Characterization of gtf, a Glucosyltransferase Gene in the Genomes of Pediococcus parvulus and Oenococcus oeni, Two Bacterial Species Commonly Found in Wine

Abstract

“Ropiness” is a bacterial alteration in wines, beers, and ciders, caused by β-glucan-synthesizing pediococci. A single glucosyltransferase, Gtf, controls ropy polysaccharide synthesis. In this study, we show that the corresponding gtf gene is also present on the chromosomes of several strains of Oenococcus oeni isolated from nonropy wines. gtf is surrounded by mobile elements that may be implicated in its integration into the chromosome of O. oeni. gtf is expressed in all the gtf⁺ strains, and β-glucan is detected in the majority of these strains. Part of this β-glucan accumulates around the cells forming a capsule, while the other part is liberated into the medium together with heteropolysaccharides. Most of the time, this polymer excretion does not lead to ropiness in a model medium. In addition, we show that wild or recombinant bacterial strains harboring a functional gtf gene (gtf⁺) are more resistant to several stresses occurring in wine (alcohol, pH, and SO₂) and exhibit increased adhesion capacities compared to their gtf mutant variants.

“Ropiness” or “oiliness” is one of the major bacterial alterations in wine. It has no impact on human health. However, the viscosity of the spoiled wines prevents their commercialization (43, 54, 55). Many microbial species have been isolated from ropy wines: Streptococcus mucilaginous and Leuconostoc and Lactobacillus species (15, 18, 34, 36, 53, 55). However, the species most often incriminated is Pediococcus parvulus (15, 31, 48, 53). Different strains have been isolated from red and white ropy wines from the Bordeaux region in France or from Basque country ropy ciders and were then shown to cause ropiness when grown in model media. These strains were first called Pediococcus cerevisiae (31), later classified as Pediococcus damnosus by DNA/DNA hybridizations (15, 32), and then finally classified as Pediococcus parvulus based on 16S RNA sequencing (53). The increase in viscosity of wine or model media caused by P. parvulus is due to the production of a high-molecular-weight β-glucan. This fibrillar polymer consists of a trisaccharide repeating unit with a β-1,3-linked glucosyl backbone and branches made up of single β-1,2-linked d-glucopyranosyl residues (16, 17, 28). The level of production of soluble β-glucan by pediococci is low (200 mg/liter at most) and is only slightly modified by external growth parameters (49, 50). Polymer production is controlled by a single transmembrane glucosyltransferase, Gtf, a 567-amino-acid, 65-kDa protein that polymerizes glucosyl residues from UDP glucose (53). Gtf exhibits significant homology with Tts (32% identical amino acids), an inverting glucosyltransferase produced by Streptococcus pneumoniae type 37, which synthesizes a branched β-1,3 β-1,2 glucan whose structure is very close to that of the glucan causing ropiness (1, 29). Actually, the antibodies targeting the type 37 polysaccharide capsule also agglutinate ropy pediococci (51). The gtf gene encoding Gtf is highly conserved (99.9% identity) in wine- and cider-spoiling bacteria, as it is located on the following: (i) a small plasmid of 5.5 kb in pediococci found in wine (pF8801 [GenBank accession no. AF196967] [22, 32]), (ii) a 35-kb plasmid, pPP2 (GenBank accession no. AY999683) in P. parvulus 2.6 isolated from cider (18, 53), and (iii) on a 5.5-kb pLD1 plasmid (GenBank accession no. AY999684) in ropy Lactobacillus diolivorans G77 isolated from cider (18, 53).

Different primer sets were proposed for the detection of the gtf gene by PCR (9, 22, 51, 53). They were used to analyze various bacterial strains isolated from wines (spoiled, ropy, or not spoiled or ropy) and causing an increase or not causing an increase in model medium viscosity (ropy or nonropy strains). All the ropy strains were isolated from spoiled beverages, and most of them displayed gtf orthologs. They belong to the species P. parvulus, P. damnosus, Lactobacillus diolivorans, L. suebicus, and L. collinoides. However, some of these ropy strains did not give positive gtf internal fragment amplification, suggesting the existence of different genetic determinants of ropiness (51, 53).

Interestingly, the presence of gtf orthologs was detected in Oenococcus oeni strains isolated from wine and cider. A partial gtf gene sequence of 950 bp was amplified from genomic DNA of O. oeni IOEB 0205, even though this bacterial strain was isolated from nonspoiled champagne (51). The occurrence of a gtf gene among O. oeni populations was confirmed by Werning et al. (53) in a survey of 20 O. oeni strains isolated from spoiled ropy (4 strains) or nonspoiled ciders (16 strains). The O. oeni I4 strain, isolated from ropy cider, harbored a gtf ortholog (GenBank accession no. AY999685) which displayed 98.8% identity with PF8801 gtf. The authors did not specify if other microorganisms were present in this cider, and the role of O. oeni I4 in the spoilage remained unclear. An additional collection of 80 IOEB O. oeni strains, all isolated from nonropy wines (32 white wines, 41 red wines, and 7 rosés or others), was screened for the presence of gtf by PCR: the genomic DNA of 18/80 strains enabled amplification of an internal fragment of gtf (117 bp) (9, 10).

O. oeni drives malolactic fermentation in most wines and is commercialized as a malolactic starter. The presence of gtf in the genome of this useful species is worrying, as it suggests that it can represent a vector for ropiness in wine. Although the species has never been blamed, the role of O. oeni in cider ropiness can be suggested from the results of Ibarburu et al. (26) showing that O. oeni I4 produces β-glucan in a model medium and induces an increase in viscosity. Closer characterization of the gtf gene in the wine O. oeni strains harboring it is now necessary to (i) specify the O. oeni gtf gene sequence and location and (ii) assess the gene expression and the protein functionality but also to (iii) analyze the possible role of gtf in bacterial strain persistence in wine.

MATERIALS AND METHODS

Bacterial strains, growth parameters, and plasmids.

The bacterial strains used in this study are presented in Table 1. The nonropy mutant P. parvulus IOEB 0206 was obtained from P. parvulus IOEB 8801 by plasmid curing experiments (51).

TABLE 1.

