Vibrio cholerae O1 El Tor strains linked to global cholera show region-specific patterns by pulsed-field gel electrophoresis

Abstract

Vibrio cholerae O1 El Tor, causative agent of the ongoing seventh cholera pandemic, is native to the aquatic environment of the Ganges Delta, Bay of Bengal (GDBB). Recent studies traced pandemic strains to the GDBB and proposed global spread of cholera had occurred via intercontinental transmission. In the research presented here, NotI-digested genomic DNA extracted from V. cholerae O1 clinical and environmental strains isolated in Bangladesh during 20042014 was analyzed by pulsed-field gel electrophoresis (PFGE). Results of cluster analysis showed 94.67% of the V. cholerae strains belonged to clade A and included the majority of clinical strains of spatio-temporal origin and representing different cholera endemic foci. The rest of the strains were estuarine, all environmental strains from Mathbaria, Bangladesh, and occurred as singletons, clustered in clades B and C, or in the small clades D and E. Cluster analysis of the Bangladeshi strains and including 157 El Tor strains from thirteen countries in Asia, Africa, and the Americas revealed 85% of the total set of strains belonged to clade A, indicating all were related, yet did not form an homogeneous cluster. Overall, 15% of the global strains comprised multiple small clades or segregated as singletons. Three sub-clades could be discerned within the major clade A, reflecting distinct lineages of V. cholerae O1 El Tor associated with cholera in Asia, Africa, and the Americas. The presence in Asia and the Americas of non-pandemic V. cholerae O1 El Tor populations differing by PFGE and from strains associated with cholera globally suggests different ecotypes are resident in distant geographies.

1. Introduction

Cholera, with seven pandemics reported to date, represents a significant chapter in human history and infectious disease. The acute form of diarrhea caused by Vibrio cholerae is related to production of a toxin that triggers the characteristic water loss and severe dehydration of cholera (Harris et al., 2012). Cholera remains a threat in many countries, notably where access to safe drinking water is limited. Bangladesh is a developing country where cholera is endemic, with two annual peaks in some regions of the country (Alam et al., 2011). The estimated global burden of cholera is 2.86 million cases and 95,000 deaths. In Bangladesh alone, approximately 100,000 new cases are diagnosed each year, resulting in 3,272 deaths (Ali et al., 2015). V. cholerae strains are widely distributed globally in many coastal, estuarine, and brackish water ecosystems as free-living bacterial cells or associated with zooplankton, namely copepods (Colwell and Huq, 1994; Huq et al., 2005; Lutz et al., 2013). Brackish waters in coastal areas support bacterial populations, with environmental stimuli favorable for bacterial growth prompting cholera outbreaks (Huq et al., 2005).

V. cholerae O1 is classified into two biotypes, classical and El Tor, based on phenotypic and genetic differences (Kaper et al., 1995; Nair et al., 2002, 2006). The seventh and ongoing pandemic is attributed to the El Tor biotype of V. cholerae O1 (Hu et al., 2016; Weill et al., 2017). Beginning in 1961 in Indonesia, the seventh pandemic of cholera included Africa in 1970, Latin America in 1991, and more recently Haiti and Yemen (Choi et al., 2016; Domman et al., 2017; Eppinger et al., 2014; Qadri et al., 2017). After the initial cases occurred, most of these regions continued to suffer episodes of cholera. There is debate whether the recurrent outbreaks of cholera in Africa and the Americas are caused by distinct intercontinental introduction or resurgence of indigenous clones. A few recent investigations investigated the genetic homogeneity of the 7th pandemic strains and connected both origin and recurrent transmission to a single source, the Bay of Bengal (Mutreja et al., 2011). However, the co-existence of several local lineages, along with the pandemic clones and their regional evolution, has painted a very complex picture of bacterial population dynamics, especially in and out of endemic settings (Codeço, 2001). Thus, monitoring pandemic and local strains became a priority for some investigators, as a means of developing an effective public health model and strategy for controlling the current pandemic and preventing future pandemics of cholera.

