点击显示 收起
Faculty of Medicine and Health Sciences, University of Malaysia Sarawak
Sarawak State Department of Health, Kuching
Sabah State Department of Health, Kota Kinabalu, Malaysia
Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London, United Kingdom
Background.
The population genetic structure of Plasmodium falciparum differs between endemic regions, but the characteristics of a population recently fragmented by effective malaria control have been unknown.
Methods.
Genotypic analysis of 10 microsatellite loci widely separated in the parasite genome was conducted on 288 P. falciparum isolates from 8 foci in Malaysian Borneo, a region in which malaria incidence has been progressively reduced.
Results.
Within all P. falciparum foci, moderate levels of allelic diversity were found, but levels of multilocus linkage disequilibrium were extremely variable. The population with the highest proportion of mixed-clone infections also had the highest allelic diversity and nonsignificant linkage disequilibrium. In contrast, several populations showed evidence of clonal expansion, and one offshore island population had exceptionally high levels of linkage disequilibrium. Genetic differentiation between many populations was very high and strongly associated with the geographical distance between them.
Conclusions.
High levels of differentiation and contrasting population structure among P. falciparum populations in Malaysian Borneo indicate that they are genetically independent. This supports the feasibility of individually eradicating the remaining P. falciparum foci.
The protozoan parasite Plasmodium falciparum is the most important eukaryotic pathogen in humans, being responsible for 1 million deaths each year [1]. The parasite is haploid for most of its life cycle but reproduces sexually during a brief diploid stage, after which meiotic recombination occurs with a high crossover rate [2]. Globally, P. falciparum is known to exhibit a mosaic of population genetic characteristics, apparently correlated with local levels of endemicity and transmission intensity [3]. For example, levels of allelic diversity, parasite outcrossing, and gene flow are generally high in African populations, low in South American populations, and intermediate in Southeast Asian populations [36]. The majority of malaria cases reported in Malaysia occur in Sabah and Sarawak, 2 states separated from Peninsular Malaysia by the South China Sea and located in the northern part of the island of Borneo [7] (figure 1). The most prevalent species of human malarial parasite in Malaysia is P. falciparum, for which the reported annual incidence in Sabah has been reduced drastically (from 30,999 reported cases in 1995 to 3052 reported cases in 1999), whereas in Sarawak the number of cases has been more stable but low (<1000 cases/year) ([7] and Sarawak Vector-Borne Diseases Control Programme, unpublished data). Data on the entomological inoculation rate in Sabah or Sarawak are currently unavailable. Malaria has been controlled in Malaysian Borneo by use of a combination of insecticide residual spraying, distribution of insecticide-treated bed nets, reduction of mosquito breeding sites, and employment of trained microscopists in remote endemic regions to diagnose and treat malaria cases. Thus, P. falciparum now tends to occur in discrete foci in Malaysia. The low level of human migration between regions of endemism, because of the mountainous terrain covered with primary rain forest, is likely to restrict parasite gene flow. Moreover, the rapidly declining endemicity may lead to a more fragmented population structure with greater genetic isolation between endemic foci. Parasites with alleles associated with pyrimethamine and sulfadoxine resistance have been reported in both Peninsular Malaysia and Malaysian Borneo, but pyrimethamine-resistance allele frequencies show stark differences between the 2 regions [8]. Here, we use 10 microsatellite loci to analyze P. falciparum population structure in diverse endemic foci within Malaysian Borneo and consider the implications for the spread of antimalarial drug resistance and other epidemiological aspects of the infection.
MATERIALS AND METHODS
P. falciparum sample collection and study sites.
Two hundred eighty-eight P. falciparum isolates were analyzed from 8 separate populations in Malaysian Borneo3 in Sarawak and 5 in Sabah (figure 1). The locations, samples sizes, and years of collection are as follows: Serian (n = 35), collected in 2001, 2002, and 2003; Lundu (n = 54), collected in 2000 and 2001; Bau (n = 16), collected in 2001; Banggi Island (n = 40), collected in 2001 and 2002; Telupid (n = 45), collected in 2001; Lahad Datu (n = 33), collected in 2001; Kunak (n = 30), collected in 2001 and 2002; and Malinsau (n = 35), collected in 1997. The Malinsau sample was previously genotyped as part of a global survey of microsatellite frequencies [6, 9]. Populations for sampling were selected in areas with relatively high incidences of P. falciparum, to ensure that sample sizes sufficient for population genetic analysis could be compiled, but these populations do not represent all of the P. falciparum foci that exist in Malaysian Borneo.
