Molecular typing of Shigella sonnei isolates circulating in Nanjing, China, 2007–2011

Introduction: Shigellosis is a major public health concern worldwide. This study intended to assess the baseline genotyping data among local Shigella sonnei strains spanning over five years. Methodology: Fifty non-repeat clinical strains of S. sonnei isolated from stools of patients in different hospitals in Nanjing, China, were studied. Three subtyping tools, pulsed-field gel electrophoresis (PFGE), multi-locus sequence typing (MLST), and multi-locus variablenumber tandem-repeat (VNTR) analysis (MLVA), were used for routinely subtyping local S. sonnei. Results: DNA sequencing only identified two sequence types (STs) among the 50 isolates in the MLST profiles, whereas PFGE and MLVA both showed suitable discriminatory power and yielded 19 and 30 different patterns, respectively. The major PFGE pattern comprised 21 strains isolated from different years. A total of four complexes were identified by MLVA, with the isolates differing by a single locus (singlelocus variants). Conclusions: The S. sonnei strains circulating in Nanjing, China, in 2007–2011 originated from different clones with a degree of diversity. Most of the clones were closely related to each other. Overall, the strains were distinguishable by PFGE and MLVA. MLVA based on eight selected VNTR loci represented a more favorable degree of discrimination than did PFGE and may be a reliable complement for PFGE for routine subtyping of S. sonnei. The problems of MLST in subtyping regarding S. sonnei were also demonstrated.


Introduction
Shigellosis continues to be a major public health concern worldwide, predominantly affecting children under five years of age in developing countries [1]. Shigellosis is a notifiable disease in China. Although recent surveillance data showed that the overall trend of shigellosis in China is declining, the disease remains more common than in developed countries and presents itself as a serious condition. Shigellosis can be caused by S. dysenteriae, S. flexneri, S. boydii, or S. sonnei. The dominant Shigella species may be related to the location as well as to the stage of economic development. Among the Shigella species of concern, S. flexneri accounts for most of the cases in the developing world, and S. sonnei is most prevalent in the developed world [1,2]. However, a shift in Shigella dominance from S. flexneri to S. sonnei has been observed in certain newly industrialized areas in Asia, such as Taiwan and South Korea [3][4][5]. Similarly, increasing rates of S. sonnei cases were recorded in economically developed provinces in China between 2001 and 2010 [6].
Although S. sonnei is becoming increasingly important as a causative agent of shigellosis in China, there is limited knowledge on the molecular epidemiology of the pathogen. Several molecular typing methods with relatively high discriminatory power, such as pulsed-field gel electrophoresis (PFGE), multi-locus sequence typing (MLST), and multi-locus variable-number tandem-repeat (VNTR) analysis (MLVA), have been used to characterize the strains of Shigella species [7][8][9]. Of these, PFGE is the most widely used and standardized approach for molecular typing of bacterial isolates [10]. MLST was first proposed in 1998 as a definitive and portable typing approach for analyzing bacteria [11]. MLVA analysis of VNTR is a newly devised typing method for characterizing bacterial pathogens, including S. sonnei [9]. Molecular characterization of Shigella strains originating from defined geographical regions provides important information to understand its epidemiology and microbiology, and the pathogenesis of Shigella infection. This prompted us to further characterize local S. sonnei strains at the molecular level for both clinical and epidemiological purposes.

Bacterial isolates
A total of 50 S. sonnei isolates were collected from hospitals in Nanjing, China, between 2007 and 2011. All the strains were isolated from stool specimens of patients with gastroenteritis. Strain names are specified as location, year, and sample number in sequence. These include five strains from an outbreak in 2011 [12]. Individual isolates were analyzed by standard biochemical tests. Confirmation was made using both API-20E biochemical test strips (bioMérieux SA, Marcy-l'Etoile, France) and a slide agglutination test using Shigella group antisera (Denka Seiken Co., Ltd., Tokyo, Japan) according to the manufacturer's instructions.