Whole genome typing methods, e.g., multi-locus sequence typing (MLST), multi-locus variant analysis (MLVA), ribotyping, random amplification of polymorphic DNA (RAPD), and pulsed- field gel electrophoresis (PFGE), have been employed to differentiate strains and monitor transmission routes (Leal et al., 2004; Maiden et al., 1998; Okada et al., 2012; Rahaman et al., 2015). PFGE is a DNA fingerprinting method that can discriminate bacterial strains. Before introduction of whole genome sequencing (WGS), epidemiological studies of cholera relied on PFGE (Rahaman et al., 2015). An earlier study highlighted the value of PFGE in revealing clonality among strains from two well-defined cholera outbreaks in Malaysia (Mahalingam et al., 1994). Intrinsic limitations include restriction digestion being skewed by mobile elements, hence restricted value for phylogeny. Although PFGE does not provide as high resolution as WGS, its stability and reproducibility allow rudimentary, yet comprehensive, analysis of ancestry (Rahaman et al., 2015).

In the study reported here, the objective was to understand both regional diversity and global distribution of V. cholerae O1 El Tor representing seventh pandemic lineage. Therefore, strains in Bangladesh isolated over a decade were compared with strains from thirteen countries across Asia, Africa, and the Americas. Both environmental and clinical strains were included since the local environment can influence strains, persisting and becoming epidemic, as well as providing access to autochthonous strains of V. cholerae El Tor.

2. Materials and methods

2.1. Geographical profile of strains

A total of 169 strains were collected between 1991 and 2014 from four endemic sites in Bangladesh: Mathbaria (n = 99); Dhaka (n = 38); Chhatak (n = 31); and Matlab (n = 1). Of these, 119 and 50 were of clinical and environmental origin, respectively. To investigate phylogenetic relationships, an additional 157 strains (150 clinical and 7 environmental) were collected from 13 countries across Asia, Africa, and Latin America [Nepal 39, Thailand 32, Vietnam 15, Pakistan 3, India 2, Sri Lanka 1, Zambia 9, Zimbabwe 12, Mozambique 2, Mexico 34, Brazil 2, Peru 3, and Haiti 3] (Table 1). Despite variations in the gene contents, all strains were V. cholerae O1 biotype El Tor as confirmed by the phenotypic characteristics (Kaper et al., 1995) and biotype-specific genes (Nair et al., 2002, 2006). Detailed information for the strains is provided in Tables 1 and 2, and Supplementary Table S1.

Table 1.

Geographic source of V. cholerae O1 El Tor strains included in this study.

Country	No. of strains	Source		Serotype
		Environmental	Clinical	Inaba	Ogawa
Bangladesh	169	50	119	33	136
Brazil	2	-	2	2	-
Haiti	3	-	3	-	3
India	2	-	2	2	-
Mexico	34	-	34	17	17
Mozambique	2	-	2	-	2
Nepal	39	6	33	-	39
Pakistan	3	-	3	1	2
Peru	3	-	3	-	3
Sri Lanka	1	-	1	-	1
Thailand	32	1	31	17	15
Vietnam	15	-	15	-	15
Zambia	9	-	9	-	9
Zimbabwe	12	-	12	4	8
	n = 326

Table 2.

