Genotypes and Toxin Gene Profiles of Staphylococcus aureus Clinical Isolates from China

A total of 108 S. aureus isolates from 16 major hospitals located in 14 different provinces in China were characterized for the profiles of 18 staphylococcal enterotoxin (SE) genes, 3 exfoliatin genes (eta, etb and etd), and the toxic shock syndrome toxin gene (tsst) by PCR. The genomic diversity of each isolate was also evaluated by pulsed-field gel electrophoresis (PFGE), multilocus sequence typing (MLST), and accessory gene regulator (agr) typing. Of these strains, 90.7% (98/108) harbored toxin genes, in which tsst was the most prevalent toxin gene (48.1%), followed by sea (44.4%), sek (42.6%) and seq (40.7%). The see and etb genes were not found in any of the isolates tested. Because of high-frequency transfer of toxin gene-containing mobile genetic elements between S. aureus strains, a total of 47 different toxin gene combinations were detected, including a complete egc cluster in 19 isolates, co-occurrence of sea, sek and seq in 38 strains, and sec and sel together in 11 strains. Genetic typing by PFGE grouped all the strains into 25 clusters based on 80% similarity. MLST revealed 25 sequence types (ST) which were assigned into 16 clonal complexes (CCs) including 2 new singletons. Among these, 11 new and 6 known STs were first reported in the S. aureus strains from China. Overall, the genotyping results showed high genetic diversity of the strains regardless of their geographical distributions, and no strong correlation between genetic background and toxin genotypes of the strains. For genotyping S. aureus, PFGE appears to be more discriminatory than MLST. However, toxin gene typing combined with PFGE or MLST could increase the discriminatory power of genotyping S. aureus strains.


Introduction
S. aureus is one of the most common clinical and foodborne pathogens. It can cause a wide variety of infections ranging from skin and tissue infections, toxin-mediated diseases, pneumonia, and bacteremia. The infections of S. aureus can be acquired through both hospital and community settings including food poisoning. It has been reported that S. aureus colonizes the skin and mucosal surface of 20 to 40% of humans, implying a high risk of developing the infections of this organism [1]. A recent study found that 47% of meat and poultry in US grocery stores were contaminated with S. aureus, and 52% of those bacteria were resistant to antibiotics [2]. The high prevalence and rapid spread of drug-resistant S. aureus strains in food supplies have increased the risk of S. aureus infections and posed a great threat to food safety and public health.
S. aureus can produce a wide variety of virulence factors. Depending on the strain, some can produce exotoxins such as toxic shock syndrome toxin (TSST), a causative agent of toxic shock syndrome, and exfoliative toxins (ETs), responsible for staphylococcal scalded skin syndrome. Some strains express heatstable enterotoxins (SEs), a major cause of staphylococcal food poisoning [3]. To date, there are 21 identified SE and staphylococcal-like enterotoxin (SEl) genes, including sea to see, seg to sev. Among them, SEA is the most common enterotoxin found in food and is frequently associated with staphylococcal food-poisoning outbreaks worldwide [4]. All of the SEs share sequence and structural similarities and are typically encoded by the genes located on mobile genetic elements (MGEs), such as plasmids, prophages, transposons, or pathogenicity islands. These MGEs are effective vehicles for spreading virulence and drug resistance genes between S. aureus strains through horizontal gene transfer, which often changes the ability of the pathogen to cause disease and has significant impact on the evolution of the organism [5]. It is known that the sequences of certain SE genes are grouped together on MGEs. For example, the enterotoxin gene cluster (egc), comprising seg-sei-sem-sen-seo and sometimes seu, is located on the genomic island vSAb [6]. The genes of sea-sek-seq are typically present together in phage Q3 [5]. A family of pathogenicity islands carries seb-sek-seq on SaPI in S. aureus COL [6], tsst-sec-sel on SaPI2 (SaPln1/SaPlm1) in strain N315/Mu50 [6,7], or sec-sel on SaPI3 in strain MW2 [6,8]. In addition, sed-sej-ser are encoded on plasmids, and seh is linked to the staphylococcal cassette chromosome mec (SCCmec) elements, a determinant of methicillin resistance [5,9].
As major virulence factors in S. aureus, TSST, ETs, and SEs have been implicated in host colonization, invasion of damaged skin and mucus, gastrointestinal infection, and evasion of host defense mechanisms. Therefore, it is important to determine the toxin gene profiles of S. aureus strains from different clinics to understand the genetic and pathogenic relatedness, as well as the epidemiology of S. aureus. While Wu et al. [10] have recently reported SE gene profiles of community-acquired methicillinresistant S. aureus isolates (CA-MRSA) from Chinese children, few reports have described enterotoxin and exotoxin gene profiles of clinical S. aureus strains from different geographical areas in China, especially in methicillin-sensitive S. aureus (MSSA). Expression of most virulence factors in S. aureus is controlled by the accessory gene regulator (agr) locus. Based on the amino acid sequence polymorphisms of the agr-encoding autoinducing peptide and its corresponding receptor, S. aureus strains can be divided into 4 major agr groups (I -IV) [11]. It would be interesting to determine the prevalence of agr groups in S. aureus isolates from various hospitals and to investigate a possible relationship between agr groups and the occurrence of toxin genes.
For genetic typing of bacterial strains, pulse-field gel electrophoresis (PFGE) is known to be a highly discriminatory technique and is frequently used for characterizing genetic diversity and outbreak investigations of microbial pathogens. PFGE is a whole genome typing method based on DNA fragment patterns generated by restriction digestion, so its stability may be insufficient for reliable studies of the evolution and phylogenetic relationships of bacterial strains [12]. Despite recent improvements on standardized protocols and interpretation criteria of PFGE data, comparison of the results from different laboratories remains difficult. Multilocus sequence typing (MLST), which involves sequencing 7 housekeeping genes in each S. aureus genome and then comparing them with the established sequence information in MLST database (http://www.mlst.net), could be more accurate and reproducible. However, housekeeping genes are relatively stable by nature, and changes in these genes accumulate slowly over time, making MLST less discriminatory than PFGE [13]. Therefore, it would be advantageous to use both techniques for accurate and discriminatory genotyping of bacterial populations.
In this study, we examined 108 clinical S. aureus isolates from different regions of China for the patterns of 22 toxin genes by PCR. Genotypic information of the strains was determined by PFGE, MLST, and agr typing. The relationships between the genotypes and toxin gene profiles, genetic background and geographical distributions of the strains were analyzed. Finally, toxin gene typing was compared to PFGE and MLST methods for discriminatory ability in genotyping S. aureus strains.

