Development of a sequence typing scheme for differentiation of Salmonella Enteritidis strains.

A DNA sequence typing scheme based on the caiC and SEN0629 loci was developed for differentiation of Salmonella Enteritidis strains and validated using a diverse collection of 102 isolates representing 38 phage types from different sources, year of isolation, geographical locations and epidemiological backgrounds. caiC encodes a probable crotonobetaine/carnitine-CoA ligase, and SEN0629 is a pseudogene. Our system allowed for discrimination of 16 sequence types (STs) among the 102 isolates analysed and intraphage type differentiation. Our findings also suggested that the stability of phage typing may be adversely affected by the occurrence of phage type conversion events. During a confirmatory phage typing analysis performed by a reference laboratory, 13 of 31 S.  Enteritidis strains representing nine phage types were assigned phage types that differed from the ones originally determined by the same reference laboratory. It is possible that this phenomenon passes largely unrecognized in reference laboratories performing routine phage typing analyses. Our results demonstrate that phage typing is an unstable system displaying limited reproducibility and that the two-loci sequence typing scheme is highly discriminatory, stable, truly portable and has the potential to become the new gold standard for epidemiological typing of S . Enteritidis strains.


Introduction
Salmonella Enteritidis is a major cause of human salmonellosis worldwide (Rodrigue et al., 1990). Epidemiological surveillance of this bacterium is principally based on the use of phage typing and genotyping methods. Phage types are generally considered to be stable and definitive epidemiological markers, but this is in contrast with several studies reporting various mechanisms of phage type conversions. For example, Frost et al. (1989) reported the conversion of S. Enteritidis PT4 to PT24 based on the acquisition of a plasmid belonging to the incompatibility group N (IncN). Likewise, Threlfall et al. (1993) have shown interrelationships between PT 4,7,7a,8,13,13a,23,24 and 30 caused by the loss or acquisition of an IncN plasmid. Subsequently, Rankin & Platt (1995) reported that the use of temperate phages 1, 2, 3 and 6 from the phage typing scheme of Ward et al. (1987) enabled conversion of PT4, 6a, 6a, 13 and 15 to PT8, 4, 7, 13a and 11, respectively. They were also able to convert PT1 to PT20, and PT15 to PT11. Chart et al. (1989) reported that conversion of S. Enteritidis PT4 to PT7 involved the loss of the lipopolysaccharide layer with a concomitant loss of virulence. Brown et al. (1999) demonstrated that transfer of a plasmid belonging to the incompatibility group X (IncX) into 10 isolates of S. Enteritidis belonging to 10 different phage types (PT1,2,3,4,8,9,9b,10,11 and 13) resulted in phage type conversion in 8 of the 10 strains (PT1,2,4,8,9,9b,10 and 11).
Phage typing requires specialized phage collections and bacterial strains for their propagation and for this reason is only performed in a few reference laboratories. Furthermore, the fact that most isolates of S. Enteritidis belong to a limited number of phage types highlights the lack of discriminatory power of the phage typing system. The predominance of certain S. Enteritidis phage types within certain geographical locations further underlines the need for high-resolution typing systems. In the United States, the predominant phage types are PT8 and PT13a (Hickman-Brenner et al., 1991), except for the west coast particularly in California, where PT4 emerged as the predominant phage type (Kinde et al., 1996;Patrick et al., 2004). PT4 has been most observed in Western Europe (Nygard et al., 2004).
DNA sequence-based approaches are highly discriminatory methods of characterizing bacterial isolates in a standardized, reproducible and portable manner (Maiden et al., 1998). Each isolate is defined by the alleles at each of the gene fragment loci and isolates with the same allelic profile can be assigned as members of the same clones (Maiden et al., 1998;Spratt, 1999).
Key advantages of DNA sequence-based typing methods over banding pattern-based subtyping techniques are that they are unambiguous and can be readily compared between laboratories, thus facilitating global, large-scale surveillance (Maiden et al., 1998;Wiedmann, 2002). Sequence data can be stored in a shared central database to provide a broader resource for epidemiological studies (Lemee et al., 2004).
Here, we report the development of a DNA sequence typing scheme for differentiation of S. Enteritidis strains based on the caiC and SEN0629 loci. The scheme was validated using a variety of S. Enteritidis isolates from different sources, year of isolation, geographical locations and representing a wide range of phage types and epidemiological backgrounds. Furthermore, we demonstrate that phage typing is an unstable system displaying limited reproducibility.

DNA extraction
Salmonella Enteritidis genomic DNA was extracted using the GenElute Bacterial Genomic DNA Kit (Sigma, St Louis, MO) according to the manufacturer's instructions.

PCR amplification and DNA sequencing
Primers used for PCR amplification of caiC and SEN0629 locus fragments are listed in Table 2. PCR was carried out in a PTC 100 Peltier Thermal Cycler (GMI, Ramsey, MN). PCR amplification was performed using the Ready-Mix Taq PCR Reaction Mix (Sigma) following the manufacturer's instructions. PCR was carried out in a final volume of 50 lL using 25 lL of the ReadyMix, 0.3 lM of each primer, 1 lL of DNA extract and sterile water to make up the final volume. The PCR thermal cycling conditions included an initial denaturation at 94°C for 5 min, 35 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 40 s, extension at 68°C for 60 s and final extension at 68°C for 5 min. PCR products were purified using a QIAquick PCR purification kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions, and analysed by 1.5% agarose gel electrophoresis. Purified PCR products were sent to the University of California DNA sequencing facility at the University of California (Davis, CA) along with PCR primers for direct sequencing. Sequencing was performed in both directions to ensure accuracy.

