Genomic Diversity in Staphylococcus xylosus^▿

Pulsed-field gel electrophoresis.

Genomic DNA of S. xylosus strains was prepared in agarose plugs as described previously (21). Plugs were restricted with the endonuclease I-CeuI (New England Biolabs, United Kingdom) or SmaI enzyme (Promega, Charbonnière, France) according to the manufacturers' instructions. Fragments were separated by PFGE in 1% agarose gels in 0.5× Tris-borate-EDTA buffer on a CHEF-DR III apparatus (Bio-Rad, Ivry, France). Electrophoretic conditions were 50 s to 100 s for 6 h and 10 s to 30 s for 18 h or 50 s to 90 s for 22 h and then 1 s to 12 s for 13 h at 14°C at a constant voltage of 6 V·cm⁻¹ and an angle of 120° for the determination of specific subtracted fragment location. Lambda DNA concatemers were used as molecular size markers (Promega). After migration, gels were stained in ethidium bromide, digitized under UV light (Gel Doc 2000; Bio-Rad), and analyzed using the Quantity One quantitation software (Bio-Rad). Similarities between the PFGE profiles, based on band position, were derived by the Dice correlation coefficient with a maximum position tolerance of 2%. A hierarchical unweighted pair group method analysis was used to generate a dendrogram.

Chromosomal DNA and plasmid preparation.

Chromosomal DNA from S. xylosus was isolated by the method of Marmur (18). To isolate plasmid DNA from S. xylosus, cells were lysed with 16 μg/ml lysostaphin, and the QIAprep Miniprep kit (Qiagen, Courtaboeuf, France) was subsequently applied.

Suppressive-subtractive hybridizations and differential screening.

Three subtracted DNA libraries were generated following the recommendations for the Clontech PCR-Select bacterial genome subtraction kit (BD Biosciences Clontech, Erembodegem, Belgium). Genomic DNAs of S. xylosus strains S04002, S04009, and 00-1747 were used as the testers and that of S. xylosus C2a was used as a driver. To increase the proportion of chromosomal subtracted DNA, the endogenous plasmids of the tester strains were also used as drivers. Genomic and plasmid DNAs were digested by the appropriate RsaI enzyme. Suppression-subtractive hybridizations were carried out as recommended by the supplier except that hybridization temperatures were set at 61°C instead 63°C due to the low G+C content of Staphylococcus DNA. The pool of subtracted fragments was amplified by PCR using the Advantage cDNA polymerase (BD Biosciences Clontech). PCR products generated in each library were cloned into the pGEM-T Easy vector (Promega) according to the supplier's recommendations and transformed into Escherichia coli TG1 competent cells, which were then cultured on Luria-Bertani agar plates containing 100 μg/ml ampicillin, 50 μg/ml 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside, and 0.5 mM isopropyl thiogalactoside. Resulting transformants were individually grown in Luria-Bertani medium containing 100 μg/ml ampicillin in 96-well microtiter plates and stored in 20% glycerol at −70°C. Subtracted library clones were randomly selected as templates for PCR amplification using nested primers 1 and 2R, provided in the genome subtraction kit. PCR products were analyzed by agarose gel electrophoresis to verify the size of the inserts and the presence of one distinct amplified product. PCR products of each subtracted library were screened for tester-specific fragments using dot blot hybridization. They were arrayed in duplicate onto positively charged nylon membranes (Hybond N⁺; Amersham Biosciences, Orsay, France). One μg of RsaI-restricted genomic DNA from the driver and each tester DNA was randomly labeled with the Dig-High Prime system (Roche Applied Science, Neuilly sur Seine, France) and then used as probes to separately hybridize the blots. The digoxigenin-labeled probes were detected by the Dig color detection kit (Roche).

The distribution of tester-specific fragments was also evaluated among S. xylosus strains. Genomic DNAs from 20 strains (Table ​(Table1)1) were used as probes, as described above. The distribution of tester-specific fragments was analyzed by a complete linkage hierarchical cluster analysis using Microsoft Excel software with the XLStat add-in. A distance matrix was calculated by using Sokal and Sneath's (1) similarity coefficient to take into account positive (occurrence) and negative (nonoccurrence) matches.

DNA sequencing and analysis.