Results
Toxin gene content in S. aureus isolates By PCR amplification of sequence-specific regions in 22 toxin genes, the occurrence of the SE, ET and TSST genes in 108 S. aureus isolates was determined. It was found that 98 strains (90.7%) contained at least one toxin gene and 80 strains (74.1%) carried two or more toxin genes (Fig. 1A). Although the average number of toxin genes per isolate was 4, up to 12 distinct toxin genes were detected in two given strains. The prevalence of each SE gene, eta, etb, etd, and tsst in the isolates is shown in Fig. 1B. The most abundant toxin gene was tsst presented in 52 (48.1%) isolates, followed by sea (44.4%), sek (42.6%), and seq (40.7%). The see and etb genes were not detected in any of the isolates. Consistently, see was reported to be the least frequently identified SE gene in S. aureus [14].
In 98 toxin gene-positive strains, there were 47 different toxin genotypes, of which 6 genotypes contained a single toxin gene and the rest were comprised of multiple toxin genes (Table 1). Notably, the combinations of SE genes were conserved in these strains. The sea-sek-seq combination including sea-sek-seq co-existed with additional toxin genes was the most common, harbored by 38 isolates. The seb-sek-seq combination was in 5 isolates, and a complete set of egc clusters (seg, sei, sem, sen and seo) was found in 19 isolates, of which 4 contained an additional SE gene (seu). The sej-ser and sedsej-ser genes were present in 15 and 5 isolates, respectively, and secsel was found in 11 isolates. Previously, sea-sek-seq and seb-sek-seq, carried by staphylococcal phage Q3 and pathogenicity island SaPI1, respectively, were reported to coexist in S. aureus [10]. In agreement, this study detected sea-seb-sek-seq together in two of the isolates. Moreover, different combinations of sea-sek-seq, sec-sel, sedsej-ser, and egc clusters were found in these strains. The high prevalence and conserved combinations of SE genes in these clinical strains suggest a high-frequency transfer of MGEs encoding various SE genes, which could significantly contribute to pathogenicity of these strains.
Except for seb and seu found only in MSSA, the rest SE genes were present in both MRSA and MSSA. Interestingly, all of the MRSA strains were determined to be SE gene-positive, suggesting these MRSA strains could be more pathogenic than the MSSA strains.

