Genetic diversity and population structure of genes encoding vaccine candidate antigens of Plasmodium vivax

Blood samples from patients were collected during years 2006-2007 from endemic areas in the Peruvian Amazon region of Loreto Department using human use approved protocols (NMRCD.2005.0005). In Peru, P. vivax corresponds to more than 80% of the malaria cases and the majority of them are found in the Amazonic Department of Loreto [41]. Fifty-seven samples were obtained from Padrecocha, in Punchana district, a village situated five km from Iquitos (Loreto's capital) and forty-nine samples were collected from a village located in the San Juan district just south of Iquitos. All samples were confirmed to be positive for P. vivax by microscopic analysis on Giemsa-stained blood smears.

DNA was purified from 200 μL of whole blood using the QIAamp DNA Blood mini kit (QIAGEN, Valencia, CA). PCR amplification was performed in a 50 μL reaction mixture containing 0.2 μM each of forward and reverse primers, 200 μM each dNTP, 0.6 units of DNA polymerase recombinant (Invitrogen, Carlsbad, CA), 3 μL of 10× PCR buffer, 1.5 mM of MgCl₂ and 5 μL of DNA extracted from blood. To assess antigenic variability of Pv CSP, the Central Repetitive Region, Region I and Region II were amplified. The PCR conditions were: 5 min at 94°C; 1 min at 94°C, 1 min at 58°C, 2 min at 72°C for 35 cycles; and a final extension of 72°C for 2 min, using the primers VSC-OF (5'ATGTAGATCTGTCCAAGGCCATAAA3') and VSC-OR (5' TAATTGAATAATGCTAGGACTAACAATATG 3') [42]. The amplification of the highly polymorphic regions II and III of Pv TRAP [22] was obtained using the primers Pvsspf: 5' GTCGATTTATACCTCCTAGTTGACGG 3' and Pvsspr: 5' CAGCTTGAACGTCGACCTCCGC 3' with the following conditions: 5 min at 94°C; 1 min at 94°C, 1 min at 50°C, 2 min at 72°C for 35 cycles; and a final extension of 72°C for 2 min. The amplification of the central region of the Pv DBP_II domain was done using primers Liz-1 (5' TGGGGAGGAAAAAGATG 3') and Liz-2 (5' AACGGAACAAACGCATAAC 3') [43]. The PCR conditions used were as follows: 4 min at 94°C; 1 min at 94°C, 2 min at 58°C, 2 min at 72°C for 35 cycles; and a final extension of 72°C for 3 min. Pv MSP-1 primers described by Espinosa et al. (HARB1_f: 5' GGCAACAACGATGACGAC 3', HARB1_r: 5' CTTCTCAAGTCGAGCAG 3', HARB2_f: 5' TCACTGTAAAGAAATTGCAG 3' and HARB2_r: 5' ATACATGTGTGCTCGGAG 3') were used for amplifying the two HARB containing regions: Pv MSP-1₂₀ and Pv MSP-1₁₄ kDa [31] with the following conditions: 5 min at 94°C; 1 min at 94°C, 1 min at 50°C, 1 min at 72°C for 35 cycles; and a final extension of 72°C for 5 min for the first region and 5 min at 94°C; 1 min at 94°C, 1 min at 45°C, 1 min at 72°C for 35 cycles; and a final extension of 72°C for 5 min for the second region. Finally, a nested PCR was used to amplify domain I of Pv AMA-1 as described by Cheng and Saul [44]. PCR products were purified using the QIAquick PCR purifications kit (QIAGEN, Valencia, CA, USA) and sequenced using BigDye^® Terminator v3.1 sequencing kit (Applied Biosystems, Foster City, CA, USA), with the primers previously mentioned and an ABI 3100 automated sequencer.

