Correspondence: Maria Grazia Fortina, Dipartimento di Scienze e Tecnologie Alimentari e Microbiologiche, Sezione di Microbiologia Industriale, Università degli Studi di Milano, via Celoria 2, 20133 Milan, Italy. Tel: +39 02 50319131; fax: +39 02 50319191; e-mail: email@example.com
The diversity of a collection of 49 Lactococcus garvieae strains, including isolates of dairy, fish, meat, vegetable and cereal origin, was explored using a molecular polyphasic approach comprising PCR-ribotyping, REP and RAPD-PCR analyses and a multilocus restriction typing (MLRT) carried out on six partial genes (atpA, tuf, dltA, als, gapC, and galP). This approach allowed high-resolution cluster analysis in which two major groups were distinguishable: one group included dairy isolates, the other group meat isolates. Unexpectedly, of the 12 strains coming from fish, four grouped with dairy isolates, whereas the others with meat isolates. Likewise, strains isolated from vegetables allocated between the two main groups. These findings revealed high variability within the species at both gene and genome levels. The observed genetic heterogeneity among L. garvieae strains was not entirely coherent with the ecological niche of origin of the strains, but rather supports the idea of an early separation of L. garvieae population into two independent genomic lineages.
In the last two decades, foodborne diseases have been emerging as an important and growing public health concern. Among the causes, changes in agricultural and zootechnical practices, increase in international trade, changes in consumer lifestyles, and increasing number of old and/or immune-compromised consumers have been mentioned. Furthermore, new pathogens or new biovars of known bacterial species are frequently being reported (Mor-Mur & Yuste, 2010). The impact of most new pathogens on specific ecosystems and their pathogenicity are not known. Indeed, the traditional food inspection systems are insufficient, because knowledge of the emerging pathogens is incomplete. Furthermore, the new techniques based on DNA analysis are not always applicable, in the absence of genetic data on these new biotypes. Therefore, to determine the concept of healthy food, it is crucially important that we expend efforts to comprehensive study of new emerging pathogens present in food products.
Lactococcus garvieae is a pathogen that causes septicemia in fish and serious damage to fish aquaculture worldwide (Vendrell et al., 2006). However, the host range of L. garvieae is not limited to aquatic species. The pathogen has been found in domestic animals, in cows with mastitis and in various artisanal cheeses made with goat and cow raw milk, sometimes as a major component (Fortina et al., 2003; Foschino et al., 2006; Fernández et al., 2010). In addition, clinical cases associated with L. garvieae infection have been reported in humans (Li et al., 2008).
Despite the growing importance of L. garvieae in both human and veterinary medicine, little research data are available on this pathogen in food matrices other than fish products. The literature is mostly about epidemiological studies on fisheries and, in summary, as regards L. garvieae stressed two serotypes based on the presence of a capsule, which plays an important role in pathogenicity, and on a potential ability to produce intra- and extracellular toxins (Vendrell et al., 2006). High biodiversity also occurred depending on the geographical origin of the pathogen (Vela et al., 2000; Schmidtke & Carson, 2003; Eyngor et al., 2004).
Over the last few years, we collected a significant number of L. garvieae strains from different artisanal Italian raw milk cheeses (Fortina et al., 2003), which we compared with those isolated from fish (Fortina et al., 2007, 2009). The results emphasized a genetic difference among strains from the two ecological niches, particularly the presence in all dairy strains of the phospho-beta galactosidase gene (lacG), which was lacking in fish isolates. Recently, L. garvieae has been isolated from human, ruminant, and water sources (Aguado-Urda et al., 2010); in these strains, lacG seemed heterogeneously scattered. Lactococcus garvieae has also been isolated from different types of food, such as vegetables (Kawanishi et al., 2007) and meat (Santos et al., 2005) but only poorly characterized.
In the present work, we monitored the population structure of L. garvieae strains from dairy products and fish included in original works (Fortina et al., 2007, 2009), in addition to newly isolated strains collected from different food matrices, namely meat, vegetables, and cereals. DNA fingerprinting analyses consisting of random amplification of polymorphic DNA (RAPD), (GTG)5-PCR, and BOXA1R-PCR, ribotyping, and a multilocus restriction typing (MLRT) were performed.
