, we searched for metabolic and symbiotic islands in the genome of L. lactis NCDO 2118 using Lactococcus lactis subsp. cremoris MG1363 and Lactococcus garviae Lg2 genomes as subjects. After, we consolidated and manually curated the results. The choice of metabolic and symbiotic islands was made based on the lifestyle of L. lactis NCDO 2118, a strain isolated from vegetables, and its metabolic importance.

All the analyses were performed using GENBANK files and default parameters. The results were exported in tabulated format and used in BRIG (Blast Ring Image Generator) to generate circular genome comparative views [91]. Finally, the prophage prediction was performed using the GENBANK file and the software Phast [92], and the results were exported in table format and used as input in BRIG.

Bacteriocin prediction.

The bacteriocin prediction was performed in BAGEL software using the.fna file from L. lactis NCDO 2118. Briefly, the software works with a curated dataset of bacteriocins, in which the input data are evaluated based on a Hidden Markov Model. The genetic information is analyzed based on combinations of PFAM domains [42]. For the putative bacteriocin predicted on L. lactis NCDO 2118 (NCDO2118_1768), we used the Transporter Classification Database (TCDB) [93] with an e-value of e-07.

Circular comparison map of genomic sequences.

To create circular genome comparisons, we used the software BRIG and all genome sequences in the.fna format; we created the figure with L. lactis NCDO 2118 as reference strain. Additionally, we added the coordinates of the genomic islands and phage regions to the figure to visualize genome plasticity events. Finally, all genomes underwent BLAST analyses against the reference strain to create the circular comparison map.

Metabolic pathway prediction.

A genome sequence in.fasta and a genome annotation in the.gbk format were used for reconstructing the Lactococcus species metabolic pathways. Posteriorly, the Pathway/Genome Databases (PGDB) for each of the 16 strains were computationally predicted using Pathway Tools software version 16.5 [94], developed by SRI International. The MetaCyc, a highly curated and non-redundant reference database of small-molecule metabolism, was used as a reference database for the PathoLogic component of the Pathway Tools software [95]. The metabolic pathways of L. lactis NCDO 2118 were used as a reference for the comparative analysis using the following comparisons: i) L. lactis NCDO 2118, L. lactis KF147 and Lactococcus lactis subsp. lactis IL1403, ii) non-pathogenic strains of L. lactis (L. lactis subsp. lactis and cremoris), and iii) all strains in this study.

Identification of the secretome.

The prediction of the putative subcellular localizations of L. lactis NCDO 2118 proteins was performed in silico using SurfG+. This software contains such tools as SignalP, LipoP and TMHMM for the identification of motifs [46]. Interestingly, SurfG+ uses the size of the membrane wall to better differentiate the membrane (MEM) and potentially surface exposed (PSE) proteins. Here, the measurements of the membrane wall were performed with electron microscopy with EM10A equipment (Zeiss), as previously described [96].

L. lactis NCDO 2118 was grown at 30°C for 18 h in M17 medium (Difco) containing 0.5% glucose [18] and then centrifuged. The resulting precipitate (~500 mL) was placed in an Eppendorf tube, fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) for 6 h at 8°C and washed three times with 0.1 M sodium cacodylate buffer (pH 7.2). After washing, the sample was post-fixed in 1% osmium tetroxide in 0.1 M sodium cacodylate buffer (pH 7.2) + 1.5% potassium ferrocyanide for 90 minutes, washed with 0.1 M with sodium cacodylate buffer (pH 7.2), dehydrated in a graduated ethanol series (50% EtOH, 70% EtOH, 95% EtOH, and 100% EtOH), and incorporated in Eponate–Araldite resin. Ultrathin sections were obtained using uranyl acetate and lead citrate and then verified by Zeiss-EM-10A [97]. The micrograph was obtained by one CCD Mega View camera. The thickness of the L. lactis NCDO 2118 wall was determined from the image analysis micrograph in ImageJ software (available at imagej.nih.gov/ij/).

To measure the wall, we used at least five micrographs of L. lactis NCDO 2118 with magnifications of 50,000 and 100,000 times. We calculated the mean size of the cell walls, and the average number of amino acids for the obtained wall thickness was ~55 amino acids. This value was added to the SurfG+ software together with the.fasta sequence of amino acids (.faa) exported from the strain of interest.

After this process, we used OrthoMCL tool to predict the orthologous and paralogous genes between L. lactis NCDO 2118 and L. lactis IL1403.

Bacterial strains and growth conditions.

For in vitro analyses, we used the probiotic strain L. lactis NCDO 2118 [18] and the pathogenic strains Salmonella enterica serovar Typhimurium ATCC 14028, Escherichia coli ATCC 25723, Staphylococcus aureus ATCC 29213, Bacillus cereus ATCC 11778, Listeria monocytogenes ATCC 15313, Enterococcus faecalis ATCC 19433, and Pseudomonas aeruginosa ATCC 25853.

