Appl Environ Microbiol 77(3): 954-965

PMC: PMC3028724

PMID: 21131523

Genomic Diversity of Lactobacillus salivarius<a href="#fn2" rid="fn2" class=" fn">▿</a> <a href="#fn1" rid="fn1" class=" fn">†</a>

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The L. salivarius strains used in this study are listed in Table Table1.1. L. rhamnosus GG (54) was also employed as a positive control for biofilm formation in this study. The strains were routinely cultured at 37°C under microaerophilic conditions (5% CO₂) in de Man-Rogosa-Sharpe (MRS) medium (Oxoid Ltd., Basingstoke, Hampshire, United Kingdom). Bacterial strains were maintained as frozen stocks at −80°C in 25% glycerol.

TABLE 1.

L. salivarius strains used in this study

Strain	Origin	Source or reference^a
UCC118	Human ileal-cecal region	12,61
UCC119	Chicken cecum	51
AH4231	Human ileal-cecal region	61
AH4331	Human ileal-cecal region	61
AH43310	Human ileal-cecal region	61
AH43324	Human ileal-cecal region	61
AH43348	Human ileal-cecal region	61
JCM1040	Human intestine	42
JCM1042	Human intestine	42
JCM1044	Human intestine	42
JCM1045	Human intestine	42
JCM1046	Swine intestine	42
JCM1047	Swine intestine	42
JCM1230	Chicken intestine	42
CCUG47825	Human blood, 55-year-old female	CCUG
CCUG47826	Human blood, 55-year-old female	CCUG
CCUG45735	Human blood	CCUG
CCUG43299	Human blood	CCUG
CCUG38008	Human gall, 73-year-old male	CCUG
CCUG47171	Human tooth plaque	CCUG
CCUG44481	Bird	CCUG
CCUG27530B	Human abdomen, abscess	CCUG
DSM20554^T	Human saliva	49
DSM20555^T	Human saliva	49
DSM20492	Human saliva	DSM
NCIMB8816	Human saliva, Italy	NCIMB
NCIMB8817	Turkey feces	NCIMB
NCIMB8818	St. Ivel cheese	NCIMB
NCIMB702343	Unknown	NCIMB
LMG14476	Cat with myocarditis	LMG
LMG14477	Parakeet with sepsis	LMG
01M14315	Human gallbladder pus	64
L21	Human feces	-^b

^{CCUG, Culture Collection University Göteborg; DSM, Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH; JCM, Japan Collection of Microorganisms; LMG, Laboratorium voor Microbiologie, Universiteit Gent; NCIMB, National Collections of Industrial Food and Marine Bacteria.}

^{Provided by Professor Gerald Tannock, University of Otago, Otago, New Zealand.}

Microarray description, labeling, and hybridization.

Genomic DNA (gDNA) isolation was performed as described previously (38). The array-CGH platform was a customized high-definition microarray manufactured by Agilent Technologies, as described by Fang et al. (21). Briefly, the L. salivarius array contained 60-mer oligonucleotides corresponding to 2,184 genes (including annotated pseudogenes) in the genome of L. salivarius UCC118. A maximum of four probes (21 replicates) for each gene were designed with eArray (Agilent Technologies) from each open reading frame (smaller genes had fewer probes) and were spaced throughout the coding region. A total of 1,500 Agilent quality control spots were also included on the array. The gDNA to be analyzed was fragmented to an average size of 100 to 600 bp by sonication at 10 A for a total of 20 cycles (one cycle equals 30 s on, 30 s off) in iced water. Fragmentation was confirmed by agarose gel electrophoresis. Test and reference gDNAs were labeled using randomly primed fluorescent polymerization reactions. Briefly, 2 μg of template DNA and 3 μg of random nonamers (MWG) were combined in a reaction volume of 41.5 μl, denatured at 95°C for 5 min, and snap cooled on ice. Deoxynucleoside triphosphate (dNTP) solution (1 μl; 5 mM each dATP, dGTP, and dTTP, plus 2 mM dCTP), 5 μl of 10× exo-Klenow buffer, 1 μl of exo-Klenow (3 to 9 U/μl), and 1.5 μl of the fluorescent nucleotide analog cyanine 3-dCTP or cyanine 5-dCTP (Amersham Pharmacia Biotech) were added to the labeling mixture containing either test or reference template DNA. Labeling reaction mixtures were incubated at 37°C for 120 min, followed by incubation at 75°C for 15 min to stop the reaction. Labeled test and reference target DNAs were purified with a Qiagen MiniElute Purification kit (Qiagen, Valencia, CA) according to the manufacturer's instructions, with the following modifications: labeled test and reference DNA samples were individually purified and quantified using the NanoDrop ND-1000 UV-Vis spectrophotometer (NanoDrop Technologies, Rockland, DE). The labeled samples were then mixed with the components of an Agilent Oligo CGH/chip-on-chip hybridization kit, according to the manufacturer's instructions, with the addition of 0.54 μg of salmon sperm DNA prior to denaturation. Hybridizations were performed in an Agilent hybridization oven (G2545A) at 65°C for 24 h. Each array was washed in a 50-ml conical polypropylene tube using 50 ml of wash buffer 1 (Agilent Technologies, CA) for 5 min at room temperature, 50 ml of wash buffer 2 (Agilent) prewarmed to 37°C for 1 min, 50 ml acetonitrile (Sigma; >99.5% analytical grade) at room temperature, and then stabilization and drying solution (Agilent) for 30 s at room temperature. Two dye swap replicate hybridizations were performed for each strain tested.

Microarray data acquisition and analysis.

Slides were scanned using the Agilent Microarray Scanner System (G2505B) with Agilent scan control software version 7.0 for the 44k microarray at a resolution of 5 μm and red and green photomultiplier tube (PMT) settings at 10. Agilent Feature Extraction software version 9.1 was used for feature extraction. Default settings were employed, except that linear and locally weighted scatterplot smoothing (Lowess)-based normalization was performed on the data set and background-subtracted signals were extracted for both the red (Cy5) and green (Cy3) channels. Low-signal spots were removed using the software-defined flag IsWellAboveBG, and control spots were omitted from further analysis. Microarray data outliers were removed with the Grubbs test (22), and the mean of replicate probe values was calculated. P values were calculated according to the Cyber T test (4). Normalized signal ratios were transformed to their base 2 logarithm value, log₂ (test/reference). The distribution of the log₂-transformed signal ratios for each hybridization reaction was analyzed separately, and the mode of the normal distribution of the ratio values fitting the main peak was calculated. The log₂ ratio value of each averaged replicate probe was then modified in order to shift the main peak of each hybridization reaction to center around zero. The method, modified from Chen et al. (11), was performed to further reduce the effects of hybridization efficiency variation on the data set.

The genes used for MLST analysis in this study were used to validate the normalization method for the data set. Cutoff values for genes presumed to be “present,” “divergent,” and “absent” genes in our analysis were chosen based on the BLASTN (3) alignments resulting from a comparison of the probe sequences for each open reading frame on the L. salivarius array to the draft genome sequence of the type strain, DSM20555⁽http://www.ncbi.nlm.nih.gov/sites/entrez?db=genome&cmd=Retrieve&dopt=Overview&list_uids=6258). The resulting BLASTN Bit scores and E values were compared to the empirically determined signal ratios, and the following log₂ cutoff values were chosen: highly conserved ≥ −1.5 ≥ conserved ≥ −2.4 ≥ divergent ≥ −4.5 ≥ highly divergent ≥ −5.8 ≥ absent. Genes that gave low signal intensity values in comparison to the reference strain, UCC118, were considered to contain sufficient sequence divergence so that strong hybridization did not occur under the conditions of stringency used. Genes that gave an amplified signal in the test strain in comparison to the reference strain were considered to be present in additional copies in the test strain. As a second method to benchmark CGH data and cutoff values, we used the previously determined pattern of hybridization signals from Southern blotting of the same L. salivarius strain collection, with three genes for carbohydrate-metabolizing enzymes and two bacteriocin genes (37).

Clustering of the CGH data was performed using consensus present and divergent (coupled) calls and highly divergent and absent (coupled) calls for all CGH experiments were converted to the integers 1 (present) and absent (0). CGHdist (14) software was then used to estimate distance matrices based on the gene content derived from the results of the gene calls. A neighbor-joining tree was then generated using SplitsTree 4.8 software (24). Clustering of the strains based on CGH results was performed by hierarchical clustering using the complete linkage clustering method implemented in Genesis software (55).

MLST.