Materials and methods
Sampling, selection, and identification of L. garvieae isolates
A total of 40 food samples, purchased from different supermarkets or collected from different mills of Northern Italy, were analyzed for the presence of L. garvieae. The products consisted of raw meat (beef, poultry, and turkey), processed meat products (salami and sausages), several vegetables, and cereals (wheat flour, wheat bran, soybean, and barley; Table 1). All samples were aseptically collected and transported in isothermal boxes to the laboratory. For L. garvieae isolation, food samples (25 g) were enriched in 1 : 9 (w/w) M17 broth (Difco, Detroit, MI) supplemented with 1 g L−1 glucose (M17-G) at 37 °C for 24 h. After enrichment, total DNA was extracted as reported below and the presence of L. garvieae was established through a species-specific PCR assay, as reported by Zlotkin et al. (1998). For each sample positive to the species-specific amplification, L. garvieae selection was attempted on M17-G agar. Appropriate dilutions in 0.1% peptone solution of positive-enriched cultures were plated and incubated at 37 °C for 24 h; after incubation, randomly selected colonies were purified and then submitted to taxonomic identification, as reported previously. Strains were maintained in M17-G broth; serial transfer was minimized to prevent the occurrence of mutations as a result of adaptation to laboratory medium and conditions. Stock cultures were maintained at −80 °C in M17-G with 15% glycerol.
Table 1. Prevalence of Lactococcus garvieae in different food products and strains analyzed in this study
strains previously isolated
strains kindly provided by Dr Prearo (Experimental Institute for Zooprophylaxis, Torino, Italy) and by Dr. Amedeo Manfrin (Venetian Experimental Institute for Zooprophylaxis, National Reference Lab. for Fish Diseases, Legnaro, Italy).
For strains grown in pure culture, DNA was extracted as previously described by Fortina et al. (2003). For the extraction of DNA from food samples, the Ultraclean™ Microbial DNA Isolation Kit (Mo Bio Laboratories Inc., Carlsbad, CA) was used according to the manufacturer's instructions. The concentration and purity of the DNAs were determined using a UV-Vis spectrophotometer (SmartSpec™ Plus, Bio-Rad, Milan, Italy).
Gene targets and PCR amplification
Internal fragments of seven loci, atpA (α-subunit of ATP synthase), tuf (bacterial elongation factor EF-Tu), dltA (D-alanine-D-alanyl carrier protein ligase), als (α-acetolactate synthase), gapC (glyceraldehyde-3-phosphate dehydrogenase), galP (galactose permease), lacG (phospho-β-galactosidase) were amplified using primers and conditions previously described or developed in this study on the basis of the available nucleotide sequences reported in GenBank databases. The specific primers and conditions used and their amplification products are reported in Table 2, with relevant references. PCRs were performed in a 25 μL reaction mixture contained 100 ng of bacterial DNA, 2.5 μL of 10× reaction buffer (Fermentas, Vilnius, Lithuania), 200 μM of each dNTP, 2.5 mM MgCl2, 0.5 μM of each primer, and 0.5 U of Taq polymerase (Fermentas). After incubation for 2 min at 94 °C, samples were subjected to 35 cycles of 60 s at the annealing temperature (Table 2), followed by 1 min at 72 °C; the reaction was completed by 7 min at 72 °C and kept at 4 °C using a PCR-Mastercycler 96 (Eppendorf, Hamburg, Germany). Amplification products were separated on a 1.5% agarose gel stained with ethidium bromide in 1× TAE (40 mM Tris-acetate, 1 mM EDTA, pH 8.2) buffer and photographed.
Table 2. PCR primers and conditions used for the detection of genes in Lactococcus garvieae strains
Primer pair (5′-3′)
Annealing temperature (°C)
Nucleotide accession number
atpA (α-subunit of ATP synthase; Naser et al., 2005)
Products from each amplified locus were tested to select a suitable discriminating restriction enzyme, that is, a panel of two or five enzymes that cut frequently along each of the amplified fragments was examined to clearly identify allelic variations. Overnight restriction digestion was carried out at 37 °C in a 20 μL reaction mixture containing 4 μL of the PCR product, 2 μL of 10× incubation buffer and 10 U of each enzyme (Amersham Pharmacia Biotech., Milan, Italy). Restriction digests were subsequently analyzed by agarose electrophoresis (2% agarose gel).
