Clonal Outbreak of Plasmodium falciparum Infection in Eastern Panama

Abstract

Identifying the source of resurgent parasites is paramount to a strategic, successful intervention for malaria elimination. Although the malaria incidence in Panama is low, a recent outbreak resulted in a 6-fold increase in reported cases. We hypothesized that parasites sampled from this epidemic might be related and exhibit a clonal population structure. We tested the genetic relatedness of parasites, using informative single-nucleotide polymorphisms and drug resistance loci. We found that parasites were clustered into 3 clonal subpopulations and were related to parasites from Colombia. Two clusters of Panamanian parasites shared identical drug resistance haplotypes, and all clusters shared a chloroquine-resistance genotype matching the pfcrt haplotype of Colombian origin. Our findings suggest these resurgent parasite populations are highly clonal and that the high clonality likely resulted from epidemic expansion of imported or vestigial cases. Malaria outbreak investigations that use genetic tools can illuminate potential sources of epidemic malaria and guide strategies to prevent further resurgence in areas where malaria has been eliminated.

Malaria transmission has been greatly reduced in Panama over the past several decades, suggesting that only a limited number of parasite types may remain among infected individuals. After 3 decades of successful malaria control, Panama experienced an outbreak of epidemic proportions between 2001 and 2005, accounting for >60% of all cases reported over the past 35 years [1, 2]. Study of P. falciparum isolates collected during the outbreak provides an opportunity to study the population structure of Panamanian parasites and to understand genetic signatures associated with reemerging parasites during an elimination campaign.

Genetic diversity in parasite populations has been linked to transmission intensity [3, 4], with limited genetic diversity and extensive linkage disequilibrium [4–6] observed in low-transmission settings. South America shows evidence of decreased outcrossing and genetic drift primarily resulting from inbreeding and asexual replication [7–9]. Population bottlenecks, low transmission, inbreeding, and epidemic expansions can all give rise to clonal lineages [3, 9–15]. We hypothesize that, with fewer and fewer parasite types, limited opportunities for outcrossing during the mosquito phase of the lifecycle results in transmission of highly related or even clonal parasites. Genetic traits, such as drug resistance, are inherited from one generation to the next and, thus, are predicted to be homogeneous among highly related parasites. Furthermore, the epidemic rise of malaria transmission during 2001–2005 in Panama similarly predicts that these parasites would be highly related or even genetically identical.

A molecular barcode has recently been deployed in Senegal, where data on barcoded parasites revealed increases in and persistence of clonality coincident with increased deployment of malaria control interventions [5]. Uncommon malaria cases or outbreaks [16] in settings of low malaria endemicity provide an ideal opportunity to ascertain whether simple genetic fingerprinting tools like the molecular barcode can reveal clonal parasite population structure. Using the molecular barcode and a panel of assays for drug resistance loci, we detected highly related clonal and parasite lineages in Panama, consistent with this hypothesis. Deployments of genotyping tools that reveal parasite population structure in low-transmission settings are critical for successful malaria elimination campaigns.

METHODS

Ethics Statement

Parasite samples were obtained from studies previously approved by the Gorgas Memorial Institute of Health Studies (ICGES) institutional review board, as part of a nested study conducted by the Panama Ministry of Health (MINSA) and the malaria surveillance and vector control program at the ICGES Central Public Health Laboratory, or by the institutional review board of the Malaria Vaccine and Drug Development Center (MVDC), as part of a study conducted by the Centro Latino Americano de Investigación en Malaria (CLAIM) [17] in Colombia. DNA extracted from P. falciparum field isolates (ie, parasites in donor blood specimens) was collected anonymously and is not linked to the identity of the blood donor.

Study Sites and P. falciparum Isolates

We examined DNA from 37 P. falciparum isolates collected during 2003–2008 from individuals in malaria-endemic provinces in eastern Panama (Panamá, n = 17; Kuna Yala, n = 15; and Darien, n = 5) and DNA from 20 P. falciparum isolates collected during 2011–2012 from healthcare facilities in 3 malaria-endemic sites in Colombia (Buenaventura, n = 5; Tumaco, n = 5, and Tierra Alta, n = 10). Samples were collected at the corregimiento or department level (the smallest political division), together with geographic, demographic, and epidemiological malaria data.

