Probable Migratory Route Of Chloroquine Resistant Haplotypes

Published Date: 02 Nov 2017

Results from this study provide important information about the impact of chloroquine usage and transmission intensity on the population structure of Indian P. falciparum parasites.

4.1. Parasite diversity across Indian P. falciparum populations

P. falciparum populations from high transmission areas (HTA) generally shows higher genetic variation than those from low transmission areas (LTA) (Anderson et al., 2000) and maintain multiclonal infections (i.e., multiple genotypes per infected person) in such areas (Paul et al., 1998). We observed that genetic diversity within both neutral microsatellite loci and pfcrt-flanking microsatellite loci varies according to transmission intensity. The extensive microsatellite diversity within HTA corroborates earlier studies (Garg et al., 2007; Joshi, 2003), which demonstrated extensive polymorphism in surface antigens akin to high levels of multiple infections in HTA. Higher values for many measures of genetic diversity (AR, PR and He) within HTA indicate high rates of recombination and less inbreeding in parasite populations (Conway et al., 1999). In this study, a reduction in all measures of genetic variation was observed at pfcrt-flanking loci of resistant pfcrt-haplotypes, when compared with the pfcrt-flanking loci of wild-type (Table 1, 2). However, such reduction was not observed in resistant pfcrt-haplotypes at neutral loci, suggesting that there may be selection around the pfcrt.

4.2. Trend of chloroquine selection

There was a significant reduction in genetic diversity (He) within the pfcrt-flanking loci of the resistant pfcrt-haplotypes relative to wild-type, which indicates chloroquine selective pressure on the pfcrt. Reduced width of the selective sweep (pfcrt-flanking loci with reduced variation) around the resistant CVIET, in comparison to the resistant SVMNT (Fig. 1), is possibly due to the higher recombination rates associated with HTA, while inbreeding and reduced recombination rates are associated with LTA (Anderson et al., 2000; Conway et al., 1999; Nash et al., 2005). An approximate two-fold higher He value for the flanking SVMNT haplotypes was observed in HTA, when compared to He value of flanking loci of the SVMNT haplotype from LTA (Table 2), which reiterates higher recombination events in populations at HTA and a varied strength of selection at different geographic region cannot be ignored, i.e., weak selection strength at HTA in comparison of LTA. Another reason may be varied time points of colonization of different resistant haplotypes at various regions. The CVIET was possibly introduced earlier than SVMNT in India and may have evolved multiple times due to recombination events adding genetic diversity to population under HTA epidemiology. When reduction of heterozygosity was assessed at different locations, it provides a selective sweep around -20kb to +6kb in all sites under high transmission conditions, whereas sites exposed to low transmission conditions showed wider selective sweep around -24kb to +22kb. The exception in LTA was Madhya Pradesh, where a considerably higher amount He was observed. In a selective sweep, reduction of heterozygosity was also accompanied with large regions of elevated linkage disequilibrium around resistant haplotypes. The observed LD value supports above narrower selection valley as it got reduced with addition of a distant loci at upstream (+106kb) onwards. The decline in LD is slightly lower for CVIET (IAS =0.08) in HTA than for SVMNT (IAS =0.10) in LTA, consistent with weaker selection event at HTA, as discussed above. Hence, a different pattern of genotypes had aroused in various regions (broadly HTA and LTA), which can translated into the detection of geographical structure in relation to CQR among Indian P. falciparum.

The difference in relative fitness of pfcrt-haplotype within different epidemiological conditions may contribute to the differential hitchhiking patterns and also to the variable time of fixation for resistant haplotypes. The data presented here indicates a molecular basis of CQR that differs with changing transmission intensities; this scenario likely contribute to the observation of varied levels of CQR in different pfcrt-haplotypes of Indian isolates documented in a recent report (Lumb et al., 2012).

4.3. Inference of population structure among Indian P. falciparum

Three neutral loci located on three different chromosomes were used to identify the population structure of Indian P. falciparum. After STRUCTURE analysis, two discrete clusters associated with transmission intensity were identified, where Neutral cluster1 was significantly comprised of isolates from HTA and Neutral cluster2 was significantly comprised of isolates from LTA. Furthermore, there was minimal admixture between HTA and LTA, which supports earlier reports of the association between the genetic structure of P. falciparum and patterns of transmission (Anderson et al., 2000). The STRUCTURE results also corroborate the differential measurements of genetic diversity at HTA and LTA. Further pairwise FST analysis confirmed the population differentiation provided by STRUCTURE, estimating substantial levels of genetic differentiation between populations from HTA and LTA (Table S3). This genetic differentiation could be due to genetic drift; however, the observed value of Nm ≥ 1 between all of the subpopulations does not support genetic drift (Slatkin, 1987).