PFGE
In 108 S. aureus isolates typed by PFGE, 25 clusters were classified by using 80% similarity as a cutoff. Each cluster was designated by a number from 1 to 25 (Fig. 2). In each cluster, the number of strains varied substantially, e.g., cluster 1, 2, 22 and 24 included only one strain, whereas cluster 17 contained 21 strains. The correlation between PFGE genotypes and strain origins was insignificant. Some isolates from the same source and location were typed into different PFGE clusters, e.g., 16 strains isolated from sputum from the First Hospital of Jinlin University were distributed to 9 different clusters. In contrast, some isolates from different origins and sources were grouped into the same PFGE clusters, e.g., 21 isolates from 5 hospitals and 3 different sources all belonged to cluster 17. Among the 25 PFGE clusters, 18 clusters included isolates from two or more different hospitals and 17 clusters had isolates from two or more sample sources. Furthermore, 39 MRSA isolates belonged to 6 different clusters and 69 MSSA strains were distributed to 19 different clusters, suggesting no association between methicillin resistance and PFGE type in these strains.
By MLST, 108 S. aureus isolates were grouped into 25 sequence types (STs) which were assigned to 16 clonal complexes (CCs) including two new singletons ( Table 2). Among these, 11 STs (191511 STs ( -192411 STs ( , 1926 were submitted as new registrations to the MLST database (http://www.mlst.net) because of the sequence information from these new alleles. With the exception of ST120 and new STs, the rest of STs included more than one isolate. CC239 was a major MLST type which consisted of ST239 and two new STs (ST1923 and ST1924), and included 35 isolates. In agreement with the results in MLST database, all of the 35 isolates in CC239 were MRSA. The remaining 4 MRSA strains belonged to CC5. In contrast, 69 MSSA strains were distributed in 14 different CCs, including 9 new STs. In this study, not only were11 new STs identified, but also 6 known STs (ST6, 7, 15, 25,120 and 630) were first discovered in S. aureus isolates from China.

agr typing
Four agr groups (I -IV) were detected in 108 S. aureus isolates by multiplex PCR based on the amplicon size difference. The results showed that agr group I was the most predominant, detected in 60.2% (65/108) of the strains. The groups of agr III, II, and IV were less common and found in 19.4%, 16.7%, and 2.8% of the isolates, respectively. One isolate repeatedly yielded negative result for any of the agr types tested.

Association of toxin genes with genotypic background
When analyzing the correlation between toxin gene profiles and genotypes of PFGE, MLST, and agr typing of the strains, 47 toxin gene patterns were found in 108 S. aureus isolates classified into 25 PFGE clusters, 16 MLST CCs, and 4 agr groups (Fig. 2). Interestingly, diverse toxin gene patterns were found in the strains belonging to the same PFGE clusters, STs, and agr groups. For each cluster, the number of toxin genes could vary significantly from 1 to 12. For example, an MSSA strain belonging to CC30, PFGE cluster 23 and agr III carried an unusually large number (n = 12) of toxin genes. Likewise, 10 isolates in CC5, agr II and 3 different PFGE clusters were found to have 8 different SE gene patterns with high numbers of toxin genes (mean, 8.5; range, 6 to 12), all of which carried a complete egc cluster. However, the most common MLST type (C239, n = 35), present in agr I and 5 divergent PFGE clusters, had only 4 distinct SE genes patterns with lower numbers of toxins (mean, 3.8; range, 2 to 6).
Moreover, the most prevalent SE gene combination was sea-sekseq, carried by 38 S. aureus isolates with 32 strains in CC239 and 6 strains in CC1. The combination of seb-sek-seq was found in 5 strains belonging to CC59 (3), CC1 (1) and CC30 (1), 6 of the 11 strains containing sec-sel belonged to CC5, and the remaining strains were in CC1 (2), CC30 (2) and singleton1922. All of the 5 isolates containing sed-sej-ser belonged to CC5. The tsst gene coexisted with different kinds of SE genes and was detected in various CCs (Table 1). Even though 28/52 tsst positive and 7/9 etd containing strains were found in agr I, the correlation between the agr types and occurrence of tsst or etd is insignificant because a majority (.60%) of the isolates belongs to agr group I.
Taken together, these data suggest that the toxin gene patterns do not strongly associate with strain genotypes determined by PFGE, MLST or agr typing, except that a majority (71.1%) of seasek-seq positive isolates were from two hospitals (B &C) in northern China and typed into CC239, agr I and PFGE clusters 17-21. Since more than one distinct toxin gene combination was found in each cluster, CC and agr group, toxin gene typing could be more discriminatory than PFGE, MLST or agr typing.