Preparation of crude bacterial DNA
Crude DNA template was prepared as follows: a pure culture of S. sonnei was plated onto Luria-Bertani agar and incubated overnight at 37°C. A loopful of bacterial culture was removed from the plate, suspended in 100 μL of sterile deionized water, boiled for five minutes, and immediately cooled on ice. After centrifugation at 12,000 g for three minutes, the supernatant was transferred into a new tube for use. Crude DNA template was used for MLST and MLVA.

Pulsed-field gel electrophoresis (PFGE)
The standardized PulseNet protocol of the Chinese Center for Disease Control and Prevention was used for PFGE typing (http://www.cdc.gov/pulsenet/pathogens/shigella.html) . Briefly, genomic DNA was digested with XbaI (Takara Bio, Shiga, Japan), and separated in 1% agarose gels with a contour-clamped homogeneousfield apparatus (CHEF-DRIII system, Bio-Rad, Hercules, USA). XbaI-digested Salmonella serotype Braenderup H9812 was used as the molecular weight standard. The electrophoretic profiles were visualized by UV illumination, and analyzed using BioNumerics software version 5.1 (Applied Maths, Kortrijk, Belgium). Banding patterns were compared based on the Dice similarity coefficient, and a dendrogram with a setting of 1.5% for optimization and position tolerance of 1.5% for band comparison was constructed by UPGMA (unweighted pair-group method with arithmetic averages) algorithm.

Multi-locus sequence typing (MLST)
Fifteen housekeeping genes were amplified using the primers and polymerase chain reaction (PCR) conditions described below for MLST analysis: aspC, clpX, fadD, icdA, lysP, mdh, uidA, arcA, aroE, cyaA, dnaG, grpE, mtlD, mutS, and rpoS. Loci and primers from the EcMLST database (http://www.shigatox.net/ecmlst) are shown in Table  1. The PCR mixture contained 0.2 μM each of primer, DNA template, 10x PCR buffer 5 μL (Takara Bio, Shiga, Japan), and dNTP 200 μM (Takara Bio, Shiga, Japan) in a volume of 50 μL. Cycling conditions were as follows: initial denaturation at 94°C for 10 minutes, 30 cycles of amplification (92°C for 1 minute, 58°C for 1 minute, 72°C for 30 seconds), followed by 72°C for 5 minutes. The PCR products were sequenced at Shanghai Sangon Biotech. Sequences were aligned and analyzed using DNAStar analysis programs (DNAStar, Madison, USA). The results were submitted to the EcMLST database. Different sequences of a given locus were given allele numbers, and each allelic profile defined a sequence type (ST). New sequence types would be assigned a number for the purpose of the analysis.

Multilocus variable-number tandem-repeat analysis (MLVA)
Eight VNTR loci, SS1, SS3, SS6, SS9, SS10, SS11, SS13, and SS23 reported by Liang et al. [9] were used. The genomes of S. sonnei strain Ss046 (GenBank Accession No. CP000038) were previously explored for VNTR loci identification [13]. The primers and conditions are indicated in Table 2. Simpson's indices of diversity for MLVA were calculated according to Hunter and Gaston [14]. Three multiplex PCR combinations were carried out: mPCR1 consisted of primers for SS3, SS6, SS9, and SS23; mPCR2 consisted of primers for SS1, SS10, and SS11; mPCR3 consisted of primers for SS13. Each 20 μL PCR mixture contained 10× PCR buffer 2 μL (Takara, Japan), 0.2 mM of each dNTPs, 0.2 μM of each primer, 1 unit of TaqDNA polymerase (Takara, Japan), and 1 μL of DNA template. Cycling conditions were slightly modified: initial denaturation at 94°C for 5 minutes, 30 cycles of amplification (94°C for 45 seconds, 55°C for 50 seconds, 72°C for 60 seconds), followed by 72°C for 10 minutes.  The resulting PCR products were sent to Beijing TSingKe Technology for separation by capillary electrophoresis on an ABI 3730XL genetic analyzer with a GeneScan 500 LIZ size standard. Data were collected, and the lengths of the amplicons were determined with GeneMapper version 4.0 (Applied Biosystems, Foster City, USA). Amplicons of different lengths from each locus were subjected to nucleotide sequencing to verify the repeat sequence. The primers used for sequence determination were the same as those used for PCR but without the dye label. Amplicon sizes were converted to copy numbers using BioNumerics software version 5.1 (Applied Maths, Kortrijk, Belgium), and the data was analyzed as described previously [15]. A dendrogram was constructed by UPGMA clustering based on categorical coefficient. Non-approximated confidence interval (CINA) was calculated according to formulas described previously [16].