Bacterial strains used in this study

Species	Growth conditionsa	Strain	Reference or origin	Genotypeb
Escherichia coli	LB, 37°C, agitation	DH5α		gtf mutant
Lactococcus lactis	M17, 30°C	IL-1403		gtf mutant
	M17 + Ery (5 μg·ml−1), 30°C	IL-1403(pGK13)		gtf mutant
	M17 + Ery (5 μg·ml−1), 30°C	IL-1403(pESoo8)	This study	gtf+
Pediococcus parvulus	MRS, 25°C	IOEB 8801	Bordeaux red wine	gtf+
		IOEB 0206	Laboratory mutant (51)	gtf mutant
Oenococcus oeni	MRS, 25°C	IOEB 8413	Bordeaux red wine	gtf mutant
		IOEB Sarco 1491		gtf mutant
		IOEB Sarco 436a		gtf mutant
		IOEB Sarco 450	Commercial starters	gtf mutant
		IOEB Sarco 277		gtf mutant
		PSU1		gtf mutant
		IOEB 0205	Champagne	gtf+
		IOEB Sarco 389	Chablis white wine	gtf+
		IOEB Sarco 390		gtf+
		IOEB Sarco 392		gtf+
		IOEB Sarco 393		gtf+
		IOEB Sarco 397b	White wine	gtf+
		IOEB Sarco 410	Bordeaux white wine	gtf+
		IOEB Sarco 413		gtf+
		IOEB Sarco 421	Jura white wine	gtf+
		IOEB Sarco 422		gtf+
		IOEB Sarco 423		gtf+
		IOEB Sarco 448	Chablis white wine	gtf+
		IOEB Sarco 449		gtf+
		IOEB Sarco 454	Bourgogne white wine	gtf+
		IOEB Sarco 455		gtf+
		IOEB Sarco 456		gtf+
		IOEB Sarco 457		gtf+
		IOEB Sarco 459		gtf+

^aEry, erythromycin.

^bgtf⁺, strain displaying gtf gene; gtf mutant, strain whose genomic DNA did not enable the amplification of the expected fragment internal to gtf (primer sets FQ1/FQ2 and PF1/PF8).

P. parvulus and O. oeni were propagated at 25°C, without agitation, in liquid MRS (12) containing the following ingredients (in g·liter⁻¹): glucose, 20; yeast extract, 4; beef extract, 8; Bacto peptone, 10; sodium acetate, 5; trisodium citrate, 2; K₂HPO₄, 2; MgSO₄·7H₂O, 0.2; MnSO₄·H₂O, 0.1; Tween 80, 1 ml. The pH of MRS was adjusted to 5.0.

Lactococcus lactis strains were grown at 30°C, without agitation, in M17 broth (47) containing 0.5% glucose. Escherichia coli DH5α cells were grown at 37°C (150 rpm) in Luria-Bertani (LB) medium (37). Plasmid pGK13 was used for the cloning experiments. It replicates in E. coli and L. lactis. Recombinant strains of E. coli and L. lactis were grown in the presence of 100 μg·ml⁻¹ and 5 μg·ml⁻¹ of erythromycin, respectively.

For exopolysaccharide (EPS) synthesis studies, strains were grown in dialyzed medium in order to avoid the presence of polysaccharides in the initial medium. A 10× concentrated medium was prepared and dialyzed against the desired final volume of distilled water for 24 h at +4°C, using 6,000- to 8,000-Da-molecular-size-cutoff membranes. The pH was adjusted to 5.0, and the medium was sterilized for 20 min at 121°C. The inoculum represented 2% of the total volume. The fully filled bottles were incubated at 25°C without agitation except just before sampling. Beads were added to compensate for the loss of liquid after sample removal and limit the volume of air in the bottle.

Molecular biology techniques. (i) DNA and RNA extraction.

Total genomic DNA from lactic acid bacteria was purified using the Wizard Genomic DNA purification kit (Promega, Madison, WI). For total RNA extraction, cells were harvested by centrifugation (6,000 × g, 15 min), suspended in the RNA isolation reagent Tri reagent (Sigma), and disrupted with glass beads (0.1 mm) in a Fast prep FP120 instrument at 4°C for 45 s each time for a total of six times at 6,500 × g. Cell debris was eliminated by centrifugation, and RNA was purified from the supernatant by chloroform extraction. RNA was then precipitated using isopropanol, washed with 80% ethanol, and finally resuspended in diethyl pyrocarbonate-treated water. RNA concentration was calculated from the absorbance measured at 260 nm (SmartSpec Plus spectrophotometer; Bio-Rad). Samples were treated with DNase as indicated by the manufacturer (DNaA-free; Ambion). The absence of chromosomal DNA was controlled by PCR using primers PF1 and PF8. The quality of RNA samples was checked on a 1% formaldehyde-agarose gel. The cDNA was then synthesized using the iScript cDNA synthesis kit (Bio-Rad), as recommended by the manufacturer.

(ii) PFGE and Southern blotting.

Bacteria grown in MRS broth were harvested during the exponential growth phase, washed twice with TE buffer (10 mM Tris-HCl, 1 mM EDTA [pH 8]), resuspended in T100E buffer (10 mM Tris-HCl, 100 mM EDTA [pH 7.5]), and embedded in 1% agarose slices. DNA was extracted by incubating the gel slices for 8 h at 37°C in T100E buffer containing 10 mg of lysozyme per ml, followed by 16 h at 37°C in TE buffer supplemented with 1.5% N-lauryl sarcosine and 2 mg of pronase per ml. The gel slices were subsequently transferred into T100E buffer and stored at 4°C until use. To obtain NotI digests, the gel slices were washed four times with TE buffer, rinsed with water, and incubated for 16 h at 25°C in 120-μl reaction mixtures containing 150 U of NotI (New England Biolabs) according to the manufacturer's instructions. Pulsed-field gel electrophoresis (PFGE) was performed in a 1% agarose gel using the CHEF-DRIII system (Bio-Rad) with pulse times of 1 to 25 s for 20 h at 6 V/cm and 15°C in 0.5× TEB buffer (45 mM Tris-OH [pH 8], 45 mM boric acid, 1 mM EDTA). DNA was transferred to a Hybond-N+ membrane (Amersham Biosciences) and hybridized by the method of Maniatis et al. (37). The DNA probe corresponded to a 950-bp internal region of the gtf gene amplified by PCR from total DNA from O. oeni IOEB 0205 cells using primers PF1 and PF8 (see Table 2). The probe was labeled with digoxigenin-11-dUTP by using the digoxigenin DNA labeling kit (Roche), and detection was done by chemiluminescence with an antidigoxigenin antibody and CDP-Star (Roche).

