Phenotypic and genotypic characterization of Canadian clinical isolates of. Vibrio parahaemolyticus to 2009

Transcription

1 JCM Accepts, published online ahead of print on 22 January 2014 J. Clin. Microbiol. doi: /jcm Copyright 2014, American Society for Microbiology. All Rights Reserved. 1 2 Phenotypic and genotypic characterization of Canadian clinical isolates of Vibrio parahaemolyticus to Swapan K. Banerjee 1 *, Ashley K. Kearney 2, Celine A. Nadon 2,5, Christy-Lynn Peterson 2, Kevin Tyler 1, Laurene Bakouche 1, Clifford G. Clark 2, Linda Hoang 3, Matthew W. Gilmour 2,4,5, Jeffrey M. Farber 1 Bureau of Microbial Hazards, Food Directorate, Health Canada, Ottawa, Ontario 1 ; National Microbiology Laboratory, Public Health Agency of Canada, Winnipeg, Manitoba 2 ; BC Centre for Disease Control, Vancouver, British Columbia 3 ; Diagnostic Services of Manitoba, MS673C Health Sciences Centre, Winnipeg, Manitoba 4 ; Department of Medical Microbiology and Infectious Diseases, University of Manitoba, Winnipeg, Manitoba 5 * Corresponding author: Mailing address: Bureau of Microbial Hazards, Food Directorate, Health Canada, 251, Sir F. D. Banting Driveway, Ottawa, Ontario K1A 0K9, CANADA. Tel: ; swapan.banerjee@hc-sc.gc.ca 1

2 19 ABSTRACT Vibrio parahaemolyticus is the leading bacterial cause of foodborne illness due to the consumption of contaminated seafood. The aim of the present study was to determine the population of subtypes and establish a better understanding of the various types of Vibrio parahaemolyticus strains that are causing human illness in Canada. Subtypes were obtained for 100 human-clinical isolates of V. parahaemolyticus collected between 2000 and 2009 by performing serotyping, ribotyping, pulsed-field gel electrophoresis and multilocus sequence typing. Within this panel of strains, there was a high level of diversity (between 22 to 53 subtypes per method), but the presence of predominant clones with congruent subtypes between the various methods was also observed. For example, all 32 isolates belonging to sequence type (ST) 36, were serogroup O4, while 31 of them were ribotype EcoVib , and 24 of the 32 were SfiI PFGE pattern VPSF With regard to the presence of known virulence genes, 74 of the 100 isolates tested positive for the presence of the thermostable direct hemolysin (tdh) by polymerase chain reaction, and 59 of these 74 strains also contained the second virulence marker, the tdh-related hemolysin (trh). The detection of trh was more predominant (81%) among clinical isolates and only four (4%) clinical isolates tested negative for the presence of both tdh and trh. This database comprising 100 clinical isolates of V. parahaemolyticus strains from Canada forms a baseline understanding of subtype diversity for future source attribution and other epidemiologic studies. 2

3 43 INTRODUCTION Vibrio parahaemolyticus, a member of the genus Vibrio, is ubiquitous in the marine environment around the world. It can be found naturally in estuarine environments with species distribution being driven by the climatic conditions and eutrophication of the regional waters. Whereas choleragenic vibrios are predominant in tropical regions, halophilic vibrios, such as V. parahaemolyticus, are abundant in the estuaries of the temperate zone (39). Several Vibrio species have caused human illness in Canada, including V. parahaemolyticus, and this has led to an interest in the epidemiology of Vibrio species (17, 26). Although the incidence of vibriosis reported to Canada s National Enteric Surveillance Program (NESP) is low, i.e., around 35 cases per year in this millennium to date, there has been an increasing trend from 18 cases in 2000 to 51 cases in 2012, of which the majority were caused by V. parahaemolyticus (32). Under-reporting likely occurs because of methodology issues, a lack of awareness of the disease, and the fact that the illness is usually self-limiting and ill individuals would likely not be visiting their physicians. However, V. parahaemolyticus infections can have severe consequences in immunocompromised individuals, or persons with certain medical conditions (30). For Vibrio spp., strain delineation is an important strategy in understanding the epidemiology of the disease and thus in preventing outbreaks and sporadic illnesses. The presence of hemolysin genes, serogroup and urease production in V. parahaemolyticus strains isolated from human patients, helped to identify the source of contaminated seafood in an earlier publication (24). In other studies, additional typing methods have been used for analyses linking foods associated with illness (15). Additionally, ribopattern and serotype can cluster both environmental and clinical V. parahaemolyticus isolates, implying that these typing methods are useful for source-tracking and identification of strains that can cause human illness (16). 3

4 Herein, we describe the use of pulsed-field gel electrophoresis (PFGE), multilocus sequence typing (MLST), ribotyping and serotyping, in addition to virulence gene detection by PCR, to study in detail the molecular epidemiology of V. parahaemolyticus strains isolated from Canadian patients (19, 23, 26, 40). A retrospective descriptive laboratory analysis of clinical V. parahaemolyticus isolates from across Canada (primarily British Columbia), and available from the inventory of the National Microbiology Laboratory, Public Health Agency of Canada, was conducted. Additional objectives of this work were to use the data to generate a comprehensive database, and to improve awareness and reporting of illnesses caused by Vibrio species in Canada, particularly V. parahaemolyticus. MATERIALS and METHODS Strain collection All human clinical isolates originating from provincial public health laboratories and submitted to the National Microbiology Laboratory (Public Health Agency of Canada), British Columbia Centre for Disease Control (BCCDC) or the Bureau of Microbial Hazards (BMH, Health Canada) between 2000 and 2009 were selected for this study. Additional clinical isolates were obtained from the inventory of the Canadian Food Inspection Agency (CFIA, Burnaby, BC). The isolates were sourced to the following provinces: British Columbia (BC), Alberta (AB), Saskatchewan (SK), Ontario (ON), New Brunswick (NB) and Prince Edward Island (PEI). Standard strains of pathogenic V. parahaemolyticus, ATCC17802 (American type culture collection strain from our BMH collection, see ref. 3) and NY477 (38), were used as appropriate controls, as required. Biochemical analyses 4