The nucleotide sequences of the intragenic regions of all of the following genes were used for MLST analysis: parB, pstB, rpsB, pheS, ftsQ, nrdB, and rpoA (see Table S1 in the supplemental material). These genes were chosen on the basis of the essential nature of their gene products, their sizes and chromosomal locations, and, in some cases, their effective use, or that of genes with similar functions, in previous Lactobacillus MLST studies (5, 8, 19, 50). Primers were designed using Primer 3 software (32). An approximately 900-bp internal fragment of each gene was amplified to allow accurate sequencing of a 700- to 800-bp fragment within each amplicon, using the primers specified in Table S2 in the supplemental material, on both the forward and reverse strands. DNA was sequenced by MWG Biotech (Ebersberg, Germany). Different allelic sequences, with at least one nucleotide difference per allele, were assigned arbitrary numbers. A combination of seven alleles defined the allelic profile of each strain, and a unique allelic profile was designated with a sequence type (ST). Strains with the same ST are considered to be members of a single clone or lineage. A similarity matrix was generated from the ST data using a Web-based version of SplitsTree 4 (http://pubmlst.org/perl/mlstanalyse/mlstanalyse.pl?site=pubmlst). Split decomposition analysis of the allelic profile data and individual alleles was performed using SplitsTree 4.8 (24). Network-like structures revealed incompatible phylogenetic signals at the pheS and ftsQ loci. Subsequent phylogenetic analysis was carried out using in-frame concatenation of the genes parB, rspB, rpoA, pstB, and nrdB for each strain. A 3,862-bp-long concatenated artificial sequence (CAS) was created for each strain by in-frame concatenation of the sequences of the five gene fragments. MEGA software version 4.8 (57) was used to perform multiple-sequence alignments and phylogenetic tree generation of both the single gene and the CAS of each strain. Neighbor-joining trees were generated using the Kimura two-parameter method (29) for each of the five genes, and a supertree of the CAS was generated using the same parameters. The reliability of the groups was evaluated by bootstrap testing with 1,000 resamplings. All of the MLST data from this study have been deposited in the L. salivarius MLST database (http://pubmlst.org/lsalivarius/).

EPS isolation and quantification.

Semidefined medium (SDM) was used to assess the capacity of L. salivarius to produce EPS when grown on different carbon sources. The SDM was modified from that of Kimmel and Roberts (28) (see Table S3 in the supplemental material) and contained either glucose (glu-SDM), galactose (gal-SDM), or sucrose (suc-SDM) as the carbon source (28). Ten milliliters of each of the media was inoculated with each of the test strains, which had been precultured in the relevant SDM. Cultures were harvested for EPS extraction following static incubation at 30°C till they reached early stationary phase. Differentiation of bound EPS (EPS-b) from released EPS (EPS-r) was done by the method of Tallon et al. (56) with minor modifications. EPS-r was precipitated from the culture supernatant of pelleted cells (15,000 × g; 15 min; 4°C) with 2 volumes of cold (−20°C) ethanol, while EPS-b was first extracted from phosphate-buffered saline (PBS)-washed pelleted cells by overnight incubation of the PBS-washed cells at 4°C in 0.05 M EDTA prior to ethanol precipitation of the resulting culture supernatant. The EPS precipitates were centrifuged (6,000 × g; 30 min; 4°C), dried briefly at 50°C, and resuspended in double-distilled sterile water. EPS samples of L. salivarius strains UCC118 and CCUG44481 were subsequently dialyzed (cutoff, 6,000 to 8,000 Da) against 1 liter of distilled water for 24 h with three water changes per day. The total amount of carbohydrate in the EPS was determined using the phenol-sulfuric acid method (20) with glucose as a standard. The results were expressed in microgram equivalents of glucose per milliliter of growth medium. The concentration of EPS was determined in triplicate for each strain.

Confocal imaging of biofilms.

Glass coverslips (22 by 22 mm) were surface sterilized and placed horizontally into six-well tissue culture plates (Nunc, Roskilde, Denmark). Overnight cultures of L. salivarius strains and L. rhamnosus GG were harvested, and PBS-washed cells were adjusted to an optical density at 600 nm (OD₆₀₀) of 1.0. One hundred microliters of each OD-adjusted culture was inoculated into 4 ml of medium (glu-SDM, suc-SDM, or AOAC). The inoculated medium was dispensed (2 ml) into each well so that the coverslip was fully submerged, with independent biological duplicate measurements of each test strain and medium carried out. Uninoculated media were used as negative controls in the described experimental setup. Following 72 h of static incubation at 30°C, the coverslips were washed three times by immersion and agitation in PBS solution and stained with Syto 9 (7.5 mM/ml in Ringer's solution) for 30 min under light-limiting conditions at 4°C. The coverslips were then washed a further three times in PBS solution. Confocal imaging was performed with a multiphoton confocal laser scanning microscope (Zeiss LSM 510 inverted microscope). The objective lens was a Plan-Apochromat (63×/1.4 oil) (Carl Zeiss Microimaging, Inc., Thornwood, NJ). An argon laser with a maximum-emission line at 488 nm was used as the excitation source, and a long-pass filter was applied at 505 nm. Z-stack sections were collected at 0.5-μm intervals. Three-dimensional (3D) reconstitutions of the biofilms were generated by Zen MicroImaging software (2009). Background subtraction of images was performed using IMARIS software (Bitplane, Zürich, Switzerland), and biofilm thickness was quantified using Comstat2 (23) under the ImageJ shell. Thicknesses of biofilms were expressed as the mean of replicate measurements ± the standard deviation.

Statistical analysis.

A one-way analysis of variance was performed in order to test the significance of differences in the thickness of biofilms when grown in different media. Differences in measurements were considered significant when the P value was <0.05.

Microarray data accession number.

The microarray data can be found at EMBL-EBI Array Express under accession number E-MEXP-3036.

Bacterial strains and culture conditions.

TABLE 1.

L. salivarius strains used in this study

Strain	Origin	Source or reference^a
UCC118	Human ileal-cecal region	12,61
UCC119	Chicken cecum	51
AH4231	Human ileal-cecal region	61
AH4331	Human ileal-cecal region	61
AH43310	Human ileal-cecal region	61
AH43324	Human ileal-cecal region	61
AH43348	Human ileal-cecal region	61
JCM1040	Human intestine	42
JCM1042	Human intestine	42
JCM1044	Human intestine	42
JCM1045	Human intestine	42
JCM1046	Swine intestine	42
JCM1047	Swine intestine	42
JCM1230	Chicken intestine	42
CCUG47825	Human blood, 55-year-old female	CCUG
CCUG47826	Human blood, 55-year-old female	CCUG
CCUG45735	Human blood	CCUG
CCUG43299	Human blood	CCUG
CCUG38008	Human gall, 73-year-old male	CCUG
CCUG47171	Human tooth plaque	CCUG
CCUG44481	Bird	CCUG
CCUG27530B	Human abdomen, abscess	CCUG
DSM20554^T	Human saliva	49
DSM20555^T	Human saliva	49
DSM20492	Human saliva	DSM
NCIMB8816	Human saliva, Italy	NCIMB
NCIMB8817	Turkey feces	NCIMB
NCIMB8818	St. Ivel cheese	NCIMB
NCIMB702343	Unknown	NCIMB
LMG14476	Cat with myocarditis	LMG
LMG14477	Parakeet with sepsis	LMG
01M14315	Human gallbladder pus	64
L21	Human feces	-^b

^{Provided by Professor Gerald Tannock, University of Otago, Otago, New Zealand.}

Microarray description, labeling, and hybridization.

Microarray data acquisition and analysis.

MLST.

EPS isolation and quantification.

Confocal imaging of biofilms.

Statistical analysis.

Microarray data accession number.

The microarray data can be found at EMBL-EBI Array Express under accession number E-MEXP-3036.

RESULTS

Comparative genome hybridization reveals unusually high-level diversity in L. salivarius.

Using an array based on the genes annotated in the L. salivarius UCC118 genome, we performed CGH on 32 additional strains with diverse origins. For the initial data analysis, the smaller plasmids pSF118-20 and pSF118-44 and the megaplasmid pMP118 were not excluded. A heat map constructed from hybridization signals (Fig. (Fig.1)1) clearly illustrates the presence of 18 regions at which genomic diversity is concentrated. These functions involve transposases, bacteriophage genes, CRISPR loci, EPS biosynthesis, and carbohydrate metabolism, which have been recognized as being encoded by hypervariable regions in other lactobacilli (5).These hypervariable regions did not always align with regions of anomalous (G+C) mol% content (Fig. (Fig.1),1), suggesting that these regions were not acquired by horizontal gene transfer but may have been inherited from the ancestral L. salivarius genome and subsequently lost over time. The putative conjugation region of the megaplasmid pMP118, previously noted as being nonfunctional, was also highly divergent. Based upon hierarchical clustering, three major divisions (Fig. 1A to C) were distinguished and strains were assigned to clusters that occurred at the first major branching point of the dendrogram. One of these contained five out of nine animal isolates, but no other discrete strain clusters were identified, apart from the grouping of recent local intestinal isolates (Fig. (Fig.1).1). Among these, strains AH43310 and AH43324 showed complete conservation of all loci tested, except those on the 20-kb plasmid.