Low population differentiation was detected between falciparum subpopulations within HTA except WBG, which has moderate differentiation when compared with other regions. Corroborating earlier reports of genetically similar surface antigens (msp1 and msp2) between Assam and Eastern India (Joshi et al., 2007) indicate a pattern of current and/or historical migration. CN and Eastern India were also found to be genetically similar, indicating that there is considerable parasite gene-flow between mainland and island parasite populations, which supports previously reported data based on surface antigen (ama-1) (Garg et al., 2007). In LTA, MP showed low genetic differentiation with RAJ, GOA and TN (FST = 0.05, P>0.094). (RAJ-GUJ-UP) and (GOA-TN) showed high genetic differentiation between them. Thus, the Southwest (GOA-TN) part of India has the most genetically isolated population from rest of India except for MP. The three-level hierarchical analysis of genetic differentiation (AMOVA) supported the division of Indian P. falciparum population to four geographical groups, as maximum variations observed were within the subpopulation rather than among subpopulations of a particular geographical group. A significant positive correlation between genetic variation and geographic distance was obtained by Mantel test on all the populations. This indicates a genetic isolation by geographic distance in the falciparum population of India.

Geographical distance, alone, does not adequately explain the divergent parasite populations across India, especially given that populations of ASM-MAH, MAH-CN, and ORS-CN are greater than 1200 km apart and showed no genetic differentiation between them. However, the analysis of the Mantle test revealed a significant isolation by distance (IBD) at LTA only, i.e., genetic differentiation depends on geographical distances between studied regions of LTA. No such correlation between genetic distances and geographic distances in HTA was observed. This again supports the division of Indian P. falciparum population to four geographical groups. Further Mantel testing involving all populations excluding the Southwest group revealed no correlation between genetic distances and geographic distances. It is likely that the genetic differentiation between the Southwest region and all other geographically distinct groups is responsible for the significant IBD observations.

4.4. Spread of hitchhiking in Indian P. falciparum population

Collectively, the results (STRUCTURE, FST, AMOVA and IBD) indicate subdivided populations of P. falciparum isolates in India. A deviation from a single random mating population may impact the amount of genetic hitchhiking around a beneficial mutation in a population. Interesting patterns of population differentiation were observed while comparing pairwise FST between neutral loci and pfcrt-flanking loci (Table 4a) at the four groups of population inferred above; Southwest and Northeast-East-Island show highly differentiated populations at both neutral and pfcrt-flanking loci indicating heterogenous population and low gene-flow between them; Central and Northeast-East-Island or Northwest show low genetic differentiation at both neutral and pfcrt-flanking loci indicating homogenous population and continuous gene-flow between them; Central and Southwest show moderate differentiation at both neutral and pfcrt-flanking loci indicating a limited gene-flow between populations; Southwest and Northwest show high genetic differentiation at neutral loci but low genetic differentiation at pfcrt-flanking loci indicating that evolutionary process (i.e., strength of selection) is faster than rate of migration. In a subdivided population, genetic hitchhiking can introduce population differentiation (large FST) in an initially homogeneous population (small FST) (Bierne, 2010; Slatkin and Wiehe, 1998) and the similar pattern is displayed between Northwest and Northeast-East-Island groups (Table 4a). The observed gradient of heterozygosity for a selective sweep may be used to infer the geographical movement of the beneficial mutation in different subpopulation, as reduction in expected heterozygosity initially accumulates in subpopulations where beneficial mutation are introduced and subsequently increase with the spread across neighboring subpopulations (Kim and Maruki, 2011). Finally, our data also show a gradual increase of heterozygosity in the flanking loci of SVMNT from the Southwest to Northeast-East-Island part of India, which suggests a probable introduction of the SVMNT haplotype in southern India and then spread towards other regions.

4.5. Probable migratory route of chloroquine-resistant haplotypes

The STRUCTURE analysis of pfcrt-flanking loci showed 3 discrete clusters associated with the different pfcrt-haplotypes (CVIET, SVMNT, and CVMNK). This in turn reflects the earlier observation of varied level of selective sweep associated with regional variation in chloroquine pressure (Lumb et al., 2012). The pairwise FST estimate at pfcrt-flanking loci show high genetic differentiation between East-Island (ORS, JHK, CHG, CN), with that of Northwest (UP, RAJ, GUJ) and strong differentiation between East-Island and Southwest (GOA&TN), indicating a limited migration of resistant haplotypes between East-Island with that of Northwest and Southwest. In turn, the Northeast region may have indeed, been a major migratory route of the resistant haplotypes, likely originating from the Southeast Asia region and then spreading into Eastern India and consequently to other parts of India. Meanwhile, the moderate genetic differentiation (FST = 0.09-0.15) observed between the Central (MP, MAH) group and all other groups indicate genetic interaction between the resistant parasites in central and other parts of India.

