Molecular Description and Industrial Potential of Tn6098 Conjugative Transfer Conferring Alpha-Galactoside Metabolism in Lactococcus lactis

Abstract

A novel 51-kb conjugative transposon of Lactococcus lactis, designated Tn6098, encoding the capacity to utilize α-galactosides such as raffinose and stachyose, was identified and characterized. Alpha-galactosides are a dominant carbon source in many plant-derived foods. Most dairy lactococcus strains are unable to use α-galactosides as a growth substrate, yet many of these strains are known to have beneficial industrial traits. Conjugal transfer of Tn6098 was demonstrated from the plant-derived donor strain L. lactis KF147 to the recipient L. lactis NZ4501, a derivative of the dairy model strain L. lactis MG1363. The integration of Tn6098 into the genome of the recipient strain was confirmed by Illumina sequencing of the transconjugant L. lactis NIZO3921. The molecular structure of the integration site was confirmed by a PCR product spanning the insertion site. A 15-bp direct repeat sequence (TTATACCATAATTAC) is present on either side of Tn6098 in the chromosome of L. lactis KF147. One copy of this sequence is also present in the L. lactis MG1363 chromosome and represents the sole integration site. Phenotypic characterization of all strains showed that the transconjugant has not only acquired the ability to grow well in soy milk, a substrate rich in α-galactosides, but also has retained the flavor-forming capabilities of the recipient strain L. lactis MG1363. This study demonstrates how (induced) conjugation can be used to exploit the beneficial industrial traits of industrial dairy lactic acid bacteria in fermentation of plant-derived substrates.

Lactic acid bacteria (LAB) are widely applied in food fermentations, where they contribute to the organoleptic, preservation, and health properties of the fermented food stuffs. Because of their industrial relevance, numerous LAB genome sequencing initiatives have been conducted (13, 17, 32). Comparative genomics of LAB has revealed the existence of a large inter- and intraspecies genomic diversity (4, 5, 9, 18, 26, 27, 35, 39). Horizontal gene transfer (HGT) appears to be a primary driver in LAB evolution and the creation of diversity. HGT is facilitated by the widespread occurrence of plasmids, phages, and mobile genetic elements such as insertion sequence (IS) elements and conjugative transposons, which serve as effective vectors for the spread of genes and traits within and over the species barrier (20, 40).

Lactococcus lactis is a primary constituent of many starter cultures used for the manufacturing of fermented dairy products, especially cheese and butter. Because of its tremendous industrial importance, it has long been the primary model organism for LAB research revealing molecular insight into many traits relevant for industrial applications, including the production of flavor compounds, vitamins and other nutraceuticals, and exopolysaccharides relevant for texture development (3, 8, 39).

L. lactis occupies two distinct niches. The occurrence of the species in dairy substrates has long been described, and L. lactis strains are readily isolated from artisanal and spontaneous fermentations. More recently there has been increasing interest in strains found in (fermented) plant material (21). Due to the different selective pressures in plant-related habitats, such strains typically have different phenotypes and they are often phylogenetically distinct from dairy isolates (26). Recently, the genomes of two plant isolates were sequenced, revealing extensive adaptation to the plant ecological niches, which is illustrated by the presence of several gene clusters encoding proteins for the degradation of plant carbohydrates (33, 34). In these studies, strain L. lactis subsp. lactis KF147 was found to be capable of growth on several α-galactosides, and in its genome, a complete gene cluster for α-galactoside uptake, breakdown, and subsequent d-galactose conversion by the Leloir pathway was found.

Considering the above characteristics, it is clear that the diversity reservoir present in the pangenomes of lactic acid bacteria is huge and remains largely unexploited for industrial application. Several approaches can be envisioned for harvesting this diversity. The first relies on the phenotypic or genome-based selection of strains expressing desired phenotypes. Examples for L. lactis may include the selection of strains capable of producing specific vitamins or expressing enzymes involved in the production of flavor compounds (2, 3). Alternatively, the transfer of desired traits into strains with proven industrial performance can be an effective means to build on this genetic potential. For instance, the introduction of a heterologous α-galactosidase from Lactobacillus plantarum into L. lactis allowed the selective removal of nondigestible α-galactosides from plant-derived foodstuffs such as soy (15). As consumer and regulatory demands strongly favor the application of food-grade and/or non-genetically modified organism (GMO) approaches, transfer of desired traits is preferably achieved through non-GMO strategies. A classic example is represented by the conjugative transfer of the transposon encoding nisin production and resistance, which is extensively applied to develop industrial starter cultures that inhibit blooming of spoilage microbes (11, 28, 29, 31).