TABLE 2.

Primers used in this study

Primer	Sequence (5′→3′)	Purposea
C1	ACACCCTTAATTCATAGTTCACTGGGCTAC	gtf island analysis
C2	CTCGTGACGCTGTCATTCGATATACTAC
D	ATGAAGGAACTGGATATCCACTCCGT
E	GTTTTAAGAGACATTCAGAACCGCGT
F	GCATTAAATCAGGCGGCCATTTG
G	CGGTCGTGAATTGATGGCACG
GspB	TTTACCCTAGCAGCTCCAATAATTGC
GspC	GTCCAGCGTTGCACATTCCACTAG
GspD	TTGGCAGAACTAGCGAAAGTACGC
GspE	TCCACTAATGTGATTACACTGAAGAATGC
H	GAGTCCGTACACCCATCTT	gtf qPCR
FQ1	CTGTGTCTGGCGTTTCTGTAGG
FQ2	GCCACCGCATAGGGTATTTGTC
LDH1	GCCGCAGTAAAGAACTTGATG	ldh qPCR
LDH2	TGCCGACAACACCAACTGTTT
PF1	GATTGTAATAAAATAAAAAGACCC	Checking within gtf
PF8	CATATGATAACACGCAGGGC
Start	GGAATTCCGGAATGTAGTATAAATGTTAAATG (EcoRI site underlined)	gtf cloning
Stop	TCACGCTATCTGCCATGGAAGCGAAC (NcoI site underlined)

^aqPCR, quantitative PCR.

(iii) Sequencing of the gtf locus.

The sequence of the gtf locus of O. oeni IOEB 0205 was determined by the linker-mediated PCR strategy with the Topo-Walker kit (Invitrogen). The sequences of the primers used are shown in Table 2. Purified DNA from O. oeni IOEB 0205 was used in order to identify restriction enzyme cleavage sites located between 1 and 5 kb from the 950-bp internal fragment of the gtf gene amplified with primers PF1 and PF8. Two primers located inside this fragment and directed toward the upstream (GspC) or downstream (GspD) regions were used to elongate DNA fragments containing an EcoRI site and a BamHI site, respectively, at their extremities (Fig. 1). The Topo linker was added to the extremities of the elongated DNA by topoisomerase-mediated ligation, providing two DNA templates that were amplified by PCR with the primers for upstream GspB and downstream GspE and a primer located inside the Topo linker. The two PCR products were gel purified and sequenced (Milligen, France). Subsequent PCR experiments using primers designed from orf1 orthologs in lactic acid bacteria and primer GspE enabled the complete region shown in Fig. 1 to be amplified.

FIG. 1.

Genetic organization of various DNA regions containing the gtf gene in pediococci and O. oeni. (A) Plasmid pF8801 (5.5 kb) from P. parvulus IOEB 8801 (GenBank accession no. AF196967) and plasmid pPP2 (35 kb) from P. parvulus 2.6 (GenBank accession no. AY999683). (B) O. oeni IOEB 0205 chromosome (this study). (C) Orthologous genes in other lactic acid bacteria. (Map 1) Transposases in Lactobacillus casei ATCC 334 (three copies of CP_000423), in L. casei ATCC 393 (AF445084), and in bacteriophage phiAT3 (YP_025041). (Map 2) Transposase IS30 in Pediococcus pentosaceus ATCC 25745 (CAA83666), Lactobacillus plantarum (ISLpL2 [GenBank accession no. AF459445]), O. oeni PSU1 (three copies of the complete IS [GenBank accession no. CP_000411]), L. casei ATCC 334 (four copies of the complete IS [GenBank accession no. CP_000423]), and Lactobacillus brevis ATCC 367 (five copies of the complete IS [GenBank accession no. CP_000416]). (Map 3) Transposase in L. casei ATCC 334 (four copies of the complete IS [GenBank accession no. CP_000423]). The large arrows indicate genes, while the small horizontal arrows (→) indicate the positions of the primers for PCR experiments and the small vertical arrows (↓) indicate the restriction sites. The boxes and the gray-shaded sections link the DNA regions exhibiting significant sequence identity.

(iv) PCR analysis.

PCR mixtures (25 μl) contained 0.5 μl of each primer (50 pmol), 0.5 μl of genomic DNA, and 2 μl of custom-made PCR master mix 12.5X (Qbiogene, France). The reaction mixture was preheated for 5 min at 95°C, and 35 cycles (denaturing step of 30 s at 95°C, annealing step of 30 s at 55°C, and extension step of 1 min at 72°C) were carried out. Using primers GspB and C2 enabled the amplification of a 1,277-bp fragment, while primers GspE and C1 generated a 997-bp fragment from the DNA from gtf⁺ strains. The primer sets E/D, F/D, and G/H were used to search for the genes surrounding gtf in DNA from various O. oeni gtf mutant strains. The relative positions of the primers are indicated in Fig. 1, and their sequences are shown in Table 2.

(v) Reverse transcription-PCR (RT-PCR) analysis.

The 50-μl reaction mixture contained 25 μl of the 2× Sybr green PCR supermix (Bio-Rad), 5 pmol each of primers FQ1 and FQ2 for gtf amplification and LDH1 and LDH2 for d-lactate dehydrogenase (ldhD) amplification, and 1 μl of cDNA in the appropriate dilution (to ensure a final concentration of 0.1 ng/μl). The reaction mixture was preheated for 5 min at 95°C, and 35 cycles (denaturing step of 30 s at 95°C, annealing step of 30 s at 63°C for gtf and 30 s at 60°C for ldhD, and extension step of 30 s at 72°C) were carried out. After the last cycle, a melting curve analysis was performed using the iCycler iQ (Bio-Rad) to check PCR specificity. The results were analyzed using the comparative critical threshold method with the O. oeni ldhD gene as an internal calibrated target, as proposed by Desroches et al. (13) for this microorganism.

(vi) Cloning of gtf.