At all collection sites, blood slides and finger-prick blood samples on filter paper were collected from individuals presenting with fever at hospitals and health centers. In Banggi Island, samples were also collected during mass blood surveys. Blood samples were screened for malaria parasites by use of both light microscopy and a nested polymerase chain reaction (PCR) malaria detection assay [10]. Isolates found to be positive for P. falciparum by PCR were used in the study.
DNA extraction and microsatellite genotyping.
DNA was extracted from dried finger-prick blood-spot samples, by use of methods detailed elsewhere [11]. DNA samples were stored in 96-well microarray plates at -20°C. Positive controls obtained from samples from previous studies that were known to contain P. falciparum, as well as negative controls obtained from uninfected individuals, were also included in each plate. The following 10 microsatellite loci were genotyped (P. falciparum chromosome assignments are given in parentheses): TA1 (chromosome 6), TA42 (chromosome 5), TA81 (chromosome 5), TA87 (chromosome 6), TA109 (chromosome 6), ARAII (chromosome 11), Pfg377 (chromosome 12), PfPK2 (chromosome 12), Poly- (chromosome 4), and TA102 (chromosome 12). The first 9 loci are a subset of the 12 loci described by Anderson et al. [12] for typing P. falciparum finger-prick blood samples, and the amplification conditions for the last locus (TA102), which were also developed by T. J. C. Anderson, are described elsewhere [6].
Microsatellites were PCR-amplified using the heminested PCR protocol of Anderson et al. [12] with the primer dye labels described by Conway et al. [6]. PCR products were analyzed on an ABI377 sequencer and were sized using Genescan (version 3.1; Applied Biosystems) and Genotyper (version 2.5; Applied Biosystems) software. All electropherograms were visually inspected, and, for each isolate, the predominant allele at each locus was scored. Any secondary alleles that were >30% of the height of the predominant allele on the electropherogram were also scored as minority alleles (presumed to be a result of additional parasite clones in an infection).
Statistical analyses.
Two estimates of the proportion of mixed-clone infections in each population were used. First, any isolate with a minority allele at any locus was scored as a mixed-clone infection. Second, the proportion of all individual locus genotype scores that contained minority alleles (mixed within an isolate) was calculated. This gives a more conservative estimate of the degree of mixed-clone infection but is more robust to the effects of occasional genotyping error.
To determine allele frequencies in each population, the predominant allele at each locus within each isolate was counted (the additional alleles in mixed-clone infections were not counted for this purpose). The genetic diversity in each population was assessed by calculating both the mean number of alleles (A) and the mean expected heterozygosity (H) across loci in each population. H values were calculated using the FSTAT program [13] (version 2.9.3.2; available at: http://www2.unil.ch/izea/softwares/fstat.html); H was calculated for each locus as H = [n/(n - 1)][1 - p], where n is the number of isolates sampled and pi is the frequency of the ith allele. The interpopulation variance in allele frequencies (FST) was assessed using the estimator of Weir and Cockerham [14] for each pairwise population comparisonthat is, 28 comparisons in totalas implemented in FSTAT [13]. Each FST value was tested to determine whether it was statistically significantly different from 0 (involving 1000 random permutations of the data).
Each population was also examined for evidence of a recent genetic bottleneck (i.e., a severe population size decrease). After a bottleneck, both the number of alleles and the expected heterozygosity in a population are predicted to decrease. Rare alleles are, however, purged more rapidly, which causes a more rapid decline in the number of alleles than in heterozygosity. Therefore, by use of the computer program BOTTLENECK [15], expected heterozygosity values can be compared with heterozygosity values calculated under a mutation-drift equilibrium model, to detect whether there is a significant number of loci with an excess of heterozygosity. In addition, allele frequency distributions were visually inspected for evidence of a deficit in rare allelesthat is, a mode shiftas expected after a bottleneck event [16].