5 (i) API 20E strips: Biochemical identification and confirmation of presumptive isolates were accomplished by using API 20E diagnostic strips (34), following the procedure described by the manufacturer (biomérieux Canada Inc., Saint-Laurent, QC, Canada). A cell suspension of the presumptive isolate was prepared by re-suspending one isolated colony from a TSA-2N plate into 5 ml saline (0.85% NaCl), and sometimes repeated with 2% NaCl for better identification. The bacteria were distributed in the cupules of the strip, as instructed by the manufacturer, and incubated overnight at 35 o C. (ii) BBL DrySlide Oxidase test: Using a sterile loop (plastic), isolated colonies were picked and smeared directly onto the reaction area of the BBL DrySlide Oxidase (Becton Dickinson Microbiology Systems, Sparks, MD, USA), as described by the manufacturer. Each reaction area can accommodate up to four tests. A dark purple color within 30 sec indicated a positive reaction. (iii) Vibriostat O/129 test: The sensivities of V. parahaemolyticus isolates to vibriostat O/129 (2,4-diamino-6,7-diisopropylpteridine) was determined by spreading uniform lawns of bacteria, approximately 10 7 cells per ml of saline, with a cotton-tipped swab, on TSA-2N, and aseptically placing discs of both 10 and 150 μg of the O/129 (Oxoid Ltd., Hampshire, England) on the plates. The plates were incubated at 35 o C for 18 to 24 h, and a clear zone of inhibition of growth around the discs indicated susceptibility to O/129 at the observed concentration. Polymerase chain reaction (PCR) analyses The primers used to confirm the identity of V. parahaemolyticus were as follows: forward, 5'-AAA GCG GAT TAT GCA GAA GCA CTG-3', and reverse, 5'- GCT ACT TTC TAG CAT TTT CTC TGC-3', to generate an amplicon of 450 bp of the thermolabile hemolysin (tlh) gene (5). To detect the presence of pathogenic V. parahaemolyticus, the following primers 5

6 were used: forward, 5'- GAA GTA CCG ATA TTT TGC - 3', and reverse, 5' - ATG TTG AAG CTG TAC TTG - 3', to generate an amplicon of 385 bp of the thermostable direct hemolysin (tdh) gene (33). For the tdh-related hemolysin (trh), the primers used were as follows: forward, 5 -TTG GCT TCG ATA TTT TCA GTA TCT-3, and reverse, 5 -CAT AAC AAA CAT ATG CCC ATT TCC G-3, yielding an amplicon of 484 bp (4, 13). Conditions used in the PCR were as reported by the authors for each primer pair. Serotyping Confluent (overnight at 35 o C) growth of the experimental bacteria on TSA-2N Petri dishes were lifted using a 10 μl sterile (disposable plastic) loop and re-suspended in 5 ml of 2% NaCl containing 5% glycerol. Usually, five to six loopfuls were obtained from the confluent growth on the plate to acquire a thick suspension. The bacterial suspensions were autoclaved (121 o C) for 90 min, or 120 min (if needed), cooled, and washed twice by centrifugation (12,000 g, 10 min) to discard the glycerol, and the final pellet was re-suspended in 2% NaCl. Other steps, including slide agglutination tests, were as described by the manufacturer [DENKA SEIKEN, Japan, and available worldwide from the Mast group ( and elsewhere (25). The antisera kits were distributed by Mast Diagnostics (Mast Group Ltd., Merseyside, U.K.) and purchased from Alere Inc., Ottawa, Ontario, Canada. Strains with untypeable (UT) antigenic markers were designated as OUT ( O untypeable) or KUT ( K untypeable). Automated ribotyping Ribopatterns were generated using an automated RiboPrinter (DuPont Qualicon, Wilmington, Delaware) with accessories from the same manufacturer, which included membrane-processing packs and all the reagents for cell lysis, deproteinization, restriction digestion, and hybridization. The EcoRI kit (DuPont Qualicon) was used for restriction of the 6

7 V. parahaemolyticus genome, following the instructions supplied by the manufacturer. The resulting ribopatterns were imported into BioNumerics (Applied Maths, Austin, TX, USA), to generate dendrograms (Dice coefficients, 1.5% tolerance). A description of the procedure from cell lysis to image analysis was reported earlier (6, 7). Pulsed-field gel electrophoresis (PFGE) PFGE was performed according to the standardized PulseNet protocol with the restriction enzymes SfiI and NotI ( and ref. 2). PFGE patterns were designated by using BioNumerics (Applied Maths Inc., Sint-Martins-Latem, Belgium) for analysis and comparison (DICE coefficient of 1.5% tolerance). Multi-locus sequence typing (MLST) Template DNA was prepared using the Puregene DNA extraction kit (Qiagen Inc, Toronto, ON) according to manufacturer s directions. MLST analysis was conducted via the V. parahaemolyticus MLST website and database ( developed by Keith Jolley (22). Oligonucleotide primer sequences used for both DNA amplification and sequencing are shown in Table 1. PCR was performed with Invitrogen Platinum Taq DNA polymerase (Life Technologies Inc, Burlington, ON) following the manufacturer s directions. PCR products were purified using the Agencourt AMPure XP PCR purification kit (Beckman Coulter Genomics, Danvers, MA) and sequenced with an ABI3730 sequencer (Life Technologies Inc.) DNA sequence data was analyzed and a minimum spanning tree was generated using BioNumerics. Index of diversity Simpson s Index of diversity (SID) was used to determine the probability by which the typing system could differentiate consecutive samples randomly picked from a pool of 7