The gene encoding Gtf was amplified by PCR from total DNA from O. oeni IOEB 0205, with the primers Start and Stop designed to introduce EcoRI and NcoI sites, respectively, at the boundaries (Table 2). The insert, recovered from the PCR product after gel purification and digestion with NcoI and EcoRI, was cloned into the corresponding restriction sites of the pGK13 vector; the constructed plasmid, pESoo8 was introduced into E. coli DH5α, and positive clones were selected at 37°C in the presence of erythromycin (100 μg/μl). The plasmid insert was sequenced (Milligen, France) to ensure that no mutations occurred in the gtf gene. The pESoo8 plasmid was subsequently introduced into L. lactis IL-1403 cells by electrotransformation, and positive clones were selected at 30°C in the presence of erythromycin (5 μg/μl) (7).

Analytical methods. (i) Substrate and product analysis.

Glucose and lactate concentrations were measured using enzymatic in vitro tests (Boehringer Mannheim, Germany). For EPS concentration measurements, the whole medium was centrifuged (10,000 × g, 15 min, 4°C), and 5 volumes of 96% ethanol containing 5% HCl (1 N) were added to the supernatant to precipitate the polysaccharides. The tubes were left to stand for 24 h at 4°C. The tubes were then centrifuged (10,000 × g, 15 min, 4°C), and the pellet was washed with 80:20 ethanol-water, centrifuged again, dried for 20 min at 65°C, and dissolved in distilled water. The amount of neutral polysaccharides was determined using the phenol-sulfuric acid method (14), with glucose as the standard. The determination was carried out on three replicate samples.

(ii) Size exclusion chromatography.

The whole medium (100 ml) was centrifuged (10,000 × g, 20 min, 4°C), and macromolecules from the supernatant were precipitated with 5 volumes of 95% ethanol containing 5% HCl (1 N). After 24 h at 4°C, the pellet was recovered by centrifugation (10,000 × g, 20 min, 4°C), washed with 80% ethanol, resuspended in water, and freeze-dried. The freeze-dried powder (30 mg) was resuspended in 3 ml of elution buffer (50 mM K₂HPO₄, 150 mM NaCl [pH 7.0]) and analyzed using a Sephacryl S400 HR column (1.6 by 83 cm) (Amersham Biosciences), eluted at a 1.2-ml/min flow rate with a Waters 515 pump. Detection was done with a Waters 2414 refractometer. Column calibration was carried out with commercial dextrans having molecular masses ranging from 15,000 to 2 × 10⁶ Da.

(iii) Polysaccharide monomer composition analysis.

Fractionation of the culture supernatant by ultrafiltration was also performed. A volume of 560 ml of dialyzed MRS culture supernatant was filtered on a 100-kDa-cutoff Amicon membrane. The monomer composition of polysaccharides (the freeze-dried powder described above or the high-molecular-weight fraction obtained by ultrafiltration) was determined after acid hydrolysis (2 N H₂SO₄ for 6 h at 100°C). The neutral monomer composition was determined by gas-liquid chromatography of alditol acetate derivatives using inositol as the internal standard. Sugar analysis was performed with an Agilent 6850 series gas chromatograph system equipped with an ESP2380 macrobore column (25 m by 0.53 mm).

(iv) ¹H and ¹³C NMR analysis.

The ¹³C nuclear magnetic resonance (NMR) spectra of the glucans were recorded with a Bruker AC 300 spectrometer operating at a frequency of 75,468 MHz. Samples were examined as solutions in D₂O (10 to 15 mg in 0.35 ml of solvent) at 70°C in 5-mm-diameter spinning tubes (the internal standard was ¹³CH₃ at 31.5 ppm relative to tetramethyl silicon). Quantitative ¹³C spectra were recorded using the INVGATE Bruker sequence, with 90 pulse length (6.5 ms), 15,000-Hz spectral width, 8,000 data points, 0.54-s acquisition time, and a relaxation delay of 1.5 s, and 100,000 scans were accumulated.

(v) Immunological analysis.

Agglutination tests were performed using S. pneumoniae type 37-specific antisera as previously reported (51). Four microliters of antiserum was spotted on a slide with 20 μl of culture broth and incubated for 30 min at 4°C before observation using phase-contrast microscopy.

Stress challenges.

The bacteria were grown to the early stationary phase, harvested by centrifugation (6,000 × g, 5 min, 4°C), and resuspended in either fresh medium (M17 agar plus 5 μg/ml erythromycin for L. lactis strains and MRS for the other bacteria) containing the stressor (alcohol or sulfur dioxide or modified pH) or filtered (0.2-μm-cutoff membrane) red or white wine. The red wine used was of the merlot variety. The white wine was from a blend of different varieties of white grapes. The ethanol content was 12%, the pH was 3.81 (red) or 3.25 (white), and the sulfur dioxide content was 20 mg/liter. After 3 h at 25°C (or 1 h at 30°C for L. lactis strains), serial 10-fold dilutions in 0.9% NaCl were plated on MRS agar (or M17 agar containing 5 μg/ml erythromycin for L. lactis strains) and incubated at 25°C (or 30°C for L. lactis). Colonies were counted 3 days to 1 week later. Survival ratios were calculated from CFU in the stressed medium cell suspension compared to the supernatant cell suspension.

Adhesion assays.

The ability to form a biofilm on an abiotic surface was quantified by a method adapted from O'Toole et al. (42). Each bacterial strain was inoculated in the appropriate medium, and 16 200-μl samples of cell suspension were deposited in a sterile 96-well polystyrene microtiter plate. After 1 or 5 days (L. lactis) or 12 or 30 days (O. oeni and P. parvulus) at 25°C, the wells were gently washed three times with 200 μl of 0.9% NaCl, dried in an inverted position, and stained with 1% crystal violet. The wells were rinsed again, and the crystal violet was solubilized in 100 μl of ethanol-acetone (80:20, vol/vol). The absorbance at 595 nm was determined using a microplate reader (Molecular Devices). Three independent assays were performed.

Statistical analysis.

The statistical significance of differences between means was calculated using analysis of variance followed by Tukey post hoc comparisons (P < 0.05).

Nucleotide sequence accession number.

The nucleotide sequence data reported in this study have been deposited in the DDJB/EMBL/GenBank database under the accession number EU556433.

