The pandemic generating lineage is the most abundant Vibrio cholerae genotype in Dhaka’s water system
Dhaka, one of the most densely populated cities in the world (>21 million residents as of June 2020), is located within the Ganges delta. It is an inland city, with a water system primarily consisting of freshwater rivers and canals. Cholera is endemic there and shows biannual peaks in reported cases during the spring and fall, i.e. before and after the monsoons [38]. However, little is known about the abundance and distribution of V. cholerae in natural waterbodies and if it correlates with environmental and human factors. To track the abundance of total V. cholerae and pandemic generating (PG) lineage V. cholerae, we used culture-based detection as well as qPCR analysis. A 272 bp hypervariable stretch of the viuB marker gene, which is present in a single copy in V. cholerae [21] was amplified and sequenced from fortnightly samples taken in seven different water reservoirs in and around Dhaka city from October 2015 to Mach 2016 (Fig. 1). Presence of V. cholerae could be detected by both culture-based isolation and qPCR throughout the sampling period. Abundance of total V. cholerae did not show large fluctuations with fortnightly sampling, ranging from 2 x 105 to 5 x105 genome copies/L (Figure 2). V. cholerae O1 could only be detected in ~8% of the samples using culture-based isolation followed by identification with conventional PCR (Table S2). However, qPCR for the rfbO1 and ctxA genes using DNA extracted from water detected toxigenic PG V. cholerae O1 throughout the sampling period, with monthly average abundance of this genotype of 1.2 x 105 to 2.5 x 105 genome copies/L (Figure 2).
Using viuB amplicon sequencing, twenty-five viuB alleles were found in total, differing from each other by two or more single nucleotide polymorphisms. Each unique allele represents a specific V. cholerae lineage, roughly the equivalent of a multilocus sequence typing (MLST) clonal complex [21, 39]. Amongst them, 12 viuB alleles (each with > 20,000 sequence reads) were included for further analysis of the V. cholerae community composition, while the other 13 alleles representing < 1% of the total population were excluded following the strategy adopted previously [21]. The viuB-73 allele corresponding to the PG lineage [21], and specifically to 7th pandemic O1 El Tor strains in the Dhaka environment, was present at all seven sites throughout the six months of sampling (Figure 2). That this allele dominates V. cholerae communities in the water bodies of Dhaka is not surprising, given the endemicity of cholera in the city and the contamination of water bodies with human waste. Interestingly, viuB-73 displayed high abundance based on viuB sequencing throughout the six months sampling period, as opposed to only occasional detection of strains corresponding to this genotype using culture-based techniques [15, 38]. Sustained presence of PG V. cholerae detected by both qPCR and amplicon sequencing underscores the importance of culture-independent methods of tracking pathogenic V. cholerae in nature. Although regular cholera episodes occur year-round, the bacterium is usually only culturable during the two seasonal peaks [38], presumably being present in its viable but nonculturable (VBNC) state at other times [17]. In subspecies-level population analysis, viuB-73 remained the predominant allele in Dhaka, representing 24% to 92% of the total V. cholerae population. In a study of the natural V. cholerae population in cholera-free Oyster Pond on the US east coast, viuB-73 was found sporadically at lower relative abundance and other viuB alleles dominated the V. cholerae population [40]. This study shows that PG V. cholerae was present even in the so called ‘non-epidemic’ months of the year (December-February) and was the predominant genotype throughout the sampling period in the surface water of cholera endemic Dhaka.
Subspecies level diversity correlates with variation in environmental parameters
Studies conducted to look into populations of V. cholerae in natural habitats have shown that different environmental factors might influence their abundance [41, 42]. But how variation in these parameters influence subspecies level diversity of Vibrio cholerae is not known. In this study, several environmental variables were measured in situ (pH, dissolved oxygen, conductivity, total dissolved solids, salinity and water temperature) to gain insight on their effect on the absolute abundance of the species and relative abundance of the different viuB alleles found in the water reservoirs in Dhaka. Salinity, pH and temperature showed little variation at the seven sites sampled, ranging from 0-0.8 ppt, 6.5-8.0 and 30-32 °C, respectively. Conductivity (134.5 -1608 μs/cm) and TDS (67.2-804 mg/L) changed noticeably with sites and time of sampling. Dissolved oxygen (DO) also changed across the time and space ranging from 0.1-5.05 mg/L. Correlation of the environmental parameters with the diversity (richness) and population composition were assessed to see which parameter(s) could influence those. Diversity (richness) of V. cholerae alleles was significantly correlated with salinity, conductivity, total dissolved solids (TDS) and dissolved oxygen (DO) (Fig. 4). Salinity, conductivity and TDS were found to be positively correlated with allele richness (Pearson correlation coefficient 0.44, 0.41 and 0.40 respectively; P < 0.01), while DO (mg/L) showed a negative correlation (Pearson correlation -0.30; P<0.05) with that population characteristic. In previous studies, various physicochemical variables have been found to be associated with the abundance and persistence of V. cholerae in aquatic environments and with the risk of cholera outbreaks [43, 44]. Temperature and salinity have been observed to influence planktonic populations, which is a well-known habitat for V. cholerae in the aquatic ecosystem [41]. Adaptation to a wide range of salinity levels also facilitates V. cholerae’s survival in various aquatic environments (from coastal to inland water) [42, 43]. Previous studies suggest that abundance of V. cholerae decreases with increasing salinity and that they are most abundant in salinities ranging from 0 to 10 ppt [42, 43].