Multilocus linkage disequilibrium was assessed using the standardized index of association (ISA), calculated using the LINUX-based computer program LIAN 3.1 [17] (available at: http://kiwi.ice.mpg.de/lian/). ISA is independent of the number of loci analyzed and has been employed in previous population studies of P. falciparum microsatellites, allowing comparison between studies. One thousand random permutations of the data were used to calculate probabilities of the observed ISA values. For each isolate, a single infection haplotype was constructed using the predominant allele at each locus, as described by Anderson et al. [3]. LIAN 3.1 cannot analyze isolates containing any missing data, so only those haplotypes with allelic data for all 10 loci were used (n = 239). Isolates containing >1 clone, as determined using the proportion of samples with >1 allele at any locus measure, were removed from the analysis, and linkage disequilibrium was reassessed (n = 146). The results give the single-clone measure of multilocus linkage disequilibrium. In addition, all haplotypes represented more than once in each population data set were removed, leaving only a single representative of each haplotype, and multilocus linkage disequilibrium was reassessed (n = 188).
An isolation-by-distance model was tested by assessing the relationship between genetic and geographical distance, by use of the method of Rousset [18]. The natural log of the distance in kilometers was plotted against the genetic distance (FST) transformed as follows: FST/(1 - FST); the significance was then assessed with a partial Mantel test of matrix correlation. The test was implemented using the computer program GENEPOP [19], available as a Web interface at http://wbiomed.curtin.edu.au/genepop/.
RESULTS
Genetic diversity within populations.
Of the 288 isolates studied, allelic data for all 10 microsatellite loci were obtained for 239 (83%). For the remaining 49 isolates, allelic data were obtained for 79 microsatellite loci (genotyping failed for 13 loci in these isolates). A full tabulation of allele frequencies in each of the populations is given in table 1. The genetic diversity of each population was measured using A and H. Values of A were within a relatively narrow range, from 3.9 in Bau to 5.1 in Malinsau (table 2). Values of H spanned a somewhat greater range, from 0.44 in Lundu to 0.63 in Malinsau, but showed a pattern of diversity between populations similar to that of A (table 2). The proportion of isolates with >1 allele at any locus varied greatly between populations, ranging from 17% of isolates in Kunak to 71% of isolates in Malinsau (table 2). The proportion of all individual locus scores within a population with >1 allele ranged between 4% in Telupid and 19% in Malinsau (table 2). When all of the above measures were used, the population from Malinsau was found to be the most diverse.
Multilocus linkage disequilibrium.
ISA among alleles at multiple loci (a measure of multilocus linkage disequilibrium) did not differ significantly from 0 in Malinsau (table 3). In all of the other populations, significant linkage disequilibrium was found when all P. falciparum isolates (without any missing data) were included (table 3). The degree of significant linkage disequilibrium in these populations was, however, highly variable, being lowest in Lundu (ISA = 0.02) and highest in Banggi Island (ISA = 0.19) (table 3). Although the lowest linkage disequilibrium was seen in the population with the highest proportion of mixed-clone infections (Malinsau), there was only a weak negative correlation between these indices among all 8 populations (Spearman's R = -0.29), and the correlation was not significant.
When isolates containing mixed-clone infections were removed from the analysis (to exclude the possibility of composite false haplotypes), patterns of linkage disequilibrium were not greatly affected, except that there were too few isolates for informative analysis in Bau (table 3). When duplicate haplotypes in a population were represented only once in the analysis (to exclude any effect of recent epidemic expansion of discrete multilocus types), there was a decrease in the level of linkage disequilibrium in all P. falciparum populations except that in Banggi Island (table 3). This was particularly notable in the populations from Telupid and Kunak, where linkage disequilibrium decreased to nonsignificant levels.
Genetic and geographical distances between populations.