The conjugative transfer of raffinose (10) and sucrose (11) metabolism has been demonstrated in Lactococcus lactis. Many L. lactis strains isolated from sprouted seeds were able to ferment raffinose, and nearly all of these were able to conjugally transfer this property to the L. lactis MG1614 recipient (10). Here we present a molecular analysis of the transfer of α-galactoside metabolism between L. lactis strains. It is demonstrated that in the nondairy strain L. lactis subsp. lactis KF147, isolated from mung bean sprouts, this trait is encoded on a 51-kb genomic fragment, which can be efficiently transferred by conjugation to the well-known dairy strain L. lactis subsp. cremoris MG1363. The 51-kb conjugative transposon and its integration site were identified and characterized. Moreover, further phenotypic characterization of the transconjugant showed that it has acquired the ability to grow well in soy milk while retaining flavor-forming capabilities of the recipient strain.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

L. lactis NZ9000 is a derivative of L. lactis subsp. cremoris MG1363 in which part of the pepN gene has been replaced by the nisR and nisK genes by double crossover, exactly as described for the construction of L. lactis NZ3900 from NZ3000 (6, 14). L. lactis NZ4501, a derivative of L. lactis NZ9000 carrying an erythromycin resistance marker (noxA::ery double crossover) (14), was used as the recipient strain in conjugation experiments. Lactococcus lactis subsp. lactis KF147 (10, 33, 34) and Lactococcus raffinolactis ATCC 43920 were used as donor strains. In this study, we generated strain Lactococcus lactis NIZO3921, which is a derivative of L. lactis NZ4501 that has received a DNA fragment from L. lactis KF147 by conjugation (Fig. 1). All bacteria were grown without aeration at 30°C in M17 medium (Merck, Darmstadt, Germany) supplemented with 0.5% glucose (GM17) or anaerobically (incubation in an anaerobic jar with an Anaerocult A sachet [Merck] to ensure an oxygen-free environment) in chemically defined medium (CDM) for L. lactis supplemented with 1% raffinose, sucrose, or l-arabinose (23). When appropriate, the antibiotic erythromycin (Ery) was added to the media, at a concentration of 5 μg/ml.

FIG. 1.

Schematic representation of the conjugative transfer of Tn6098 from the dairy isolate L. lactis KF147 to Lactococcus lactis NZ4501, a derivative of the dairy model strain Lactococcus lactis MG1363. Ery^r, erthromycin resistant; Raf+, grows on raffinose; Suc+, grows on sucrose. Positions of the target insertion sequence TTATACCATAATTAC are indicated. A part of Tn6098 is enlarged and shows genes encoding α-galactoside uptake and metabolism.

Transfer of α-galactoside metabolism.

The recipient strain (L. lactis NZ4501) and donor strain (e.g., L. lactis KF147) were grown overnight in GM17 (with the appropriate antibiotic if necessary) at 30°C. The overnight cultures were inoculated at 2% in fresh GM17 medium (with appropriate antibiotic) and incubated at 30°C until an optical density at 600 nm (OD₆₀₀) of approximately 0.5 was reached. Then a mating mixture was prepared by adding 5 ml culture of both the recipient and donor strain together (with 10 ml each of the recipient and donor strain alone taken along as controls). The cells of the mating mixture were harvested by centrifugation (10 min at 2,000 × g). The pellet was resuspended in 10 ml GM17 and incubated at 37°C for 45 min. The resulting cells were washed twice with GM17 and subsequently resuspended in 1 ml GM17. Subsequently, 0.5 ml of this resuspension was transferred to 1.5 ml fresh GM17 medium and incubated at room temperature for 20 min. Next, 5 ml GM17 was added and the cells were harvested by centrifugation. The pellet was resuspended in 1 ml GM17 and incubated at 30°C for 90 min. The cells were harvested by centrifugation and washed five times with PFZ (peptone physiological salt solution; Tritium Microbiologie, Eindhoven, Netherlands), and appropriate dilutions (10⁻¹ to 10⁻⁷) were plated on selective CDM plates with 1% carbon source (e.g., raffinose) and 5 μg/ml erythromycin. The plates were incubated anaerobically at 30°C for 3 to 5 days, after which colonies were counted and further study of the transconjugants was performed.

To determine the necessity of all steps in the transfer protocol, appropriate dilutions were plated on selective CDM plates after every step in the protocol. The plates were incubated anaerobically at 30°C for 3 to 5 days, after which colonies were counted. DNase I could have a possible negative effect on the transfer frequency, by hampering uptake of extracellular DNA; it was therefore added to all media and wash solutions at a concentration of 100 μg/ml.

To determine if the transfer could also be achieved with a commonly applied conjugation protocol, several conjugation experiments were conducted with L. lactis NZ4501 as the recipient strain and L. lactis KF147 as the donor strain. Plate-mating experiments were done as described by Kelly et al. (10), and filter-mating experiments were performed (7). A 0.45-μm-pore-size Millipore filter (25-mm diameter; type HAWP02500 [Millipore, Billerica, MA]) was used in the experiments.

Stability of L. lactis NIZO3921.