Plasmid DNA containing tester-specific inserts was isolated from individual clones using the NucleoSpin Plasmid QuickPure kit (Macherey-Nagel, Hoerdt, France). Sequencing reactions were performed using the BigDye terminator as recommended by the supplier (Applied Biosystems, Courtaboeuf, France) and analyzed by using an ABI 310 automated DNA sequencer. Bioinformatic analyses were performed from Web-based servers. The sequences were analyzed by BLASTX and/or BLASTN searches against the NCBI GenBank database. Based on the analysis, redundant sequences were identified and sequence assembly was done with the CAP3 program (http://bioinformatics.iastate.edu/aat/sas.html). ORF Finder and BLASTX analyses were performed on contigs. When no similarity was found, a PSI-BLAST analysis was performed.

Determination of chromosomal locations of subtracted sequences.

For Southern blotting, digested chromosomal DNAs of S. xylosus S04002, S04009, and 00-1747 strains separated by PFGE were passively transferred onto positively charged nylon membranes (Hybond-N⁺; Amersham Biosciences). Equal amounts of PCR amplicons of each tester-specific insert were pooled and randomly labeled using the Dig-High Prime system (Roche Applied Science). After hybridizations, detection was performed using the Dig chemiluminescence detection kit following the supplier's instructions (Roche Applied Science). To confirm the localization of tester-specific inserts, chromosomal DNA restriction fragments of interest were excised from the PFGE gel, rinsed three times for 30 min with 10 mM Tris-HCl, pH 8.5, and digested with RsaI (Promega) according to the supplier's recommendations. DNA fragments were thus extracted from agarose and purified before labeling with the AlkPhos direct labeling kit (Amersham Biosciences) and hybridized to the dot blot containing the subtracted unique sequence as described for the differential screening.

Suppressive-subtractive hybridizations of S. xylosus C2a from S04002, S04009, or 00-1747.

Following SSH, three DNA subtraction libraries were created containing 192, 384, and 320 clones of putative specific DNAs of S. xylosus S04002, S04009, and 00-1747, respectively. The number of clones corresponded to the proportion of chromosome size differences of the three tester strains evaluated in our previous study (5). After PCR amplification of the cloned inserts, clones giving no amplicon or multiple amplicons were discarded. To check the specificity of cloned inserts of the three libraries, PCR products of the inserts were arrayed and hybridized with the genomic DNA probes of the three respective tester strains or the driver C2a strain. From S. xylosus S04002, S04009, and 00-1747 DNA subtraction libraries, a total of 75, 105, and 92 DNA fragments were found to be specific to the tester strains, respectively. All these DNA fragments were two-end sequenced, and 40, 43, and 39 different sequences were identified, respectively (Table ​(Table2).2). A total of 29.67, 22.78, and 25.42 kb were extracted from the S. xylosus S04002, S04009, and 00-1747 genomes, respectively. The sizes of these specific DNA fragments varied from 125 to 1,715 bp, with an average size of 700 bp. Their G+C contents were 30.3% for those of the S04002 subtracted library, 32.8% for the S04009 subtracted library, and 31.1% for the 00-1747 subtracted library. These GC percentages were at the lower limit of S. xylosus G+C content, estimated to be between 30 and 36% (31). SSH has thus been used successfully for the extraction of the differential genetic content of S. xylosus strains arising from various niches.

Identification of specific sequences of S. xylosus S04002, S04009, and 00-1747 and a large proportion of specific metabolism sequences.

The 122 tester-specific fragments were analyzed by BLASTX and PSI-BLAST searches and led to the identification of 149 open reading frames (ORFs), as shown in Table ​Table2.2. These ORFs were assigned to one of seven classes: metabolism, regulatory, cell membrane/surface structure, resistance, transport, phage-related structure, and IS elements and unknown functions. Fourteen ORFs resembled hypothetical bacterial genes of unknown function, while two fragments had no significant matches with known sequences. The other ORFs had significant similarity to sequences with known functions largely described in the Staphylococcus genus. However, for the S. xylosus 00-1747 strain, 20 of its 47 detected ORFs corresponded to genes from other bacterial genera (Table ​(Table2).2). A majority of ORFs seemed to be related to specific metabolisms as biotin, myo-inositol, and sorbitol, to amino acid and fermentative metabolisms, and to pentose and glucuronate interconversions (Table ​(Table22).