Individual FST values for the 10 microsatellite loci ranged from 0.051 (for TA102) to 0.405 (for TA42) in an analysis across all populations. Averaged across all loci, the genetic differentiation among the different populations was high (FST = 0.239, indicating that 24% of genetic diversity was partitioned between populations). Pairwise comparisons of genetic differentiation and geographical distances between all P. falciparum populations studied are shown in table 4. The FST values ranged from 0.038 (between the closely situated populations at Lundu and Bau) to 0.376 (between Lundu and Kunak), and all were statistically significantly greater than 0 (P < .01). Level of genetic differentiation was strongly associated with geographical distance between populations (P < .001, Mantel test of matrix correlation). Under the conventional island model of neutral differentiation, a linear relationship is expected between the transformed index FST/(1 - FST) and the natural log of the geographical distance [18]. The data for the populations here show a good fit to the expected pattern (figure 2).
Testing for genetic bottlenecks.
The frequency distribution of alleles over all loci was examined for each population, by pooling alleles into frequency bins (10 evenly spaced categories) for visual inspection. All populations showed an L-shaped frequency distribution, with the vast majority of alleles being in the rarest class (frequency, <0.10). This is as expected in populations at mutation-drift equilibrium and contrasts with reduced proportions of rare alleles in populations that have experienced a bottleneck [3, 16]. When analyses that considered all loci under an infinite alleles model and 5 of these loci under a stepwise mutation model were used, expected heterozygosity at each locus did not show any consistent departure from expectations under mutation-drift equilibrium, again providing no evidence of genetic bottlenecks.
DISCUSSION
Levels of P. falciparum diversity in each of the populations in Malaysian Borneo are similar to or slightly lower than those reported in mainland Southeast Asia and Papua New Guinea [3, 5]. In Malaysian Borneo, the incidence of P. falciparum infection has decreased over several decades, with Sabah (the most endemic state in Malaysia) experiencing the most marked decline in recent years [7]. There is a high level of genetic differentiation between most of the P. falciparum populations sampled here. Many of the pairwise FST values of genetic differentiation are similar to those reported between P. falciparum populations from different continents [3, 6] and are higher than those between Thai and Papua New Guinean populations described previously [3].
The only low pairwise FST values in the present study were for comparisons between closely situated populationsnamely, the cluster of 3 populations sampled in Sarawak and the Lahad Datu and Kunak populations sampled in Sabah. The most likely explanation for this is that human migration between endemic sites separated by large distances is rare in Malaysian Borneo, thus restricting parasite gene flow between the populations. This is plausible, since endemic areas tend to be in inland regions of Borneo, with most travel from these regions occurring along rivers connected with the coast, whereas most roads and flight connections are between coastal urban areas in which there is little or no malaria. The strong relationship between genetic and geographical distances among the populations in Borneo supports a model of isolation by distance.
Genetic differences between populations were not related to geographical distances in a recent study of multiple populations in the Brazilian Amazon region [20], and this probably reflects the very different epidemiological histories of P. falciparum in Borneo and Brazil. There has been a more stable endemic situation until recently in Borneo, and the current allele frequency distributions probably reflect a past equilibrium between genetic drift and migration that has not yet been erased by recent successful malaria control. In contrast, malaria has emerged to become common in different parts of the Brazilian Amazon region in recent decades, after colonization by new communities engaged in farming and mining, and the genetic structure of the parasite population appears to reflect sporadic introductions and subsequent population expansions.
Within local P. falciparum populations, the proportion of mixed-clone infections in a human population is an important parameter, because only mixed-clone infections permit cross-fertilization and recombination between parasite genomes in the mosquito vector [21]. Therefore, a negative correlation between the proportion of mixed-clone infections and the degree of linkage disequilibrium in any given population is expected, because linkage disequilibrium is rapidly broken down by recombination [9]. There was a very wide range of multilocus linkage disequilibrium (ISA) among the 8 P. falciparum populations sampled from Malaysian Borneo, ranging from virtually 0 to levels that were as high as or higher than those that have been reported from anywhere in the world [36]. This suggests that recombination within the different populations in Malaysian Borneo takes place at very variable frequencies.