8 unrelated strains (20, 21). This concept was used to assess the discriminatory power (D) of the method. A value of 1.0 indicates that the typing method is able to distinguish one strain from all other members of that bacterial population, and a value of 0.0 represents an identical population (clonality). Congruence in diversity A measure of congruence in diversity, namely, the adjusted Wallace coefficient (AW) and respective 95% confidence intervals (CI), was used to assess the directional agreement between typing methods. This was calculated using an online tool available at This framework compares and relates multiple typing methods applied on diverse isolates of the same species (8, 36, 37). The online software calculates CI based on large sample approximation by using the resampling (jackknife pseudo-values) method, and also compares it with the original Non-Approximated Confidence Interval (CINA). A high value (close to 1.00) of AW indicates strong concordance between the resulting clusterings and provides statistical support to confirm clonality of the diverse isolates. RESULTS Characterization of V. parahaemolyticus isolates. Biochemical analysis using API20E and DrySlide oxidase tests identified the 100 isolates as V. parahaemolyticus. These analyses produced 23 different profiles, with 96 of the isolates displaying over 90% similarity to the species, and 74 isolates being identified with 99% confidence as V. parahaemolyticus. The other four isolates with weak identification by API20E were confirmed and listed as V. parahaemolyticus following positive test results from molecular (PCR) and various typing 8

9 analyses. Additionally, 27% of the strains were positive for urease activity by API20E (see Table S1 in the supplemental file). Overall, 99 of the 100 clinical V. parahaemolyticus isolates tested positive for the presence of a species-specific (tlh) amplicon by PCR (see Table S1 in the supplementary file). Conventional pathotypes of V. parahaemolyticus, harbouring only tdh gene sequences were less prominent among the 100 tested clinical strains: 15% of the isolates were positive only for tdh sequences by PCR, 22% were positive for the presence of trh sequences only, and 59% were positive for both tdh and trh. Four clinical strains (4%) of V. parahaemolyticus isolated between 2004 and 2009 were negative (by PCR) for both virulence markers, suggesting the presence of unknown or emerging pathogenic traits (see Table S1). Notably, 54 of 81 isolates which tested positive for trh sequences by PCR did not test positive for the presence of urease activity by API20E, i.e., only 27 isolates were positive for both trh sequences and urease activity (Table S1). Serotyping results. The top four serogroups which were found in 90% of the clinical V. paramaemolyticus isolates, included O4 (42%), O1 (30%), O3 (12%) and O6 (6%). A total of 27 (27%) isolates belonging to the O4 serogroup showed weak agglutination with the polyvalent antisera K group II, but failed to show convincing agglutination with any of the 7 monovalent antisera in that group, even after repeated testing with different batches of the antisera. These isolates, representing the most dominant serovar, were designated as O4:KII* (O4:KgroupII). The other prominent serovars, in descending order, were O1:KUT (16%), O3:K6 (8%), O4:KUT (8%) and O1:K56 (7%). This was in addition to 17 unique serotypes comprising various combinations of the somatic and capsular antigens. Overall, 22 serotypes among the 100 clinical isolates of V. parahaemolyticus (Table S1a, sorted by serotype) displayed a discriminatory power (SID 100 ) of 0.883, and when 18 potentially related isolates were excluded from the analysis (Table S1-1), the SID 82 value increased to (Table 2). Of the 22 serotypes, 7 were detected in multiple years, 2 in two consecutive years, 3 in two 9

10 non-consecutive years, and 10 were detected only once during the 10-year period (Table S1a). Serovar (O4:KII*) was very persistent during 2000 to 2009, and peaked in 2006, when 11 out of 20 clinical V. parahaemolyticus isolates belonged to this group. Ribopattern analysis. Ribotyping, which was done using an automated (DuPont) riboprinter, generated 28 ribotypes with high SID values (SID 100 = and SID 82 = 0.915, Table 2). These values reflect smaller groupings as a whole for each type, with the EcoVib S-4 type comprising 23% of all isolates. Seventeen ribopatterns were unique, i.e., they were found only once (Table S1b, sorted by BMH ribopattern). The two standard strains (NY477 and ATCC17802) also had unique ribopatterns (Fig. S1). Both our internal (BMH) and DuPont ribopattern designations were compared in this analysis. In all, 80 (80%) of our in-house (BMH) ribopatterns found matching designations in the DuPont ribopatterns database for V. parahaemolyticus (DUP-Vp). The other 20 isolates were given internal BMH ribopattern identities which clustered into 9 different types, and were listed as new Vp (new1vp to new9vp) in the DuPont column (see Table S1), awaiting new pattern designations from the company. Using the DuPont designations, 24 distinct types or ribopatterns were generated, five of which (DUP-20491Vp, DUP-20492Vp, DUP-20494Vp, DUP-6626Vp, new2vp) were found in 70% of the V. parahaemolyticus population (Table S1c, sorted by DuPont ribopattern). These large, dominant groups accounted for good SID values (SID 100 = and SID 82 = 0.881, Table 2) associated with ribotyping. The two top clusters of BMH ribopatterns, EcoVib S-4 and EcoVib S-3, corresponded with the DuPont ribopatterns, DUP-20492Vp and DUP-6626Vp, and these two clusters showed strong concordance with the serogroups O4 and O3, respectively. PFGE patterns. Among 100 isolates tested, 50 distinguishable SfiI patterns were observed, 35 (70%) of which were associated with single isolates (Table S1d). There was one large group of 24 isolates associated with pattern VPSFI.0001 followed by VPSFI.0003 with 9 isolates (Fig.S2a,c). Despite the presence of one large group, the SID values were high 10