L. lactis NIZO3921, L. lactis NZ4501 (the derivative of Lactococcus lactis MG1363) and L. lactis KF147 were grown in GM17 for 24 h at 30°C. Appropriate dilutions were plated on GM17 plates with 5-bromo-4-chloro-3-indolyl-α-d-galactopyranoside (X-α-Gal) as an α-galactosidase indicator substrate (MP Biomedicals, Amsterdam, Netherlands), which were incubated overnight at 30°C. The 24-h cultures were also used for inoculation at 0.5% in fresh GM17 medium, which was once more incubated for 24 h at 30°C. This process was repeated for 7 days. The α-galactosidase-positive donor and recipient clones were quantified by counting blue colonies on X-α-Gal-containing plates.

Soy milk fermentation.

Overnight-grown cultures of L. lactis NIZO3921, L. lactis NZ4501, and L. lactis KF147 were used for inoculation at 1% in soy milk (in triplicate; Benesoy conventional soy milk powder; Devansoy, Carroll, IA). Soy milk cultures were grown for 24 h at 30°C, with samples taken after 6, 12, and 24 h of growth for analysis of the sugars and organic acids by high-performance liquid chromatography (HPLC) analysis. The final pH of all cultures was also determined. Furthermore, soy milk cultures were grown in closed vials for 24 h at 30°C, after which determination of the volatile metabolites was performed by gas chromatography-mass spectrometry (GC-MS) analysis. In the latter experiment, the L. lactis MG1363 derivative was grown in soy milk alone and also in soy milk with 1% glucose.

Metabolic analysis.

Extracellular metabolites present in the soy milk fermentations were measured by HPLC as previously described (37). For volatile analysis, a gas chromatography and mass spectrometry (GC-MS) setup was used. Two milliliters of phosphate buffer (pH 6.7) with 0.5 ppm 1,2,3-trichloropropane as the internal standard was added to each 5-ml batch culture in soy milk (vials closed with a Teflon septum). Headspace volatiles equilibrated at 60°C for 20 min were concentrated by solid-phase dynamic extraction followed and focused by cryofixation at −50°C. Subsequently the compounds were separated on a FactorFour VF-1ms column (30 m by 0.25 mm; Varian, Palo Alto, CA). The initial temperature of the GC column was held at 40°C for 5 min and then increased at 15°C/min to a final temperature of 250°C, at which it was held for 5 min. Next, the separated compounds were detected on a mass spectrometer by scanning a mass range of 25 to 250 in 0.25 s in the full-scan electron impact (EI) ionization mode (70 eV). Data acquisition and processing were performed with the Xcalibur software, and volatiles were identified with the NIST MS Library.

A total of 25 metabolites were identified and quantified in the different samples. In order to investigate the relationship between sample type and metabolite profile, hierarchical bi-clustering was performed. Data were normalized first for the samples and then for the metabolites by generating z-scores. In this data transformation, the average for each sample is 0 and the standard deviation is 1. Bi-clustering was performed in the Genesis software (38), using average linkage clustering as an agglomeration rule.

Genomic characterization.

Genomic DNA of the colonies was obtained by using InstaGene Matrix (Bio-Rad, Veenendaal, Netherlands). Subsequently, a PCR was performed to confirm the transfer of the α-galactoside metabolism from L. lactis KF147 to L. lactis NZ4501. The primers RM0009 (sense; 5′-TCACTTGATTTCATTTGATTTGACTTC) and RM0010 (antisense; 5′-ATGACACTAATCACATTTGATGAAAGC) were designed to verify the presence of the α-galactosidase gene of L. lactis KF147 (LLKF_2267) in the transconjugants. The integration site of the KF147 fragment in the MG1363 chromosome was determined by PCR using primer RM0013 (5′-GGTCAGAGCCTCGTCTGATTTCC; forward) in gene llmg_2319, preceding the putative integration site in MG1363 DNA, and primer RM0014 (5′-CTAGTTCGTTTGTGATGCACTCG; reverse) in the 51-kb fragment from KF147 (in gene LLKF_2229).

Full genome resequencing of L. lactis NIZO3921 with Illumina technology was performed by GATC-Biotech (Konstanz, Germany). In order to determine the sequence of the transferred region, the reads were compared to the reference genomes of L. lactis KF147 (GenBank code CP001834) (33) and L. lactis MG1363 (GenBank code AM406671) (41), as well as to the recently published L. lactis NZ9000 (GenBank code CP002094) and revised L. lactis MG1363 genome sequence (16). Alignment of the Illumina reads was done with a tool developed in house, RoVar (S. A. F. T. van Hijum et al., unpublished), which uses BLAT, version 34 (12), to align reads to a repeat-masked reference sequence. Repeat masking of the reference sequence was done by (i) creating 30-bp fragments, (ii) aligning these fragments to the reference sequence by using BLAT, and (iii) masking regions to which fragments align perfectly in multiple positions in the reference sequence. Next, Illumina sequence reads were aligned to the reference genome sequences by BLAT. Alignment events of reads to the reference genomes were allowed, provided that nucleotide substitutions (single nucleotide polymorphisms; SNPs) or gaps (small insertions or deletions; INDELs) were at least 4 bp from either end of the reads. RoVar provides detailed information regarding structural variations (SNPs and INDELs) present in the reads compared to the reference sequences. The following criteria were used for selecting structural variations: (i) the region of the structural variation should be present only once in the reference sequence; (ii) no perfect match reads should align; and (iii) the reads should unambiguously represent at least 10 unique sequences, which in effect represents 10 or more reads supporting each structural variation.