In the S. xylosus S04009 strain genome, fragments corresponding to five genes involved in myo-inositol metabolism as well as the Na⁺:myo-inositol cotransporter were subtracted (Table ​(Table2).2). Myo-inositol is a hexitol which is a constituent of virtually every living cell; it is synthesized by plants and is essential for humans and animals. Different bacterial species (Klebsiella spp., Salmonella spp., Pseudomonas spp., Bacillus subtilis, etc.) are able to use myo-inositol as carbohydrate source, but genes implicated in this catabolism have been described in few species (16). In Bacillus subtilis, two iol operons were described, including an operon of 10 genes, iolABCDEFGHIJ, and an operon of two genes, iolRS (40). Until now, these genes had never been described in a Staphylococcus genus.

Five ORFs of the S. xylosus 00-1747 strain were related to lysine and glycine metabolism genes, none of which corresponded to staphylococcal genes (Table ​(Table2).2). Sarcosine oxidase and betaine aldehyde dehydrogenase were shown to be implicated in metabolism of glycine betaine, an effective osmoprotectant (1, 4). Only betaine aldehyde dehydrogenase and dihydropicolinate synthase were identified in actual staphylococcal sequenced genomes, but these genes showed no significant match at the DNA level with these ORFs. Sarcosine oxidase and saccharopine dehydrogenase have never been observed in staphylococcal genomes. The fact that the S. xylosus 00-1747 strain contained numerous nonstaphylococcal homology ORFs might be explained by its origin. This strain was isolated from mouse dermatitis in Korea. The particular origin of this strain suggests that ecological and/or geographical divergences in S. xylosus species may exist.

From the S. xylosus meat starter S04002 subtraction, fragments corresponding to fermentative metabolism were found: an acetolactate synthase which transforms pyruvate in acetolactate, its regulator AlsR, and an acetoin dehydrogenase which transforms acetoin into diacetyl, as well as a lactate permease also implicated in anaerobic metabolism (Table ​(Table2).2). Acetoin and diacetyl are aroma compounds. Their production in fermented meat products has been characterized with a dairy product odor (20). Some S. xylosus strains produce these aroma compounds more than other staphylococcus species (34). This property is variable between S. xylosus strains, and a correlation between strain persistence in ripened meat and the capacity of strains to produce acetoin has been shown (9). In lactic acid bacteria isolated from fermented sausages, the production of acetoin is stimulated by a low pH and a low sugar concentration, conditions that could be found at the end of the ripening process (9). Recently, Fuchs et al. (8) showed that in S. aureus the production of acetoin and lactate was stimulated by anaerobic growth conditions.

The large number of differential genes predicted to be involved in metabolism strongly suggested a wide adaptation of S. xylosus strains to their environment and to the availability of the different substrates.

Horizontal transfer in S. xylosus S04009.

The S. xylosus S04009 strain isolated from cow mastitis showed 12 specific subtracted fragments corresponding to phage-related sequences (Table ​(Table2).2). The presence of at least three different bacteriophages, 77, 2638A, and EW, was deduced, and an IS605/IS200 transposase was also detected. Three phage-related proteins were found in S. xylosus S04002 and one in 00-1747 subtracted fragments. Phages in S. aureus are well studied: 37 phage genomes are reported in the GenBank database (3, 17). To our knowledge, there are few studies of phages from coagulase-negative staphylococci, and only two phages from Staphylococcus epidermidis have been completely sequenced and molecularly characterized (3). Other putative evidence of horizontal transfer in S. xylosus S04009 is the presence of specific sequences corresponding to antibiotic resistance as methicillin, fosfomycin, and macrolide resistance proteins (Table ​(Table22).

Particular S. xylosus surface structures.