An external file that holds a picture, illustration, etc.
Object name is zam9991017380001.jpg

FIG. 1.

CGH analysis of 33 L. salivarius strains. The CGH data are ordered according to the organization of the UCC118 genome, with replicons ordered left to right as follows: chromosome, pMP118, pSF118-20, and pSF118-44. The color legend corresponds to the log₂ values of normalized hybridization signal ratios (test strain/reference strain) on the right. The gradient goes from black to blue to yellow to depict the absence, conservation, or overrepresentation of a gene in the test strain. The dendrogram shows the relationship of the test strains compared to the UCC118 genome, using hierarchical clustering of CGH data with a Euclidean distance. The GC% of the UCC118 sequence is mapped onto the concatenated replicons under the genomic diversity map. The numbers 1 to 18 at the top indicate hypervariable genomic regions in L. salivarius: 1, CRISPR genes; 2, carbohydrate metabolism; 3, Sal2; 4, hypothetical proteins; 5, transposases; 6, Sal1; 7, EPS cluster 1; 8, Sal4; 9, mucus-binding protein; 10, hypothetical proteins; 11, hypothetical proteins; 12, EPS cluster 2; 13, Sal3; 14, mannose phosphotransferase system (PTS); 15, ABC transporter; 16, conjugation region; 17, bacteriocin locus; 18, small plasmids.

The genomic diversity revealed by CGH is summarized in Table Table2,2, which shows the conservation levels of genes or regions of interest in the test strains relative to UCC118. Conservation in pseudogene numbers varied among the tested isolates, with some strains lacking more than 20 of the pseudogenes identified in UCC118. This may be indicative of genome decay and may be an indication of ongoing adaptation within the species L. salivarius. Only one strain, other than the two nearly identical AH isolates, harbored a complete bacteriophage identical to that in UCC118. Strain NCIMB8818 harbored Sal1; this strain is a cheese isolate from the United Kingdom and is unlikely to be clonally related to the AH or UCC strains. All other strains substantially lacked Sal1 and Sal2 prophages. Restriction-modification (RM) systems act as a barrier to bacteriophage infection (25); the UCC118 genome includes an unusual shufflon that provides potential for encoding multiple type I RM systems. Where CGH indicated divergence of any gene within the RM locus, as occurred in 20 out of 32 strains, this divergence was invariably in the gene encoding the specificity-determining substrate. A further 4 strains completely lacked this RM shufflon, including DSM20555; this strain has recently been sequenced by the Human Microbiome Consortium and may represent a useful transformation recipient.

TABLE 2.

Strain-specific characteristics of 33 L. salivarius strains as determined by CGH analysis

Origin	Strain	Characteristics
		Pseudogene		Prophage		RMS^c	bsh1^d	Abp118^e	2-C-R-S		Regulators		Mannose PTS		MBP^lspA
		HD	Abs	SalI	Sal2	RMS^c	bsh1^d	Abp118^e	HD	Abs	CS	MP	A	B	MBP^lspA
Human	UCC118	0	0	+	+	+	+	+	0	0	53	7	+	+	+
	AH43310	0	0	+	+	+	+	+	0	0	53	7	+	+	+
	AH43324	0	0	+	+	+	+	+	0	0	53	7	+	+	+
	AH4231	3	12	∼	∼	∼	+	+	0	0	49	4	+	−	+
	AH4331	15	0	∼	∼	∼	+	+	0	0	50	4	+	∼	+
	AH43348	8	6	∼	∼	∼	+	+	0	0	50	4	+	∼	+
	L21	3	19	∼	∼	∼	+	+	0	0	50	4	+	∼	−
	DSM20554	4	18	∼	∼	∼	+	−	0	1	48	4	−	−	−
	DSM20555	6	12	∼	−	−	+	∼	0	1	49	5	+	∼	+
	DSM20492	1	18	∼	∼	∼	+	−	0	1	47	4	−	−	−
	NCIMB8816	2	19	∼	∼	∼	+	∼	0	0	48	3	−	−	−
	CCUG47171	5	16	∼	∼	∼	+	∼	1	0	47	4	−	−	∼
	JCM1040	2	20	∼	∼	+	+	−	0	1	51	3	+	−	−
	JCM1042	3	18	∼	−	∼	+	∼	1	0	48	3	−	−	−
	JCM1044	3	14	∼	−	∼	+	∼	1	0	48	3	−	−	−
	JCM1045	6	13	∼	∼	∼	+	∼	1	0	51	4	+	∼	+
	CCUG47825	11	14	∼	∼	∼	+	+	0	0	51	4	+	∼	−
	CCUG47826	4	19	∼	∼	∼	+	+	0	0	50	4	+	−	−
	CCUG45735	4	12	∼	∼	∼	+	+	0	0	49	4	+	−	+
	CCUG43299	4	17	∼	∼	+	+	−	0	1	51	3	+	−	−
	CCUG38008	3	20	∼	∼	∼	+	∼	0	0	50	3	+	−	−
	CCUG2753OB	2	9	∼	∼	+	+	+	0	0	52	4	+	∼	+
	O1M14315	3	19	∼	∼	+	+	−	2	2	51	3	+	−	−
Animal	NCIMB8817	16	8	∼	∼	∼	+	∼	1	0	49	5	+	∼	+
	JCM1046	5	14	∼	∼	+	+	∼	1	0	51	3	+	∼	−
	JCM1047	6	21	∼	−	∼	+	∼	0	0	47	3	−	−	+
	LMG14477	4	15	∼	∼	∼	∼	∼	0	0	46	3	−	−	+
	LMG14476	4	16	∼	∼	−	+	∼	0	0	47	5	+	−	+
	CCUG44481	7	18	∼	∼	∼	+	∼	0	0	47	5	−	−	∼
	UCC119	2	27	∼	∼	−	+	+	0	1	45	4	−	−	∼
	JCM1230	5	26	∼	−	∼	−	∼	0	0	45	1	+	−	+
Food	NCIMB8818	2	12	+	∼	+	+	+	0	0	52	4	+	−	+
Unknown	NCIMB 702343	4	22	∼	∼	−	+	∼	1	0	49	4	+	−	∼

^{∼, highly divergent and absent locus/region; −, absent locus/region; +, conserved locus/region. Pseudogenes were compared to 64/72 pseudogenes of UCC118 that are represented on the microarray. HD, highly divergent; Abs, absent; CS, genes/loci associated with the chromosome of UCC118; MP, genes/loci associated with pMP118. All calculations excluded genes associated with pSF118-20 and pSF118-44.}

^{Conservation of intact prophage SalI (40,288 bp) and Sal2 (12,529 bp).}

^{Conservation of type 1 restriction modification system (LSL_0915-LSL_0920) encoded by UCC118.}

^{Bile salt hydrolase gene (}bsh1; LSL_1801).

^{Region associated with bacteriocin production (Abp 118, LSL_1909-LSL_1920).}

^{Two-component response regulatory systems encoded by UCC118 (encodes 9).}

^{Mucus-binding protein,}lspA (LSL_0311).

Many of the variable traits (Table (Table2)2) are related to niche adaptation or survival. Consistent with PCR-based screening (21), all L. salivarius strains harbored a gene for bile salt hydrolase, except for JCM1230 (a chicken gut isolate) and LMG14477 (a parakeet isolate). The megaplasmid-encoded structural genes for bacteriocin Abp118 production were highly conserved in 20 of the strains tested. However, of these 20 strains, 8 lacked one or the other of the two genes associated with bacteriocin export (LSL_1909 and LSL_1910) and 1 strain showed divergence of the regulator gene of the 2-component regulatory system that governs transcriptional regulation of Abp118. This diversity corroborates the observed lack of bacteriocin production in many strains despite their harboring many of the associated genes (37).

The ability to sense and respond to environmental cues is an important survival trait for many bacteria. Although the repertoire of two-component systems is relatively conserved across the strain panel, genes for individual transcriptional regulators are very divergent or absent. The presence of genes associated with mannose uptake has been associated with intestinal persistence in L. johnsonii (17). Only a minority of strains lacked either of the two mannose utilization systems, and the candidate probiotic strains had at least one, and sometimes two, mannose utilization loci. However, some strains described as being intestinal in origin lacked genes for mannose utilization. Both in vitro (7, 62) and bioinformatics (6) analyses have indicated that residence in the human gut may be promoted by expression of mucin-binding proteins. Mutation of the lspA gene of UCC118 significantly reduced adhesion to HT29 cells (62). The lspA gene was conserved in all of the AH strains, but also in almost all of the animal isolates, so its role in intestinal persistence warrants further investigation.