Assembly of remaining Illumina sequence reads, not aligning with either reference genome sequence, into contigs was done using ABYSS version 1.0.14 (36), with the following parameters: 36 bp in length and 28 k-mer in size. Alignment of contig sequences with either L. lactis MG1363 or KF147 was done by BLAT, in which alignment events were considered acceptable where contigs aligned with either or both reference sequences, with at least 85% of the contig nucleotides matched.

RESULTS

Transfer of α-galactoside metabolism.

We initiated a study on the transfer of genomic loci between L. lactis strains, using a genome-shuffling strategy based on protoplast fusion (22), focusing on transfer of sugar utilization genes from the plant isolate L. lactis KF147 to a derivative of L. lactis strain MG1363, carrying antibiotic resistance markers. During preliminary experiments, transfer of sugar metabolism was indeed observed (results not shown), but this also occurred in control experiments in which no protoplasts were prepared. This indicated that genome shuffling was not the cause of the transfer, but that transfer of DNA was induced by our protocol which included several incubation and centrifugation steps. The applied protocol led to transfer of the α-galactoside metabolism of L. lactis KF147 to a derivative of L. lactis MG1363, resulting in the transconjugant L. lactis NIZO3921 (Fig. 1). Transfer of α-galactoside metabolism was not detected when the type strain of L. raffinolactis, ATCC 43920, was used as the donor strain. When transconjugant libraries were tested for clones with the ability to ferment α-galactosides (raffinose), melibiose, sucrose, or l-arabinose, only transfer of the α-galactoside metabolism could be observed. Transfer of the α-galactoside metabolism occurred at a frequency of 10⁻⁵ transconjugants per donor cell. Investigation of the necessary steps within the transfer protocol showed that all incubation steps and nearly all centrifugation steps were essential for induction of the transfer of the galactose-positive phenotype. The addition of DNase I to all media and wash solutions did not have an effect on the transfer frequency. In addition, a classic conjugation protocol using filter mating did not result in transfer of the α-galactoside metabolism, whereas plate-mating experiments resulted in 10⁻⁹ transconjugants per donor cell, which is a much lower transfer frequency than that observed with the described transfer protocol.

Stability of the transconjugant strain L. lactis NIZO3921.

The stability of the transferred fragment in L. lactis NIZO3921 was examined by following the presence of the α-galactoside metabolism during sequential transfers under nonselective growth conditions. After 24 h of growth in GM17, 0.3% of cells of both the transconjugant L. lactis strain NIZO3921 and the donor strain L. lactis KF147 had lost the ability for α-galactoside metabolism. When the transconjugant L. lactis NIZO3921 and the donor strain L. lactis KF147 were grown under nonselective conditions (GM17) for 7 days by daily sequential transfer, only 1 to 2% of cells had lost α-galactoside metabolism, demonstrating the stability of the transconjugant. The recipient strain L. lactis MG1363 did not generate any blue colonies after 7 days of sequential transfer.

Genomic characterization of the transconjugant.

The genome of the transconjugant L. lactis NIZO3921 was sequenced with Illumina technology. Initially, only the published genome sequences of the donor strain L. lactis KF147 (33) and of the recipient strain L. lactis MG1363 (41) were used as references to determine the sequence of the transferred fragment. Using our RoVar algorithm, 4,808,276 Illumina sequence reads (length, 36 nucleotides [nt]) could be aligned on the L. lactis MG1363 genome, resulting in an average coverage of 60-fold, and another 117,876 reads mapped to the L. lactis KF147 genome, with 70-fold coverage (see below). The entire chromosome of L. lactis MG1363 was covered by reads, except for the repeat-masked regions (about 28 kb) and an 1.8-kb region (bp 305041 to 306862) corresponding to the truncated pepN gene of L. lactis NZ9000.

Next, reads that did not satisfy RoVar alignment criteria to L. lactis MG1363 were assembled into contigs with ABYSS. A total of 150 contigs with an n50 of 6,014 could be assembled. The entire 51-kb transposon of strain KF147 is represented by 17 contigs of over 200 nt (size range, 213 to 15,949 nt) (see Table S1 in the supplemental material for details). Interestingly, one of these 17 contigs of KF147 sequence (2,081 nt, containing 6,226 reads, 108-fold coverage) links the left and right ends of the 51-kb transposon of strain KF147, with a single copy of the 15-bp repeat (see below) in the middle, suggesting circularization of the transposon in a small subset of the population. The remainder of the assembled contigs were all <200 nt and map to several positions on the KF147 and MG1363 genomes, indicating that they are short repeats.