Phylogenetic analysis
Sequences obtained in this study and those retrieved from GenBank were aligned using CLUSTALW integrated in the freely available ARB software package (Ludwig et al., 2004). Alignments were trimmed to a uniform length (corresponding to nucleotide positions 82788-83514 for caiC and 696231-697280 for SEN0629 on the genome sequence of S. Enteritidis str. 125109, accession no. AM933172). The trimmed alignments were used to construct a concatenated alignment. Phylogenetic trees based on the neighbour-joining method (Saitou & Nei, 1987) were constructed from the individual alignments as well as from the concatenated alignment using MEGA version 4.0 package (Tamura et al., 2007). Evolutionary distances were calculated by Kimura's two-parameter model of substitution (Kimura, 1980). Bootstrap confidence values were generated using 1000 repeats of bootstrap samplings (Felsenstein, 1985).

Nucleotide sequence accession numbers
The nucleotide sequences determined in this study have been deposited in GenBank under accession numbers JN546231-JN546434. Full alignments of all 16 sequence types displaying all bases as well as differences to sequence type 1 were deposited as a popset in GenBank.

Results and discussion
Characteristics of the caiC and SEN0629 marker loci caiC encodes a probable crotonobetaine/carnitine-CoA ligase and the fragment analysed ranged from position 82788 to 83514 on the genome sequence of S. Enteritidis str. P125109, accession no. AM933172. SEN0629 is a pseudogene and the fragment analysed ranged from position 696231 to 697280 on the genome sequence of S. Enteritidis str. P125109 (Table 3). Its homolog on the genome sequence of S. Typhimurium LT2 accession no. NC_003197 is located at the STM0660 locus and encodes a cytoplasmic protein. caiC and SEN0629 display a GC content of 54.2% and 55.2%, respectively. The combined use of caiC and SEN0629 sequences for typing 102 S. Enteritidis strains representing 38 phage types enabled the identification of 16 sequence types and intraphage type discrimination (Table 1, Fig. 1). Isolates kept their initial sequence type after being resequenced, thus indicating the high stability of caiC and SEN0629 as marker genes for S. Enteritidis subtyping.
Instability and limited reproducibility of the phage typing system A total of 31 S. Enteritidis strains representing phage types 1, 4, 6, 6a, 6b, 8, 13, 13a, 14b were initially phage typed by NVSL and later sent to the same institution for a second phage typing. Of the 31 S. Enteritidis strains, 13 presented phage types that differ from the ones determined originally (Fig. 1, Table 1). One ATCC strain (ATCC 13076) was initially typed as PT1 and subsequently typed as RDNC. Three strains were originally typed as PT6b and subsequently typed as PT5a, PT5a and untypeable. Two other strains were initially typed as PT4 and were later typed as PT1a and RDNC. Three strains were first typed as PT6a and in the second analysis phage typed as PT5a, PT6b and RDNC. Two other strains originally typed as PT13 were subsequently typed as PT6a. Two strains with original phage types 6 and 8 were subsequently phage typed PT4 and RDNC, respectively. Surprisingly, one strain had converted from a less common phage type PT6 to the most predominant phage type PT4 in Europe and vice versa, and strains of more prevalent phage types 4, 8 and 13 had converted to less prevalent phage types 1a, RDNC and 6a (Table 1). It should be noted that strains ID 502 and ID 1387 were initially phage typed as PT13 and subsequently phage typed as PT6a, thus appearing to become a different clonal lineage. These observations underline the major limitations encountered while using phage typing for epidemiological investigation and severely restrict its value for monitoring the epidemic spread of S. Enteritidis. Our findings confirm previous studies reporting the occurrence of phage conversions. Frost et al. (1989) reported the conversion of strains of PT4 to strains of PT24 in S. Enteritidis based on the acquisition of IncN plasmids. Inter-relationships were shown between strains of several phage types based on the lost or acquisition of an IncN plasmid (Threlfall et al., 1993). Conversion of PT4 to PT7 and PT1, 4, 6 to PT7 by loss of lipopolysaccharide has been described (Chart et al., 1989). Temperate phages 1, 2, 3, and 6 were used to convert PT4 to PT8, PT6a to PT4, PT6a to PT7, PT13 to PT13a and PT15 to PT11 (Rankin & Platt, 1995). Transfer of a plasmid belonging to the IncX into 10 isolates of S. Enteritidis belonging to 10 different phage types (PT1,2,3,4,8,9,9b,10,11 and 13) resulted in phage type conversion in 8 of the 10 strains (PT1,2,4,8,9,9b, 10 and 11) (Brown et al., 1999).