Changes in subspecies level population composition were also correlated with variation in environmental factors. Even though no strong correlation was found with the abundance of individual V. cholerae lineages and environmental conditions, there were moderate correlations observed in case of some viuB genotypes. Based on Pearson correlation co-efficient values, abundance of viuB-73, viuB-06 and viuB-07 were positively correlated with salinity, conductivity and TDS, whereas that of viuB-39 was negatively correlated with these variables (Supplementery sheet 1). This suggests that changes in the abundance of these four genotypes drove the changes in population diversity. In a study of the V. cholerae population in cholera free Oyster Pond coastal ecosystem (Falmouth, MA, USA), the viuB-73 allele was rarely found in the ocean [21], whereas it was present in the brackish pond and lagoon water connected to the ocean, suggesting that high salinity might represent an environmental barrier to the dispersal and range of cholera. In this study, salinity appeared to be the most significant parameter to explain the distribution of viuB alleles (Figure 3). In the NMDS plot (stress 0.13 and P < 0.01), generated by plotting the abundance of viuB alleles against the direction of change in the environmental variables, It appears that the environmental gradient of salinity significantly impacts the community structure. Among the major alleles, salinity gradient was positively associated with viuB-07 and negatively associated with viuB-73.
The negative correlation between viuB-73 abundance and salinity (Pearson correlation coefficient -0.31) observed in this study also suggest that PG V. cholerae (viuB-73) might be adapted to low salinity . The PG lineage represented by viuB-73 might have higher tolerance of rapid variations at low salinities, as it was consistently predominating in all but one site of Dhaka where salinity levels fluctuated between 0-0.8 ppt. The only site at which viuB-73 was not dominant was site 7 (Fig. 5), in which salinity was more stable and only varied from 0.4 ppt to 0.5 ppt, and never dropped below 0.4 ppt (Fig. 3). This suggests that even at the subspecies level, small environmental fluctuations could have an influence on population composition of V. cholerae. In the low-lying Ganges delta, salinity intrusion is considered a major threat due to the changing climate. Reduced upstream discharge, sea level rise and other catastrophic events such as cyclones can lead to increase in the salinity of inland water bodies [43]. A salinity increase of about 26% was recorded in coastal regions in Bangladesh over the last 35 years [45]. Such a shift in salinity could be affecting the composition of V. cholerae populations, possibly changing the distribution and abundance of various lineages, some of which could pose a threat to human health.
Dissolved oxygen (DO), a measure of free non-compound oxygen present in an aquatic system, is another variable influencing microbial communities [46]. A study in Hood Canal, Washington, USA, demonstrated that there was a strong negative correlation between bacterial richness and DO [47]. So far, there are no studies describing the correlation between the intraspecies diversity of V. cholerae and DO, but it was demonstrated earlier that V. cholerae is most abundant in low DO environments [48, 49]. In our analysis, DO was found to impact the V. cholerae diversity negatively in the Dhaka environment (Fig. 4). DO could also be one of the factors contributing to the differential adaptation at the subspecies level. Among the sampling locations, site 7 had the lowest average DO (0.90 mg/L). Average DO in site 1 was also low at 0.95 mg/L, but the other five sites (site 2-6) had noticeably higher average DO concentration ranging from 2.7 to 3.0 mg/L. Conductivity, salinity and TDS usually have a negative relationship with DO, which might have caused the lower DO observed in site 7 and site 1. Site 7 was the only one where viuB -73 was outcompeted in abundance by other viuB alleles (viuB-05 and viuB-06). V. cholerae being a facultative anaerobe, can adapt to the low-oxygen conditions by utilizing alternative energy-producing pathways (i.e. nitrate utilization) [50]. It is possible that some lineages are advantaged at lower oxygen concentrations, giving them the ability to co-exist with other lineages that usually outcompete them, leading to increased diversity at lower DO.