For Panamanian samples, isolates were eligible for analysis if they were collected from specimens that were obtained either by active or passive surveillance during the MINSA malaria surveillance program or from regional hospital clinical cases reported to the ICGES Central Public Health Laboratory and were positive for P. falciparum monoinfection (ie, only the species P. falciparum was present) by malaria microscopy, polymerase chain reaction (PCR), and the OptiMAL dipstick test (DiaMed, Cressier, Switzerland). For the Colombian samples, isolates were eligible if they were collected by the MVDC and were positive for P. falciparum monoinfection by microscopy and PCR. Samples containing Plasmodium vivax or mixed infections were excluded from the study.

Sample Extraction and Preparation

Genomic DNA was extracted from blood specimens directly or from blood specimens spotted onto Whatman FTA filter paper, using the QIAmp DNA Blood Mini Kit (Qiagen, Valencia, CA) according to manufacturer specifications. DNA samples were stored at the ICGES at −20°C or at the MVDC in Colombia until genotyping analysis at the Harvard School of Public Health (HSPH).

Genotyping

TaqMan Genotyping

Extracted DNA was preamplified with a pooled assay master mix containing forward and reverse primers [18] and with TaqMan PreAmp Master Mix, according to the manufacturer's instructions (Life Technologies, Grand Island, NY; part no. 4384256). Preamplified samples were genotyped as described elsewhere [19]. Samples were genotyped with both high-resolution melting (HRM) and TaqMan technologies.

High-Resolution Melting (HRM) Genotyping

Extracted DNA was preamplified using pooled forward and reverse primers for HRM drug resistance assays as described above [18] and was assayed across 23 drug resistance loci [20]. In addition, we developed and optimized HRM assays for the molecular barcode loci, based on previous design criteria [20]. Samples were amplified by PCR by using a gradient thermocycler (Bio-Rad C1000), followed by HRM analysis on the LightScanner 384 system (BioFire Defense, Salt Lake City, UT) to optimize amplification conditions. All PCR amplifications were performed using 2.0 µL of Lightscanner Master Mix (BioFire Defense), 2.5 µL of a 1:100 dilution of preamplified template, and 0.5 µL of primers and probes, overlaid with 10 µL of mineral oil. Genomic DNA from cultured P. falciparum strains with known genotype profiles representative of major and minor alleles for the panel of single-nucleotide polymorphisms (SNPs) in the molecular barcode were used for assay validation and as genotyping controls for all reaction plates. Optimal annealing temperatures were determined using gradient thermal cycling for asymmetric PCR [20] in ratios of 5:1. Molecular barcode assays 10, 11, 13, 21, and 24 were performed optimally under asymmetric forward to reverse primer ratios of 5:1; all other assays required a 1:5 primer asymmetry. Final primer concentrations for a 5-µL total reaction volume were 0.5 µM excess primer, 0.1 µM limiting primer, and 0.4 µM of the 3′-blocked probe. Amplification conditions were 95°C denaturation for 2 minutes, 50 cycles of 94°C for 5 seconds and 66°C for 30 seconds, and a premelt cycle of 5 seconds each at 95°C and 37°C. The change in fluorescence was recorded as samples were melted from 40°C to 90°C on the LightScanner 384 system and analyzed using the Call-It module within the LightScanner software. Assay sequences are shown in Supplementary Table 1.

Statistical Analysis

Samples were genotyped for 24 SNP markers in the molecular barcode, using both TaqMan and HRM assays, and for SNP variant calls, based on overall concordance between 3 and 5 replicates per sample. Assay calls were imported into GenAlEx 6.5 for statistical and population genetic analysis [21]. Samples showing strong fluorescence signals for both alleles in TaqMan assays or shared peaks in HRM assays were marked as “N” and treated as missing data in downstream genetic analysis. Two or more N's among the 24 SNPs assayed was taken to indicate that >1 P. falciparum genome was present. Summary statistics were calculated in GenAlEx 6.5, and a genetic distance matrix was used to generate principal components analysis (PCA) plots by using the haploid genetic distance between polymorphic alleles. When indicated, 40 random barcoded samples from Senegal were evaluated with samples from the Colombian and Panama populations to test for population structure in the presence of a group outside the clade of interest.