11 (SID 100 = 0.932, and SID 82 = 0.971, Table 2). Similarly, 53 NotI patterns were observed overall, with patterns VPNTI.0001 (19 isolates) and VPNTI.0003 (9 isolates) being the most common (Fig.S2b,d). A total of 41 isolates had unique NotI types (Table S1e). PFGE using NotI was slightly more discriminatory than SfiI, yielding SID values of 0.95 (SID 100 ) and (SID 82 ), respectively (Table 2). A total of 57 pattern combinations were observed when the results from NotI and SfiI were combined. Pattern combinations VPSFI.0001/VPNTI.0001 (18 isolates) and VPSFI.0003/VPNTI.0003 (9 isolates) were the most common. All remaining pattern combinations were observed in 5 or less isolates, including 45 pattern combinations that were characteristically unique (Table S1f, sorted by [SfiI+NotI]PFGE). The resulting SID values were high, (SID 100 ) and (SID 82 ), respectively (Table 2). Distributions of the PFGE patterns are provided in the supplementary data (Fig. S2a for the SfiI-patterns, and Fig.S2b for the NotI-pattern), and the corresponding dendrograms are also in the same file (Fig. S2c,d), including those generated by the specific O groups and the primary (SfiI) enzyme (Fig. S2e). As expected, the major serogroups, O4 (n=42) and O1 (n=30) yielded significant diversity within the individual groups with SID values of 0.9 and acquired from 23 and 22 PFGE patterns, respectively. The other prominent serogroup, O3 (n=12), generated 11 patterns with a SID value of (Fig.S2e, Table S2 in the supplemental file). MLST analysis. Based on the seven house-keeping genes of V. parahaemolyticus, 27 sequence types (STs) were observed among the 100 isolates. The most common and significant STs among this Canadian study set were ST36 (n=32), ST417 (n=15), ST43 (n=13) and ST3 (n=9). There were 19 unique designations. A Minimum Spanning Tree (MST) of the STs was overlayed with the other subtype information, such as PFGE-patterns, ribopatterns and serotypes, to summarize the findings and display the congruence of the various fingerprints (Fig. 1). 11

12 All isolates in the ST36 cluster belonged to serogroup O4, 27 of which were O4:KII* serovars detected during the 10-year period, including 23 isolates displaying the same BMH ribopattern (EcoVib S-4) and matching DuPont ribopattern (DUP-20492Vp). Incidentally, 11 strains belonging to the serovar O4:KII* were isolated in 2006 and seven of them displayed identical patterns in each of the six typing schemes; their PCR profiles were also identical, and all seven were from BC. Four O4:KUT serovars of V. parahaemolyticus clustered with this ST36 and shared minimally divergent ribopatterns and PFGE patterns with the predominant serovar (O4:KII*) of this group. In addition, one isolate of V. parahaemolyticus O4:KUT (strain C143) was allotted a new designation, ST639; it was closely related to ST36 and formed a new clonal complex. This strain (C143) shared the same ribopattern with O4:KII* in ST36 (Fig. 1). The new MLST group, ST417 (n=15), included isolates from serogroup O1 only, including 10 of the 16 belonging to the pandemic serovar, O1:KUT, 8 of which were in the same clusters by all the typing methods tested (Fig.1). However, this ribopattern (EcoVib S1) was not found in the DuPont database, and was therefore designated as new2vp (Table S1). Apparently clonal, 6 of these isolates were from BC and 2 from AB. The other 6 isolates of O1:KUT scattered into 6 different STs (ST3, ST15, ST43, ST199, ST546 and ST635). The other significant group, ST3 (n=9), included 6 of the 8 pandemic O3:K6 V. parahaemolyticus strains, with an identical ribopattern (EcoVib S-3, DUP6626Vp) causing illness from 2003 to The other two O3:K6 strains generated slightly different ribopatterns (strain# with EciVib S-1 in ST3, and strain# with EcoVib S-3 in ST429) but clustered in the clonal complex (ST3 and ST429). The new sequence type, ST429, in the pandemic complex, shared the same ribopattern (DuPont designation, DUP-6626Vp) and serotype (O3:K6) with the predominant isolates in ST3, but differed in PFGE profiles (Table S1g). 12

13 Among the other groups, four different serovars fell into ST43 (n=13), O1:K56 (n=7), O1:KUT, O4:K63 (n=4) and O4:KUT, with 10 isolates having an identical ribopattern (EcoVib S-5 or DUP20491Vp), while ST50 (n=5) contained only serogroup O6 strains, all with an identical ribopattern (EcoVib S-8 or DUP6626Vp), and 4 of which were O6:K18 (Table S1g). Detection of clusters of V. parahaemolyticus using multiple molecular typing methods. There were two definitive clusters detected, each with indistinguishable BMH Ribopatterns and their corresponding DuPont designations, MLST sequence types (STs), SfiI PFGE types, and NotI PFGE types (Table S1h, sorted by year). A well-defined cluster of 8 cases from BC in 2006 yielded V. parahaemolyticus isolates with the combined type O4:KII*, EcoVib S-4, Dup20492Vp, ST36, VPSFI.0001, VPNTI One of the isolates differed by expressing the O4:KUT serotype instead of O4:KII*. The 12 other isolates from BC in 2006 were dissimilar by two or more types (Table S1). A single isolate with an indistinguishable combined type was identified in 2004 (1/11 isolates total for that year). Similar types were found in 2000 (2/5 isolates with indistinguishable ribopatterns and MLST, but differing in both PFGE types), in 2001 (3/4 isolates with indistinguishable serotypes, ribopatterns, and MLST, but differing in at least one PFGE type), in 2002 (3/5 isolates with indistinguishable serotypes, ribopatterns, and MLST, but differing in at least one PFGE type), in 2005 (1/11 isolates differing from the 2006 cluster only by NotI PFGE type), and 2008 (1/13 isolates with indistinguishable types from the 2006 cluster, 2/13 with indistinguishable serotype, ribopatterns, and MLST type, but having different PFGE types). A potential 2008 cluster in patients from BC and AB had common ribopatterns and a common ST (ST417), but differed in PFGE types (Fig. 1, and Table S1h, sorted by year). A second very well-supported cluster of cases attributed to V. parahaemolyticus was detected in 2009 in 8 patients from BC and AB. The combined type associated with the cluster was: O1:KUT, EcoVib S-1, new2vp, ST417, VPSFI.0003, VPNTI An additional 13