RESULTS

Pediococcus parvulus is the most described vector of oiliness in wines and ciders. The metabolic pathways involved and the dedicated genetic determinants (glucosyltransferase gene gtf) leading to ropiness have been studied in several strains. Molecular detection tools based on PCR have been designed, and they were used to analyze a collection of 80 O. oeni IOEB strains all isolated from nonropy wines. The expected amplicon was amplified in a total of 18/80 strains (9, 10, 51). These 18 O. oeni gtf⁺ strains were geographically unrelated, since they originated from wines produced in different geographic areas (Table 1). However, it is worth noting that they all originated from white wine. Their NotI-digested total DNA was compared by PFGE (data not shown) to assess their genetic relatedness. All the strains displayed distinct PFGE profiles, except for the IOEB Sarco 422 and IOEB Sarco 421 isolates, which could not be distinguished on the basis of their PFGE profiles. Both isolates came from Jura white wine.

Genetic localization of gtf in O. oeni gtf⁺ strains.

Native and NotI-digested total DNAs from O. oeni IOEB 0205 were analyzed by PFGE, and after Southern blotting, the membrane was hybridized with a probe targeting an internal fragment of gtf. Prior to digestion, hybridization was observed with chromosomal DNA (Fig. 2B, lane 1). After digestion by NotI, the probe hybridized to a 242-kb fragment (Fig. 2B, lane 2). Analysis of the other O. oeni gtf⁺ strains also demonstrated a chromosomal location on a NotI-digested fragment whose size varied from 220 to 250 kb, depending on the strain (not shown).

FIG. 2.

(A) PFGE analysis of O. oeni IOEB 0205. Undigested genomic DNA (lane 1) or DNA digested by NotI (lane 2). chDNA, chromosome DNA. (B) Southern hybridization analysis of total DNA prepared for panel A, using an internal fragment of the gtf gene as a probe.

A 3,528-bp nucleotide sequence containing gtf was determined by a linker-mediated strategy, starting from the internal 950-pb region obtained by PCR (primers PF1 and PF8) on O. oeni IOEB 0205 genomic DNA. The G+C content of this region was 39%, which is quite similar to the 37% G+C content reported for the two available O. oeni genomes (GenBank accession no. CP_000411 and NZ_AAUV00000000). The genetic organization of the sequence is depicted in Fig. 1B. The O. oeni IOEB 0205 gtf gene displayed 1,701 nucleotides, as with the orthologous genes described in the literature (22, 53). The identity with reported gtf genes was very high: 97.8% with the pF8801 gtf gene, 97.9% with the pLD1 gtf gene (L. diolivorans), 98% with the pPP2 gtf gene (P. parvulus 2.6), and 99.1% with the O. oeni I4 gtf gene (53). The 26 nucleotides upstream of gtf containing the ribosome binding sequence were 100% conserved between plasmids pPP2, pF8801, and the chromosome of O. oeni IOEB 0205. The upstream sequence of gtf genes in O. oeni I4 and L. diolivorans G77 were not available for comparison. Downstream of gtf, 10 nucleotides were conserved between the P. parvulus pPP2 and pF8801 plasmids, but these nucleotides were not present at the end of the O. oeni IOEB 0205 gtf gene (Fig. 1).

Comparison of the 3,528-bp DNA sequence with the National Center for Biotechnology (NCBI) database revealed interesting homologies indicating a mosaic structure (Fig. 1). In O. oeni IOEB 0205, gtf is surrounded by two putative transposases. The first one (ORF1), 621 nucleotides (nt) (251 amino acids) long, had a protein sequence 99% identical (89.6% to 100% coverage) to that of transposases (COG 2801) from Lactobacillus casei ATCC 334, L. casei ATCC 393, and bacteriophage phiAT3. The region downstream of ORF1 (nt 621 to 747) was 98% identical to the downstream region of this transposase in L. casei or bacteriophage phiAT3. Furthermore, the region ΔIS from nt 747 to 1020 was 91% identical to the nucleotide sequence (25% coverage) of a transposase IS30 encountered in several lactic acid bacteria, including O. oeni PSU1. The second open reading frame (ORF) (ORF2), 646 nt long and divergently transcribed from gtf, had a protein sequence 99 to 100% identical (64% coverage) to that of several putative IS30 transposases (COG2826) from L. casei ATCC 334.

Next we examined the genetic environment in other gtf⁺ strains. The primer sets GspB/C2 and GspE/C1 were used and generated the same amplification profile in all the strains tested (Fig. 3). The amplicons from four additional strains were sequenced, revealing 99.8% identity between IOEB strains Sarco 393, Sarco 410, Sarco 421, and 0205 and 99.9% identity between IOEB strains Sarco 397 and 0205.

FIG. 3.

Multiplex PCR analysis on genomic DNA using primers GspB/C2 (997-bp amplicon) and C1/GspE (1,277-bp amplicon). Lanes: 1, negative control (no DNA); 2, P. parvulus IOEB 8801; 3, P. parvulus IOEB 0206; 4, positive control (O. oeni IOEB 0205); 5, negative control (O. oeni ATCC BAA-1163). Lanes 6 to 22 contain O. oeni IOEB Sarco strains 389 (lane 6), 390 (lane 7), 392 (lane 8), 393 (lane 9), 397b (lane 10), 410 (lane 11), 413 (lane 12), 421 (lane 13), 422 (lane 14), 423 (lane 15), 448 (lane 16), 449 (lane 17), 454 (lane 18), 455 (lane 19), 456 (lane 20), 457 (lane 21), and 459 (lane 22).

In addition, PCR experiments were carried out in order to amplify these regions from genomic DNA from several O. oeni gtf mutant strains (IOEB Sarco strains 450, 277, 1491, and 436a and IOEB 8413). The primer sets D/E, D/F, and G/H (positions shown in Fig. 1B) did not enable any amplicons to be obtained (not shown). Last, the DNA sequences upstream and downstream of gtf were not present in the O. oeni PSU1 and ATCC BAA_1163 genomes (GenBank accession no. CP_000411 and NZ_AAUV00000000). Only the ΔIS sequence was found. However, multiple IS30-related elements exist in O. oeni PSU1.

Functional analysis of gtf and Gtf with O. oeni IOEB 0205.