Reported limitations of existing S. Enteritidis genotyping systems
PFGE that is currently the gold standard technique for subtyping S. Enteritidis isolates is laborious, requires precise standardization and displays limited subtyping potential (Hudson et al., 2001;Liebana et al., 2001). Ribotyping is a laborious procedure that includes multiple steps such as DNA isolation, restriction, electrophoresis, Southern blotting, probe preparation and hybridization (Landeras & Mendoza, 1998). Thong et al. (1995) analysed a total of 61 isolates of S. Enteritidis using PFGE and ribotyping and came to the conclusion that the close genetic similarity observed between epidemiologically unrelated and outbreak-related isolates of S. Enteritidis suggests that both PFGE and ribotyping are of limited value in the epidemiological analysis of these particular isolates.
PCR-based methods such as RAPD lack the ability to separate artefactual variation and true polymorphism (Tyler et al., 1997;Landers et al., 1998). The application of RAPD requires the identification of primers capable of recognizing DNA polymorphisms among isolates; however, it is not possible to predict which primers will be useful to differentiate strains of a species or serotype (Landeras & Mendoza, 1998).
MLVA is based on the detection of short sequence repeats that vary in copy number (i.e. variable number of tandem repeats or VNTR) in the microbial genome at various regions. Recently, Parker et al. (2010) reported that PFGE, MLVA and DNA microarray-based comparative genomic indexing failed to discriminate between S. Enteritidis PT30 strains related to outbreaks from unrelated clinical strains or between strains separated by up to 5 years. In that study, 20 of the 21 S. Enteritidis PT30 strains analysed had identical alleles at each of the nine VNTR loci that were examined and all of the S. Enteritidis PT30 and the S. Enteritidis PT9c strains analysed failed to amplify the SE3 VNTR locus. Boxrud et al. (2007) concluded in their study that while data portability is facilitated by the use of sequence-based subtyping methods, their use of fragment analysis to assess VNTR polymorphisms is subject to some of the same limitations seen for other gel-based systems.
The value of plasmid profiling as an epidemiological tool for S. Enteritidis is limited by the prevalence of the targeted plasmids in the strains being investigated (Maslow et al., 1993). In the study by Martinetti & Altwegg (1990), plasmid profiling of S. Enteritidis showed limited potential because the plasmid identified occurred at a relatively low frequency. Plasmid profiling has been shown to be of limited use for the subdivision of S. Enteritidis PT4, as many strains carry a single 38-MDa plasmid (Threlfall et al., , 1994. IS200 profiling and microarray analysis grouped the majority of S. Enteritidis phage types only into two fragments and two major lineages, respectively (Stanley et al., 1991;Porwollik et al., 2005).

Genotypic diversity among clonally related S. Enteritidis strains
Most phage types tested in this study formed a major cluster (ST1-ST13) on the phylogenetic tree. Strains of phage types 14, 16 and 27 (ST14-ST16) were distantly related to each other and clustered apart from the major cluster. The phylogenetic tree constructed based on concatenated nucleotide sequences of caiC and SEN0629 showed distinct subclusters of strains. Two of the subclusters included most of the phage types reported to belong  Table 3. Bootstrap values higher than 50% from 1000 replicates are shown. The evolutionary distances were computed using the Kimura twoparameter method and are in the units of the number of base substitutions per site. The bar represents 0.5% evolutionary sequence divergence. Strains displaying a second phage type that differed from the initial phage type are highlighted in yellow. Designation of sequences (strains) is as follows SE strain ID, source, specimen, year of isolation, first phage type, second phage type and sequence type (see Table 1 for details). NA, not available; ND, not determined; Ref., reference; Sw., swab; Un., untypeable; Env., environment; Peric., pericardium; Effl., effluent; Intest., intestine; MEX, Mexico; SWI, Switzerland; FRA, France. to either the PT4 or the PT8 clonal lineage. This is consistent with previous studies, which have shown that S. Enteritidis can be divided into two clonal lineages based on IS200, and whole genome microarray analysis (Stanley et al., 1991;Porwollik et al., 2005). Morales et al. (2005) reported that no DNA hybridization differences were found between a wild-type S. Enteritidis PT13a strain and a biofilm-forming S. Enteritidis PT13a strain; however, our scheme was able to differentiate the two strains and assigned them two distinct alleles.
Likewise, Olson et al. (2007) analysed more than 11 300 base pairs of sequence for each of seven S. Enteritidis PT13 strains but did not detect a single polymorphic site. Our two-loci sequence typing scheme was able to assign three sequence types to the four PT13 isolates analysed. Our data concur with previous studies (Guard et al., 2011;Shah et al., 2012), suggesting that genotypic differentiation may correlate with the phenotypic properties of the analysed strains.

Conclusion
In conclusion, we have developed a sequence typing system for S. Enteritidis a major food-borne pathogen. The high discriminatory ability of our system allows the differentiation of S. Enteritidis strains, including strains within the same phage type. Furthermore, our results demonstrate that the two-loci sequence typing scheme is stable, truly portable and has the potential to become the new gold standard for epidemiological typing of S. Enteritidis strains.
The results presented here also demonstrate that phage typing is unstable, incoherent and displays limited reproducibility.