Structure (v2.3.3) [22] analyses were run with 100 000 burn-in steps, followed by 10 000 iterations, and repeated 20 times. Structure Harvester [23] was used to identify the most likely number of subpopulations (K) based on the delta-K criterion [24], and CLUMPP [25] and Python 2.7.3 were used for data visualization. Phylogenetic trees were constructed using the Bayesian Markov chain Monte Carlo method [26]. TreeAnnotator v1.7.5 and TreeStat v1.7.5 were used to generate a maximum clade credibility tree, and the tree was visualized with FIGTREE v1.3.1 [27].

RESULTS

Genotyping Reveals 3 Clonal Parasite Populations Among P. falciparum From Eastern Panama

Using a molecular barcode that assesses 24 neutral, unlinked SNPs with a high minor allele frequency, we identified 3 clonal parasite populations from among 37 P. falciparum samples collected in 2 provinces (Panamá and Darien) and mostly within 3 Amerindian reservations (Kuna Yala, Madugandi, and Embera-Wounan) in eastern Panama between 2003 and 2008 (Figure 1). These SNP data were then used to group the parasites into 3 clusters of essentially genetically identical parasites (Figure 2). These 3 clonal clusters included isolates mainly from Kuna Yala (n = 15; group I), Darien (n = 5), and Panamá (n = 15) provinces, along with 2 from Kuna Yala (PE01 and PF02; group II) and 2 coastal samples (PF002 and KY44; group III).

Figure 1.

Study sites in Panama and Colombia where Plasmodium falciparum isolates were collected during 2003–2008. Sites are color coded by clonal groups identified by molecular barcoding (group I = blue, group II = green, group III = red, and Colombian = orange), and province and country boundaries are denoted by gray lines. The diameter of each circle corresponds to the number of samples collected at each site. The base map for this figure was generated with Python 2.7.3, using shapefiles (http://www.diva-gis.org/gdata).

Figure 2.

Epidemiologic and genotyping data for Panamanian and Colombian isolates. Isolate sample names are color coded by collection site (left: Darien = purple, Panamá = green, and Kuna Yala = blue). Epidemiologic data are shown with genotypes for each sample. Groups I (blue), II (green), and III (red) and the Colombian group (orange) are indicated on the right.

Twelve of the 24 alleles among the Panamanian isolates had polymorphisms delineating 3 groups, yielding 4096 theoretically possible genotypes. Analysis of molecular variance (AMOVA) by using the molecular barcode revealed the percentage of molecular variance within and between groups; 68% of the variation was explained by collection site (P = .001), compared with 96% (P = .001) explained by clonal groups, which we define as clusters of essentially genetically identical parasites, regardless of geographic site of isolation. These data imply that clone correction was more consistent with these data; consequently, the clonal groups were used for additional genetic analysis.

Clonal Subpopulations of Eastern Panama Are Related to Parasites From Neighboring Colombia That Have Greater Genetic Diversity

Given the very low prevalence of P. falciparum infection in Panama before and after this epidemic in 2001–2005, we sought to test whether these parasites were related to parasites from neighboring Colombia, which exhibits a higher incidence of malaria. Twenty P. falciparum samples collected from Buenaventura and Tumaco (on the Pacific coast) and Tierra Alta (in the northwestern region) in Colombia during 2011–2012 (Figure 1) were similarly genotyped by barcode and compared with samples from Panama. Structure analysis revealed 2 clusters with significant likelihoods of K = 2 (ΔK = 108; Figure 3A) and K = 5 (ΔK = 61; Figure 3B). Structural analysis recovered the highly clonal populations in Panama and found evidence of 2 additional populations in Colombia (group IV and group V; Figure 3B).

Figure 3.