Admittedly, more neutral loci may be needed to make concrete conclusions about geographic structure in Indian P. falciparum, but even with these three neutral loci, there is evidence of a genetic structure that is strongly linked with the patterns of malaria transmission. Additionally, resistant haplotypes are also differentially structured; probably due to above found geographic structure or varied chloroquine usage in these locations. While determining F-statistics estimations, a small amount of genetic exchange between populations is enough to prevent the accumulation of large genetic differences between them. Thus, the significant strong genetic differentiation found between Southwest and Northeast-East-Island part of India implies limitations in direct dispersal of resistant haplotypes and supports the movement via Central India.

Human migration between malaria-endemic regions plays an important role in the movement of resistant alleles in distant populations (Hume et al., 2003; Lqbal et al., 2002). For example, two recent studies based on pfcrt-flanking microsatellite markers (Mixson-Hayden et al., 2010; Rawasia et al., 2012) reported that the resistant SVMNT found in India and Pakistan migrated from PNG. Evidence for this was also found by clusters of neutral microsatellite markers that were reported earlier between PNG (Pacific region) and Thailand (Southeast Asia) (Anderson et al., 2000). In this study, the microsatellite profile of pfcrt-flanking loci associated with the resistant SVMNT and CVIET were similar to that of microsatellite profile reported from PNG and SEA, respectively (Mixson-Hayden et al., 2010; Wootton et al., 2002). A recent study (Awasthi et al., 2011) examining the allele frequency of various pfcrt-haplotypes postulates a probable route of migration for different pfcrt-haplotypes in India and reported that the resistant SVMNT arrived India from PNG via SEA. Here, the population differentiation, gene-flow and isolation by distance observed between different regions of India stimulate to postulate another probable route for dispersal of resistant SVMNT across India through Sri Lanka or southern India (Fig. 3).

Human migration related to labor and tourism had been associated with prevalence of CQR in India, particularly laborers traveling from eastern part of India to the western states (Sethi et al., 1990; Sharma and Sharma, 1988; Sharma, 2000). The study site in TN (Rameswaram Island) has been associated with continuous human migration with Sri Lanka and being a holy place the study site also receives a large number of pilgrims from all over India. In addition a large number of tourists stay here before or after visiting Sri Lanka. These human migrations within this region have likely helped to facilitate the spread of chloroquine-resistant parasites between both country and throughout India (Rajagopalan et al., 1986). This, in turn, may have given chance for immigration of SVMNT from Sri Lanka to India as a recent study reported ubiquitous appearance of SVMNT in Sri Lanka (Zhang et al., 2011) and we also observed similar frequency of SVMNT from this study site. The predominance of similar mosquito vectors (Anopheles culicifacies; sibling B,E) in both locations (Surendran et al., 2000) provide an added foundation for this hypothesis.

It is well known about human migration between southern India, Sri Lanka and Indonesia for trading and the cultural contacts was much greater than hitherto been imagined (Karafet et al., 2005). Similarly, Pakistan shares international borders with Rajasthan and Gujarat, which reports the fixation of the SVMNT haplotype with similar pfcrt-flanking loci in both countries (Mallick et al., 2012; Rawasia et al., 2012). On the other hand, countries sharing a border with northeast India, such as Nepal, Bangladesh, and Myanmar, reports higher frequency of CVIET (Banjara et al., 2011; Kawai et al., 2011; Mohapatra et al., 2005). It is worth putting the South Asia-Southeast Asia migration corridor in perspective. The above discussed active routes of human migration lead us to propose immigration of the SVMNT from PNG to India via Sri Lanka or southern India. Finally, the results of median joining NETWORK construction reveal two distinct clusters of resistant CVIET and SVMNT, indicating accumulation of resistant haplotypes on a distinct genetic background. However, clustering of multi-locus haplotype in SVMNT supports a more recent origin than that of the CVIET cluster, which seems to have undergone a large number of mutational steps over time. The maintenance of diversity over time in HTA suggests a large effective population size. Thus, the observed strong geographic structure may be governed by transmission intensity. It is also possible that multiple selective sweeps have occurred independently in different geographic locations due to local adaptation.