When the CGH data were analyzed by clusters of orthologous groups (COG) assignment (see Table S4 in the supplemental material), the widest variation was seen in COG category G for carbohydrate transport and metabolism. Genes resident on the megaplasmid of UCC118 are required to complete the pentose phosphate pathway (PPP) for heterofermentation. CGH data indicated that all strains harbored the chromosomally encoded PPP-related genes, but the pMP118-located PPP-related genes were variably present, emphasizing the importance of the megaplasmid as a reservoir for contingency metabolism genes.

Concordant CGH clustering and MLST phylogeny of L. salivarius.

To further investigate L. salivarius strain relatedness, we performed MLST. The resulting phylogeny was compared to CGH-based clustering, from which the smaller plasmids were excluded (Fig. (Fig.2).2). The MLST-based tree (Fig. (Fig.2A)2A) was robust, supported by high bootstrap values. Three major clades were evident, one of which (clade C) included five of the animal isolates. Although the topology and primary nodes of the CGH tree were not identical, the fine grouping of strains was broadly concordant, with 15 strains sharing the same grouping pattern in both the MLST and CGH trees, as shown by the numbering (numbers 1 to 6) in Fig. Fig.2,2, and in addition, a further 7 strains show similar grouping patterns in both trees. In the MLST tree, all but one blood isolate (CCUG43299) clusters with AH candidate probiotic isolates of intestinal origin, and this blood isolate also clusters with JCM1040, which is of human intestinal origin. Thus, L. salivarius strains that are isolated from blood or tissue (e.g., gallbladder or pus) are not genetically distinct based upon the tested comparisons.

An external file that holds a picture, illustration, etc.
Object name is zam9991017380002.jpg

FIG. 2.

(A) SplitsTree (v.4.8) (24) was used to generate a neighbor-joining tree of maximum-likelihood-based distances generated by CGHdist (v.1) (http://cbr.jic.ac.uk/dicks/software/cghdist/index.html, based on CGH). The scale bar represents the number of gene differences (present or divergent/absent) per gene site. (B) Supertree generated from the concatenation of 5 MLST gene fragments. The tree was generated using the Kimura two-parameter method and neighbor-joining algorithm. Bootstrap values (1,000 replicates) over 60% are shown at the nodes. The scale bar represents the number of substitutions per site. See the text for an explanation of major clades, indicated alphanumerically.

Separation of L. salivarius strains by EPS gene content.

The overall CGH analysis (Fig. (Fig.1)1) indicated that the two clusters for EPS production in the UCC118 genome were highly divergent among the tested isolates. EPS has a number of biologically significant roles in commensal lactobacilli, including stress resistance, adhesion, and interaction with the immune system (34). The distribution of genes in EPS clusters 1 and 2 was therefore examined in greater detail (Fig. (Fig.3).3). Based upon cluster 1, four groups of strains were distinguished. Group D contained 6 of the 9 animal isolates and essentially lacked the entire EPS cluster 1. Groups A and B contained most of the human isolates. The genetic diversity in groups B and C was concentrated in two regions encoding functionally related sets of genes, both involving genes that govern EPS sugar content. Two of the AH candidate probiotic strains in group A displayed gene profiles identical to those of UCC118, as did the food isolate NCIMB8818 and the clinical isolate CCUG2753OB. These five strains also showed complete conservation of EPS gene cluster 2 (Fig. (Fig.3B).3B). Even though the remaining strain groupings were characterized by divergence or total absence of the central 17 genes in EPS cluster 2, strain group F corresponded closely to strain group B for EPS cluster 1. This strain group is most proximal to the one (group A/E) that includes UCC118 and two of the AH strains. The strain group including the animal isolates was most distant. Thus, the distribution and relatedness of genes for EPS biosynthesis is rationally related to probiotic potential and/or human intestinal origin.

An external file that holds a picture, illustration, etc.
Object name is zam9991017380003.jpg

FIG. 3.

CGH data for the EPS cluster 1 (A) and cluster 2 (B) regions of UCC118. A dendrogram was generated from the EPS cluster 1 and EPS cluster 2 CGH data using hierarchical clustering with a Euclidean distance metric. The color legend corresponds to the log₂ values of normalized hybridization signal ratios (test strain/reference strain) on the right. The gradient goes from black to blue to yellow to depict the absence, conservation, or overrepresentation of a gene sequence within the test strain. Each four-digit number corresponds to the locus identity of the represented gene, which includes the prefix LSL_.

EPS production varies independently of CGH-based groups.

Production of EPS has been well characterized for commensal lactobacilli, but not for L. salivarius. To search for correlations with CGH-based strain groups, we screened cell-bound and released EPS production levels in the panel of strains. This was performed in the presence of three different sugars, since the available carbon source can limit EPS production (26). The EPS levels for all 33 strains are shown in Fig. S1 in the supplemental material; data for eight strains are shown in Fig. Fig.4.4. These strains were selected to cover the major CGH strain groupings plus additional isolation sources. For all strains, the EPS production level varied depending on the available carbon source, and no single carbon source supported high-level EPS production in all strains. The levels of cell-bound and released EPS varied independently by strain. Perhaps surprisingly, since EPS production might be considered a probiotic-related trait, UCC118 and a closely related group A/E strain of food origin, NCIMB8818, both produced relatively low levels of EPS on the sugars tested (Fig. (Fig.4).4). The highest levels of bound EPS were produced by a blood isolate (CCUG47826) and a saliva isolate (DSM20492). The bird isolate CCUG44481 produced dramatically higher levels of released EPS than any other strain, 28 times higher levels than the next most highly producing strain, CCUG47826. The released EPS levels for the 33 strains had been determined by the described screening method. Subsequently, the CCUG44481 EPS-r was purified by dialysis to remove residual monosaccharide carryover (resulting from the described screening method) from the fermentation media, which may artificially inflate EPS values. Dialysis reduced the measured EPS level by 23% in CCUG44481.

An external file that holds a picture, illustration, etc.
Object name is zam9991017380004.jpg

FIG. 4.

EPS-r (A) and EPS-b (B) production levels of 8 L. salivarius strains representative of the groups illustrated in Fig. Fig.33 and the origins from which the strains were isolated. The strains were cultured in three growth media, gal-SDM, glu-SDM, and suc-SDM. NCIMB8816 did not grow in gal-SDM, and therefore, analysis of the strain in that medium was not included. All cells were incubated at 30°C until early stationary phase. EPS is expressed in μg/ml culture. The error bars represent the standard deviations of three replicate experiments.

Effect of EPS production on biofilm formation.

It has been suggested that EPS production might confer adhesion ability on commensal bacteria (63). In contrast, it has also been proposed that EPS may have a shielding effect on surface components that are responsible for the adhesive properties of some strains. Indeed, a recent study has demonstrated that an EPS-deficient mutant of L. rhamnosus GG has reduced capacity to produce biofilms (35). We therefore tested biofilm formation in L. salivarius strain UCC118 and strain CCUG44481, the highest EPS producer. We benchmarked against L. rhamnosus GG (Fig. (Fig.5A),5A), which has been previously shown to produce monospecies biofilms (36). We tested the three strains in suc-SDM, a major modifier of EPS production levels in L. salivarius (this study); glu-SDM, which contains the carbon source that is regularly used to culture L. salivarius; and AOAC medium, which allowed direct comparison with the published biofilm production data for strain LGG (36). In addition to being a strain-dependent trait within the species L. salivarius, the capacity to form biofilms was also highly medium dependent. Of the three strains examined, UCC118 consistently formed the thickest biofilms in the three media tested. A significant reduction (P = 0.001) in biofilm thickness was seen when UCC118 was grown in AOAC medium (4.13 ± 0.75 μm) in comparison to suc-SDM (8.2 ± 0.97 μm) or glu-SDM (6.75 ± 1.2 μm). In contrast to UCC118, CCUG44481 preferentially formed biofilms in AOAC medium (3.8 ± 0.52 μm) over either suc-SDM (2.0 ± 0.7 μm) or glu-SDM (2.25 ± 0.65 μm), in which it formed thin biofilms containing many voids and hollows. L. rhamnosus GG also formed thin biofilms on the glass substrate, with depths similar to those of CCUG44481 on both suc-SDM (2.5 ± 0.7 μm) and glu-SDM (2.0 ± 0 μm). There was no significant difference in the depths of the biofilms formed by strain LGG in the three media tested, and similarly to CCUG44481, voids and hollows were present throughout the LGG biofilms. Prominent 3D structures (Fig. (Fig.5B)5B) were visible within the architecture of the L. salivarius biofilm. These features were evident for both L. salivarius strains when the medium used was capable of supporting biofilms with a minimum depth of 3.8 μm. These structures, which resembled flat-topped mushrooms, were completely absent from the LGG images. The biofilms formed by strain LGG were more easily dislodged by the washing steps employed than L. salivarius biofilms. This may indicate that strain LGG has a reduced affinity for the glass substrate employed in this study than the L. salivarius strains. The EPS production levels of the strains tested could not be quantified accurately when grown in the AOAC medium. However, biofilm formation by the two L. salivarius strains was inversely proportional to released-EPS levels when grown on sucrose and glucose.