14 case from BC in 2009 differed only in ribopatterns, while four ST417 isolates associated with different ribopatterns and PFGE types were recovered in ST417 was not detected in any other isolates during the study period. Other possible clusters were smaller or were not as well supported, differing by one or more types, e.g., a cluster of human cases was found in the province of BC during the years 2005 (2/11 cases) and 2006 (3/20 cases) with the combined type EcoVib S-6, Dup20494Vp, ST36, VPSFI.0001, VPNTI Congruence analyses and Simpson s Index of Diversity (SID). The adjusted Wallace (AW) coefficient with 95% confidence intervals (95% CI) was calculated for the entire dataset of 100 isolates (Table 3). The analysis is directional, and influenced by the SID values (Table 2) of the typing methods. As seen in Table 3, AW PFGE(SfiI) MLST is 0.997, whereas in the reverse direction, AW MLST PFGE(SfiI) is 0.442; AW PFGE(NotI+SfiI) Serotype is 0.762, while AW Serotype PFGE(NotI+SfiI) generated a low value of DISCUSSION In this study, we performed serotyping, biotyping and molecular typing on a diverse set of 100 Canadian clinical V. parahaemolyticus isolates collected during the period from 2000 to This data will provide valuable baseline information for future studies initiated by the newly formed Vibrio Reference Service of Canada. Our analysis revealed substantial genetic diversity among the isolates tested, indicating that many of the illnesses may have been sporadic. Outbreaks of V. parahaemolyticus have been reported in several USA locations bordering Canada. In the summer of 1997, an outbreak associated with shellfish harvested from California, Oregon and Washington in the US (and BC in Canada) resulted in more than 200 illnesses (9, 14, 27, 35). The most common serovars among the implicated V. parahaemolyticus isolates were O4:K12 and O1:K56. In a 14

15 follow-up study, Kaufman et al. (24) reported that the V. parahaemolyticus strains potentially virulent to humans during the US Pacific Northwest outbreaks, belonged to serogroups O1 and O4. In our analyses, the majority (72%) of the clinical V. parahaemolyticus isolates from Canada belonged to the same two serogroups. Most of the strains belonging to serogroups O1 and O4 also tested positive for the hemolysin genes, trh or tdh, or both, by PCR (Table S1). Jones et al. (23) examined clinical isolates from across North America (July 2006 to November 2007) as well as oyster isolates collected across the US in 2007, and found that the predominant serovars were the pandemic strain O1:KUT, followed by O4:K12, the most common serotype observed in human illnesses from the Pacific Northwest (16, 23, 40). Among the Canadian clinical isolates of V. parahaemolyticus isolated from 2000 to 2009, serovar O4:KII* (O4:K12 is a serotype of this group) was the most common, followed by the pandemic serovar O1:KUT. Furthermore, the molecular typing data and congruence analysis presented here indicate that the O1 and O4 serogroups were associated with large clusters of cases that could have been identified as outbreaks, had they been followed up by epidemiological investigations. There were some interesting correlations of ST with geographical distribution. While STs 36, 417, 43, and 50 contained isolates from patients only in BC and AB, ST3 aggregated isolates from patients in BC, AB, and ON; it was the only grouping that contained isolates from widely divergent regions. All other isolates from provinces other than AB and BC had quite unique types. These differences suggest that MLST data may be useful for epidemiological investigations into the origins and distribution of V. parahaemolyticus. Regardless of the correlation observed between serogroup and ST, like O4 and ST36, O1 and ST417, O6 and ST50, there were some STs that contained a number of different serogroups. Significant among them were ST43 (O1 and O4), ST3 (O1 and O3), ST8 (O1 and O4) and ST141 (O3, O8 and OUT) as seen in Fig.1. 15

16 Notably, all 27 isolates belonging to serogroup O4:KII* fell into ST36. This ST is reported to be a genetically exclusive sequence type of a clinical nature (40), and ecologically persistent in the Pacific coasts of the US and Mexico (1, 23, 40). Based on the PubMLST database, the distribution of ST36 appears to be endemic to the Pacific coast (1, 40) and clonally related to the O4:K12 serovar. One illness involving V. parahaemolyticus O4:K12 was known to have occurred in Japan in 1998 (16). Since all of our O4:KII* isolates of V. parahaemolyticus belonged to this ST36 group, it is possible that they are in the process of serovar-transition, and thus were not being specifically typed by our stringent criteria for agglutination. The pandemic O3:K6 serotype of V. parahaemolyticus is known to have serovariants based on similarities of molecular fingerprinting studies (28, 31). These reports have identified several serotypes, including O1:KUT and O6:K18, which have spread around the world and which may have diverged from O3:K6 by alteration of the O:K antigens. MLST data have further supported the finding that multiple serotypes can occur in a single genetic lineage (12, 28). Since V. parahaemolyticus is prone to recombination events (18), it is likely that divergence is a strategy to survive the host defense mechanisms. This could explain why some of the STs clustered in multiple serogroups. These serovariants with slightly altered fingerprinting patterns could become important strains for follow-up studies examining the mechanisms of pathogenesis in V. parahaemolyticus, including association with, and impact of, the human microbiome and global warming. In 2004, a follow-up investigation after an outbreak of V. parahaemolyticus infection in a cruise ship linked to the consumption of raw oysters from Alaska, identified a total of 62 cases (29). The majority of the clinical V. parahaemolyticus isolates were identified as serotype O6:K18. In our data, three out of 9 clinical V. parahaemolyticus isolates from Canada in 2003 were identified as serotype O6:K18, with identical STs (ST50) and ribopatterns (EcoVib S-8, DUP6626Vp), (Table S1). Further investigation with genotypic analysis is required to find out if there is any association between the Canadian cases and the Alaskan outbreak. 16