Results
PFGE analysis of XbaI-digested chromosomal DNA of the 50 S. sonnei strains yielded 19 reproducible PFGE patterns, ranging in size from approximately 20.5 to 668.9 kb. As shown in Figure 1, one major PFGE pattern, designated type 5, was shared by 21 of the 50 analyzed strains (42%). Type 4 (8%), type 11 (8%), type 7 (6%), and type 9 (6%) were relatively prevalent among the remaining isolates. The above five types comprised 70% of the isolates with similarities of 93.00%, suggesting that most of the strains were closely related to each other despite being isolated in different years. The strains isolated during the same year exhibited similar PFGE patterns. The similarities of the five strains associated with an outbreak in 2011 were more than 97.00%, suggesting a very close relatedness of the epidemiologically related strains. Based on 96% similarity, 10 clusters (A-J) were observed, with the largest containing 60% of the isolates (Figure 1). The similarity index for all groups (A-J) was 83%.
MLVA based on eight VNTR loci was performed to further characterize the isolated S. sonnei strains. Overall, the 50 S. sonnei strains were discriminated into 30 different MLVA types as shown in Figure 2. Most MLVA types have a distance of two to four loci from each other, mainly in SS1, SS3, SS6, and SS9. A total of four complexes were identified by MLVA where the isolates differed by a single locus (singlelocus variants). The five outbreak-related strains were grouped to complex 3 (type 19 and 20, differing at SS3). Single-locus variants occurring during an outbreak were common [9].   Two main clusters, cluster CI including type 1 and 2, and cluster CII consisting of the remaining types, were observed from the dendrogram generated. S. sonnei is one of the major bacterial pathogens causing travelassociated diarrhea [17]. Therefore, the results of this study were compared with MLVA profiles of 40 strains from two other regions in China to determine whether these strains were related. Ten strains obtained from a foodborne outbreak in Anhui province in 2011 and thirty strains isolated from sporadic cases in Beijing between 2010 and 2011 were included. As shown in Figure 2, six isolates from sporadic gastroenteritis cases in Nanjing displayed an identical MLVA pattern, type 22. This particular MLVA pattern was also exhibited by nine strains obtained from patients in a foodborne outbreak in Anhui, a province adjacent to Nanjing (Figure 3). In this outbreak in Anhui, the single strain recovered from leftover food showed a MLVA pattern that was highly related to MLVA 22, differed by one locus at SS6. Nevertheless, the MLVA profiles of the Nanjing isolates differed at two to six VNTR loci from those of Beijing strains. Overall, the combination of the eight VNTR loci with high and low D values was adequate for molecular subtyping of the S. sonnei isolates.