Membership fraction plot of eastern Panamanian and Colombian subpopulations. Structure analysis showing K = 2 (A) and K = 5 (B) clustering. The genotype of each sample is represented by a single vertical line partitioned into segments in proportion of the estimated membership in the different subpopulations. The uppermost level of structure among Panamanian isolates (ie, the most likely number of clusters [K = 2]; A) and the second-most-likely number of clusters (K = 5; B) are shown. The sample name is color coded by region (Darien = purple, Panamá = green, Kuna Yala = blue, coastal = red, and Colombia = orange).

We further explored parasite population structure by using a hierarchical clustering approach and PCA. To test the robustness of the clustering, we created an “outgroup,” defined as a population outside the clade of interest, using additional genotypes from Senegal. In the absence of the outgroup, group II isolates clustered together in one branch of the phylogenetic tree (Figure 4) and were separated from the majority of Colombian samples and group I isolates. The 2 Colombian isolates (C0048 and C0027) that shared fractional membership with group 1 shared the same branch of the phylogenetic tree. Remaining Colombian samples formed their own branch distinct from the Panamanian samples. The majority of this structure between Panamanian and Colombian groups was retained with the addition of a continentally distinct geographic outgroup (Supplementary Figure 1), and greater genetic variation between Senegalese isolates was evidenced by higher tree heights among the genotypes from this population.

Figure 4.

Maximum clade credibility tree of Plasmodium falciparum isolates from eastern Panama and Colombia. Isolates are color coded at terminal branch arms on the basis of study site. Groups I (blue), II (green), and III (red) and the Colombian group (orange) are distinguished by triangles. Branch nodes are labeled by the median height at each node for recalculated genetic distances, based on 10 000 bootstrap replicates.

Results of PCA supported this clustering analysis and revealed a separate structure for groups I, II, and III, as well as the Colombian samples, which were separated by the first 2 principal components (PC1; Figure 5A). Additional variation attributable to differences among Colombian samples (PC2; Figure 5B) was observed, and inclusion of a continental outgroup from Senegal showed differentiation between these 2 continental populations (Figure 5A), while still distinguishing group II from the American parasite population (Figure 5B).

Figure 5.

Two-dimensional plots of principal components analysis (PCA) of Panamanian Plasmodium falciparum isolates are shown, using 24 alleles of the molecular barcode for the Panamanian and Colombian isolates alone (A), with PC1 explaining 39% of the variance and PC2 explaining 18% of the variance, or with the Senegalese outgroup (B), with PC1 explaining 26% of the variance and PC2 explaining 14% of the variance. Samples are color coded by group and region (group I = blue, group II = green, group III = red, the Colombian group = orange, and the Senegalese group = gray).

Further examination of the 2 parasite populations clustered by structure analysis (K = 2) revealed that 1 group contained only Panamanian samples (from Darien and Panamà; group II) and that the other contained both Panamanian (group I and group III) and Colombian isolates. The separation of group II parasites was supported by findings of both PCA and phylogenetic tree analysis, which showed a distinct parasite population. The mixed Panamanian and Colombian cluster contained several highly related parasites. For example, 2 Colombian samples from Tierra Alta (C0027 and C0048) shared 50% of their membership fraction with group I, and 5 Colombian samples from Tumaco and Buenaventura (C90021, C9009, C9030, C3148, and C3004) shared ≥50% of their membership fraction with group III. The 2 coastal samples (KY44 and PF002; group III) were more closely related to Colombian parasites than to other Panamanian parasites (Figure 3), based on population structure analysis. Parallel analysis on a subset of monogenomic samples (containing <2 N's among the 24 SNPs assayed) yielded similar results (Supplementary Figures 2–5).

The Clonal Panamanian Parasite Population Is Supported by Nearly Identical Drug Resistance Haplotypes

Drug pressure historically applied to P. falciparum selects drug-resistant parasites. To test whether parasites with identical barcodes shared drug-resistant loci profiles, we assayed 32 SNPs across six genes associated with drug resistance (Supplementary Table 2). All Panamanian isolates harbored resistance mutations in pfcrt, pfmdr1, pfdhps, and pfdhfr, and, excluding mixed calls for pfmdr1 in 2 samples (KY44 and PF002), all had identical drug resistance haplotypes, irrespective of their clonal group assignment. These 2 samples (KY44 and PF002) that contained evidence of mixed calls for pfmdr1 were collected off the Panamanian coast (coastal; group III) and lacked 2 additional changes, including D1246 in PfMDR1 and A437 in PfDHPS. The only other genetic variation among these drug resistance loci noted were mixed allele calls from 2 Darien isolates, suggesting the presence of both the mutant and wild-type alleles at amino acid positions 86, 1042, and 1246 (DA29) or residue 1246 (DA04) of PfMDR1. Overall, there was significantly greater diversity among drug resistance loci from Colombian parasites, compared with Panamanian parasites (Supplementary Table 2).