At one extreme, the P. falciparum population in Malinsau is notable for having no significant linkage disequilibrium, the highest level of allelic diversity, and the highest proportion of mixed-clone infections. It should be noted that the samples from Malinsau were collected in 1997, whereas samples from all of the other populations were collected between 2000 and 2003. In recent years, P. falciparum incidence has declined markedly in Sabah, mainly because of an effective malaria control program [7], and the samples from Malinsau may represent a population sampled before this period of particularly rapid decline.
An elevation of linkage disequilibrium caused by multiple representations of the same clone can be interpreted as evidence of epidemic population structure [22]. Epidemic population structures have been reported in South American and Southeast Asian populations of P. falciparum that have been tested using the ISA among microsatellite loci [3]. In the present study, the populations in Kunak and Telupid showed the strongest evidence of epidemic population structure, since most multilocus linkage disequilibrium did not persist when duplicate haplotypes were represented only once in the analysis.
In contrast, the population on Banggi Island is exceptional, in that it has the highest ISA value of any P. falciparum population tested to date, but the ISA did not decrease when duplicate haplotypes were removed. The proportion of mixed-clone infections in the Banggi Island samples was low but not quite the lowest, consistent with a probable low rate of effective recombination. This population may also be interesting for investigation of the factors affecting the strength of linkage disequilibrium among closely linked loci on a single chromosome. In African populations of P. falciparum, the rate of decline in linkage disequilibrium with distance between loci is known to be very rapid, even within a gene [9, 23]. However, in regions with lower levels of transmission, the rate of decline is not as rapid, as is shown by a 3-fold lower estimated recombination parameter in a Thai population compared with that in a Nigerian population [24]. Populations with a low effective recombination rate, such as that on Banggi Island and potentially others in Malaysian Borneo, could be useful in initial genomewide allelic association studies to map particular phenotypic traits in P. falciparum, as long as the traits of interest were common enough and sufficiently large sample sizes could be achieved.
The findings of the present study also have local implications for the epidemiology and control of P. falciparum. For example, a low level of gene flow between many of the populations within Malaysian Borneo suggests that parasite alleles may spread slowly between populations. If drug resistance occurred as a major problem in one area, it would not necessarily spread immediately to other areas. We therefore recommend that residual foci of P. falciparum in areas of declining endemicity should be monitored independently for the occurrence and frequency of drug resistance. As malaria control becomes more effective in reducing incidence throughout Southeast Asia, it may become increasingly important to understand the fragmented genetic structure of residual parasite populations.
Acknowledgments
We thank the staff of the Sabah Vector-Borne Diseases Control Programme and the nursing and laboratory staffs of Lundu Hospital, Serian Hospital, Bau Hospital, Lahad Datu Hospital, Kunak Health Clinic, and Telupid Health Clinic, for collecting the blood samples used in this study; and Dr. Nirmal Singh, Deputy Director of Health (Public Health), Sabah State Department of Health and Medical Services, for assistance in organizing the collection of blood samples in Sabah. We also thank Dr. Yvonne-Marie Linton (The Natural History Museum, London) for comments on the manuscript.