Discussion
We confirmed the existence of diverse S. sonnei clones and the usefulness of PFGE in local epidemiological studies. We also concluded that most of the isolates were closely related to each other in the period of time studied. Strain 2008-01 assigned to ST 115 by MLST was grouped into cluster D (type 11) in the PFGE profiles. Compared to PFGE profiles, which revealed 19 distinct genetic patterns and 10 clusters, it is apparent that MLST lacks the ability to provide a satisfactory level of discrimination. This low resolution versus PFGE has been recently reported with Shigella species [18,8] and with some other species of bacteria as well [19]. Studies suggest that MLST may not be powerful for characterizing closely related strains within a specific serotype due to high sequence conservation of the housekeeping genes [19,20].
MLVA may not be appropriate as a subtyping tool for phylogenetic analysis among different bacterial species or serotypes. Nevertheless, MLVA is useful to establish genetic relatedness among strains of a monomorphic species, such as S. sonnei [21]. Though applied in a limited collection of S. sonnei isolates, our study suggests that these eight loci were sufficient for the subtyping of S. sonnei. Strains isolated five years apart with highly similar MLVA patterns were observed; for example, strain NJ2007-10 in type 9 and NJ2011-32 in type 10 were closely related to each other (99.0% similarity), revealing the existence of a circulating S. sonnei clone with minor genetic changes during the period. Moreover, a number of strains in this study were related to S. sonnei from the neighboring Anhui province, and to lesser extent, to those from Beijing. MLVA type 22 was identified in a foodborne outbreak in Anhui province, suggesting that this genotype has been circulating in Nanjing and its neighboring regions since 2011. Moreover, the outbreak strains may be among the most adaptive strains that are prone to spread [22]. MLVA type 19 and 20 are comprised of five outbreak-related strains in Nanjing, and type 22 is comprised of nine outbreakrelated strains in Anhui. This indicates more adaptation and greater spreading ability of certain closely related MLVA genotypes.
The results of typing based on MLVA and PFGE were compared. Overall, MLVA identified more types than PFGE among the 50 isolates, and the genetic diversity was higher than that observed in PFGE. Three strains (2011-07, 2011-33, and 2011-22) exhibited identical PFGE patterns (type 9) and were also classified into an identical MLVA pattern (type 1). Nine strains were unique by both PFGE and MLVA. However, those strains that belonged to the same PFGE profile (PFGE type 1, 4, 5, 7, and 11) were distinguishable by MLVA. For example, 21 strains included in one major PFGE pattern (type 5) were further discriminated into 15 minor MLVA genotypes (MLVA type 2,17,3,13,4,15,19,20,24,22,18,25,30,26,29), suggesting that MLVA typing exhibited more detailed differences between S. sonnei that had similar PFGE patterns. However, the opposite was also found, since isolates with identical MLVA types could correspond to different PFGE types. For example, five isolates belonging to five PFGE patterns (type 1, 2, 3, 14, 16) were clustered into a single MLVA pattern (type 5). Different subtyping methods assessed the genetic characterization in different parts of the chromosome. Variation at the restriction sites may result in differences in PFGE profiles, while gene mutations may affect MLVA repeat locations. A combination of PFGE and MLVA analysis may yield more information about the clonality of Shigella sonnei.

Conclusions
In this study, we presented epidemiological trends of Shigella sonnei isolates in Nanjing for a period of time and correlations among three different typing methods. Although PFGE and MLVA both showed suitable discrimination in Shigella sonnei subtyping, MLVA exhibited a higher discriminatory ability than did PFGE. Our results were somewhat consistent with the reports that PFGE may not be sufficiently discriminative for clonal research of S. sonnei strains that have evolved over very long time periods [9,21]. MLVA would be useful for phylogenetic analysis of PFGE-indistinguishable S. sonnei strains. MLST of 15 housekeeping genes showed it to have low discriminatory ability as a subtyping tool for local S. sonnei isolates. The observations suggested that the S. sonnei circulating in Nanjing belonged to different clones that may be disseminated with minor genetic variations. Most of the clones were closely related to each other. Although our results may not reveal the precise clonal nature of local S. sonnei due to the small sample size, the observation has nevertheless described the genetic background of S. sonnei circulating in Nanjing in a five-year period. The PFGE banding patterns of these strains have been recorded into the national S. sonnei PFGE database, and the information will serve as the starting point for the molecular surveillance of shigellosis in the Nanjing area. MLVA may help to further differentiate subtypes due to its higher discrimination. Maintaining a local S. sonnei database may be useful to not only monitor changes in the genetic diversity of isolates over years, but also to detect diffuse outbreaks. Moreover, if patterns of more outbreak strains are supplemented, it is possible to confirm certain subtypes that are prone to spread and are associated with multiple unrelated outbreaks. The identified clones may pose higher risks of causing illness or outbreaks, referred to as a public health risk. Overall, the data will provide a useful typing resource which will therefore help address clinical and epidemiological issues regarding S. sonnei in this area.