An external file that holds a picture, illustration, etc.
Object name is zam9991017380005.jpg

FIG. 5.

L. salivarius and L. rhamnosus GG biofilms examined by confocal microscopy. (A) All strains were cultured statically in three media, AOAC, glu-SDM, and suc-SDM, for 72 h at 30°C, as indicated on the left. Bacterial cells (green) were stained with Syto 9. The colored bars on the edges of the images represent the orientation of the view point; red indicates the x axis, yellow indicates the y axis, and turquoise indicates the z axis. Representative images from two independent experiments are shown as single and stacked optical sections of each strain in each medium at ×630 magnification. Biofilm thickness measurements (± standard deviation among replicated measurements) are represented for each condition tested in the top left corner of each image. (B) (1) Zoomed architecture (×200) of a UCC118 biofilm when grown in suc-SDM. (2) 3D reconstitution of the surface view and cross section of a UCC118 biofilm grown in suc-SDM. The color legend corresponds to the depth of the biofilm in μm.

Comparative genome hybridization reveals unusually high-level diversity in L. salivarius.

FIG. 1.

TABLE 2.

Strain-specific characteristics of 33 L. salivarius strains as determined by CGH analysis

Origin	Strain	Characteristics
		Pseudogene		Prophage		RMS^c	bsh1^d	Abp118^e	2-C-R-S		Regulators		Mannose PTS		MBP^lspA
		HD	Abs	SalI	Sal2	RMS^c	bsh1^d	Abp118^e	HD	Abs	CS	MP	A	B	MBP^lspA
Human	UCC118	0	0	+	+	+	+	+	0	0	53	7	+	+	+
	AH43310	0	0	+	+	+	+	+	0	0	53	7	+	+	+
	AH43324	0	0	+	+	+	+	+	0	0	53	7	+	+	+
	AH4231	3	12	∼	∼	∼	+	+	0	0	49	4	+	−	+
	AH4331	15	0	∼	∼	∼	+	+	0	0	50	4	+	∼	+
	AH43348	8	6	∼	∼	∼	+	+	0	0	50	4	+	∼	+
	L21	3	19	∼	∼	∼	+	+	0	0	50	4	+	∼	−
	DSM20554	4	18	∼	∼	∼	+	−	0	1	48	4	−	−	−
	DSM20555	6	12	∼	−	−	+	∼	0	1	49	5	+	∼	+
	DSM20492	1	18	∼	∼	∼	+	−	0	1	47	4	−	−	−
	NCIMB8816	2	19	∼	∼	∼	+	∼	0	0	48	3	−	−	−
	CCUG47171	5	16	∼	∼	∼	+	∼	1	0	47	4	−	−	∼
	JCM1040	2	20	∼	∼	+	+	−	0	1	51	3	+	−	−
	JCM1042	3	18	∼	−	∼	+	∼	1	0	48	3	−	−	−
	JCM1044	3	14	∼	−	∼	+	∼	1	0	48	3	−	−	−
	JCM1045	6	13	∼	∼	∼	+	∼	1	0	51	4	+	∼	+
	CCUG47825	11	14	∼	∼	∼	+	+	0	0	51	4	+	∼	−
	CCUG47826	4	19	∼	∼	∼	+	+	0	0	50	4	+	−	−
	CCUG45735	4	12	∼	∼	∼	+	+	0	0	49	4	+	−	+
	CCUG43299	4	17	∼	∼	+	+	−	0	1	51	3	+	−	−
	CCUG38008	3	20	∼	∼	∼	+	∼	0	0	50	3	+	−	−
	CCUG2753OB	2	9	∼	∼	+	+	+	0	0	52	4	+	∼	+
	O1M14315	3	19	∼	∼	+	+	−	2	2	51	3	+	−	−
Animal	NCIMB8817	16	8	∼	∼	∼	+	∼	1	0	49	5	+	∼	+
	JCM1046	5	14	∼	∼	+	+	∼	1	0	51	3	+	∼	−
	JCM1047	6	21	∼	−	∼	+	∼	0	0	47	3	−	−	+
	LMG14477	4	15	∼	∼	∼	∼	∼	0	0	46	3	−	−	+
	LMG14476	4	16	∼	∼	−	+	∼	0	0	47	5	+	−	+
	CCUG44481	7	18	∼	∼	∼	+	∼	0	0	47	5	−	−	∼
	UCC119	2	27	∼	∼	−	+	+	0	1	45	4	−	−	∼
	JCM1230	5	26	∼	−	∼	−	∼	0	0	45	1	+	−	+
Food	NCIMB8818	2	12	+	∼	+	+	+	0	0	52	4	+	−	+
Unknown	NCIMB 702343	4	22	∼	∼	−	+	∼	1	0	49	4	+	−	∼

^{Conservation of intact prophage SalI (40,288 bp) and Sal2 (12,529 bp).}

^{Conservation of type 1 restriction modification system (LSL_0915-LSL_0920) encoded by UCC118.}

^{Bile salt hydrolase gene (}bsh1; LSL_1801).

^{Region associated with bacteriocin production (Abp 118, LSL_1909-LSL_1920).}

^{Two-component response regulatory systems encoded by UCC118 (encodes 9).}

^{Mucus-binding protein,}lspA (LSL_0311).

DISCUSSION

The current study shows that up to 23.6% of the gene content of L. salivarius strains is variable compared to the UCC118 genome. The most conserved strains were human GIT isolates, while the greatest divergence occurred in animal-associated isolates. The extent of diversity in L. salivarius is higher than that revealed by CGH in L. plantarum (up to 20% gene divergence [43]) and L. casei (up to 19% gene divergence [9]), despite these two species inhabiting a wider range of environments and having considerably larger genomes than L. salivarius (31, 40). L. johnsonii (47) is primarily a GIT-associated organism, and it also displayed a higher level of gene conservation (17% divergence [5]) than L. salivarius using the gene conservation parameters applied in this study. The most conserved genes in L. salivarius were those associated with information storage and processing, while the most divergence was noted in hypothetical proteins, mobile elements, and pseudogenes. This high proportion of pseudogenes may indicate that L. salivarius is subject to an ongoing process of genome degradation for niche adaptation, as reported for other LAB (40).

Two L. salivarius strains (AH43310 and AH43324) of human intestinal origin had gene contents nearly identical to that of UCC118, indicating their potential as useful probiotic candidates. Also included in group A was one strain isolated from a diseased subject and a food isolate. The next closest group (group B) contained the remaining AH strains and, surprisingly, five septicemia isolates. Some isolates from human infections were also shown by MLST to be closely related to the group B AH strains (Fig. (Fig.1),1), and clinical isolates did not form a separate MLST cluster. This suggests that the rare cases of septicemia caused by L. salivarius are due to compromised host barriers or defenses, rather than specialized “pathogenic strains,” as has been shown for other Lactobacillus species (52). We are also currently exploring the possibility that disease isolates harbor additional genes required for enhanced pathogenic potential, which would not be identified in the CGH data.

Group C as defined by the CGH analysis included the majority of the animal isolates, five salivary isolates, and three human intestinal isolates. Although the primary CGH separation of group C is due to the small plasmids, this group was also evident from the MLST data, indicating that it reflects the evolutionary history of the whole genome. The proximity of some strains of human origin to animal-derived strains emphasizes the fact that L. salivarius strains of human origin cannot be universally expected to exert probiotic effects in humans.