L. lactis NCDO 2118 was grown at 37°C in MRS medium (Difco) without agitation for 18 hours. L. monocytogenes was cultured in TSB-YE for 24 hours at 28–30°C. The pathogenic strains were grown in BHI medium (BD) for 24 hours at 37°C. To prepare the solid and semi-solid culture media, we added 1.5% and 0.2–0.75% of agar, respectively.

L. lactis gastric juice susceptibility.

L. lactis NCDO 2118 stationary phase cells were suspended in either 0.9% saline solution (pH 7) or simulated gastric juice (NaCl 2 g/L, pepsin 3.2 g/L, adjusted to pH 2.5 with concentrated HCl) and incubated at 37°C for 3 h. Solutions were centrifuged, the supernatant was discarded, and the pellets were suspended in MRS broth. Bacterial growth was evaluated by inoculating MRS broth with 2% v/v of control cells in saline and artificial gastric juice-treated cells onto microplates in triplicate, before incubating them in a Microplate Spectrophotometer System SpectraMax 340 (Molecular Devices Inc., Sunnyvale, CA, USA) at 37°C for 18 h. The OD_620nm (optic density) was recorded at 30 min intervals. The percentage of growth inhibition was calculated as (1 –areaAGJ/areaCT) x 100, where areaAGJ and areaCT are the areas under the growth curve for the simulated gastric juice and control, respectively. The total area under the curve was calculated by definite integration using the OriginPro 8.5 program (OriginLab Corporation, Northampton, MA, USA). The results were based on the average of three independent assays.

Susceptibility to bile salts.

The susceptibility of L. lactis NCDO 2118 to bile salts was evaluated according to the method of Silva et al., (2013) [98]. For this, the L. lactis NCDO 2118 strain was grown in MRS medium at optical density of 0.6 and transferred (2% v/v) to MRS medium supplemented or not with 0.3% of Oxgall (Oxoid Ltd., Basingstoke, UK). The OD_620nm was recorded at 30 min intervals while incubating at 37°C for 18 h in a microplate reader. The percentage of growth inhibition was calculated as (1 –areaBS/areaCT) x 100, where areaBS and areaCT are the areas under the growth curve for bile salt and control cells, respectively. The percentage of bacterial viability was determined in a Microplate Spectrophotometer System SpectraMax 340 (Molecular Devices Inc., Sunnyvale, CA, USA) in the same manner as described above. The results were based on an average of three independent assays.

Cell surface hydrophobicity.

MATS was measured to evaluate the bacterial cell surface hydrophobicity [99]. Measurement of the cell surface hydrophobicity of L. lactis NCDO 2118 was performed with xylene using the MATS method. Bacterial stationary phase cultures were centrifuged, washed twice and adjusted to an OD_600nm of 0.6 with 0.1 M KNO₃, pH 6.2 (A₀). Then, xylene was added in suspension 16% (v/v) and maintained for 10 minutes at room temperature. The tube was agitated vigorously, and after 30 minutes, the aqueous phase was collected for optical density OD_600nm measurement. The reduction percentage of optical density was calculated. The results were based on the average of three independent assays.

Antagonistic activity.

Bacterial isolates were cultured in MRS broth for 24 h at 37°C within an anaerobic chamber. A 5 μL aliquot of the culture was then spotted onto MRS agar. After incubation at 37°C for 48 h under anaerobic conditions, the cells were killed by exposure to chloroform for 20 min. Residual chloroform was allowed to evaporate, and Petri dishes were overlaid with 3.5 mL of a soft agar containing brain heart infusion (Acumedia, Neogen Co., Lansing, MI, USA), tryptone soy broth (Difco) supplemented with 0.5% yeast extract (Acumedia), or Ellinghausen–McCullogh–Johnson–Harris with Leptospira enrichment EMJH (Difco) inoculated with 0.2 mL of a 24 h culture of Staphylococcus aureus ATCC 29213, Enterococcus faecalis ATCC 19433, Pseudomonas aeruginosa ATCC 25853, Bacillus cereus ATCC 11778, Escherichia coli ATCC 25723, Salmonella enterica serovar Typhimurium ATCC 14028, Leptospira interrogans serovar Icterohaemorrhagiae, or Listeria monocytogenes ATCC 15313. After incubating at 37°C for 24 h under aerobic or anaerobic conditions, depending on the indicator strain, the antagonistic activity was determined based on the presence of a growth inhibition zone, using the method of Tagg as modified by Branco et al., (2010) [100].

Antibiotic susceptibility.