References
1. Korenromp EL, Williams BG, Gouws E, Dye C, Snow RW. Measurement of trends in childhood malaria mortality in Africa: an assessment of progress toward targets based on verbal autopsy. Lancet Infect Dis 2003; 3:34958. First citation in article
2. Su X-Z, Ferdig MT, Huang Y, et al. A genetic map and recombination parameters of the human malaria parasite P. falciparum. Science 1999; 286:13513. First citation in article
3. Anderson TJC, Haubold B, Williams JT, et al. Microsatellites reveal a spectrum of population structures in the malaria parasite Plasmodium falciparum. Mol Biol Evol 2000; 17:146782. First citation in article
4. Durand P, Michalakis Y, Cestier S, et al. Significant linkage disequilibrium and high genetic diversity in a population of Plasmodium falciparum from an area (Republic of The Congo) highly endemic for malaria. Am J Trop Med Hyg 2003; 68:3459. First citation in article
5. Ferreira MU, Nair S, Van Hyunh T, Kawamoto F, Anderson TJC. Microsatellite characterization of Plasmodium falciparum from cerebral and uncomplicated malaria patients in southern Vietnam. J Clin Microbiol 2002; 40:18547. First citation in article
6. Conway DJ, Machado RLD, Singh B, et al. Extreme geographical fixation of variation in the Plasmodium falciparum gamete surface protein gene Pfs48/45 compared with microsatellite loci. Mol Biochem Parasitol 2001; 115:14556. First citation in article
7. Singh B, Cox-Singh J. Parasites that cause problems in Malaysia: soil transmitted helminths and malaria parasites. Trends Parasitol 2001; 17:597600. First citation in article
8. Cox-Singh J, Zakaria R, Abdullah MS, Rahman HA, Nagappan S, Singh B. Short report: differences in dihydrofolate reductase but not dihydropteroate synthase alleles in Plasmodium falciparum isolates from geographically distinct areas in Malaysia. Am J Trop Med Hyg 2001; 64:2831. First citation in article
9. Conway DJ, Roper C, Oduola AMJ, et al. High recombination rate in natural populations of Plasmodium falciparum. Proc Natl Acad Sci USA 1999; 96:450611. First citation in article
10. Singh B, Cox-Singh J, Miller AO, Abdullah MS, Snounou G, Rahman HA. Detection of malaria by nested polymerase chain reaction of dried blood spots on filter papers. Trans R Soc Trop Med Hyg 1996; 90:51921. First citation in article
11. Cox-Singh J, Mahayet S, Abdullah MS, Singh B. Increased sensitivity of malaria detection by nested polymerase chain reaction using simple sampling and DNA extraction. Int J Parasitol 1997; 27:15757. First citation in article
12. Anderson TJC, Su X-Z, Bockaire M, Lagog M, Day KP. Twelve microsatellite markers for characterisation of Plasmodium falciparum from finger prick blood samples. Parasitology 1999; 119:11325. First citation in article
13. Goudet J. FSTAT (version 1.2): a computer program to calculate F statistics. J Hered 1995; 86:4856. First citation in article
14. Weir BS, Cockerham CC. Estimating F statistics for the analysis of population structure. Evolution 1984; 38:135870. First citation in article
15. Piry S, Luikart G, Cornuet JM. BOTTLENECK: a computer program for detecting recent reductions in the effective population size using allele frequency data. J Hered 1999; 90:5023. First citation in article
16. Cornuet JM, Luikart G. Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 1996; 144:200114. First citation in article
17. Haubold B, Hudson RR. Lian 3.0: detecting linkage disequilibrium in multilocus data. Bioinformatics 2000; 16:8478. First citation in article
18. Rousset F. Genetic differentiation and estimation of gene flow from F statistics under isolation by distance. Genetics 1997; 145:121928. First citation in article
19. Raymond M, Rousset F. GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. J Hered 1995; 86:2489. First citation in article
20. Machado RL, Povoa MM, Calvosa VSP, et al. Genetic structure of Plasmodium falciparum populations in the Brazilian Amazon region. J Infect Dis 2004; 190:154755. First citation in article
21. Walliker D, Babiker H, Ranford-Cartwright L. The genetic structure of malaria parasite populations. In: Sherman IW, ed. Malaria: parasite biology, pathogenesis, and protection. Washington, DC: ASM Press, 1998:23552. First citation in article
22. Maynard Smith J, Smith NH, O'Rourke M, Spratt BG. How clonal are bacteria Proc Natl Acad Sci USA 1993; 90:43848. First citation in article
23. Polley SD, Conway DJ. Strong diversifying selection on domains of the Plasmodium falciparum apical membrane antigen 1 gene. Genetics 2001; 158:150512. First citation in article
24. Polley SD, Chokejindachai W, Conway DJ. Allele frequency based analyses robustly identify sites under balancing selection in a malaria vaccine candidate antigen. Genetics 2003; 165:55561. First citation in article