It is not clear if L. salivarius is truly autochthonous in the human GIT (i.e., if it colonizes and forms a self-sustaining population [48]). The anti-infective (51) and bacteriocin-producing (37) abilities of L. salivarius are strain specific. Together with the variable presence of genes that encode and regulate bacteriocin production, CGH also revealed considerable divergence in genes for survival in the harsh physical environment of the human and animal GITs. The widespread presence of a bile salt hydrolase gene (LSL_1801) homolog in all but one of the L. salivarius strains tested indicates positive biological selection for this gene function, as described by Fang et al. (21). The presence of such a gene in strains from nonintestinal sources suggests that all or most L. salivarius strains are adapted for periods of GIT transit as some part of their long-term ecological lifestyles. Reinforcing this idea, the genes associated with mucus-binding proteins in UCC118 were similarly not exclusively found in GIT isolates. A recent study by Oh et al. found that, with few exceptions, L. reuteri populations are composed of host-specific ecotypes that have coevolved with specific vertebrates (45). In contrast, the grouping of L. salivarius strains by niche association was not unequivocally supported by the gene content of the strains, as revealed by CGH analysis, although a cleaner separation of, e.g., animal strains was provided by MLST. This does not exclude the possibility that additional niche-specific genes may be present in L. salivarius strains, which are not detectable by a CGH approach.

Despite the lack of grouping of strains based on their overall genomic profiles, most of the animal-associated isolates clustered by hierarchical analysis of EPS cluster 1 genes, but not cluster 2. However, for both clusters 1 and 2, a limited number of human-derived strains showed significant conservation of the UCC118 EPS gene complement. Berger et al. (5) demonstrated a complex patchwork pattern of EPS genes in L. johnsonii strains, with well-conserved regions alternating with regions of sequence diversity (5). The pattern of EPS gene diversity in L. salivarius is more systematic, with two readily visible blocks of diversity in EPS gene cluster 1 (Fig. (Fig.3).3). These genes are predicted by annotation to be responsible for the sugar decoration of the EPS molecule. Their variation, against a backdrop of conservation of the core genes for EPS production, suggests environmental selection or adaptation in EPS sugar content. The pattern of conservation of EPS genes and loci in L. salivarius warrants further investigation from a functional perspective and partly motivated our phenotypic analysis.

Until a recent report (39), the EPS production level of L. salivarius had not been described. L. salivarius BRC 14759 (DSM20555/JCM1231) produced a combined EPS concentration of 45.3 mg/liter in chemically defined medium supplemented with 5 g/liter lactose at 40°C (39). This is significantly lower than the levels observed for strain DSM20555 (121.48 mg/liter) in the current study. This probably reflects the importance of culture conditions used for of EPS production but may also be a feature of the screening method used in this study for handling a large number of strains. The amount of released EPS produced by CCUG44481 (8.15 g/liter) when grown in sucrose was significantly larger than those for all other strains tested on any other carbon source. EPS production levels across multiple strains did not appear to be uniformly related to either the source of isolation or the conservation of EPS cluster 1 or EPS cluster 2 but were highly dependent on the culture conditions. The production of EPS in L. salivarius strains cannot currently be attributed to either EPS cluster without functional characterization of these regions, informed by chemical analysis of purified EPS. In addition, the EPS-producing phenotype could also be the result of an additional functional EPS-related cluster of genes that are present in a subset of the L. salivarius strains but that are absent in UCC118 and therefore absent from the current CGH data.

Adherence and biofilm formation can increase the gut residence time of commensal strains, as well as promote pathogen exclusion, host-cell interaction, and immune stimulation (53). Biofilm formation has previously been shown to be a strain-specific trait in lactobacilli (36), and biofilm formation in L. salivarius was clearly strain dependent and, in addition, medium dependent (Fig. (Fig.5A).5A). It has previously been reported that the L. salivarius type strain, JCM1231 (DSM20555 in this study), does not form substantial biofilms (33). It was therefore significant that the probiotic strain UCC118 showed the highest capacity for biofilm formation among the strains tested in this study. In suc-SDM, UCC118 formed biofilms over three times thicker than those of LGG. CCUG44481 was generally incapable of forming substantial biofilms. Biofilm formation was inversely related to EPS-r, and the theory that cell surface molecules other than EPS are responsible for UCC118 biofilm formation warrants further investigation. The observation of mushroom-like protrusions from the surface of the L. salivarius biofilm mat is the first description of its kind for a Lactobacillus biofilm and thus provides insight into the architecture of biofilms of commensal lactobacilli. The measured depths of the biofilms formed by UCC118 reported here are similar to those of the probiotic strain L. reuteri ATCC 55730 (7 ± 2 μm) (27). It is interesting that UCC118 also formed thicker biofilms in vitro than those of L. reuteri 100-23 (ca. 2.5 μm after 32 h in vitro), although the variation in experimental procedure may have plausibly caused the observed difference (58). This strain has been shown to form biofilms (ca. 20 μm) on the epithelial surface of the forestomach of previously Lactobacillus-free mice (60). It is possible that the substantial biofilms formed by UCC118 may contribute to the desirable host interaction properties of the strain.

This study has established a phylogenomic framework for L. salivarius and has shown limited clustering of strains from human intestinal, animal, or blood sources. Correlation with complex phenotypes, such as EPS production, is challenging. We are currently overlaying this genomic framework with additional genome sequences and with data from a number of traits related to host interaction to facilitate biodiversity-based screening of the relevant genes.

Supplementary Material

[Supplemental material]

Click here to view.

Department of Microbiology,^{Alimentary Pharmabiotic Centre, University College Cork, Cork, Ireland,}^{Dipartimento di Biotecnologie, Università degli Studi di Verona, Verona, Italy}³

^{Corresponding author. Mailing address: Department of Microbiology and Alimentary Pharmabiotic Centre, Food Science Bldg., Western Road, University College Cork, Cork, Ireland. Phone: (353) 21 490 3997. Fax: (353) 21 490 3101. E-mail:}ei.ccu@elootowp

Received 2010 Jul 14; Accepted 2010 Nov 21.

Abstract

Strains of Lactobacillus salivarius are increasingly employed as probiotic agents for humans or animals. Despite the diversity of environmental sources from which they have been isolated, the genomic diversity of L. salivarius has been poorly characterized, and the implications of this diversity for strain selection have not been examined. To tackle this, we applied comparative genomic hybridization (CGH) and multilocus sequence typing (MLST) to 33 strains derived from humans, animals, or food. The CGH, based on total genome content, including small plasmids, identified 18 major regions of genomic variation, or hot spots for variation. Three major divisions were thus identified, with only a subset of the human isolates constituting an ecologically discernible group. Omission of the small plasmids from the CGH or analysis by MLST provided broadly concordant fine divisions and separated human-derived and animal-derived strains more clearly. The two gene clusters for exopolysaccharide (EPS) biosynthesis corresponded to regions of significant genomic diversity. The CGH-based groupings of these regions did not correlate with levels of production of bound or released EPS. Furthermore, EPS production was significantly modulated by available carbohydrate. In addition to proving difficult to predict from the gene content, EPS production levels correlated inversely with production of biofilms, a trait considered desirable in probiotic commensals. L. salivarius displays a high level of genomic diversity, and while selection of L. salivarius strains for probiotic use can be informed by CGH or MLST, it also requires pragmatic experimental validation of desired phenotypic traits.

Abstract

Lactobacillus spp. are lactic acid bacteria (LAB) that display phylogenetic, phenotypic, and ecological heterogeneity that is reflected in their taxonomic diversity (13). Lactobacilli have complex nutritional requirements that are reflected in the diverse, carbon-rich habitats in which they are found (59). Lactobacilli have been studied extensively because of their importance for the production of fermented foods and beverages (59). Some well-characterized lactobacilli are generally regarded as safe (GRAS), and in more recent times, they have been used as probiotics and vaccine carriers (30). Administration of probiotic cultures benefits the host through a wide variety of mechanisms that are increasingly recognized as being species and strain specific (34). Knowledge of the genetic basis for strain diversity in potentially probiotic species is thus called for. Comparative genomics has emerged as a powerful approach in this era of high-throughput sequencing technologies, and it provides a technological platform to identify strain-specific traits (13).

Lactobacillus salivarius (38) is part of the indigenous microbiota of the gastrointestinal tract (GIT) and oral cavity of humans and hamsters (49). The species has also been isolated from human breast milk (41) and from the intestinal tracts of swine (10) and chickens (1). There has been a recent increase in the number of studies in which the probiotic utility of diverse L. salivarius strains was explored (44). However, there is no detailed information on the genomic variability of the species to serve as a reference for identifying strain-specific properties. In this study, we examined the diversity of L. salivarius by applying multilocus sequence typing (MLST) and comparative genomic hybridization (CGH) to a collection of strains. They were derived from a range of ecological niches and were diverse in plasmid content and phenotypic traits (21, 37, 38). The panel included the strain UCC118, whose genome has been sequenced (12) and which has been extensively studied for its probiotic properties (44). MLST is a powerful sequence-based typing method that has been applied to more than 48 bacterial taxa (2). It utilizes the internal nucleotide sequences of multiple housekeeping genes to infer the genetic relatedness of bacterial strains and species. MLST has been applied to industrially relevant LAB strains (15) and to Lactobacillus species, including L. plantarum (16), L. casei (8, 18) and L. sanfranciscensis (46). CGH facilitates the comparison of unsequenced strains on a genome-wide level and can enable the correlation of phenotypic patterns within a species with the genomic content. Horizontal gene transfer (HGT) is often associated with niche adaptation and was detected by CGH studies of other Lactobacillus species, which have revealed strain-specific traits, including carbohydrate utilization and bacteriocin and exopolysaccharide (EPS) production (5, 43). One of the aims of the current study was therefore to investigate whether niche adaptation or probiotic potential was evident in the general L. salivarius population.