V. cholerae genotypes based on ctxB, rstR, and tcpA genes.

Countries	ctxB			rstR				tcpA
	Classical	El Tor	Negative	Classical	El Tor	Both	Negative	Classical	El Tor	Negative
Bangladesh	167	2	-	7	162	-	-	-	169	-
Nepal	39	-	-	-	39	-	-	-	39	-
Thailand	27	-	5	9	18	-	5	-	32	-
India	2	-	-	-	2	-	-	-	2	-
Pakistan	3	-	-	-	3	-	-	-	3	-
Sri Lanka	1	-	-	-	1	-	-	-	1	-
Vietnam	15	-	-	-	15	-	-	-	15	-
Zambia	9	-	-	-	9	-	-	-	9	-
Mozambique	2	-	-	2	-	-	-	-	2	-
Zimbabwe	12	-	-	-	12	-	-	-	12	-
Mexico	5	16	13	-	16	5	13	1	26	3
Brazil	-	2	-	-	2	-	-	-	2	-
Peru	-	3	-	-	3	-	-	-	3	-
Haiti	3	-	-	-	3	-	-	-	3	-

^*PCR results for four of the Mexican V. cholerae O1 El Tor strains included in this study were inconsistent, and hence, their tcpA type could not be determined conclusively and not included in this table.

2.2. Pulsed-field gel electrophoresis (PFGE)

Whole agarose-embedded genomic DNA was prepared from each strain. PFGE was carried out using a contour-clamped homogeneous electrical field (CHEF-DRII) apparatus (Bio-Rad), following procedures described previously (Cameron et al., 1994). Conditions for separation were as follows: 2 to 10s for 13 h, followed by 20 to 25 s for 6 h. An electrical field of 6 V/cm was applied at an included field angle of 120°. Genomic DNA of the test strains was digested by NotI restriction enzyme (Gibco-BRL, Gaithersburg, MD) and Salmonella enterica serovar Braenderup was digested using XbaI, with fragments employed as molecular size markers. Restriction fragments were separated in 1% pulsed-field- certified agarose in 0.5× TBE (Tris-borate-EDTA) buffer. Post- electrophoresis gel-treatment included gel staining and de-staining. DNA was visualized using a UV transilluminator and images were digitized using a one-dimensional gel documentation system (Bio-Rad).

2.3. Image analysis

The fingerprint pattern in each gel was analyzed using a computer software package, Bionumeric (Applied Maths, Belgium). After background subtraction and gel normalization, the fingerprint patterns were typed according to banding similarity and dissimilarity, using the Dice similarity coefficient and unweighted-pair group method employing average linkage (UPGMA) clustering, as recommended by the manufacturer. The results were graphically represented as dendrograms.

3. Results

The NotI restriction enzyme digested genomic DNA of the test strains into 20 to 23 fragments and the molecular sizes of the DNA fragments ranged from 20 to 350 kb. Digested genomic DNAs of different spatiotemporal origin and resulting different biotype categorizations were subjected to PFGE. The resulting band patterns were analyzed by Dice similarity coefficient and UPGMA clustering methods to determine genetic and ancestral relatedness.

3.1. Local diversity and distribution of the Bangladesh strains

In a dendogram obtained by UPGMA analysis of DNA band patterns, the Bangladesh strains comprised five different clades, A, B, C, D, and E (Figs. 1 and 2). Of the 169 strains, 160 clustered in clade A, suggesting a single lineage. A single clinical strain, collected in 1991 from Matlab, clustered with clade A, a clade of predominantly clinical strains from endemic sites, including estuary villages in Bangladesh. While a few environmental strains were found to cluster in clade A, other environmental strains collected in 2012 from Mathbaria exhibited different PFGE patterns and did not join clade A (Supplementary Fig. S1). Strains with different pulsotypes included three singletons and a small clade [B (1); C(1); D(6); E(1)] (Fig. 1). Singletons in clades B, C, and E were strains from the aquatic environment. It should be noted that clade D included strains of both clinical and environmental origin. All clade D strains possessed rstR classical biotype, a characteristic limited to this clade (Supplementary Table S1). Most of the Chhatak strains (27 of 31) had the same pulsotype closely related to V. cholerae strains from Dhaka and Mathbaria (Supplementary Fig. S1).

Fig. 1.

Clonal diversity and geographical distribution of strains from Bangladesh.

Strains belonging to clade A were found in all four areas. Clade A contained both clinical and environmental strains. In addition to clade A, strains of other clades were found, but only in Mathbaria.