Temperature is another factor affected by climate change that is known to have an impact on V. cholerae populations. While this species is found at a wide range of temperatures (10 to 30 °C), the highest abundance is observed at >20 °C [42]. However, no significant correlation between temperature and viuB allele diversity (Pearson correlation coefficient 0.1) were observed in Dhaka (Fig. 4). Water temperatures remained within the range of 27.4 °C to 30.8 °C throughout the sampling period, a pattern that differs dramatically from regions where shifts in temperature are noticeable during summer and winter [51]. In a temperate region, attachment to particles and hosts has been shown to increase when temperature increases above 22 °C, contributing to changes in the lineage composition of a V. cholerae population when a seasonal change occurs in a temperate climate [21]. It is likely that at conditions with consistent high temperatures such as those found in a tropical region like Bangladesh, V. cholerae is not overly responsive to this parameter, either in terms of its growth rate or particle attachment behavior.
High human population density correlates with changes in the V. cholerae population of Dhaka reservoirs
Dhaka is one of the most densely populated areas in the world, with a density of 23,234 people per square kilometer within a total area of 300 square kilometers. This huge burden of human population has significant impact on the ecology and evolution of V. cholerae and consequently epidemiology of cholera in this region [52, 53]. Based on the demographic records of Dhaka city, the area surrounding sampling site 7 (Kamrangir char) is the most densely populated (100,000 people/km2) among the seven sites studied, whereas population density at the other six sites ranged from 10,000/km2 to 60,000/km2 (Fig. 1) [54]. Thus, human impacts on the water reservoir in this area is expected to be much higher than other sites, with a higher level of fecal contamination and industrial waste mostly from tannery industrial units. This area is mostly inhabited by a dense low-income population where people use shared hygiene facilities such as showers and toilets [55]. Open defecation has been reported from poorly maintained shared facilities. Additionally, high population density in this area frequently causes an overload of septic tanks, which results in the overflow of untreated effluent to the water reservoir [56]. The open drainage system commonly causes mixing between sewage and fresh water, increasing the possibility of V. cholerae transmission between the water reservoir and local population. Amongst the seven different sites studied, most of them had TDS concentration of < 300mg/L on average, except sites 1 and 7, where TDS varied from 308 to 472 mg/L and from 476 to 575 mg/L, respectively (Table S3). TDS value comes from the combination of the disassociated electrolytes and other compounds such as dissolved organic matter [57]. Typically, natural bodies of water have dissolved solids due to the dissolution and weathering of rocks and soil. However, human activities also influence the concentration of TDS in water reservoirs, which is especially likely in Dhaka’s inland water bodies because of urban runoff as well as wastewater discharge. This heavy influence of human population at Kamrangir char likely led to different population dynamics of V. cholerae at this site compared to other locations in the city. However, in our sampling sites, conductivity, salinity and TDS were found to be tightly correlated indicating inorganic ion might be the major contributor of the variation in TDS values. Hence, major cause of higher conductivity, salinity and TDS values observed in site 7 can also be indicators of increased chemical pollution there in comparison to other sites. Noticable difference in the diversity and population composition in site 7, might suggest that human population density can have a direct or indirect impact on the V. cholerae population at the subspecies level.
Although absolute abundance of total V. cholerae remained stable at all seven sites throughout the six months sampling period, striking differences appeared in case of site 7, where total V. cholerae abundance was around 28% to 43% higher than any other site (Fig. 2). It is unclear if this higher abundance at site 7 is due to a more constant input of V. cholerae from human waste, an indirect increase in numbers from nutrients linked to this waste, or the existence of a niche allowing the expansion of a particular lineage that does not compete with others (sympatry). Unlike sites 1 to 6, at which viuB-73 was more abundant than all other alleles combined (Fig. 5), three viuB alleles (viuB-05, viuB-06 and viuB-07) had greater combined abundance than viuB-73 at site 7. The trend is most pronounced during December 2015 to March 2016, coinciding with a decrease of water quality in the reservoir during December to April [58]. These lineages have so far not been found in environmental surveys outside of Dhaka. Interestingly, these three alleles (viuB-05, viuB-06 and viuB-07) have been found to be associated with strains representing a basal long branch clade in whole-genome phylogeny of known global V. cholerae strains [59]. Representative isolates of this clade were found to be indistinguishable from typical V. cholerae based on conventional phenotypic tests, but they are phylogenetically and genotypically divergent [60].