Specific surface structures were found in the S. xylosus tester strains. The meat starter strain S. xylosus S04002 possessed a specific fragment corresponding to a 67-kDa myosin-cross-reactive streptococcal antigen-like protein. This protein family is thought to have structural features in common with the beta chain of the class II antigens. In Streptococcus pyogenes it reacts with antimyosin antibodies and plays a role in the pathogenesis of streptococcal infections (13). This gene is found in other Staphylococcus sequenced genomes, such as S. aureus, S. epidermidis, or Staphylococcus haemolyticus, but its function in those bacteria is still unknown. The S. xylosus 00-1747 strain implicated in mouse dermatitis contained a specific fragment resembling a Toll-interleukin receptor (TIR). Such a eukaryotic signal-mimicking molecule has been described in Salmonella enterica serovar Enteriditis (24). It was demonstrated that it is an important virulence factor in vivo by modulating host defense mechanisms involved in regulation of NF-κB and caspase activation. Experiments in mice infected with a Salmonella strain deficient in TIR showed reduced lethality. Predicted TIR-like domains are widespread and show an unusual distribution in bacterial species. A lateral transfer of this gene has been supposed (24).

Specific fragments shared by S. xylosus S04002, S04009, and 00-1747.

The presence of common fragments in the three S. xylosus tester strain chromosomes was detected by dot blot hybridization using genomic DNA of the three strains as a probe. The three strains shared only six specific fragments. Those six common fragments corresponded to arsenic resistance genes with the arsenate reductase and the arsenical efflux pump membrane protein, a riboflavin kinase, a hypothetical phage infection protein, and two fragments which were similar to an autolysin, a mannosyl-glycoprotein endo-β-N-acetylglucosamidase (Table ​(Table22).

Among the 122 specific fragments, 66 hybridized with S. xylosus S04009 genomic DNA. This mastitis isolate contained the highest number of specific fragments in its genome.

The S. xylosus strains S04009 and 00-1747 isolated from mastitis and dermatitis, respectively, shared the highest number of specific fragments, with 29 in common. They had specific sequences similar to a hexitol-specific PTS system, probably sorbitol specific. Indeed, other fragments related to sorbitol assimilation were isolated, such as a sorbitol dehydrogenase and the putative regulator of the sorbitol operon. Those two commensal strains also shared numerous specific DNA sequences implicated in resistance to antibiotics and other toxic components, such as tellurium, but also antibiotic synthesis-related sequences, such as a monooxygenase and a lipopeptide, micrococcin P1. These two strains also had in common the feoB gene, implicated in ferrous ion transport. This gene was identified in S. aureus and S. epidermidis genomes but had not been characterized (10). The ability to acquire iron (Fe) in a host is known to be an essential attribute of most pathogenic bacteria (2).

The meat starter strain S04002 and the strain S04009 isolated from mastitis shared 20 fragments. They have in common the iolABCDE myo-inositol genes but not the iolG gene, which was present in S04002 but absent in S04009. IolG corresponds to a myo-inositol 2-dehydrogenase, which is the first enzyme of myo-inositol catabolism, degrading the incoming myo-inositol into 2-keto-myo-inositol. The iolFHIJ and iolRS genes described in B. subtilis were not detected in the subtracted libraries (40). The gene iolR corresponds to a negative regulator of the iol operon, iolF is related to a myo-inositol transporter, and iolE codes for a 2-keto-myo-inositol dehydratase, while the functions of the other iol genes are still unclear in B. subtilis (40).

The meat starter S. xylosus S04002 had only nine fragments in common with the mouse dermatitis strain 00-1747. The biotin synthesis-related sequences obtained after genomic subtraction of these two strains seemed to be drastically different at the DNA level, and none of these fragments hybridized with the two genomic DNAs.

Conservation of tester-specific fragments within S. xylosus strains.

The extent of polymorphism among S. xylosus strains for the specific fragments was assessed. The RsaI-digested genomic DNAs of 20 different S. xylosus strains, 10 used as starters in the food industry and 10 isolated from cow or goat mastitis (Table ​(Table1),1), were hybridized with the 122 specific fragment sequences of the three tester strains. Four specific fragments (D39, D129, D137, and D158) from the S. xylosus 00-1747 library and four (Mb33, Mb66, M85, and M130) from the S. xylosus S04009 library were absent in the genomic DNA of the 20 S. xylosus strains. No specific fragment was present in all the S. xylosus genomes. The subtracted fragment most commonly found, which hybridized with 21 of the 23 S. xylosus strains (including the three tester strains), was fragment D294, corresponding to a hypothetical membrane protein and a CHAP domain corresponding to an amidase domain frequently found in autolysin precursor. A hierarchic ascendant classification gathered the strains into two clusters named I and II (Fig. ​(Fig.22).