O. oeni IOEB 0205 was propagated in dialyzed MRS medium. Glucose metabolism by O. oeni IOEB 0205 led to biomass multiplication (maximum optical density at 600 nm of 2.8) and exopolysaccharide accumulation (227 mg/liter) (Fig. 4A). Throughout the whole culture period, EPS synthesis represented a loss of 1.25% of the glucose consumed by the bacteria. EPS synthesis occurred during growth and glucose consumption. No polymer degradation occurred over 60 days (Fig. 4A). The gtf gene was expressed all along the culture period with a threefold increase during the exponential growth phase (Fig. 4B). The culture medium was clearly ropy from the 8th day (Fig. 4C). Moreover, upon agitation, the resuspended cell deposit formed a long string, which is a common trait of glucan-producing pediococci. However, to the naked eye, the cell deposit was less cohesive than that observed with ropy pediococci. Most O. oeni gtf mutant cells formed chains 3 to 7 units long, while O. oeni IOEB 0205 formed chains 10 to 20 units long, indicating a specific phenotype for this strain (Fig. 5A). Immunoagglutination assays were performed at different culture times (5th, 15th, and 28th day; Fig. 4A and 5B). These clearly indicated the presence of β-glucan at the cell surface throughout the culture. In the same conditions, O. oeni gtf mutant strains never agglutinated in the presence of anti-type 37 antibodies (not shown).

FIG. 4.

(A) Growth and exopolysaccharide production by O. oeni IOEB 0205 in dialyzed MRS. •, optical density at 600 nm [OD(600)]; □, glucose; ▵, EPS. The big black arrows indicate the samples used for immunoagglutination assays. (B) gtf expression levels (relative to the lactate dehydrogenase gene ldh). (C) Visualization of ropiness induced by O. oeni IOEB 0205 in dialyzed MRS medium.

FIG. 5.

Immunoagglutination induced by anti-type 37 antibody. (A and B) O. oeni IOEB 0205 without antibody (A) and after antibody addition (B). (C and D) L. lactis IL-1403(pESoo8) without antibody (C) and after antibody addition (D). (E and F) L. lactis IL-1403(pGK13) without antibody (E) and after antibody addition (F).

The polysaccharides in the supernatant were precipitated at the end of the culture and analyzed by size exclusion chromatography. The chromatographic profile was compared with that obtained for P. parvulus IOEB 8801 (Fig. 6). Two main classes of molecules were obtained. The first group of molecules displayed a molecular mass lower than 60 kDa. These represented 45% and >95% of the polymers precipitated from P. parvulus IOEB 8801 and O. oeni IOEB 0205, respectively. The second group was composed of high-molecular-mass molecules (more than 10⁶ Da) eluted at around 80 min. These molecules represented less than 5% of the polymer precipitated in the O. oeni IOEB 0205 culture broth and 55% of the polymer precipitated in that of P. parvulus IOEB 8801. According to Llaubères et al. (28), the ropy β-glucan has a mean molecular mass of 800 kDa and should elute at this peak. The chemical analysis of the whole precipitate, as well as the analysis of the high-molecular-weight fraction obtained by size exclusion chromatography, indicated the presence of glucose, galactose, mannose, ribose, and N-acetylglucosamine. The background signal was very high on the ¹H NMR spectrum, indicating the presence of numerous oxidic bonds. In order to improve isolation of the high-molecular-weight β-glucan, macromolecules from the supernatant were concentrated by ultrafiltration without prior alcohol precipitation. A methylation analysis was then conducted and indicated the predominance of glucan in the retentate. The ¹H NMR spectrum exhibited a lower background signal and the characteristic signals of the H-1 of β-1,3 and β-1,2 linked glucopyranoses. This was confirmed by the ¹³C NMR spectra through the C-3 signal at 87.6 ppm and the three C-1 signals of equal intensity at 104 to 105 ppm (16, 17, 28). β-Glucan was thus present in this high-molecular-weight retentate. However, its isolation and analysis were much more difficult than in the case of pediococci. Similar problems have been mentioned by Ibarburu et al. (26) with O. oeni I4.

FIG. 6.

Size distribution analysis of the alcohol-precipitated polysaccharides from the dialyzed MRS culture supernatant of O. oeni IOEB 0205 and P. parvulus IOEB 8801. RI, refractive index.

O. oeni is not amenable to genetic transformation. Hence, to demonstrate the role of gtf in glucan production, the gene was expressed in L. lactis IL-1403. Prior to cloning, PCR was used to check (primer sets PF1/PF8 and FQ1/FQ2) that L. lactis IL-1403 did not display gtf. Expression of gtf in L. lactis IL-1403(pESoo8) was then checked by RT-PCR (not shown). Immunoagglutination assays clearly indicated that L. lactis IL-1403(pESoo8) was positive (Fig. 5C and D). Moreover, upon agitation, the resuspended cell deposit formed a long string. However, the L. lactis strain harboring the pGK13 plasmid (gtf mutant) did not agglutinate (Fig. 5E and F) or form a string upon agitation. The gtf gene in O. oeni IOEB 0205 enabled the synthesis of β-glucan on the surfaces of the cells of either O. oeni or L. lactis. After cultivation in the appropriate dialyzed M17 medium (Table 1), L. lactis IL-1403(pESoo8) culture supernatants contained 28 ± 5 mg/liter EPS, whereas L. lactis IL-1403(pGK13) culture supernatants contained only 13 ± 6 mg/liter EPS, suggesting that β-glucan was excreted by the recombinant strain. This was confirmed by NMR analysis (not shown). However, L. lactis IL-1403(pESoo8) did not induce an increase in ropiness of the medium.

Functional analysis of gtf and Gtf with the other O. oeni gtf⁺ strains.

RT-PCR analysis of RNA extracted from the cells of the 17 other O. oeni gtf⁺ strains sampled during exponential growth indicated the presence of gtf mRNA (relative expression level between 0.5 and 1 using lactate dehydrogenase as a reference). Immunoagglutination assays clearly indicated the presence of a β-glucan capsule around the cells of 13 strains. Strains IOEB Sarco 454, 455, 456, and 457 were negative, although three independent assays were performed on cells taken at various stages of growth (not shown). However, small amounts of β-glucan capsule (undetectable) cannot be ruled out for these four strains.