Fig. 2.

PFGE analysis of the strains collected from Bangladesh.

A total of 169 strains were analyzed which resulted in 5 clades. Clade A contained 94.67% of the strains. The rest 9 strains which were only found in Mathbaria formed the other 4 groups (B, C, D and, E).

3.2. Global distribution of clones

When the PFGE banding patterns of V. cholerae O1 strains from Bangladesh were compared with those of 157 strains collected from thirteen countries across Asia, Africa, and Latin America, the strains could be differentiated into 16 clades; A through P (Fig. 3). Clade A strains from Bangladesh clustered with 133 of the 293 strains, comprising a majority of strains from 14 countries and three continents (Figs. 3 and 4). Hence, a majority of the V. cholerae O1El Tor strains from different geographical regions revealed a similar PFGE pattern and fell into a major clade, with country-specific sub-clustering, i.e., subclades within the major clade A. The sub clades of A included V. cholerae O1 El Tor strains from Vietnam (n = 15), Zambia (n = 9), Haiti (n = 3), India (n = 2), Pakistan (n = 3), Sri Lanka (n = 1), and most of the strains from Zimbabwe (8 of 12) and Nepal (27 of 39), Fig. 4. Three subclades within clade A reflected a broader spatial distinction and 202 of 273 (74%) Asian and African strains comprising subclade Ia and 60 of 273 (22%) in subclade Ib. Latin American strains (16 of 21) in clade A comprised subclade II. Interestingly, the Latin American strains were predominantly V. cholerae O1 prototype El Tor (Supplementary Table S1). A few strains from Bangladesh, Thailand and one from Zimbabwe fell into sub- clade II (Fig. 4). Subclade II V. cholerae O1 El Tor strains from Mexico, isolated during 1992 to 1999, were located in a branch, separating them from strains collected between 2004 and 2008. Three V. cholerae O1 El Tor strains isolated during the 2010 Haitian cholera outbreak comprised subclade Ia, with Asian and African strains. Two V. cholerae O1 El Tor strains isolated in Bangladesh during 2011 joined with V. cholerae O1 El Tor strains from Peru, Brazil, and Mexico, based on PFGE, Figs. 3 and 4.

Fig. 3.

Comparison of band patterns between strains from Bangladesh and other countries. The strains from Bangladesh were compared with 157 additional strains collected from 13 other countries. The resulting phylogenetic tree represents 16 groups. Clade A contained 89.88% of the total strains including 160 strains of Bangladesh.

Fig. 4.

Global phylogeny of strains based on PFGE pattern.

Countries are represented by color. 160 strains of Bangladesh formed clade A, with 133 strains isolated from other countries. Three subclades were observed in clade A: Subclade Ia and Ib contained strains mostly of African and Asian origin. Subclade II comprised predominantly Latin American strains. Country subclusters were also observed.

Aside from clade A, the clade E strains from Bangladesh shared PFGE pattern with an El tor strain from Peru and two from Zimbabwe (Fig. 5). As with the Bangladesh strains, locally restricted diversity was observed for strains from Mexico, reflected by 11 distinct groups in addition to clade A. Groups F–P comprised strains from Mexico, notably those collected during 2000 to 2004 (Fig. 4).

Fig. 5.

Global distribution of clones.

4. Discussion

In many epidemiological studies, pulsed-field gel electrophoresis (PFGE) is used to discern source attributes of strains from different outbreaks. With the advent of whole genome sequencing and comparative genomics, the once gold standard PFGE is less appreciated as a DNA fingerprinting tool of epidemiological implication. In the study reported here, PFGE analysis of V. cholerae O1 El Tor strains associated with endemic cholera in Bangladesh and thirteen countries of Asia, Africa and the America showed the strains to be related genetically, but not homogenous globally. A majority of the strains comprised a major clade, with divergence noted for non-pandemic and environmental strains from the Bay of Bengal, Bangladesh and strains from the Gulf of Mexico. According to a recent WGS-based study of a restricted subset of V. cholerae clones, those strains were responsible for epidemic cholera worldwide (Lan and Reeves, 2002; Moore et al., 2014). In this study, PFGE was used to analyze clinical strains of V. cholerae from different geographical locations and with different genetic and phenotypic characteristics.