17 In 2006, 177 cases of V. parahaemolyticus occurred in the US, Pacific NW and New York state (10), including 122 cases that were associated with 17 clusters, all of which were linked to contaminated oysters and clams from harvest sites in Washington, USA, and BC, Canada. Subtyping of some of the V. parahaemolyticus isolates indicated the persistence of serovar O4:K12, which probably has some relevance with our finding of the O4:KII* serogroup in 11 out of the 20 clinical V. parahaemolyticus strains isolated in 2006 (Table S1). Serologically, O4:K12 is a serotype which belongs to serogroup O4:KII*, and both serovars fell into ST36 (40, this study). Deviations from the 2006 cluster detected in other years could indicate diversification of a bacterial population through genetic recombination, etc. These observations are consistent with the continued presence of a pool of closely related isolates in the environment throughout several years that resulted in 2006, in a predominant cluster of human illnesses due to this particular V. parahaemolyticus strain. The decrease in predominance of this strain after 2006, suggests that it may be in the process of being replaced (serovar-transition) by other types within environmental populations. It was found that among the strains isolated in 2009, 9 out of 14 originated from Alberta and BC and belonged to serotype O1:KUT, which is classified as a pandemic serovar. This could be an indication of a disease cluster that may have occurred and escaped detection due to a lack of reporting and follow-up investigation. A significant finding in our study was the identification of four (4%) clinical V. parahaemolyticus strains which were PCR-negative for the virulence genes (tdh and trh), indicating the existence of unknown virulence factors or possibly the emergence of new virulence traits. These four strains will be useful for further research on virulence factors contributing to human pathogenicity. A similar observation was also reported recently by Jones et al. (23) in the US, who found 27% of clinical isolates they examined to be negative for the hemolysin genes, tdh and trh. The need to investigate and identify other (novel) virulence traits continues to exist and was also realised by another group (35). 17

18 Cases of Vibrio spp. infection appear to be increasing globally. For example, since FoodNet surveillance began in the US in 1996, a significant contrast in the incidence of Vibrio infections was observed, whereby an early period up to 1998 was used for comparison with a similar period from 1999 to The relative rates of infection during this latter time period decreased for most of the foodborne pathogens, i.e., Listeria, Campylobacter, E. coli O157, but Vibrio infections increased by 47% (CI=9%-99%) (11). Co-ordinated national surveillance is needed to increase reporting of illnesses caused by non-cholera vibrios, particularly V. parahaemolyticus, so that we can try and obtain a true estimate of the burden of disease in Canada. Data from the present study has shown that the majority of V. parahaemolyticus clinical isolates came from the West coast of Canada, particularly BC (82%), indicating that awareness and reporting may be non-uniform. It should be noted that despite the regional bias, clinical V. parahaemolyticus isolates were heterogeneous in nature, consistent with the fact that the isolates were acquired as sporadic and unrelated cases of vibriosis. A combination of several sub-typing methods and testing for species-specific virulence markers can be used for epidemiological source-tracking of isolates involved in seafoodborne outbreaks, or to detect a sudden increase in sporadic cases of V. parahaemolyticus. This is based on the fact that some of the serotypes of clinical V. parahaemolyticus, presumably coming from sporadic cases, were probably sourced to a common food, and coincidentally linked to reported outbreaks in the neighboring regions, e.g., V. parahaemolyticus serovar O6:K18 in 2004 (29), O4:K12, O1:KUT and O1:K56 in 2006 (23, 29), and O1:KUT in 2009 (this study). This study reflects the possibility of unreported outbreaks associated with strains which appeared clonal in our congruence analysis and comparable with the predominant strains involved in outbreaks in the neighboring regions, particularly the Pacific-coasts of the US and Mexico (1, 23, 40). 18

19 When used as the method against which all other methods were compared, serotyping results showed fairly good congruence only with MLST data. In the reverse direction, serotyping demonstrated good congruence with PFGE and moderate congruence with MLST. MLST was very congruent with all three PFGE protocols. In the reverse order, the probability was moderate to low. As expected, all three PFGE typing protocols exhibited moderate to very high AW coefficients when compared against each of the three PFGE typing patterns. Similarly, the BMH and DuPont ribopattern designations were highly congruent with each other. It is to be noted that all the typing protocols, parse the V. parahaemolyticus isolates somewhat differently and therefore, provide additional discrimination. In summary, this study was able to characterize and identify diverse strains of Canadian clinical V. parahaemolyticus, and point to the congruence in diversity observed when the subtypes from the various methods were combined in the analysis, indicating a strategy that could be useful for source-tracking in outbreak situations. 19

20 Acknowledgement We would like to thank Dr. Brent Dixon and Dr. Franco Pagotto, of BMH research division of Health Canada for peer-reviewing the manuscript. Thanks to Ms. Jennifer Liu of CFIA, Burnaby, BC, for sharing some of the clinical strains from their stock. We are also thankful for the patience and support of all members of the Vibrio Reference Service of Canada (VRS), comprising researchers from Health Canada, Ottawa, Ontario, Public Health Agency of Canada, Winnipeg, Manitoba, and Canadian Food Inspection Agency of Canada, Burnaby, British Columbia, in addition to collaborators and experts from various academic institutions in Canada. Downloaded from on August 22, 2018 by guest 20