Molecular fingerprinting of L. garvieae strains
Lactococcus garvieae strains were typed by combined analysis of repetitive element (REP) typing using primers (GTG)5 (5′-GTGGTGGTGGTGGTG-3′) and BOXA1R (5′-CTACGGCAAGGCGACGCTGACG-3′; Versalovic et al., 1994; De Urraza et al., 2000) and random amplification of polymorphic DNA-PCR (RAPD) typing with primer M13 (5′-GAGGGTGGCGGTTCT-3′; Rossetti & Giraffa, 2005). An annealing temperature of 42, 48, 38 °C for (GTG)5, BOXA1R and M13, respectively, and an amplification protocol of 35 cycles were used. The PCR products were analyzed by electrophoresis and photographed as reported earlier. The digitized image was analyzed and processed using the Gel Compar software (Applied Maths, Kortrijk, Belgium). The value for the reproducibility of the assay, evaluated by the analysis of repeated DNA extracts of representative strains was >93%.
The DNA of L. garvieae strains (10 μg) was digested by incubation with 30 U of PstI endonuclease (Fermentas) according to manufacturer's instruction. A 20 μL aliquot of the digestion mixture was combined with 5 μL of loading buffer and the preparation was electrophoresed on 0.8% (w/v) agarose gel at 100 V for 2 h. DNA fragments were subsequently transferred to a nylon membrane (Roche Diagnostics GmbH, Mannheim, Germany) by Southern blot. Hybridization was performed at 60 °C using the 16S rRNA gene of L. garvieae DSM 20684T. The probe was amplified using the universal primers: 16SF, 5′-AGAGTTTGATCCTGGCTCAG-3′ and 16SR, 5′-CTACGGCTACCTTGTTACGA-3′. PCR cycle was 2 min at 94 °C, then five cycles of 45 s at 94 °C, 45 s at 50 °C, 1 min at 72 °C, followed by 30 cycles of 45 s at 94 °C, 45 s at 55 °C, 1 min at 72 °C, with a 7 min final extension at 72 °C. The DIG DNA Labeling and Detection kit (Roche) was used for digoxigenin labeling of the 1513 bp fragment. Prehybridization and hybridization overnight were performed in 50% (w/v) formamide at 42 °C and stringency washes in 0.1× SSC buffer at 65 °C (10× SSC is 1.5 M NaCl, 150 mM sodium citrate). The probe was detected by chemiluminescent detection using CSPD (Roche), and the signals were visualized by exposure to X-ray film for 2 h.
Banding pattern similarity was evaluated by construction of dendrograms using the NTSYSpc software, version 2.11 (Applied Biostatics Inc., NY), employing the Jaccard similarity coefficient. A dendrogram was deduced from a similarity matrix using the unweighted pair group method with arithmetic average (UPGMA) clustering algorithm. The faithfulness of the cluster analysis was estimated by calculating the cophenetic correlation value for each dendrogram.
Results and discussion
To contribute to the characterization of the natural variability of the species L. garvieae, we evaluated the genetic diversity of a collection of strains isolated from different sources. L. garvieae is mainly known for its presence in aquatic environments and as component of milk and many artisanal cheeses. In this work, we studied new isolates from other sources to give a comprehensive indication of the diversity found within the species. We focused our attention on food matrices not yet or poorly investigated for the presence of L. garvieae, particularly, meat, vegetables, and cereals. Of 40 food samples tested, 20 (50%) were found to contain L. garvieae (Table 1). Raw meat and meat products showed the highest prevalence of contamination with L. garvieae: All samples analyzed were positive for the presence of this bacterial species. A high rate of L. garvieae was also found in vegetables (31%), while only one cereals sample showed the presence of this species. From these sources, we selected 24 new ecotypes that were studied in comparison with previously isolated dairy and fish ecotypes (Table 1). All new isolates were properly identified by specific PCR, giving the expected amplification product of 1100 bp belonging to the 16S rRNA gene (Zlotkin et al., 1998).