Excision/integration site.

A 15-bp repeat sequence (TTATACCATAATTAC) is present on either side of the 51-kb transposon in the chromosome of L. lactis KF147 at positions 2295682 to 2295696 and 2347022 to 2347036, and this sequence is also present once in the L. lactis MG1363 chromosome (positions 2272817 to 2272831) (Fig. 1). The integration site was confirmed by PCR analysis. A primer combination was designed in which the forward primer anneals to the genome of MG1363 in the gene preceding the integration site and the reverse primer anneals to the 51-kb transposon of KF147. A PCR on genomic DNA of the transconjugant NIZO3921 generated a product with the expected nucleotide sequence (i.e., MG1363 sequence before the integration site and KF147 sequence after the site). Moreover, the Illumina reads confirmed the insertion event, as coverages of 64x and 67x were found across the left and right boundaries, respectively, of Tn6098 of L. lactis KF147 inserted into the MG1363 genome sequence at this position (see Table S2 in the supplemental material).

Putative structural variations.

All Illumina reads were aligned to the genomes of L. lactis MG1363 and KF147, using our tool RoVar, to identify putative structural variations: single nucleotide polymorphisms (SNPs), small insertions, and small deletions (INDELs) (see Table S3 in the supplemental material). Nearly all of the observed variations were also found in the recently published corrected genome sequence of L. lactis MG1363 and/or in strain NZ9000 (16). Therefore, we conclude that no structural variations occurred in the recipient genome during conjugation other than insertion of transposon Tn6098.

Structure of transposon Tn6098.

Tn6098 of L. lactis KF147 has a size of 51,339 bp, including a single 15-bp flanking repeat, and encompasses genes LLKF_2229 to LLKF_2287 (Table 1). One end of Tn6098 contains the genes xis (encoding an excisionase) and int (encoding an integrase), typically involved in excision and integration of conjugative transposons (24, 25, 30), while the other end encodes a tRNA-Met (LLKF_t0054). Tn6098 contains a cluster of genes encoding a sucrose phosphorylase (LLKF_2259) and a dedicated ABC transporter (LLKF_2260 to LLKF_2262), as well as enzymes for metabolism of α-galactosides, such as stachyose, raffinose (also referred to as melitose), and melibiose, which are typical plant oligosaccharides. This gene cluster is flanked by IS1216 elements. Another Tn6098 gene cluster (LLKF_2279 to LLKF_2287) is nearly identical to a cluster elsewhere (LLKF_0272 to LLKF_0284) on the chromosome in a different putative transposon (see below) and encodes cold shock proteins, an Mg²⁺/Co²⁺ transporter, and an N⁵-(carboxyethyl)-ornithine synthase. Furthermore, numerous hypothetical proteins of unknown function are encoded on Tn6098, some of which may be involved in conjugal transfer.

TABLE 1.