21 481 REFERENCES Abbot SL, Powers C, Kaysner CA, Takeda Y, Ishibashi M, Joseph SW, Janda JM Emergence of a restricted bioserovar of Vibrio parahaemolyticus as the predominant cause of Vibrio-associated gastroenteritis on the west coast of the United States and Mexico. J. Clin. Microbial., 27: Banerjee SK, Farber JM Susceptibility of Vibrio parahaemolyticus to trisdependent DNA degradation during pulsed-field gel electrophoresis. J. Clin. Microbiol., 47: Banerjee SK, Pandian S, Todd EC, Farber JM A rapid and improved method for the detection of Vibrio parahaemolyticus and Vibrio vulnificus strains grown on hydrophobic grid membrane filters. J. Food Prot., 65: Bej AK, Patterson DP, Brasher CW, Vickery MCL, Jones DD, Kaysner CA Detection of total and hemolysin-producing Vibrio parahaemolyticus in shellfish using multiplex PCR amplification of tl, tdh and trh. J. Microbiol. Methods, 36: Brasher CW, DePaola A, Jones DD, Bej AK Detection of microbial pathogens in shellfish with multiplex PCR. Curr. Microbiol., 37: Bruce JL Automated system rapidly identifies and characterizes microorganisms in food. Food Technol., 50: Bruce JL, Hubner RJ, Cole EM, McDowell CI, Webster JA Sets of EcoRI fragments containing ribosomal RNA sequences are conserved among different strains of Listeria monocytogenes. Proc. Natl. Acad. Sci. USA, 92: Carrico JA, Silva-Costa C, Melo-Cristino J, Pinto FR, de Lencastre H, Almeida JS, and Ramirez M Illustration of a common framework for relating multiple typing methods by application to macrolide-resistant Streptococcus pyogenes. J. Clin. Microbiol., 44:

22 Centers for Disease Control and Prevention Outbreak of Vibrio parahaemolyticus infections associated with eating raw oysters, Pacific Northwest, MMWR. 47: Centers for Disease Control and Prevention Vibrio parahaemolyticus infections associated with consumption of raw shellfish---three States, MMWR, 55: Centers for Disease Control and Prevention Preliminary FoodNet data on the incidence of infection with pathogens transmitted commonly through food---10 States, MMWR, 58: Chowdhury NR, Stine OC, Morris JG, and Nair GB Assessment of evolution of Vibrio parahaemolyticus by multilocus sequence typing. J. Clin. Microbiol., 42: Croci L, Suffredini E, Cozzi L Evaluation of different polymerase chain reaction methods for the identification of Vibrio parahaemolyticus strains isolated by cultural methods. J. AOAC Int., 90: Daniels NA, MacKinnon L, Bishop R, Altekruse S, Ray B, Hammond RM, Thompson S, Wilson S, Bean NH, Griffin PM, Slutsker L Vibrio parahaemolyticus infections in the United States, J. Infect. Dis., 181: Davies CR, Wingfield DL, Peak KK, Veguilla W, Amuso PT, Cannons AC, Cattani J Molecular characterization of Vibrio parahaemolyticus strains associated with foodborne illness in Florida. J. Food Prot., 70: DePaola A, Ulaszek J, Kaysner CA, Tenge BJ, Nordstrom JL, Wells J, Puhr N, Gendel SM Molecular, serological, and virulence characteristics of Vibrio parahaemolyticus isolated from environmental, food, and clinical sources in North America and Asia. Appl. Environ. Microbiol., 69: Fyfe M, Yeung ST, Daly P, Schallie K, Kelly MT, Buchanan, S Outbreak of Vibrio parahaemolyticus related to raw oysters in British Columbia. Can. Commun. Dis. Rep., 23:

23 Gonzalez-Escalona N, Martinez-Urtaza J, Romero J, Espejo RT, Jaykus LA, and DePaola A Determination of molecular phylogenetics of Vibrio parahaemolyticus strains by multilocus sequence typing. J. Bacteriol., 190: Hollis RJ, Bruce JL, Fritschel SJ, Pfaller MA Comparative evaluation of an automated ribotyping instrument versus pulsed-field gel electrophoresis for epidemiological investigation of clinical isolates of bacteria. Diagn. Microbiol. Infect. Dis., 34: Hunter P Reproducibility and indices of discriminatory power of microbial typing methods. J. Clin. Microbiol., 28: Hunter PR, Gaston MA Numerical index of the discriminatory ability of typing systems: an application of Simpson s index of diversity. J. Clin. Microbiol., 26: Jolley KA, Chan M-S, Maiden MCJ mlstdbnet distributed multi-locus sequence typing (MLST) databases. BMC Bioinformatics, 5:86 [ vparahaemolyticus/] 23. Jones JL, Ludeke CHM, Bowers JC, Garrett N, Fischer M, Parsons MB, Bopp CA, DePaola A Biochemical, serological, and virulence characterization of clinical and oyster Vibrio parahaemolyticus isolates. J. Clin. Microbiol., 50: Kaufman GE, Myers ML, Pass CL, Bej AK, Kaysner CA Molecular analysis of Vibrio parahaemolyticus isolated from human patients and shellfish during US Pacific north-west outbreaks. Lett. Appl. Microbiol., 34: Kaysner CA, DePaola A Vibrio. In: Bacteriological Analytical Manual, 8 th Edition, Revision A, May Chapter 9. Published by AOAC International, Gaithersburg. MD. 26. Khaira G, Galanis E Descriptive epidemiology of Vibrio parahaemolyticus and other Vibrio species. Can. Commun. Dis. Rep., 33: Marshall S, Clark CG, Wang G, Mulvey M, Kelly MT, Johnson WM Comparison of methods for typing Vibrio parahaemolyticus. J. Clin. Microbiol., 37:

24 Matsumoto C, Okuda J, Ishibashi M, Iwanaga M, Garg P, Ramamurthy T, Wong HC, DePaola A, Kim YB, Albert JM, and Nishibuchi M Pandemic spread of an O3:K6 clone of Vibrio parahaemolyticus and emergence of related strains evidenced by arbitrarily primed PCR and toxrs sequence analyses. J. Clin. Microbiol., 38: McLaughlin JB, DePaola A, Bopp CA, Martinek KA, Napolilli NP, Allison CG, Murray SL, Thompson EC, Bird MM, Middaugh JP Outbreak of Vibrio parahaemolyticus gastroenteritis associated with Alaskan oysters. N. Engl. J. Med., 353: Morris JG Jr Cholera and other types of vibriosis: A story of human pandemics and oyster on the half shell. Clin. Infect. Dis., 37: Nair GB, Ramamurthy T, Bhattacharya SK, Dutta B, Takeda Y, and Sack DA Global dissemination of Vibrio parahaemolyticus serotype O3:K6 and its serovariants. Clin. Microbiol. Rev., 20: National Enteric Surveillance Program (NESP). Annual reports and on-line data can be obtained by sending a request to NML.Enterics@phac-aspc.gc.ca available from the home page at: Nishibuchi M, Kaper JB Nucleotide sequence of the thermostable direct hemolysin gene of Vibrio parahaemolyticus. J. Bacteriol., 162: Overman TL, Kessler JF, and Seabolt JP Comparison of API-20E, API-RapidE and API Rapid NFT for identification of members of the family Vibrionaceae. J. Clin. Microbiol., 22: Paranjpye R, Hamel OS, Stojanovski A, and Liermann M Genetic diversity of clinical and environmental Vibrio parahaemolyticus strains from the Pacific Northwest. Appl. Environ. Microbiol., 78: Pinto FR, Melo-Cristino J, and Ramirez M A confidence interval for the Wallace coefficient of concordance and its application to microbial typing methods. PLoS One, 3:e

25 Severiano A, Pinto FR, Ramirez M, and Carrico JA Adjusted Wallace coefficient as a measure of congruence between typing methods. J. Clin. Microbiol., 49: Spite GT, Brown DF, Twedt RM Isolation of an enteropathogenic, Kanagawapositive strain of Vibrio parahaemolyticus from seafood implicated in acute gastroenteritis. Appl. Environ. Microbiol., 35: Thomson WK, Trenholm DA The isolation of Vibrio parahaemolyticus and related halophilic bacteria from Canadian Atlantic shellfish. Can. J. Microbiol., 17: Turner JW, Paranjpye RN, Landis ED, Biryukov SV, Gonzalez-Escalona N, Nilsson WB, and Strom MS Population structure of clinical and environmental Vibrio parahaemolyticus from the Pacific Northwest coast of the United States. PLoS One, 8:e Downloaded from on August 22, 2018 by guest 25

26 Fig.1 Minimum spanning tree (MST) of multilocus sequence typing (MLST) of 100 clinical V. parahaemolyticus strains grouped by sequence types (ST), and combined with other types, such as ribotypes, pulsed-field gel electrophoresis (PFGE) patterns of chromosomal DNA digests (NotI, SfiI) and serotypes (color coded). MST was created using Bionumerics V6.5 and ST numbers are indicated in the middle of the nodes. Lines infer relatedness by stating the number of differing alleles, and an unattached node has no alleles in common with the STs in tree. Node size is related to the number of isolates, and grey border around nodes indicates clonal complex (CC). Downloaded from on August 22, 2018 by guest 26

27 Table 1. Oligonucleotide primer sequences targeting seven house-keeping genes and used for both DNA amplification and sequencing Oligonucleotide Target Sequence (5 to 3 ) Product size (bp) reca-1f reca GAAACCATTTCAACGGGTTC 773 reca-1r CCATTGTAGCTGTACCAAGCACCC gyrb-1f gyrb GAAGGBGGTATTCAAGC 629 gyrb-1r GAGTCACCCTCCACWATGTA dnae-1f dnae CGRATMACCGCTTTCGCCG 596 dnae-1r GAKATGTGTGAGCTGTTTGC dtds-1f dtds TGGCCATAACGACATTCTGA 497 dtds-1r GAGCACCAACGTGTTTAGC pnta-1f pnta ACGGCTACGCAAAAGAAATG 470 pnta-1r TTGAGGCTGAGCCGATACTT pyrc-1f pyrc AGCAACCGGTAAAATTGTCG 533 pyrc-1r CAGTGTAAGAACCGGCACAA tnaa-1f tnaa TGTACGAAATTGCCACCAAA 463 tnaa-1r AATATTTTCGCCGCATCAAC 27

28 Table 2. Simpson s Index of Diversity for all 100 V. parahaemolyticus isolates (SID 100 ) with 95% confidence interval (CI) as well as non-approximated confidence interval (CINA), and SID 82 after removing 18 isolates likely to be epidemiologically related, and showing the corresponding CIs, for each method. Name PFGE (NotI + SfiI) No. of unique patterns/types SID PFGE (NotI) PFGE (SfiI) BMH RiboPattern DuPont RiboPattern Serotype MLST CI (95%) ( ) ( ) ( ) ( ) ( ) ( ) ( ) CINA (95%) ( ) ( ) ( ) ( ) ( ) ( ) ( ) SID CI (95%) ( ) ( ) ( ) ( ) ( ) ( ) ( ) CINA (95%) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 28

29 Table 3. Adjusted Wallace Coefficient (AW) and 95% CI for the dataset of 100 V. parahaemolyticus isolates. AW values in bold (grey cells) represent quite good congruence between the methods shown, and generally lie between 0.7 1, while coefficients with values below 0.4 demonstrate low congruence. The comparisons are directional, following the convention, AW Column 1 type Row 1 type, to determine the probability of congruence. Typing scheme Serotype BMH RiboPattern DuPont RiboPattern MLST PFGE (SfiI) PFGE (NotI) PFGE (NotI+SfiI) Serotype ( ) ( ) 0.59 ( ) ( ) ( ) ( ) BMH RiboPattern ( ) ( ) ( ) ( ) ( ) ( ) DuPont RiboPattern ( ) 1 ( ) ( ) ( ) ( ) ( ) 29 MLST PFGE (SfiI) PFGE (NotI) ( ) ( ) ( ) ( ) ( ) ( ) ( ) 0.36 ( ) ( ) ( ) ( ) 1 ( ) ( ) 0.26 ( ) ( ) ( ) ( ) 1 ( ) PFGE (NotI+SfiI) ( ) ( ) ( ) ( ) ( ) ( )

30