Sequences were analysed using the Sequencher™ software version 4.7 (Gene Codes Corporation, Ann Harbor, MI, USA) and sequence alignments were performed using the MEGA software version 4.1 [45]. Comparison with previously deposited P. vivax sequences in the GenBank was performed using BLAST search. The number of haplotypes (h), haplotype diversity (hd) and nucleotide diversity (Pi) values were obtained using the DnaSP version 5.0 [46]. The effect of natural selection was also evaluated by estimating the difference of non-synonymous and synonymous substitutions (Dn-Ds) using the Nei and Gojobori's method with the Jukes and Cantor correction. The null hypothesis of neutrality was rejected if Dn:Ds ≠ 1. Positive values for Dn-Ds indicated positive selection while negative values indicated negative selection [47]. Standard error was calculated using bootstrap with 1,000 replications and rates of Ds and Dn were compared by the Z-test of selection using Mega version 4.1 [45]. To test whether the haplotypes clustered according to their geographic origin, the model-based clustering algorithm implemented in the Structure 2.1 software was used [48]. This software uses a Bayesian clustering approach to assign isolates to K populations characterized by a set of allele frequencies at each locus. The number of such populations may be either previously known or not. The program was run five times at different K values (from 2 to 10) with a burn-in period of 10,000 iterations followed by 50,000 iterations. The admixture model was used in all analyses, which allows for the presence of individuals with ancestry in two or more of the K populations. However, clustering algorithms incorporate stochastic simulation as part of the inference and, as a consequence, independent analyses of the same data may result in several distinct outcomes, even though the same initial conditions were used. To facilitate the interpretation of population-genetic clustering results, the computer program CLUMPP (Cluster Matching and Permutation Program) was used. CLUMPP strips away the 'label switching' heterogeneity so that the 'genuine multimodality' can be detected and quantified [49]. In addition, distruct 1.1 was used to graphically display the clustering results [50]. The genetic difference between clusters was measured by the fixation index (Fst) calculated from the haplotype frequencies (Hst) and pair-wise DNA sequence diversity (Kst*) using DnaSP version 5.0 [46]. The nearest-neighbour statistic (Snn) was also calculated between geographic populations. Snn is a measure of how often the "nearest neighbours" of sequences are from the same locality in geographic space. This statistic is applicable when genetic data are collected on individuals sampled from two or more localities. Snn is expected to be near one when the populations at the two localities are highly differentiated and near one-half when the populations at the two localities are part of the same pan-mictic population [51].

A 1,100 bp PCR product corresponding to Pvcsp was amplified from 106 P. vivax samples from Peru. In addition, 326 complete CSP sequences from Brazil, Colombia, India, Iran and Korea were also included in the amino acid alignment. Sequences from Thailand, China and PNG were used to determine the percentage of strains circulating of each Pv CSP type (Table 1); however, they were not considered in the haplotype analysis since the sequences were not complete. Most of the strains circulating worldwide belonged to the VK210 type and in most population sets analysed (except Colombia and Thailand) a greater percentage of VK210 haplotypes was found. The amino acid alignment of the sequences used in this study showed 97 different VK210 haplotypes and 28 VK247 haplotypes circulating worldwide. Different haplotypes were observed in South American and Asian countries, except for one VK247 Peruvian haplotype, which was also found in Iran [52]. Thirty-seven VK210 and 19 VK247 haplotypes were identified in South America; while in Asia, 51 and 10 haplotypes were observed from each type.

In Peru, 23 different VK210 subtypes and three different VK247 subtypes were found (Figure 1). Pv CSP displayed a different pattern of a nine amino acid oligopeptide repetitions depending on the subtype. The VK210 subtypes contained the repeat region G(D/N)RA(A/D/G/V)G(Q/H)PA either 16, 17, 18 or 19 times, while VK247 subtypes contained the repeated region ANGA(G/D)(D/N)QPG, 17 or 18 times. Region I of Pv CSP showed a polymorphism in position 3 (L/V), whereas the 18 amino acid long linear peptide EWTPCSVTCGVGVRVRRR, located on Region II [53, 54], was highly conserved. Pv CSP analysis of Peruvian isolates also revealed high variability of VK210 and VK247 haplotypes compared to other populations from South America [55, 56]. Region II-plus, which contains dominant B-cell epitopes [57, 1] and block hepatocyte invasion by sporozoites [58], was highly conserved in Peruvian isolates, with the same amino acid sequence as the Sal I strain.

A total of 104 sequences of regions II and III of Pvtrap were obtained from the Peruvian isolates. In addition, 63 Thailand sequences from GenBank were included in the amino acid alignment. The N-terminal region (209-256aa), which has been proved to be immunogenic and to provide partial protection in Aotus monkeys [24] was highly conserved with only three amino acid substitutions in positions 216 (R/K), 219 (H/Q) and 229 (G/K). In general, more diversity was found in Thailand (h = 34, π = 0.007 ± 0.00029 and Hd = 0.961 ± 0.01) (Table 2) compared to Peru (h = 7, π = 0.00235 ± 0.00012 and Hd = 0.641 ± 0.026) and according to the Dn-Ds test, regions II and III of Pvtrap were under positive selection (Table 3).