First of all, the strains were screened for the presence of the lac operon. In previous studies (Fortina et al., 2007, 2009) carried out on dairy and fish isolates, we observed that only the isolates of dairy origin were able to utilize lactose, because they harbored a lac operon, which shares a high sequence homology to that found in Lactococcus lactis. As a conclusion, we hypothesized a gene gain by lateral gene transfer, which provided dairy L. garvieae strains of a key physiological property contributing to adaptation to milk/dairy niche. When lacG was tested on new isolates, we found that the ability to metabolize lactose was not exclusively related to dairy isolates, but was heterogeneously scattered among L. garvieae meat isolates. Indeed, three meat isolates (strains Smp2, Smp3, and Smp4) were positive for the presence of the lacG gene. The remaining strains from meat and the isolates from vegetables and cereals did not show any amplification signal. These results indicate that lac operon cannot be considered a suitable genetic marker for associating strains to their niche of isolation.
The molecular fingerprinting of L. garvieae strains was determined using RAPD and REP-PCR with BOXA1R and (GTG)5 primers. These methods, which use short arbitrary primers or primers targeting short repetitive sequences interspersed throughout the genome are an established approach for delineation of bacteria at the species and strain-level (Randazzo et al., 2009; Švec et al., 2010). The discriminatory power of these primer sets was similar, with 20 different profiles obtained by BOXA1R and (GTG)5 and 23 different profiles obtained by M13 for a collection of 49 strains. Although isolated at different times, some strains had identical fingerprints with all tested primers; on the contrary, most of the strains grouped at low similarity values. Independently from the primer used, the 49 strains grouped in two distinct clusters, which we named AT and BT (Fig. 1): one cluster (AT) contained all meat isolates (with the exception of BOXA1R experiment where the meat isolate Sa113 showed a unique fingerprint at a very low similarity value), whereas the other cluster (BT) included all dairy isolates. Unexpectedly, four of 12 strains isolated from fish (V32, V63, Lg23, and V79), always grouped with dairy isolates, whereas the others grouped with meat isolates. Likewise, strains isolated from vegetables allocated between the two main groups. The cluster analysis resulting from the combined profiles of the three primer sets employed, confirmed the existence of two major divisions, which were separated at a level of similarity of 0.13 (Fig. 1), and did not coincide with the ecological niche of isolation. In particular, the low correlation value between the two clusters suggested the existence of a marked genetic divergence.
When we tested several genes belonging to the core genome of L. garvieae, we observed again that all bacterial isolates can be shared out between two clusters, which are correlated to a low similarity level. Specifically, on the basis of conserved regions identified by sequence comparison of several housekeeping or functional genes in L. garvieae, we selected suitable primers to employ for PCR amplification (Table 2). The expected fragment length of the α-subunit of ATP synthase, elongation factor EF-Tu, D-alanine-D-alanyl carrier protein ligase, α-acetolactate synthase, glyceraldehyde-3-phosphate dehydrogenase, and galactose permease amplicons was observed for all the 49 strains studied. Restriction analysis of each of the loci tested produced one to seven different patterns consisting of one to seven bands, depending on locus, restriction enzyme and strain examined (Table 3). The cluster analysis resulting from the combined restriction profiles of the six amplicons reported in Fig. 2, revealed two distinct L. garvieae clusters at similarity level of approximately 0.12. Notably, the groups obtained were highly similar to PCR-fingerprinting clusters (AT and BT). In fact, also in this case, all meat isolates grouped together with the two salad and eight fish isolates (cluster AR), whereas dairy isolates grouped with cereal isolate and the remaining vegetables and fish isolates (cluster BR). Within the various subclusters, further discrimination reflected the polymorphism revealed by restriction analysis of the tested loci (Table 3). GapC gene resulted the most conserved among the tested strains; in fact, restriction analysis of the amplified fragment with different restriction enzymes did not reveal sequence variations among the strains, with the exception of isolate V79 from fish, which