Allelic differences and in-silico DDH of whole genome sequences show that organisms represented by these three alleles (viuB-05, viuB-06 and viuB-07) are not more closely related to each other on average than most pairs of V. cholerae strains would be (Fig. S1). Despite this lack of genetic similarity, their presence and absence are strongly correlated, as they are usually co-occurring (Fig. 6 and Fig. S1). Surprising dominance of the viuB-05,06 and 07 lineages over viuB-73 exclusively at the Kamrangir char location suggests that this group of lineages is specifically adapted to the environmental conditions found at this site. As the site also stood out from a human population density point of view, and that these lineages were most abundant in this site, it suggests a potential human link to the ecology of these lineages in Dhaka.. Indeed, strains phylogenetically related with these lineages have been isolated from human samples in different parts of the world [59]. Abundance of those lineages was positively correlated with TDS, conductivity and salinity, which can be considered as indicators of the impact of human intrusion in the water. Among the seven sampling sites, these three related parameters were highest on average at site 7. These three alleles (viuB-05, 06, 07) were most abundant at site 7 (54% of total V. cholerae), where viuB-73 had the lowest abundance among all sites (24%). This suggests that some form of competition could be taking place between different V. cholerae lineages and/or that the lineages respond differently to environmental factors present at this site. Possible link of the human population to the abundance and distribution of these interrelated lineages makes them compelling candidates for future studies looking into the ecology of human adapted genotypes of V. cholerae.
Intraspecies interaction could influence relative abundance of PG V. cholerae O1
Even though overall viuB-73 was the predominant allele over the six months sampling period, spatial and temporal analysis indicate that direct or indirect intraspecies competition among V. cholerae genotypes could play an important role in the population dynamics observed in Dhaka. Whenever viuB-73 displayed a drop in abundance, another lineage carrying a viuB-05, viuB-06, viuB-07, viuB-25, viuB-45, viuB-51 or viuB-79 allele underwent a rapid expansion. Although the abundance of these other alleles was not directly quantified by qPCR, the absolute abundance measurement of the total V. cholerae and PG O1 V. cholerae by qPCR (Fig. 2) supported the observation coming from amplicon sequencing data. Furthermore, this shift in abundances is repeated multiple times within our dataset. For example, viuB-25 increased in relative abundance when the viuB-73 abundance decreased at sites 1 and 7 (Fig. 5). Notably, a new viuB allele, viuB-79 (for which no cultured isolates have been found), also appeared when viuB-73 abundance was reduced at sites 6 and 7 (Fig. 5). All of these alleles showed an inverse correlation with viuB-73 relative abundance (Fig. 6). Hierarchical clustering analysis of viuB amplicon sequencing data (Fig. 6) showing statistically significant clustering (Hopkins statistics [H] < 0.5) only occurred between viuB alleles representing strains from the basal divergent lonb branch clade (viuB-05, viuB-06 and viuB-07) (H <0.5). viuB-73 showed significant negative correlation with most other viuB alleles (H >0.5) [61]. The only viuB alleles positively correlated with viuB-73 were viuB-10 and viuB-39 (Fig. 6), the latter being a ubiquitous allele present at low abundance across sampling sites as well as other geographical locations (Fig. 5).
The cause of these shifts in abundance of some alleles is unclear. It could be due to a differential response to environmental factors or trophic interactions. In the Oyster Pond (MA, USA) V. cholerae population, divergent responses of different lineages were observed in ocean, lagoon and pond, where ecosystem parameters varied substantially [40]. It is plausible that V. cholerae lineages responded differently to even the slight environmental variations observed at the sites sampled in Dhaka. Differential response of lineages to phage predation can also be a cause of shifts in the community composition. Effect of predation by bacteriophages on the V. cholerae population composition can be modulated by the environmental factors i.e. nutrient availability [61], which in turn can give advantage to certain lineages to outcompete others under certain environmental conditions.
Another factor influencing this differential response could be the difference in the ability to avoid predation. The abundance of V. cholerae is influenced by grazing from heterotrophic protists [41]. To overcome the grazing pressure V. cholerae executes different strategies such as morphological shift, i.e. from smooth to rugose, resulting in the production of VPS (Vibrio polysaccharide) that helps to encase themselves in biofilm and resist predation [62]. V. cholerae can also survive predation by becoming intracellular in a range of amoeba [63]. They are also able to kill grazers using T6SS [64]. This system encodes a syringe-like structure that can pierce cellular envelopes of other bacteria and some eukaryotes, injecting effector proteins that can kill the recipient if it does not process the cognate immunity protein [65]. This phenomenon could influence the population composition of V. cholerae, with different lineages having varying effectors and immunity proteins providing varying predatory success [65, 66]. Another possibility is that incompatibility between subspecies likely plays a role in the diversity and dynamic of the V. cholerae populations in Dhaka. Most of these genotypes T6SS effector and immunity protein profiles suggest they are incompatible and can kill each other on contact (Hussain et al., Unpublished). Even though the PG lineage carrying viuB-73 could seemingly outcompete strains represented by all other alleles at sites 1-6, it was outcompeted by the three co-occurring alleles corresponding to the viuB-05, 06 and 07 group at site 7. This site differs from the other six sites in terms of environmental parameters and surrounding human population density. These observations suggest that environmental conditions can play an important role in shaping the intra-species competition of V. cholerae in their natural environment to impact diversity and subspecies level population dynamics of the V. cholerae.