Discussion
In this study, a collection of 108 clinical isolates from China were investigated for the presence of 18 SE and 3 exfoliatin genes, and the toxic shock syndrome toxin gene by PCR and were also genotyped by PFGE, MLST and agr typing. High genetic diversity of the strains was demonstrated by all of the typing methods.
Although only 108 S. aureus isolates were analyzed, they were collected from patients in 16 major hospitals geographically distributed in 14 different provinces across China. We believe these results represent the prevalence of different S. aureus strains in China.
The occurrence of toxin genes in these isolates was found to be abundant and diverse as 90.7% of the isolates, including both MRSA and MSSA, carried toxin genes with a maximum number of 12 toxin genes per strain and a total of 47 unique toxin genotypes. Previous studies showed seb and ser were the most abundant toxin genes in the clinical S. aureus isolates from children in China and patients in New York State, respectively [10,14]. We found tsst was the most common toxin gene existing in 52 (48.1%) isolates. This discrepancy is most likely due to the different origins of the strains. Furthermore, the most common combination of SE genes sea-sek-seq, harbored by 38 S. aureus isolates, of which 32 in CC239 were all MRSA, is in agreement with other reports [10]. Usually, sea-sek-seq is clustered in phage QSa3mu, whereas the sebsek-seq cluster is in pathogenicity island SaPI1. In the present study, two strains in CC1 were found to contain sea-seb-sek-seq, indicating the possible co-existence of QSa3mu and SaPI1 in these isolates. In addition, sea or seb without sek-seq, and sea-sek without seq were  detected in some strains. The findings of these uncharacterized toxin gene combinations imply that new variants on the mobile elements of QSa3mu or SaPI1 or the existence of additional MGEs carrying SE genes. The egc cluster, present in the S. aureus genomic island vSAb, has been reported to be strictly linked to the clonal background regardless of geographical distribution of the strains [6]. Previously, the egc cluster has been reported to be present in CC5, CC22, and CC45 isolates but not in CC8, CC12, CC15, and CC395 [10,15]. This study found egc in 19 S. aureus isolates, with 11 of them in CC5 and the rest distributed to 5 MLST CCs including CC15 (Table 1). Based on these results, the egccontaining genomic island is not completely absent in CC15 and may transfer among different CCs. In addition, we found 2 seb-egc positive isolates, one in each of CC25 and CC30, which is different from the results of Varshney et al. [14] who reported that sebpositive S. aureus strains did not carry egc. The tsst gene, carried by pathogenicity island SaPI2, was found to co-exist with the egc cluster in 11 isolates, with 5 belonging to CC5, which is in agreement with the results of Takano et al. [16]. Also, tsst was present in 10 of 11 sec-sel positive isolates. Together, these findings support the frequent transfer and multiple occurrence of toxin gene-containing MGEs in S. aureus isolates.
The seh gene is reported to be restricted to the CC1 genomic background [5]. It was located in close proximity of the non-mecA containing SCC element harbored by a MSSA strain as well as the SCCmec type IV of S. aureus [17]. Previously in China and Taiwan, seh-positive isolates were found in CC1 and ST338 of CC59, but not in ST59 [10,18]. Our results showed 10 seh-positive isolates in CC1, 2 in ST59 of CC59, and 2 in CC239 turned out to be MRSA. It suggests that all the seh-positive isolates, except for two in ST239, carried the non-mecA containing SCC element.
The association between specific agr groups and types of toxin production was reported contradictorily. Jarraud et al. [19] reported that agr IV strains were associated with exfoliative syndrome and most TSST producing strains belong to agr group III [20]. However, recent studies indicated no statistical association between agr groups and toxin types [21,22]. Similarly, no strong correlation between agr groups and toxin types was obtained from this study. As the agr locus belongs to the core variable genome and was reported to be strongly linked with clonal lineages [6,23], in agreement with that, our results showed that each CC exclusively belongs to one agr group.
It has been suggested that combined assessment of virulence gene profiles and genetic background could increase the discriminatory ability of genetic investigations of S. aureus strains from geographically diverse locations [14,15]. Campbell et al. [24] found that S. aureus strains from different geographic regions had different profiles of virulence genes. Our results showed high diversity of toxin gene profiles and genotypes classified by PFGE and MLST. No significant correlation between the genotypes and geographic locations of the strains was shown, except that 71.1% sea-sek-seq positive strains came from two hospitals in northern China (Beijing Hospital and the First Hospital of Jinlin University) and were typed into PFGE clusters 17-21 and MLST CC239 (Fig. 2). Compared to the MLST database (http://www.mlst.net), 11 new STs and 6 known STs (6,7,15,25,120 and 630) were first reported in S. aureus strains from China, which expands S. aureus genotype information in the database.
According to the typing results in Fig. 2, several isolates belonging to the same CCs were assigned to different PFGE clusters. For example, CC239 strains were present in 5 divergent PFGE clusters, the strains in each of CC1, CC5 or CC188 were distributed to 3 different PFGE clusters, and CC59 strains were assigned to two PFGE clusters. In contrast, the strains within the same PFGE clusters generally belonged to the same MLST CCs. These results suggest that PFGE may have better discriminatory power than MLST for typing S. aureus strains. However, both PFGE and MLST were less discriminatory than toxin genotyping since more than one toxin gene pattern was detected in each PFGE cluster or MLST CC for almost all of the SE-positive strains. In agreement, Varshney et al. [14] also pointed out that MLST, spa typing, and PFGE were all less discriminatory than SE content in seb-positive strains in the U.S. There were 28 distinct combinations of SE genes found in S. aureus strain USA300, the predominant cause of CA-MRSA infections in the U.S. [25]. We found 47 unique toxin gene combinations in 108 strains belonging to 25 PFGE clusters, 16 MLST CCs, and 4 agr types. Given that most strains contained a distinct toxin gene profile even if their MLST backgrounds or PFGE patterns were clonal, acquisition or loss of MGEs containing SE genes may occur frequently within individual lineages.
In conclusion, our results showed a high genetic diversity of the clinical S. aureus isolates from China. Toxin genotypes were highly diverse in these strains, which could be used as a discriminatory method in combination of PFGE or MLST for genotyping S. aureus strains. The combination of toxin genes was found to not restrict to genetic background (CC) of the strain, suggesting frequent transfer of toxin gene-containing MGEs within or among CCs. In this study, 11 new STs were submitted to the MLST database and 6 known STs were first reported in S. aureus strains from China, which significantly expands our current knowledge of genetic background of S. aureus strains in China. The new SE gene combinations identified herein suggest the existence of variants or new types of MGEs.