The total level of EPS production by the 17 gtf⁺ strains varied from 50 to 250 mg/liter (227 mg/liter with strain IOEB 0205), but in contrast to strain IOEB 0205, none of the other gtf⁺ strains induced an increase in viscosity and ropiness of the medium.

Presence of gtf and resistance of bacteria to stressful conditions.

In order to demonstrate whether glucan formation confers a selective advantage, we compared the behavior of L. lactis harboring either pGK13 or pESoo8 and that of P. parvulus IOEB 8801 and its nonropy mutant IOEB 0206. Challenges were performed over 1 to 3 h on planktonic cells isolated during the early stationary growth phase. Strains were subjected to single stresses relevant in wine: acidic pH, ethanol, and sulfur dioxide (Fig. 7). Survival of gtf⁺ strains was improved by 1 to 2 log units in stressful conditions. This was particularly true at low pH. Compared to its nonropy variant P. parvulus IOEB 0206 (51), P. parvulus IOEB 8801 exhibited an increased survival rate when introduced into wine and protection by Gtf was 12.5 higher in white wine compared to red wine. That is to say, white wine represented a stronger stress than red wine for the gtf mutant strain but not for the gtf⁺ strain. L. lactis is not a bacterium found in wine, and the challenge in wine led to bacterial death. However, these experiments clearly demonstrate the improved survival rates of the P. parvulus and L. lactis gtf⁺ strains in stressful conditions.

FIG. 7.

Survival rates to single stresses (pH, ethanol, SO₂, or wine). (A) L. lactis IL-1403(pESoo8) (gtf⁺) (black bars) and L. lactis IL-1403(pGK13) (gtf mutant) (white bars). (B) P. parvulus IOEB 8801 (gtf⁺) (black bars) and P. parvulus IOEB 0206 (gtf mutant variant) (white bars).

The genetic transformation of O. oeni being impossible, the survival rate of wild strains of O. oeni in wine was investigated (Fig. 8). Three strains were used: O. oeni IOEB Sarco 450 (gtf mutant), O. oeni IOEB 0205 (gtf⁺), and IOEB 8413 (gtf mutant). As shown in Fig. 8, the gtf⁺ strain exhibited an intermediate survival rate in red wine and the best survival rate in white wine. As for pediococci, white wine represented a stronger stress than red wine for the gtf mutant strain but not for the gtf⁺ strain. However, this high level of resistance cannot be assigned to the sole gtf gene in this experiment because O. oeni is a highly variable species (11, 38) and because resistance to wine is a complex mechanism.

FIG. 8.

Survival rates in wine of various O. oeni strains displaying gtf: O. oeni IOEB 0205 (gtf⁺), O. oeni IOEB 450 (malolactic starter; gtf mutant), and IOEB 8413(gtf mutant).

Presence of gtf and biofilm formation.

The adhesion capacities of P. parvulus IOEB 8801 and its nonropy mutant IOEB 0206 and L. lactis harboring either pGK13 or pESoo8 were analyzed (Table 3). The strains of L. lactis and P. parvulus harboring gtf exhibited significantly increased adhesion compared to their gtf mutant variant in the early assay (1 or 12 days) as well as later (5 or 30 days). O. oeni IOEB 0205 did not exhibit any significant adherence compared to wild gtf mutant strains in the early assay, but, in the later assay (30 days), it showed significantly better adherence than the other O. oeni strains did.

TABLE 3.

Microplate attachment assays^a

Strain	Absorbance (595 nm)b of:
	Initial biofilm after 1 day	Mature biofilm after 5 days	Initial biofilm after 12 days	Mature biofilm after 30 days
L. lactis IL-1403(pESoo8)	0.11 ± 0.01**	0.23 ± 0.05**	ND	ND
L. lactis IL-1403(pGK13)	0.09 ± 0.01	0.11 ± 0.03	ND	ND
P. parvulus strains
IOEB 8801	ND	ND	0.79 ± 0.20**	0.99 ± 0.25**
IOEB 0206			0.28 ± 0.11	0.35 ± 0.12
O. oeni strains
IOEB 0205	ND	ND	0.09 ± 0.02	0.35 ± 0.12*
PSU1			0.10 ± 0.03	0.21 ± 0.02
IOEB 8413			0.09 ± 0.02	0.15 ± 0.05

^aBiofilms were stained with crystal violet after short (initial biofilm) or long (mature biofilm) incubation periods. The incubation period was varied depending on the species-specific growth rate in the conditions of the assay.

^bValues that were significantly different between the gtf⁺ strain and the corresponding gtf mutant strain(s) of the same species, after the same period of incubation, are indicated as follows: *, P < 0.05; **, P < 0.005. ND, not determined.

DISCUSSION

Several PCR experiments reported in the literature suggested the presence of gtf orthologs in O. oeni (10, 51, 53). In this study, we have shown that in O. oeni strains from wine, gtf is integrated in the chromosome, whereas in P. parvulus or L. diolivorans, gtf is located on the plasmid (18, 32, 51, 53). In addition, the gtf nucleotide sequence is highly conserved in O. oeni (99% identity between the five gtf orthologs sequenced in this study and gtf in O. oeni I4) but also among the other lactic acid bacteria (at least 97.5% identity). This high level of sequence identity may be due to high requirements in amino acid conservation to ensure glycosyltransferase activity, or it may indicate recent horizontal transfer between species (5, 25, 27, 30, 33, 38). Actually, frequent horizontal transfers are thought to be the cause of the high level of diversity within the species O. oeni (11, 39, 53) and may promote the evolution of bacterial gene loci through genomic island interexchange (5, 8, 24, 25, 30, 38, 53). Genomic islands can display very different structures, modes of acquisition, or mobility mechanisms (8, 24). Indeed, the mosaic structure and the presence of putative transposase genes (ORF1 and ORF2), as well as truncated insertion sequences (ISs) in the nucleotide sequence surrounding gtf, suggests that transfers and recombinations involving ISs may have led to the integration of gtf in the O. oeni chromosome. Furthermore, several facts indicate that gtf and the nucleotide sequence surrounding it in O. oeni (ORF1 and ORF2) could form part of a genomic island. (i) These three ORFs were systematically found in the O. oeni gtf⁺ strains analyzed, and their sequences are highly conserved. (ii) These ORFs were not encountered in the O. oeni gtf mutant strains analyzed. (iii) The gtf⁺ strains displayed distinct PFGE profiles and originate from very distant geographic zones. This decreases the probability of a common ancestor that received gtf and then evolved into different strains without significant changes in the gtf sequence, but (iv) the island might be larger than 3,528 bp. However, its exact size could not be determined in the present study due to the presence of IS sequences at the current boundaries resulting in multiple priming during PCR extension assays.