Annotation of the 51-kb transposable element Tn6098 of Lactococcus lactis F147

Locus tag	Nucleotide position		Strand	Gene	Amino acids	EC no.	Product (based on best BLAST hits)
	Start	Stop
LLKF_2229	2297100	2295829	−	int	424		Integrase, conjugative transposon
LLKF_2230	2297370	2297161	−	xis	70		Excisionase, conjugative transposon
LLKF_2231	2297920	2297429	−	ardA2	164		Conjugative transposon antirestriction protein
LLKF_2232	2298438	2297944	−		165		Hypothetical protein
LLKF_2233	2299597	2298533	−		355		N-Acetylmuramoyl-l-alanine amidase, CHAP domain family
LLKF_2234	2301658	2299613	−		682		Hypothetical protein
LLKF_2235	2304226	2301659	−		856		Conjugal transfer protein
LLKF_2236	2304591	2304226	−		122		Hypothetical protein
LLKF_2237	2304833	2304600	−		78		Hypothetical protein
LLKF_2238	2306107	2304839	−		423		Hypothetical protein
LLKF_2239	2306311	2306111	−		67		Hypothetical protein
LLKF_2240	2306941	2306681	−		87		Hypothetical protein
LLKF_2241	2307315	2307055	−		87		Hypothetical protein
LLKF_2242	2307614	2307348	−		89		Hypothetical protein
LLKF_2243	2308916	2307639	−		426		Replication initiation factor
LLKF_2244	2310916	2309297	−		540		DNA segregation ATPase, FtsK/SpoIIIE family
LLKF_2245	2311432	2310926	−		169		Hypothetical protein
LLKF_2246	2311770	2311435	−		112		Hypothetical protein
LLKF_2247	2311896	2311774	−		41		Hypothetical protein
LLKF_2248	2312146	2311955	−		64		Hypothetical protein
LLKF_2249	2312419	2312285	−		45		Hypothetical protein
LLKF_2250	2312682	2312416			89		Hypothetical protein
LLKF_2251	2313370	2312819	−		184		Hypothetical protein
LLKF_2252	2313800	2313480	−		107		Transcriptional regulator
LLKF_2253	2314169	2314327	+		53		Hypothetical protein
LLKF_2254	2314496	2314840	+		115		Transcriptional regulator
LLKF_2255	2314818	2315900	+		361		Hypothetical protein
LLKF_2256	2315964	2316416	+		151		IS3/IS911 transposase, N-terminal fragment (pseudogene)
LLKF_2257	2316470	2316754	+		95		IS1216 transposase A, N-terminal fragment (pseudogene)
LLKF_2258	2316702	2317118	+		139		IS1216 transposase A, C-terminal fragment (pseudogene)
LLKF_2259	2318669	2317224	−	sucP	482	2.4.1.7	Sucrose phosphorylase
LLKF_2260	2319583	2318756	−	agaA	276		Alpha-galactoside ABC transporter, permease protein
LLKF_2261	2320465	2319596	−	agaB	290		Alpha-galactoside ABC transporter, permease protein
LLKF_2262	2321810	2320566	−	agaC	415		Alpha-galactoside ABC transporter, sugar-binding protein
LLKF_2263	2321960	2322829	+	agaR	290		Transcriptional regulator, AraC family
LLKF_2264	2324445	2322901	−	purH2	515	2.1.2.3/3.5.4.10	Phosphoribosylaminoimidazolecarboxamide formyltransferase/IMP cyclohydrolase
LLKF_2265	2326054	2324570	−	galT2	495	2.7.7.10	Galactose 1-phosphate uridylyltransferase
LLKF_2266	2327241	2326051	−	galK2	397	2.7.1.6	Galactokinase
LLKF_2267	2329454	2327238	−	aga	739	3.2.1.22	Alpha-galactosidase
LLKF_2268	2330622	2329585	−	galR	346		Transcriptional regulator, LacI family
LLKF_2269	2332669	2330747	−	fbp2	641	3.1.3.11	Fructose 1,6-bisphosphatase
LLKF_2270	2332809	2333120	+		104		Hypothetical protein, N-terminal fragment (pseudogene)
LLKF_2271	2333148	2333432	+		95		IS1216 transposase A, N-terminal fragment (pseudogene)
LLKF_2272	2333380	2333796	+		139		IS1216 transposase A, C-terminal fragment (pseudogene)
LLKF_2273	2336160	2335306	−		285		IS981 transposase B
LLKF_2274	2336417	2336157	−		87		IS981 transposase A
LLKF_2275	2336808	2337641	+		278		Hypothetical protein
LLKF_2276	2338656	2337802	−		285		IS981 transposase B
LLKF_2277	2338913	2338653	−		87		IS981 transposase A
LLKF_2278	2339521	2340483	+		321	1.1.1.-	d-isomer-specific 2-hydroxyacid dehydrogenase, NAD dependent
LLKF_2279	2341936	2341013	−	ceo2	308	1.5.1.24	N5-(Carboxyethyl)ornithine synthase
LLKF_2280	2342925	2341975	−	corA5	317		Mg2+ and Co2+ transporter, CorA family
LLKF_2281	2343601	2343050	−	ymgG2	184		General stress protein, Gls24 family
LLKF_2282	2343810	2343622	−	ymgH2	63		Hypothetical protein
LLKF_2283	2344380	2343820	−	ymgI2	187		Hypothetical protein
LLKF_2284	2344649	2344410	−	ymgJ2	80		Hypothetical protein
LLKF_2285	2345364	2345164	−	cspB2	67		Cold shock protein
LLKF_2286	2345842	2345642	−	cspC2	67		Cold shock protein
LLKF_2287	2346173	2345973	−	cspA2	67		Cold shock protein
LLKF_t0054	2346274	2346348	−				tRNA-Met

Soy milk fermentation.

We studied how the transfer of Tn6098 carrying the α-galactosidase metabolism genes would affect the ability of the strain to grow in soy milk, an α-galactoside-rich substrate. Soy milk with an initial pH of approximately 6.9 and a sugar content of 0.4% sucrose, 0.2% raffinose, 0.2% stachyose, and residual glucose could be fermented by L. lactis KF147 (final pH of 4.25) and the transconjugant L. lactis NIZO3921 (final pH of 4.65). The dairy strain L. lactis MG1363 could acidify the soy milk slightly (final pH of 6.23) due to growth on the residual glucose. Sugars and organic acids were analyzed after 6, 12, and 24 h of growth in soy milk by HPLC analysis (Table 2). Galactose, fructose, citric acid, butyrate, formate, propionate, 3-hydroxybutanone (acetoin), acetic acid, ethanol, and pyruvate were not detected, or their values did not change during the fermentation. Table 2 shows that L. lactis MG1363 cannot convert sucrose, stachyose, and raffinose. The donor strain L. lactis KF147 clearly prefers the consumption of sucrose over the consumption of stachyose or raffinose, while the transconjugant, which is not capable of converting sucrose, consumes most of the stachyose and raffinose present in the soy milk. L. lactis KF147 produces more lactic acid than L. lactis NIZO3921 when grown on soy milk, which is also seen in the acidification of the soy milk cultures. In conclusion, these results show that transfer allowed the recipient dairy strain to efficiently harvest α-galactosides and grow in soy milk.