Drug resistance profiles were consistent with the relatedness between Colombian and eastern Panamanian samples detected by the putatively neutral markers of the molecular barcode. All Panamanian samples had amino acid changes at positions 75, 76, 97, and 220 in PfCRT, resulting in the CVMET-Q-S-N-I haplotype previously reported to originate in Colombia and subsequently identified in Venezuela. Here, Colombian samples from Tierra Alta also shared this PfCRT haplotype, indicating genetic similarity between these Colombian samples and those found in Panama. For PfMDR1, >80% of the Panamanian shared the triple mutant haplotype N-F-R-S-D-Y (with amino acid changes: Y184F, N1042D, and D1246Y), while remaining variation came from group III parasites that shared the N-F-R-S-D-D haplotype or from mixed alleles in some Darien isolates. Previous reports identified the N-F-R-S-D-D haplotype as having originated in Colombia, and it was detected among coastal Panamanian parasites (group III). Mixed allele calls in some Darien parasites that shared nearly identical molecular barcodes were suggestive of gene amplification of pfmdr1, which was confirmed by HRM analysis that detected the N86Y change among additional copies (eg, DA29; Supplementary Table 2). All isolates shared the double mutant I-C-N-I haplotype for PfDHFR, and group I and group II shared the SG-K-A-A PfDHPS haplotype. Group III again differed from other Panamanian parasites by carrying the wild-type allele at position 437. A summary of identified drug resistance–associated haplotypes is shown in Supplementary Figure 6. Collectively, analysis of drug resistance alleles was consistent, with members of group III being distinct from other Panamanian isolates and sharing some drug resistance haplotypic information previously reported from Colombia.

DISCUSSION

Application of molecular barcode genotyping to resurgent P. falciparum infections in Panama reveals clonal parasite populations that are distinct from genetically diverse P. falciparum from nearby Colombia. These clonal Panamanian parasites harbor nearly identical drug resistance loci, while the largely genetically diverse Colombian parasites exhibited correspondingly heterogeneous drug resistance loci. Collectively these data demonstrate the utility of simple molecular genotyping tools and approaches to reveal the similar or dissimilar nature of P. falciparum parasites among clinical cases for the purposes of outbreak investigation and can identify genetically similar or identical parasites from among these cases. The 24 SNP barcode is useful to differentiate parasite populations in settings of low endemicity, as most of the previous studies were done in African settings where endemicity is high. Although numerous reports from South and Central American malaria-endemic countries [1, 12, 14, 17, 28–31] have evaluated P. falciparum populations, these findings demonstrate the combination of a comprehensive panel of both molecular barcode SNPs and drug resistance–associated SNPs can fingerprint and identify the relatedness of transmitted parasites from clinical cases during resurgent malaria.

Given the highly related nature of these parasites from Panama, only 12 variant neutral SNPs were sufficient to divide them into 3 clonal subpopulations that were confirmed by AMOVA and maintained this structure after subsequent clustering and PCA interrogation. Investigation of molecular variance revealed that these clonal groups were more similar than groups formed by geographic information alone, consistent with the almost identical barcodes shared by these populations. Thus, molecular genotyping techniques provide granular information about parasite identity and relatedness that may aid in identifying the origin of new infections or tracking the spread of infections during an outbreak setting to better identify and target reservoirs of infection. These data support the hypothesis that eastern Panamanian isolates have limited genetic diversity and a clonal population structure. The persistence of clonal lineages throughout and after the epidemic is consistent with the hypothesis that, as parasite population structure becomes more pronounced, inbreeding occurs more frequently, and opportunities for outcrossing are limited by low frequency of complex infections [4]. Furthermore, epidemic malaria can seed the expansion of specific clonal types that can spread from the original site of the epidemic.