A 1,150 bp DNA fragment of Pvdbp was amplified from all 106 Peruvian isolates and 601 P. vivax sequences from Brazil, Colombia, India, Sri Lanka, Thailand, Korea, Iran and Papua New Guinea were also included in the analysis. The alignment of 510 bp from region II of dbp showed 150 different haplotypes, from which 39 were found in South America, 42 in Asia, 56 in Oceania and 13 were found in South America and Asia. In the Peruvian isolates 26 different haplotypes were found with genetic diversity values of π = 0.008 ± 0.00048 and Hd = 0.903 ± 0.014. In general, the average genetic and haplotype diversity in South America and Asia was very similar (Table 2). On the other hand, the analysis of DNA sequence polymorphisms showed a significant difference between synonymous and non-synonymous substitutions, indicating balancing selection in dbp _II (Table 3). In addition, several substitutions in the positions underlined were found in DBPII regions corresponding to peptides H1: F HR DITFRKLYLKRKL, H3: DEK AQ QRR KQWWN ES K an 1653: RD YVS ELPTEVQ KLKEK C, previously reported as immunogenic [27, 59].

According to the amino acid alignment of PvMSP-1₂₀ and PvMSP-1₁₄, a total of 28 haplotypes were found worldwide, 11 in South America, 14 in Asia and three in both continents. When comparing genetic diversity between populations, Peru (π = 0.00286 ± 0.00029, Hd = 0.665 ± 0.041) and Brazil (π = 0.00293 ± 0.00061, Hd = 0.75 ± 0.139) showed the highest levels of nucleotide and haplotype diversity, while in Korea, Thailand and Turkey these HARBs were highly conserved. In general, values of genetic diversity for Pvmsp-1 were low when compared with the other genes and no clear signature of selection was found according to the Dn-Ds test (Table 3).

The polymorphic region of Pvama-1 (Domain I of protein) was amplified and sequenced from 98 of the 106 Peruvian isolates. A total of 774 Pvama-1 partial sequences were used in the analysis, including samples from Brazil, Venezuela, India, Sri Lanka, China, Korea, Myanmar, Philippines, Thailand, Papua New Guinea and Solomon Islands. A total of 168 haplotypes were identified; 135 were exclusively found in Asia, nine in Oceania and 12 in South America. In addition, seven haplotypes were shared in America and Asia and five in Asia and Oceania. According to the analysis of polymorphisms, the dN-dS test indicated balancing selection in this domain (Table 3) with several haplotypes in low frequency. In addition, values of nucleotide diversity were very similar between geographic regions and populations (Table 2).

When analysing Plasmodium isolates, samples are usually grouped by geographic location; nevertheless, this approach assumes that geographic barriers or distance are the underlying factors that limit gene flow. It also assumes that genetic exchange within each population geographically defined is random. To better estimate the distribution of haplotypes of different P. vivax vaccine candidate antigens, a Bayesian clustering algorithm [48] was used to sub-group samples according to related SNPs haplotypes. The Structure model assumes that loci are independent within populations; however, this assumption is likely to be violated for sequence data, or data from non-recombining regions. It is a matter of debate whether it is valid to group sequences containing potentially linked polymorphisms in a single gene under balancing selection [60]. The main cost of the dependence within regions will be that Structure underestimates the uncertainty in the assignment of particular individuals. However, if there is enough independence across regions than LD within regions, then Structure may actually perform reasonably well [48]. Moreover, our main goal is not to describe the demographic history of the population but to get a graphically representation of the vaccine candidate's haplotype distribution. Genetic differentiation between clusters was calculated using Fst from haplotype-frequencies and pairwise DNA sequence diversity. The haplotype frequency-based statistics are more sensitive for small sample sizes, while the sequence-based statistic is a more sensitive method for detecting population structure in highly polymorphic loci [61, 62].

The results obtained with Pv TRAP, Pv DBP and Pv AMA-1 showed three different clusters. Significant differentiation was observed in haplotype and pair-wise comparisons between all clusters according to the Fst values (Table 4), except for Pvmsp-1, from which a significant number of clusters it was not possible to identify. Although admixed haplotypes were observed in Pvdbp and Pvama-1, the estimate of K = 3 best explained the distribution of haplotypes from all geographic populations; besides, the defined sub-groups were not geographically or even regionally restricted (Figure 2). Snn values between the different geographic populations were also calculated for each of the antigens. We observed that for most of the comparisons in DBP, AMA-1 (Table 5) and TRAP (Snn = 0.86803*), Snn values were near one, which indicates that the localities are highly differentiated. However, close geographic populations such as Sri Lanka and India in South Asia and Peru and Brazil in South America showed almost not genetic differentiation (Snn = 0.58, p < 0.01 and Snn = 0.65, p < 0.01). Thus, relative low Snn values were found between Brazil, India and Sri Lanka, since there are shared haplotypes circulating in South America and South Asia. Moreover, previous work also indicated low genetic differentiation using Fst analysis between dbpII sequences from Brazil and Iran, India or Sri Lanka [63]. On the other hand, at least seven of the pairwise comparisons made using MSP-1 resulted in non-significant values.