TABLE 2.

Soy milk fermentation by L. lactis strains

Strain	Incubation time (h)	Sugar and organic acid concn (mg/g)					Final pH
		Glucose	Sucrose	Stachyose	Raffinose	Lactic acid
Soy milk	0	0.33 ± 0.01	4.21 ± 0.04	2.47 ± 0.01	2.01 ± 0.11	0.0	6.89 ± 0.02
KF147	6	0.0	2.92 ± 0.09	2.37 ± 0.03	2.01 ± 0.19	1.35 ± 0.05
	12	0.0	1.63 ± 0.05	2.33 ± 0.03	1.86 ± 0.08	2.65 ± 0.12
	24	0.0	0.66 ± 0.07	2.32 ± 0.11	1.81 ± 0.16	3.53 ± 0.02	4.25 ± 0.03
MG1363a	6	0.0	4.06 ± 0.04	2.38 ± 0.02	1.95 ± 0.01	0.20 ± 0.00
	12	0.0	4.03 ± 0.09	2.36 ± 0.06	1.92 ± 0.03	0.20 ± 0.00
	24	0.0	3.78 ± 0.27	2.37 ± 0.01	1.85 ± 0.04	0.41 ± 0.26	6.23 ± 0.47
NIZO3921	6	0.0	3.96 ± 0.11	1.72 ± 0.07	0.44 ± 0.09	0.91 ± 0.01
	12	0.0	3.87 ± 0.06	1.02 ± 0.02	0.35 ± 0.04	1.72 ± 0.00
	24	0.0	3.82 ± 0.08	0.56 ± 0.14	0.11 ± 0.10	2.39 ± 0.03	4.65 ± 0.02
Blank	6	0.0	4.02 ± 0.07	2.39 ± 0.03	1.94 ± 0.09	0.0
	12	0.0	3.94 ± 0.10	2.36 ± 0.02	1.87 ± 0.04	0.0
	24	0.0	4.00 ± 0.08	2.37 ± 0.01	1.83 ± 0.06	0.0	6.90 ± 0.05

^aLactococcus lactis NZ4501, a derivative of Lactococcus lactis MG1363.

Subsequently, we focused on the flavor-forming capacity of the donor, recipient, and transconjugant. All strains were grown in soy milk, and glucose was added to allow growth of the recipient strain MG1363. Determination of volatile flavor compounds was performed by GC-MS analysis. In total, 25 metabolites were determined and correlation of sample type to GC-MS metabolite profile was performed. Figure 2 shows the hierarchical bi-clustering of the samples versus the metabolite profile. The metabolite measurements are highly reproducible. With the exception of the very similar profiles of L. lactis MG1363 and the blank, replicates of the same sample cluster at a closer distance than the different samples. The metabolic profiles of L. lactis KF147 are clearly separate from those of the other samples. Three metabolites are highly abundant in this strain: 2-methylbutanal, 2-methyl-1-butanol, and 3-methyl-1-butanol. In addition, 2-heptanone is of low abundance in this strain. Although the differences for the remainder of the samples are smaller, two subgroups can be distinguished: (i) the blank and L. lactis MG1363 and (ii) L. lactis MG1363 grown on soy milk with addition of glucose (MG1363glu in Fig. 2) and L. lactis NIZO3921. The latter two strains produce larger amounts of 1-pentanol and 1-hexanol. These results show that the flavor-forming capacity of the recipient dairy strain was largely maintained in the transconjugant.

FIG. 2.

Hierarchical bi-clustering of metabolite data obtained for different strains. The z-score transformed data are colored from blue (low abundance) to red (high abundance).

DISCUSSION

The conjugative transfer of raffinose metabolism, sucrose metabolism, or nisin-sucrose metabolism from plant-derived L. lactis strains to the recipient dairy strain L. lactis MG1614 has previously been observed (10, 11, 29). Full genome sequencing of the plant isolate L. lactis KF147 has revealed several large inserts (possibly transposons) compared to dairy strains (33, 34). One large insert (LLKF_1277 to LLKF_1308; 27 kb) encodes a degenerate nisin biosynthesis cluster, while three other transposon-like inserts are related to sugar utilization and are flanked by genes encoding either Tn5276-like integrases, cold shock proteins, or tRNAs. One putative transposon (LLKF_0274 to LLKF_0325, 43 kb; flanked by tRNAs) has genes for breakdown of complex glucans by removal of N-acetyl-glucosamine side chains and cleavage of the main glucan backbone (34). The second putative transposon (LLKF_0657 to LLKF_0705, 48 kb; flanked by an integrase) encodes transporters and enzymes for β-glucoside utilization and for sucrose utilization (scrRBAK). The largest putative transposon (LLKF_2229 to LLKF_2287, 51 kb; flanked by int/xis genes and tRNA and 15-bp repeats) was identified here as Tn6098 encoding α-galactoside metabolism.