Ethics statement
We did not feel that ethics approval of human participants was necessary for this study since all of the S. aureus isolates were given as a gift from the Institute of Clinical Pharmacology at Peking University. This research did not involve in any clinical samples or human participants.

Bacterial isolates and growth conditions
A total of 108 S. aureus isolates were originated from 16 hospitals located in 14 different provinces of China. Of these, 63 S. aureus strains were isolated from sputum, 10 from pus, 9 from blood, 6 from urine, and 20 from other clinical samples by the Institute of Clinical Pharmacology at Peking University ( Table 3). All of the isolates were identified as S. aureus strains using the API Staph-Ident system (bioMerieux, Shanghai, China), and confirmed by16s RNA sequencing using previously described primers: F27 (59-AGAGTTT-GATCCTGGCTCAG-39) and R1492 (59-TACGGTTACCTTGTTACGACTT-39) [26]. Phenotypic detection of methicillin resistance of the strains was performed as described by Fan et al. showed that 39 strains were resistant to methicillin (MRSA) [27]. All of the S. aureus strains were routinely grown in tryptic soy broth (Becton Dickinson, Sparks, MD) overnight at 37uC.

PCR detection of staphylococcal toxin genes
Genomic DNA of S. aureus was purified using a modified cetyltrimethylammonium bromide method [28] and then was subjected to PCR amplification of 18 SE genes, 3 exfoliatin genes, and the tsst gene by using the primers listed in Table 4. All of the PCR were performed in a singleplex platform in a GeneAmp PCR system 9700 (Applied Biosystems, Foster City, CA). For each reaction, 1 ml of genomic DNA (approx. 100 ng) was added to a 24-ml PCR mixture containing 1U of Taq DNA polymerase (Fermentas Inc., Glen Burnie, MD), 16buffer, 1.5 mM MgCl 2 , and 0.4 mM each primer. The amplified DNA fragments were separated in a 1.5% agarose gel stained with ethidium bromide. To ensure the primer specificity and PCR reliability, both positive and negative controls were included into each run of PCR. The toxin gene-positive strains used in the experiment were: S. aureus ATCC8095 positive for sea, sed, sej, sek, seq, and ser; S. aureus ATCC14458 for seb; S. aureus ATCC27664 for see; S. aureus ATCC27661 for seg, sei, sel, sem, sen, and seo. For tsst, sec, seh, sel, seu, eta, etb, and etd, S. aureus isolates confirmed by sequencing of these genes were used as positive controls. For a negative control, ddH 2 O was used instead of a DNA sample. To confirm the specificity of detecting toxin genes, PCR products from 10% of the toxin gene-positive samples were randomly selected for DNA sequencing, and the resulting nucleotide sequences were searched against the S. aureus sequence database in the GenBank using BLAST (http://www.ncbi.nlm.nil.gov).