Open in a separate window

FIG. 2.

Distribution of tester-specific fragments within S. xylosus strains.

Cluster I comprised five strains, all isolated from animal-associated opportunistic infections, sharing eight fragments: Mb26, Mb48, Mb147, M139, M149, D63, D96, and D213 (Table ​(Table2).2). These fragments corresponded to metabolic genes as PTS system specific to hexitol and its associated antiterminator Bgl, a sorbitol dehydrogenase, an acetyltransferase of unknown specificity, and a riboflavin kinase. Three of these fragments (Mb147, M139, and M149) corresponding to a hexitol assimilation system were absent in cluster II except one strain. The cluster could be divided into two subgroups named Ia and Ib. The Ia subgroup included the tester strain S. xylosus S04009 and the strain {"type":"entrez-protein","attrs":{"text":"S04013","term_id":"83885","term_text":"pir||S04013"}}S04013, both isolated from cow mastitis. These two strains had 43 of the 122 specific fragments in common, 25 of which were from the S. xylosus S04009 library, showing a highly related genetic content between these two strains. Subgroup Ib comprised the tester strain S. xylosus 00-1747, isolated from mouse dermatitis, and two strains isolated from goat mastitis, S04019 and {"type":"entrez-protein","attrs":{"text":"S04020","term_id":"80854","term_text":"pir||S04020"}}S04020. The genomes of these three strains shared 20 fragments, 9 of which were common to the genome of the tester strain S04009 of subgroup Ia.

Cluster II comprised 18 strains that shared a few fragments with cluster I, and they shared the E154 fragment, which was absent in cluster I except for one strain. This fragment corresponded to the accessory gene regulator C (agrC), which is a staphylococcal conserved regulator. It has been observed that there is a high genetic variability of the agr locus in Staphylococcus species and particularly in the agrC sequence (26). This variability was observed not only between Staphylococcus species but also between subpopulations of a single species (6). In S. aureus, agrD and agrC polymorphism permitted definition of a group broadly correlated with genotype which was correlated with pathotype (26). The existence of at least two different genotypic clusters (I and II) within the S. xylosus species was confirmed by the argC polymorphism distribution observed.

Cluster II could be divided into two subgroups named IIa and IIb. Subgroup IIa comprised the tester strain S04002 and the S01001 and {"type":"entrez-protein","attrs":{"text":"S01008","term_id":"88248","term_text":"pir||S01008"}}S01008 strains of meat origin and six other strains isolated from mastitis (Fig. ​(Fig.2;2; Table ​Table1).1). Within this heterogeneous subgroup, only two fragments were found to be in common: D294, described previously to be the most shared fragment between the 23 S. xylosus strains, and the fragment E154, already mentioned above. Subgroup IIb comprised eight S. xylosus strains used as starters (Fig. ​(Fig.2;2; Table ​Table1)1) and one S. xylosus isolated from cow mastitis S04016. Twenty-one fragments were common between all these strains. They corresponded to the six myo-inositol metabolism genes (iolABCDEG), all the fermentative metabolism genes (als, acetoin reductase, and lactate permease genes), and the lipase gene (E179). The presence of iol genes in the majority of starter S. xylosus strains and their absence in the actual staphylococcal sequenced genomes indicate a specific and fortuitous acquisition of these genes by some S. xylosus strains. The presence of fermentative metabolism may be an adaptation to anaerobic growth conditions. In S. aureus, the overexpression of the genes implicated in butanediol fermentation and lactate production were recently observed under fermentative conditions (8). Adaptation to anaerobic conditions and lipase activity could procure a technological advantage for S. xylosus used in food processing.

We thank Young-Suk Won for giving us the strain implicated in mouse dermatitis and Pascal Rainard for the strains isolated from mastitis. We thank Isabelle Lebert for helpful discussions, Sébastien Masseglia for technical assistance, Brigitte Duclos for secretarial assistance, and David Marsh for revision of the English.

Emilie Dordet-Frisoni is the recipient of a fellowship from the French Ministry of “Education Nationale et Recherche.” This work was financially supported by the ANR project “Genoferment” ANR-05-PNRA-020.