The acquisition of an island is not always sufficient to confer the associated phenotype to the bacteria (23). In the present case, we showed that gtf is expressed in all the O. oeni gtf⁺ strains tested. We also checked the activity of Gtf through β-glucan detection. (i) The molecular cloning of gtf into L. lactis led to the accumulation of β-glucan in the capsular form and in the soluble form. (ii) Immunoagglutination assays indicated the presence of a β-glucan capsule around the cells of most of the O. oeni gtf⁺ strains. (iii) O. oeni IOEB 0205 accumulated EPS in the culture supernatant, and β-glucan was present in soluble form, although it represented a very minor part of the polysaccharides in the supernatant.

The integration of gtf in the chromosome presents several advantages over retaining it on an episome: there is no further need for replicating elements, and no risk of losing the gene when selective pressure is lower. This finding, associated with the unexpected level of occurrence of gtf (18/80 strains) in the O. oeni collection studied by Delaherche (10), prompted us to investigate the selective advantage provided by the acquisition of gtf. The present study shows that, in O. oeni IOEB 0205, the EPSs produced were not degraded and did not serve as external sources of carbon or energy, as described in the literature for pediococci β-glucan and for most lactic acid bacteria EPS (2). In addition, by using the plasmid-cured P. parvulus strain and the recombinant L. lactis, we showed that the presence of gtf confers two selective advantages to these bacteria. (i) The first selective advantage is a higher rate of survival in acidic pH, high ethanol or sulfur dioxide concentrations in a planktonic state, and a higher rate of survival in red or white wine in a planktonic state. Lonvaud-Funel and Joyeux (31) had previously noticed that pediococci isolated from ropy wine exhibited a strong resistance to conditions in wine (e.g., ethanol, pH, or SO₂ content), but the results of this study have clearly demonstrated the link between gtf and this stress resistance. Gtf certainly promotes resistance by synthesizing a glucan capsule around the cells. Indeed, many capsular EPSs have been shown to play a role in the protection of the microbial cell against desiccation, phagocytosis, phage attack, antibiotics, toxic compounds, and osmotic stress (2, 21, 35, 44, 46). (ii) The second selective advantage brought by gtf is an increased adhesion in the early stages and above all in the later stages of biofilm formation. Biofilm formation is a sequential process initiated by the attachment of planktonic cells to a surface and subsequent steps where the entire community is embedded in an amorphous matrix partly composed of polysaccharides (6, 41, 42, 52). Linear and nearly fibrillar neutral polymers, such as β-glucan produced by Gtf, are thought to constitute key elements that strengthen and organize the biofilm structures (19, 45, 52).

In O. oeni, the selective advantage brought by gtf was not entirely demonstrated, due to the variable genetic background of the gtf⁺ and gtf mutant strains. O. oeni IOEB 0205 (gtf⁺) exhibited significantly higher adhesion than the other strains analyzed. It also exhibited a higher survival rate than strain IOEB8413 did. This, associated with the results obtained with L. lactis and P. parvulus, suggests that Gtf forms part of the selection of O. oeni stress resistance tools, in addition to other, better described metabolic equipment, such as H+ ATPases (20, 23), FtsH proteases, chaperones (3, 20) and multidrug resistance proteins (4). However, O. oeni IOEB 0205 (gtf⁺) was not as resistant to red wine as the malolactic starter IOEB Sarco 450 was. β-Glucan synthesis may be useful for survival in wine, but it is not sufficient to confer the highest survival rates on the gtf⁺ strains. Nevertheless, the high proportion of O. oeni gtf⁺ strains in the collection analyzed may reflect the following: (i) a higher degree of tolerance to wine (especially to white wine) thanks to the presence of a glucidic capsule around the cells, and (ii) a selective advantage for adhesion to grapes or winemaking material and a higher level of subsistence on the winery material through biofilm cohesion.

What about the risk of spoilage? None of the O. oeni gtf⁺ strains were isolated from ropy wine, and the species has never been blamed for inducing ropiness. However, in the case of O. oeni IOEB 0205, the β-glucan produced, although a minority among the EPSs produced, is abundant enough or may interact with the other polysaccharides present (40) to clearly increase the viscosity of dialyzed MRS. In contrast, the other O. oeni gtf⁺ strains isolated from wine, as well as the recombinant lactococcal strains, produced soluble polysaccharides, but no increase in viscosity was observed in the model medium. Such discrepancy between genetic equipment and slime production was observed by Bourgoin et al. (5) with Streptococcus thermophilus. It is possible that these O. oeni gtf⁺ strains produce additional polysaccharides which prevent β-glucan network formation and thus prevent an increase in viscosity (40). It is also possible that these strains never induce ropiness due to a level of β-glucan formation that is too low, which may be the consequence of the low availability of the precursor UDP-glucose (2, 27). Indeed, four of these strains did not agglutinate in the presence of antibodies targeting β-glucan, suggesting a very low level or even the absence of excreted β-glucan among the excreted EPSs. The composition of the medium can also influence polysaccharide formation and composition (2, 47). However, the carbohydrates in wine are far less abundant than in the model media used in this study (43). This suggests that β-glucan formation in wine may be even more difficult than in the model medium. Nevertheless, it is still possible that these O. oeni gtf⁺ strains, once in a convenient must or wine, abundantly produce β-glucan and lead to wine ropiness. As a result, though induction of ropiness in wine by most O. oeni gtf⁺ strains seems improbable, it cannot be completely excluded.

More worrying is the fact that gtf can form part of a functional mobile element. This, associated with the improved capacity of the gtf⁺ bacteria to form biofilms, may promote gtf transfer from one species to another and between various strains within a species. Indeed, biofilms are the place of accelerated conjugation rates (52). Assessment of gtf mobility will be an interesting focus for further studies.