In our experiments, genes encoding the raffinose metabolism were transferred by Tn6098 from the plant isolate L. lactis KF147 to a derivative of the well-known dairy strain L. lactis MG1363. Our sequencing analysis shows that the other putative transposons of strain KF147 apparently are not transferred in our conjugation experiments. This is in agreement with the results of Kelly et al. (10, 11), who observed that only raffinose metabolism could be transferred and not l-arabinose or sucrose metabolism. With the induced conjugation protocol, a transfer frequency of 10⁻⁵ transconjugants per donor cell was observed. Kelly et al. (10) detected an insert of approximately 60 kb by pulsed-field gel electrophoresis, but they could not detect an α-galactosidase gene on the insert. The present study revealed the whole sequence of the insert, a 51-kb transposon of KF147, and showed among others the presence of an α-galactosidase gene on this transposon. A 15-bp repeat sequence was detected on either side of the 51-kb transposon in the chromosome of L. lactis KF147, and this sequence is also present once in the L. lactis MG1363 chromosome. Both PCR analysis and Illumina sequencing of the transconjugant confirmed that integration takes place at this site in MG1363. We hypothesize that it is involved in a recombination event, transferring the 51-kb region from the KF147 chromosome to MG1363. The result of such a recombination event is that the single integration site in the MG1363 sequence is duplicated in the transconjugant, with insertion of the 51-kb fragment from strain KF147 between these repeats. Nucleotide sequences flanking the integration sites originate from MG1363 on one side of the integration site and from KF147 on the other side. The sequence reads and assembly suggest that some cells also contain a circular form of Tn6098 that has not yet integrated or has been excised again. This observation agrees with the finding of Kelly et al. (10), that after isolation of plasmid DNA a faint ∼40- to 50-kb band was detected in some of the transconjugants, which did not correspond to plasmid DNA in the donor strain and hence may represent the circularized form of Tn6098. Recently, it has also been shown that this transposon Tn6098 is spontaneously lost upon adaptation of L. lactis KF147 to growth in milk (1).

The exact 15-bp target sequence for Tn6098 is also present once in the genomes of L. lactis IL1403 and SK11 in the same gene context, which in MG1363 is about 50 bp upstream of llmg_2319 (snf), encoding an SWI/SNF family helicase, and about 170 bp downstream of llmg_2320, encoding a putative fibronectin-binding protein precursor. This target sequence is also present in a different gene context in genomes of Lactobacillus salivarius strains, Leuconostoc citreum KM20, and Oenococcus oeni PSU-1, albeit that in the last two genomes the sequence only matches to nucleotides 1 to 14. This suggests that it may be possible to transfer Tn6098 and α-galactoside metabolism to other lactic acid bacteria. In each case, the target sequence is found between two genes and transposon insertion could therefore affect expression of the downstream gene.

Alpha-galactosides are a dominant carbon source in many plant-derived foods. However, most dairy lactococci strains are unable to use α-galactosides as a growth substrate, while many of these strains are known to have beneficial industrial traits. Moreover, raffinose and related oligosaccharides are known to cause problems of flatulence, and their removal by fermentation with lactic acid bacteria has been suggested to improve the acceptability of legume-based products (e.g., soy milk-based products [19]). We therefore studied how the transfer of Tn6098 carrying the α-galactosidase metabolism genes would affect the ability of the transconjugant to grow in soy milk, which is rich in the α-galactosides raffinose and stachyose. We showed that the transconjugant L. lactis NIZO3921 is able to grow on the stachyose and raffinose present in soy milk, removing most of the stachyose and raffinose during this fermentation process. In addition, we determined the volatile flavor compounds formed during the fermentation of soy milk by strains KF147, MG1363 (with addition of glucose), and NIZO3921. Hierarchical bi-clustering of the samples versus metabolite profiles showed that the flavor-forming capacity of the recipient dairy strain MG1363 was largely maintained in the transconjugant. The donor strain L. lactis KF147 consumed sucrose and almost no stachyose or raffinose, and its metabolite profile was clearly different from the transconjugant metabolite profile. This demonstrates how (induced) conjugation can be used to exploit the beneficial industrial traits of industrial dairy lactococci in fermentation of plant-derived substrates, such as in soy milk. Moreover, the transconjugant could be used in soy milk fermentations to remove most of the raffinose and stachyose, which is expected to reduce the flatulence problems associated with the soy milk products.