EPS produced by LAB has been used in the dairy industry to improve the texture, viscosity, and rheological properties of fermented products (26). EPS is also credited with health-promoting properties, such as cholesterol lowering, immunomodulation, antitumorogenic effects, and prebiotic effects (26). Thus, L. salivarius strain clusters, defined by EPS gene content, were further analyzed for EPS production and surface properties.

Click here to view.

Acknowledgments

This research was supported by Science Foundation Ireland through a Research Frontiers Programme award to P.W.O. (05/RFP/GEN047) and by a Centre for Science, Engineering and Technology award to the Alimentary Pharmabiotic Centre. E.S. was supported by “Borsa di Mobilità Socrates-Erasmus 2007-08” from the University of Verona.

We are grateful to Alimentary Health Ltd., Kinsale, Ireland, for providing L. salivarius AH strains; to Aldert Zomer for valuable discussions and advice on microarray processing and data analysis; to Marlies Mooij for advice and help related to confocal laser scanning microscopy (CLSM) of biofilms and analysis of biofilm images; and to Michael Mangan for design of the microarray used in this study.

Acknowledgments

Footnotes

^{Published ahead of print on 3 December 2010.}

^{Supplemental material for this article may be found at}http://aem.asm.org/.

Footnotes

REFERENCES

References

1. Abbas Hilmi, H. T., T. Hanan, A. Surakka, J. Apajalahti, and P. E. J. Saris. 2007. Identification of the most abundant Lactobacillus species in the crop of 1- and 5-week-old broiler chickens. Appl. Environ. Microbiol.73:7867-7873.
2. Achtman, M. 2008. Evolution, population structure, and phylogeography of genetically monomorphic bacterial pathogens. Annu. Rev. Microbiol.62:53-70. [[PubMed]
3. Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990. Basic local alignment search tool. J. Mol. Biol.215:403-410. [[PubMed]
4. Baldi, P., and A. D. Long. 2001. A Bayesian framework for the analysis of microarray expression data: regularized t-test and statistical inferences of gene changes. Bioinformatics17:509-519. [[PubMed]
5. Berger, B., et al. 2007. Similarity and differences in the Lactobacillus acidophilus group identified by polyphasic analysis and comparative genomics. J. Bacteriol.189:1311-1321.
6. Boekhorst, J., O. Helmer, M. Kleerebezem, and R. J. Siezen. 2006. Comparative analysis of proteins with a mucus-binding domain found exclusively in lactic acid bacteria. Microbiology152:273-280. [[PubMed]
7. Buck, B. L., E. Altermann, T. Svingerud, and T. R. Klaenhammer. 2005. Functional analysis of putative adhesion factors in Lactobacillus acidophilus NCFM. Appl. Environ. Microbiol.71:8344-8351.
8. Cai, H., B. T. Rodriguez, W. Zhang, J. R. Broadbent, and J. L. Steele. 2007. Genotypic and phenotypic characterization of Lactobacillus casei strains isolated from different ecological niches suggests frequent recombination and niche specificity. Microbiology153:2655-2665. [[PubMed]
9. Cai, H., R. Thompson, M. F. Budinich, J. R. Broadbent, and J. L. Steele. 2009. Genome sequence and comparative genome analysis of Lactobacillus casei: insights into their niche-associated evolution. Genome Biol. Evol.1:239-257.
10. Casey, PG., et al. 2004. Isolation and characterization of anti-Salmonella lactic acid bacteria from the porcine gastrointestinal tract. Lett. Appl. Microbiol.39:431-438. [[PubMed][Google Scholar]
11. Chen, H.-IH., et al. 2008. A probe-density-based analysis method for array CGH data: simulation, normalization and centralization. Bioinformatics24:1749-1756. [Google Scholar]
12. Claesson, MJ., et al. 2006. Multireplicon genome architecture of Lactobacillus salivarius. Proc. Natl. Acad. Sci. U. S. A.103:6718-6723. [Google Scholar]
13. Claesson, M. J., D. van Sinderen, and P. W. O'Toole. 2007. The genus Lactobacillus—a genomic basis for understanding its diversity. FEMS Microbiol. Lett.269:22-28. [[PubMed]
14. Davey, R., G. Savva, J. Dicks, and I. N. Roberts. 2007. MPP: a microarray-to-phylogeny pipeline for analysis of gene and marker content datasets. Bioinformatics23:1023-1025. [[PubMed]
15. de Las Rivas, B., A. Marcobal, and R. Munoz. 2004. Allelic diversity and population structure in Oenococcus oeni as determined from sequence analysis of housekeeping genes. Appl. Environ. Microbiol.70:7210-7219.
16. de Las Rivas, B., A. Marcobal, and R. Munoz. 2006. Development of a multilocus sequence typing method for analysis of Lactobacillus plantarum strains. Microbiology152:85-93. [[PubMed]
17. Denou, E., et al. 2008. Identification of genes associated with the long-gut-persistence phenotype of the probiotic Lactobacillus johnsonii strain NCC533 using a combination of genomics and transcriptome analysis. J. Bacteriol.190:3161-3168.
18. Diancourt, L., et al. 2007. Multilocus sequence typing of Lactobacillus casei reveals a clonal population structure with low levels of homologous recombination. Appl. Environ. Microbiol.73:6601-6611.
19. Drevinek, P., A. Baldwin, C. Dowson, and E. Mahenthiralingam. 2008. Diversity of the parB and repA genes of the Burkholderia cepacia complex and their utility for rapid identification of Burkholderia cenocepacia. BMC Microbiol.8:44-54.
20. DuBois, M., K. A. Gilles, J. K. Hamilton, P. A. Rebers, and F. Smith. 2002. Colorimetric method for determination of sugars and related substances. Anal. Chem.28:350-356. [PubMed]
21. Fang, F., et al. 2009. Allelic variation of bile salt hydrolase genes in Lactobacillus salivarius does not determine bile resistance levels. J. Bacteriol.191:5743-5757.
22. Grubbs, FE. 1969. Procedures for detecting outlying observations in samples. Technometrics11:1-21. [PubMed][Google Scholar]
23. Heydorn, A., et al. 2000. Quantification of biofilm structures by the novel computer program COMSTAT. Microbiology146:2395-2407. [[PubMed]
24. Huson, D. H., and D. Bryant. 2006. Application of phylogenetic networks in evolutionary studies. Mol. Biol. Evol.23:254-267. [[PubMed]
25. Jeltsch, A. 2003. Maintenance of species identity and controlling speciation of bacteria: a new function for restriction/modification systems? Gene317:13-16. [[PubMed]
26. Jolly, L., S. Vincent, P. Duboc, and J.-R. Neeser. 2002. Exploiting exopolysaccharides from lactic acid bacteria. Antonie Van Leeuwenhoek82:367-374. [[PubMed]
27. Jones, S. E., and J. Versalovic. 2009. Probiotic Lactobacillus reuteri biofilms produce antimicrobial and anti-inflammatory factors. BMC Microbiol.9:35.
28. Kimmel, S. A., and R. F. Roberts. 1998. Development of a growth medium suitable for exopolysaccharide production by Lactobacillus delbrueckii ssp. bulgaricus RR. Int. J. Food Microbiol.40:87-92. [[PubMed]
29. Kimura, M. 1980. A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. J. Mol. Evol.16:111-120. [[PubMed]
30. Klaenhammer, TR. 2000. Probiotic bacteria: today and tomorrow. J. Nutr.130:415S-416S. [[PubMed][Google Scholar]
31. Kleerebezem, M., et al. 2003. Complete genome sequence of Lactobacillus plantarum WCFS1. Proc. Natl. Acad. Sci. U. S. A.100:1990-1995.
32. Koressaar, T., and MRemm. 2007. Enhancements and modifications of primer design program Primer3. Bioinformatics23:1289-1291. [[PubMed][Google Scholar]
33. Kumada, M., et al. 2008. Effects of Enterococcus faecium on Streptococcus mutans biofilm formation using flow cell system. J. Oral Biosci.50:68-76. [PubMed]