differentiated from the other strains when HaeIII was employed (Identification profile in Table 3 = Ip 24). Restriction analysis of the galP amplicon grouped the strains into two main clusters, within which the distribution of strains was always the same, even using different enzymes. One cluster (named ‘meat-group’, Ip 1, 4, 9, and 12) contained all meat isolates (with the exception of Sa113), two salad isolates and eight of the 12 fish isolates; the second cluster (named ‘dairy-group’, Ip 3, 5, 8, and 10) included all dairy isolates and the remaining vegetables and fish isolates. The isolate from wheat flour always grouped with strains of dairy origin. Strain Sa113 from meat products showed a unique restriction profile (Ip 2, 6, 7, and 11). Restriction analysis of the atpA gene with RsaI delineated the same two clusters obtained when galP gene was tested; in this case, Sa113 grouped whit meat isolates (Ip 16). Also using HpaII, the differentiation among strains was respected (meat-group, Ip14 – dairy-group, Ip 15) with an additional discrimination for four meat isolates (Smp1-2-3-4, Ip 13). The digestion of tuf gene with RsaI grouped two meat isolates (Po1 and Tac2) with dairy-group (Ip 19), while the use of HhaI permitted the separation of Sa113 (Ip 20) and the differentiation of dairy isolates and Po1 and Tac2 (Ip 22) from the remaining meat, fish and vegetable isolates (Ip 21). Restriction analysis of the dltA and als genes revealed further polymorphisms, and the possibility to discriminate the two salad (Ip 28) and the cereal isolates (Ip 42) from the other strains and to highlight two sub-groups within dairy isolates (Ip 32, 33).
Table 3. Restriction profiles obtained for Lactococcus garvieae strains after digestion of PCR-amplified regions of six genes analyzed
Amplicon size (bp)
Identification profile = Ip
590, 340, 90, 50
930, 90, 50
420, 400, 250
420, 300, 250, 100
380, 350, 190, 100, 50
550, 370, 150
370, 360, 340
880, 180, 120
880, 120, 120, 60
400, 200, 170, 150, 100, 100, 60
500, 200, 70, 150, 100, 60
480, 210, 180, 130
260, 220, 210, 180, 130
800, 160, 120
560, 290, 120
900, 400, 170
700, 550, 220
700, 400, 220, 150
780, 420, 270
500, 250, 220, 120
500, 220, 200, 70, 60, 50
500, 220, 200, 120, 50
580, 280, 220
680, 300, 100
460, 400, 160, 50
580, 300, 80, 70, 50
460, 300, 220, 100
720, 300, 50
570, 30, 200
PCR-ribotyping generated by digestion of total DNA with PstI, revealed the presence of nine different electrophoretic profiles, characterized by two to five bands of molecular weight varying from 4000 to more than 10 000 bp (Fig. 3). The data obtained indicate an important heterogeneity both in the copy number and in the distribution of the ribosomal operons along the chromosomal DNA, as evidenced in the corresponding dendrogram (Fig. 3). Two main groups were distinguished, at a low similarity level (0.36). The distribution of the tested strains within the main groups differed in part from that previously observed. In this case, dairy isolates, cereal, and six vegetable isolates (which always grouped together in the other experiments) and only one fish isolate (V79) showed the same ribotype, separated at similarity level of 0.80 with Sa113 from meat products and at minor similarity level with other two meat isolates. The remaining meat isolates grouped in different subgroups, all within group 2, which also included the remaining fish and salad isolates.
In conclusion, our results support the idea of an early separation of L. garvieae population into two independent genomic lineages. Subsequently, the environmental stimuli of a specific niche could have exerted a selective pressure favoring the emergence of several independent genotypes. It appears plausible that genomic flux within the dispensable genome, recombination events between genetically distinct strains during mixed colonization and/or gene (in)activation could have governed the bacterial adaptation to different habitats. Recently, we carried out the complete genome sequencing of one strain of dairy origin and one strain isolated from fish, belonging to ‘meat-group’ (Ricci et al., 2012). Whole-genome comparison between these and other L. garvieae available complete genomes, together with multilocus sequence typing (MLST) experiments are in progress in our laboratory for a deeper understanding of the evolutionary history and the global complexity of this bacterial species.
This work was supported by ‘Post genomica batterica per la qualità e la sicurezza degli alimenti’ project from the Lombardy region (Italy). We thank Dr S. Guglielmetti for a critical reading of the manuscript and for his useful suggestions.