Because parasite populations clustered into groups irrespective of study site, additional factors may have contributed to the clonal lineages observed. For example, it has been shown that the San Blas Range, which stretches along the northern coast of eastern Panama, adjacent to the San Blas Islands, may be an important physical and sociodemographic barrier [32], as has been shown for some malaria vector species [33]. One could thus speculate that these islands may serve as residual transmission reservoirs for geographically isolated parasite populations, consistent with the lack of additional drug resistance mutations and of shared alleles among barcoded parasites, compared with mainland parasites.

Use of outgroups, first from neighboring Colombia and then from continentally distinct Senegal, strengthens the conclusion that these Panamanian parasites were highly related and clonal. Although the time difference between sample collection limits comparison of these 2 populations, the addition of these outgroups in the analysis maintained the underlying population structure among Panamanian parasites and revealed the relatedness of parasites collected along the Caribbean coast to parasites from Colombia. The addition of Senegalese samples maintained the population structure of the Central and South American populations and revealed the limited diversity of the American populations by comparison. Molecular barcode data were used to identify clonal parasites in Colombia and Panama, to distinguish relatedness between nonclonal parasites, and to infer trends in changing parasite populations in low-transmission settings.

Drug resistance mutations were consistent with the observations that Colombian parasites were more diverse than those from Panama and that the group III parasites from the coastal region were more similar to Colombian parasites than the other groups of Panamanian parasites among those sampled. This was detected by the sharing of both pfcrt and pfmdr1 haplotypes among these parasites. Evidence of pfmdr1 copy number variation was detected, consistent with the introduction of mefloquine as a first-line therapy in 2004 [2] in Panama and with previously documented cases demonstrating that increases in pfmdr1 copy number modulate resistance to mefloquine [34–37]. Bacon et al proposed 2 pfmdr1 founder lineages in the Amazon Basin region and that the double mutant N-F-R-S-D-D may have given rise to 2 independent drug-resistant lineages [38]. Given the majority of the triple mutant N-F-R-S-D-Y and 2 seemingly vestigial N-F-R-S-D-D haplotypes from islands off the Caribbean coast (group III), these findings are consistent with northward dispersion of double mutant PfMDR1 haplotypes that likely have acquired additional mutations from selective drug pressure. These findings are consistent with those for PfMDR1 haplotypes found in the Amazon Basin [38], Peru [9], Ecuador [9], Venezuela [13], Bolivar [11], and Colombia [15]. The distinct nature of group III parasites was also detected when the pfdhfr and pfdhps loci were genotyped, with group I and group II parasites sharing the SG-K-A-A PfDHPS haplotype. Parasites from the outbreak most likely expanded from residual and/or vestigial populations from Panama, as parasites north of Panama harbor chloroquine susceptibility pfcrt alleles and the CMET pfcrt haplotype uniquely identifies parasites from the Panama/Colombia region. The ability to identify the source of an outbreak demonstrates the usefulness of molecular tools for monitoring outbreaks in elimination settings.

By using molecular genotyping strategies, highly clonal P. falciparum parasites were detected in Panama during an outbreak investigation in 2003–2005 that persisted after the epidemic subsided, through 2008. Interrogations of these clonal parasite populations and Colombian parasites are consistent with a northward migration of drug-resistant parasites from Colombia to Panama along the Atlantic coast and subsequent expansion of clonal lineages. The highly related genetic signatures of these Panamanian parasites are consistent with clonal expansion during the outbreak, and their persistence is possibly related to the likely lack of insufficient transmission that would normally allow for outcrossing between genetically distinct parasites or geographic isolation of distinct parasite populations. These findings are consistent with data from other low-transmission settings that resulted from whole-genome sequence analysis [39, 40]. Use of granular molecular fingerprint techniques such as the molecular barcode tool allows identification of highly related parasites during outbreaks and tracking of the potential spread of these infections, which is critical for ending outbreaks.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.