34. Lebeer, S., J. Vanderleyden, and S. C. J. De Keersmaecker. 2008. Genes and molecules of lactobacilli supporting probiotic action. Microbiol. Mol. Biol. Rev.72:728-764.
35. Lebeer, S., et al. 2009. Identification of a gene cluster for the biosynthesis of a long, galactose-rich exopolysaccharide in Lactobacillus rhamnosus GG and functional analysis of the priming glycosyltransferase. Appl. Environ. Microbiol.75:3554-3563.
36. Lebeer, S., T. L. A. Verhoeven, M. Perea Velez, J. Vanderleyden, and S. C. J. De Keersmaecker. 2007. Impact of environmental and genetic factors on biofilm formation by the probiotic strain Lactobacillus rhamnosus GG. Appl. Environ. Microbiol.73:6768-6775.
37. Li, Y., et al. 2007. Distribution of megaplasmids in Lactobacillus salivarius and other lactobacilli. J. Bacteriol.189:6128-6139.
38. Li, Y., et al. 2006. Polyphasic analysis indicates that Lactobacillus salivarius subsp. salivarius and Lactobacillus salivarius subsp. salicinius do not merit separate subspecies status. Int. J. Syst. Evol. Microbiol.56:2397-2403. [[PubMed]
39. Liu, C.-T., I. T. Hsu, C.-C. Chou, P.-R. Lo, and R.-C. Yu. 2009. Exopolysaccharide production of Lactobacillus salivarius BCRC 14759 and Bifidobacterium bifidum BCRC 14615. World J. Microbiol. Biotechnol.25:883-890. [PubMed]
40. Makarova, K., et al. 2006. Comparative genomics of the lactic acid bacteria. Proc. Natl. Acad. Sci. U. S. A.103:15611-15616.
41. Martín, R., et al. 2006. Lactobacillus salivarius CECT 5713, a potential probiotic strain isolated from infant feces and breast milk of a mother-child pair. Int. J. Food Microbiol.112:35-43. [[PubMed]
42. Mitsuoka, T. 1969. Vergleichende untersuchungen Über die Laktobazillen aus den faeces von menschen, schweinen und hühnern. Bakteriologie210:32-51. [[PubMed]
43. Molenaar, D., et al. 2005. Exploring Lactobacillus plantarum genome diversity by using microarrays. J. Bacteriol.187:6119-6127.
44. Neville, B. A., and P. W. O'Toole. 2010. Probiotic properties of Lactobacillus salivarius and closely related Lactobacillus species. Future Microbiol.5:759-774. [[PubMed]
45. Oh, PL., et al. 2010. Diversification of the gut symbiont Lactobacillus reuteri as a result of host-driven evolution. ISME J.4:377-387. [[PubMed][Google Scholar]
46. Picozzi, C., G. Bonacina, I. Vigentini, and R. Foschino. 2010. Genetic diversity in Italian Lactobacillus sanfranciscensis strains assessed by multilocus sequence typing and pulsed field gel electrophoresis analyses. Microbiology:mic. 0.037341-0. [[PubMed]
47. Pridmore, RD., et al. 2004. The genome sequence of the probiotic intestinal bacterium Lactobacillus johnsonii NCC 533. Proc. Natl. Acad. Sci. U. S. A.101:2512-2517. [Google Scholar]
48. Reuter, G. 2001. The Lactobacillus and Bifidobacterium microflora of the human intestine: composition and succession. Curr. Issues Intest. Microbiol.2:43-53. [[PubMed]
49. Rogosa, M., R. F. Wiseman, J. A. Mitchell, M. N. Disraely, and A. J. Beaman. 1953. Species differentiation of oral lactobacilli from man including descriptions of Lactobacillus salivarius nov. spec. and Lactobacillus cellobiosus nov. spec. J. Bacteriol.65:681-699.
50. Ruiz-Garbajosa, P., M. J. M. Bonten, D. A. Robinson, J. Top, et al. 2006. Multilocus sequence typing scheme for Enterococcus faecalis reveals hospital-adapted genetic complexes in a background of high rates of recombination. J. Clin. Microbiol.44:2220-2228.
51. Ryan, K. A., P. Daly, Y. Li, C. Hooton, and P. W. O'Toole. 2008. Strain-specific inhibition of Helicobacter pylori by Lactobacillus salivarius and other lactobacilli. J. Antimicrob. Chemother.61:831-834. [[PubMed]
52. Salminen, MK., et al. 2004. Lactobacillus bacteremia, clinical significance, and patient outcome, with special focus on probiotic L. rhamnosus GG. Clin. Infect. Dis.38:62-69. [[PubMed][Google Scholar]
53. Servin, AL. 2004. Antagonistic activities of lactobacilli and bifidobacteria against microbial pathogens. FEMS Microbiol. Rev.28:405-440. [[PubMed][Google Scholar]
54. Silva, M., N. V. Jacobus, C. Deneke, and S. L. Gorbach. 1987. Antimicrobial substance from a human Lactobacillus strain. Antimicrob. Agents Chemother.31:1231-1233.
55. Sturn, A., J. Quackenbush, and Z. Trajanoski. 2002. Genesis: cluster analysis of microarray data. Bioinformatics18:207-208. [[PubMed]
56. Tallon, R., P. Bressollier, and M. C. Urdaci. 2003. Isolation and characterization of two exopolysaccharides produced by Lactobacillus plantarum EP56. Res. Microbiol.154:705-712. [[PubMed]
57. Tamura, K., J. Dudley, M. Nei, and S. Kumar. 2007. MEGA4: molecular evolutionary genetics analysis (MEGA) software version 4.0. Mol. Biol. Evol.24:1596-1599. [[PubMed]
58. Tannock, GW. 2005. New perceptions of the gut microbiota: implications for future research. Gastroenterol. Clin. N. Am.34:361-382. [[PubMed][Google Scholar]
59. Tannock, GW. 2004. A special fondness for lactobacilli. Appl. Environ. Microbiol.70:3189-3194. [Google Scholar]
60. Tannock, GW., et al. 2005. Ecological behavior of Lactobacillus reuteri 100-23 is affected by mutation of the luxS gene. Appl. Environ. Microbiol.71:8419-8425. [Google Scholar]
61. Thornton, GM. 1996. Probiotic bacteria: selection of Lactobacillus and Bifidobacterium strains from the healthy human gastrointestinal tract; characterisation of a novel Lactobacillus-derived antibacterial protein. Ph.D. thesis. University College Cork, Cork, Ireland.[Google Scholar]
62. van Pijkeren, JP., et al. 2006. Comparative and functional analysis of sortase-dependent proteins in the predicted secretome of Lactobacillus salivarius UCC118. Appl. Environ. Microbiol.72:4143-4153. [Google Scholar]
63. Walter, J. 2008. Ecological role of lactobacilli in the gastrointestinal tract: implications for fundamental and biomedical research. Appl. Environ. Microbiol.74:4985-4996.
64. Woo, W. C. Y., A. M. Y. Fung, S. K. P. Lau, and K.-Y. Yuen. 2002. Identification by 16S rRNA gene sequencing of Lactobacillus salivarius bacteremic cholecystitis. J. Clin. Microbiol.40:265-267.

Genomic Diversity of <em>Lactobacillus salivarius</em><sup><a href="#fn2" rid="fn2" class=" fn">▿</a></sup> <sup><a href="#fn1" rid="fn1" class=" fn">†</a></sup>

MATERIALS AND METHODS

Bacterial strains and culture conditions.

TABLE 1.

Microarray description, labeling, and hybridization.

Microarray data acquisition and analysis.

MLST.

EPS isolation and quantification.

Confocal imaging of biofilms.

Statistical analysis.

Microarray data accession number.

Bacterial strains and culture conditions.

TABLE 1.

Microarray description, labeling, and hybridization.

Microarray data acquisition and analysis.

MLST.

EPS isolation and quantification.

Confocal imaging of biofilms.

Statistical analysis.

Microarray data accession number.

RESULTS

Comparative genome hybridization reveals unusually high-level diversity in L. salivarius.

TABLE 2.

Concordant CGH clustering and MLST phylogeny of L. salivarius.

Separation of L. salivarius strains by EPS gene content.

EPS production varies independently of CGH-based groups.

Effect of EPS production on biofilm formation.

Comparative genome hybridization reveals unusually high-level diversity in L. salivarius.

TABLE 2.

Concordant CGH clustering and MLST phylogeny of L. salivarius.

Separation of L. salivarius strains by EPS gene content.

EPS production varies independently of CGH-based groups.

Effect of EPS production on biofilm formation.

DISCUSSION

Supplementary Material

Abstract

Acknowledgments

Footnotes

REFERENCES

References