V. cholerae O1 El Tor biotype has dominated clinically over the classical biotype since 1961, the latter having last been isolated in Bangladesh in 1992 (Faruque et al., 1993; Kaper et al., 1995). Observed co-existence of the two biotypes for such a long time likely resulted in the hybrid characteristics of El Tor with classical biotype attributes, as observed in Bangladesh (Nair et al., 2002, 2006). In this study, the majority of V. cholerae O1 El Tor strains from clinical and environment sources comprised a major clade, suggesting similarity of strains from both sources and associated with epidemics in Bangladesh. Some of the environmental strains did not fall into clade A, suggesting those to be genetically divergent pulsotypes present in a diverse population existing in the environment. Environmental V. cholerae O1 El Tor in our study, with a few exceptions, were similar to clinical strains in clade A. Previous epidemiological surveillance conducted in the Bay of Bengal estuary has shown pathogenic strains can be detected in aquatic habitats, either in the culturable or non-culturable state, depending on the season (Alam et al., 2007).

A major genome-based study postulated the pandemic V. cholerae strain originated in Bay of Bengal villages of Asia and transmitted world-wide in three different waves (Mutreja et al., 2011). It was concluded that V. cholerae O1 El Tor has the ability to travel inter-continentally and adapt to its place of introduction by sharing niches with existing microflora in coastal and estuarine regions, including the Gulf coast of Mexico (Kaper et al., 1982). Notwithstanding the fact that outbreaks occurring after introduction can be attributed to V. cholerae and the pathogen can be introduced repeatedly or adapt locally, either is possible. Whole genome sequencing based studies linked epidemics in Africa and the Americas to multiple introduction events, rather than preexisting pathotypes (Domman et al., 2017; Weill et al., 2017). The PFGE banding patterns observed for the majority of V. cholerae O1 included in this study support an intercontinental transmission hypothesis (Chin et al., 2011), but only in a global context. The observation of country-based subclades indicates an independent evolution of the pandemic pathogen. Genetic changes were reported among initially homogeneous V. cholerae O1 El Tor initiating the Haitian cholera epidemic in 2010 (Chin et al., 2011). In this context, V. cholerae O1 El Tor strains associated with the Haitian cholera in 2010 were observed to be closely related to strains from other Southeast Asian countries (Chin et al., 2011).

While cholera had not been reported in the Americas for more than a century before 1991, the observed presence of V. cholerae classical biotype and diverse V. cholerae O1 El Tor lineages in Mexico was uncharacteristic for a region outside of Asia or Africa, suggesting a capricious nature of the bacterium (Boucher, 2016). Most V. cholerae O1 El Tor strains in Mexico that diverged separately from the pandemic clones lacked CTX prophage (Alam et al., 2014) and were not related to the non-toxigenic strains from Thailand (Tapchaisri et al., 2008). Previously, we had shown that CTX prophage negative strains dominated clinical cases in Mexico during 2001–2004 (Alam et al., 2014) and studies posited the strains to be ancestors of the V. cholerae responsible for the sixth and seventh pandemics (Boucher, 2016).

In conclusion, while the observed relatedness of PFGE patterns of V. cholerae O1 El Tor associated with cholera epidemics in Asia, Africa, and the Americas supports the potential for global transmission of the pandemic pathogen (Mutreja et al., 2011), the divergence of strains and their region-specific signatures also support independent evolution of V. cholerae locally. The PFGE data, as presented in this study show this technology can be effective for analysis and source-tracking of cholera outbreaks to prevent transmission of the deadly disease.