Pulsed-field gel electrophoresis (PFGE)
PFGE analysis was performed for all the S. aureus isolates. The procedures and buffers used for the preparation of chromosomal DNA, macro-restriction of the DNA, and PFGE were modified from an earlier report [29]. Briefly, 50 ml cell suspensions were mixed with equal volumes of 1.6% low-melting-point agarose (BioProducts, Rockland, ME) and then dispersed into plug molds. After solidification at 4uC, the plugs were transferred into 2 ml positive lysis buffer containing 1 mg/ml lysostaphin and 50 mg/ml lysozyme (Sigma-Aldrich, St Louis, MO), and incubated overnight at 37uC without shaking. After incubation, the positive lysis buffer was removed and replaced with 2 ml negative lysis buffer containing 50 mg/ml proteinase K (Sigma-Aldrich). The plugs were incubated overnight again at 55uC and then washed 3x with Tris-EDTA buffer. After that, the plugs were transferred into 2 ml of fresh TE buffer and stored at 4uC until further analysis.
For restriction digestion, a plug was cut into small slices and placed in a 100 ml reaction mixture containing 20 U of SmaI (New England BioLabs, Beverly, MA). After a 16 hr incubation at 30uC, digested chromosomal fragments were analyzed by loading the trimmed slices of the plug into a well of a 1% SeaKem agarose gel (FMC Corp. Rockland, ME) prepared with 0.5x TBE buffer (44.5 mM Tris-borate and 1 mM EDTA, pH 8.0). All the wells containing plug slices were sealed with 0.8% SeaPlaque agarose. Electrophoresis was performed in the CHEF-DR III electrophoresis cell (Bio-Rad, Melville, NY) under the following conditions: initial pulse of 5 sec, final pulse of 45 sec, at 6 V/cm at 14uC for 24 hrs. After staining with ethidium bromide and destaining with ddH 2 O, the gel was visualized using a UV transilluminator. Lambda DNA (New England Biolabs) was used as size standards and served as a control for the running parameters of the CHEF-DR units. The patterns of DNA fingerprint were analyzed using BioNumerics 4.0 software (Applied Maths NV, Austin, TX). The band patterns among different strains were compared using Dice coefficients with a 1.5% band position tolerance. A dendrogram of PFGE results was created using the unweighted pair group method with arithmetic averages (UPGMA). The cluster cutoff was set at 80% similarity [30].

Multilocus sequence typing (MLST)
Amplification of 7 housekeeping genes (arcC, aroE, glpF, gmk, pta, tpi and yqiL) of S. aureus was performed using the primers and PCR conditions specified at the MLST website (http://www.mlst.net). Amplified PCR fragments were visualized on agarose gels and purified using the Agencourt AMPure XP system (Beckman Coulter, Inc., Beverly, MA). Sequencing reactions were carried out in both forward and reverse directions using the same primers  for PCR and the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems) [31]. The products were cleaned of residual dyes using the Agencourt CleanSEQ system (Beckman Coulter, Inc.) following the manufacturer's instructions. Nucleotide se-quences were determined using an ABI 3730 DNA sequencer (Applied Biosystems). Results from each primer set were assembled using the Sequencher software v4.10 (Gene Codes Corp., Ann Arbor, MI). In a few cases, new sequencing reactions were performed and analyzed to resolve ambiguities. The DNA sequences of each locus were compared to the allele sequences in the MLST database. The alleles not matched to the corresponding sequences in the database were submitted as new registrations. The combination of alleles at the seven loci was defined as an allelic profile for each isolate, and each allelic profile was assigned to a sequence type (ST), either already defined in the database or designated as a new registration by the website curator. The eBURST algorithm was used to assign MLST clonal complexes (CCs) (http://eburst.mlst.net). In addition, the nucleotide sequences of new alleles were submitted to GenBank (http:// www.ncbi.nlm.nih.gov/genbank) and given accession numbers JN585753 -JN585759.