L. lactis NCDO 2118 antibiotic susceptibility was determined using antibiotic diffusion discs (Oxoid, England) on MRS plates. Bacteria were inoculated in MRS broth and incubated overnight at 37°C. Solutions of 10⁸ viable cells (McFarland scale) were prepared from the colonies in 3.5 mL of 0.9% buffered saline. The diluted culture (100 μL) was streaked onto MRS agar, and antibiotic discs were applied to the surface using an antibiotic disc dispenser. The discs included amikacin (30 μg), ampicillin (30 μg), ceftriaxone (30 μg), chloramphenicol (30 μg), erythromycin (10 μg), oxacillin (1 μg), penicillin G (10 U), tetracycline (30 μg) and vancomycin (30 μg). The results were interpreted according to Charteris et al., (1998) [101].

Bacterial strain, growth conditions and preparation of proteins from culture filtrates for proteomic analysis.

L. lactis NCDO 2118 and L. lactis IL1403 were pre-inoculated in M17 medium (Difco, New Jersey, USA) and incubated at 30°C for 16 h. The precultures were then inoculated (1:100) in fresh M17 medium supplemented with 0.5% (w/v) glucose (M17Glc) at 30°C until reaching an OD₆₀₀ = 0.8 (three independent experiments). The cultures were then centrifuged for 20 min at 2,700 x g. The supernatants were filtered using 0.22-μm filters, 30% (w/v) ammonium sulfate was added to the samples, and the pH of the mixtures was adjusted to 4.0. Next, 20 mL of N-butanol was added to each sample. The samples were centrifuged for 10 min at 1,350 x g and 4°C. The interfacial precipitate was collected and resuspended in 1 mL of 20 mM Tris-HCl pH 7.2 [102]. To perform label-free proteomic analysis, the protein extract was concentrated using a spin column with a 10 kDa threshold (Millipore, Billerica, MA, USA). The protein was denatured (0.1% RapiGEST SF at 60°C for 15 min) (Waters, Milford, CA, USA), reduced (10 mM DTT), alkylated (10 mM iodoacetamide) and enzymatically digested with trypsin (Promega, Sequencing Grade Modified Trypsin, Madison, WI, USA).

Proteomic analysis.

Qualitative and quantitative nanoUPLC tandem nanoESI-HDMS^E (Nano Electrospray High Definition Mass Spectrometry) experiments were performed using both a 1 h reversed phase gradient from 7% to 40% (v/v) acetonitrile (0.1% v/v formic acid) at 500 nL min^-1 and a nanoACQUITY UPLC 2D RPxRP Technology system [103]. A nanoACQUITY UPLC HSS T3 1.8 μm, 75 μm × 15 cm column (pH 3) was used with an RP XBridge BEH130 C18 5 μm 300 μm x 50 mm nanoflow column (pH 10). Typical on-column sample loads were 250 ng of the total protein digests for each of the 5 fractions (250 ng/fraction/load). All analyses were performed using nano-electrospray ionization in the positive ion mode nanoESI (+) and a NanoLockSpray (Waters, Manchester, UK) ionization source. The mass spectrometer was calibrated using a MS/MS spectrum of [Glu¹]-Fibrinopeptide B human (Glu-Fib) solution (100 fmol.μL^-1) delivered through the NanoLockSpray source reference sprayer. The multiplexed data-independent (DIA) scanning with additional specificity and selectivity for non-linear ‘T-wave’ ion mobility (HDMS^E) experiments were performed using a Synapt G2-S HDMS mass spectrometer (Waters, Manchester, UK).

Following the identification of proteins, the quantitative data were packaged using dedicated algorithms [104; 105] and searching against a database with default parameters to account for ions [106]. The databases used were reversed “on-the fly” during the database queries and appended to the original database to assess the false positive rate during identification. For proper spectra processing and database searching conditions, the ProteinLynxGlobalServer v.2.5.2 (PLGS) with Identity^E and Expression^E informatics v.2.5.2 (Waters) was used. UniProtKB with manually reviewed annotations was used, and the search conditions were based on taxonomy (L. lactis). The maximum allowed missed cleavages by trypsin were up to one, variable modifications by carbamidomethyl (C), acetyl N-terminal, phosphoryl (STY) and oxidation (M) were allowed, and a peptide mass tolerance value of 10 ppm was used [107]. The collected proteins were organized by the PLGS Expression^E tool algorithm into a statistically significant list that corresponded to higher or lower regulation ratios among the different groups. For protein quantification, the PLGS v2.5.2 software was used with the Identity^E algorithm using the Hi3 methodology. The search threshold to accept each spectrum was the default value in the program with a false positive value of 4%. The quantitative values were averaged over all samples, and the standard deviations at p < 0.05 were determined using the Expression software [107].