Genetic evidence for contribution of human dispersal to the genetic diversity of EBA-175 in Plasmodium falciparum

This study was approved by the Institutional Review Board of the Faculty of Tropical Medicine, Mahidol University, and the Research Ethics Committee of the School of Medicine, The University of Tokyo.

Peripheral blood samples were obtained from 203 P. falciparum-infected patients from Thailand. In addition, a dataset of nucleotide sequences including isolates of human P. falciparum, gorilla (Gorilla gorilla) Plasmodium, chimpanzee Plasmodium, and P. reichenowi was built from data obtained from [12].

Genomic DNA was extracted from pretreated peripheral blood samples from patients infected with P. falciparum using a QIAamp Blood Kit (Qiagen, Hilden, Germany). DNA fragments covering a part of the eba-175 coding sequence (regions II and III) of P. falciparum were amplified by PCR using the following sets of primers: EBA175-fragment1, 5′-ggaagaaatacttcatctaataacg-3′ (forward) and 5′-catcctttacttctggacacatcg-3′ (reverse), and EBA175-fragment2, 5′-gagactctgaaggttgaatgcaa-3′ (forward) and 5′-aggtgtattagacatatcttggtc-3′ (reverse). These primers were designed based on the eba-175 reference sequence from P. falciparum (GenBank accession no. X52524). PCR amplification was performed in a 13.0-µL reaction mixture containing 0.125 µL (0.125 µM) each of forward and reverse primers, 0.125 µL TAKARA LA Taq™ (5 units/µL), 1.25 µL 10× LA PCR™ Buffer II (Mg²⁺ free), 1.25 µL 25 mM MgCl₂, 1.25 µL 2.5 mM dNTP mixture, 0.5 µL (5 ng) of genomic DNA template, and 8.375 µL dH₂O using a GeneAmp^® PCR System 9700 (Applied Biosystems, Foster City, CA, USA). The PCR cycling conditions for each primer pair were 60 s initial denaturation at 94°C, followed by 40 cycles of 30 s denaturation at 94°C, 30 s annealing at 56°C, and 150 s extension at 72°C, and a final step of 5 min extension at 72°C. The PCR products were subsequently sequenced using an ABI Prism 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Primer sequences used for direct sequencing are available upon request. The isolates showing multiple superimposed electropherogram peaks at a single site following PCR-direct sequencing and a secondary peak greater than 30% of the predominant peak were considered to be mixed infections and excluded from further analyses. Low-quality sequences (i.e., high background noise or too weak signal) were also excluded. As a result, 131 sequences representing the single or most abundant sequence in each DNA sample were included in the analyses.

The nucleotide sequences obtained were aligned and translated into putative amino acid sequences using MEGA v.5.2 [13]. To examine the phylogenetic relations among 32 distinct eba-175 Thai P. falciparum alleles and two eba-175 P. reichenowi alleles (CBXM000000000 and AJ251848), a maximum likelihood (ML) tree was constructed based on the Hasegawa–Kishino–Yano model [14]. To obtain the ML tree, a nearest-neighbor-interchange (NNI) search was applied. In addition, a neighbor-joining (NJ) [15] tree was generated using 194 eba-175 partial region II sequences from P. falciparum worldwide and two P. reichenowi sequences (CBXM000000000 and AJ251848), based on the Nei–Gojobori (NG) model [16] and the Jukes–Cantor (JC) correction [17]. The construction of phylogenetic trees and estimation of best-fit substitution model for the ML tree were implemented in MEGA v.5.2 [13]. All the positions containing insertions/deletions were eliminated from the analyses (complete deletion option). Branch support values were computed by bootstrap analyses with 1,000 replications. A network of 194 P. falciparum and two P. reichenowi eba-175 alleles was also constructed based on synonymous substitutions using the neighbor-net method [18] in SplitsTree4 ver. 4.13.1 [19].

The time to the most recent common ancestor (tMRCA) of the eba-175 alleles from Thai falciparum was estimated from the linearized tree based on synonymous substitutions among the 32 distinct eba-175 alleles by using the MEGA v5.2 [13]. The neutral substitution rate was calculated assuming that P. falciparum and P. reichenowi diverged 6 million years ago (MYA) [5, 20]. In addition, the MCMCTree program in the PAML 4.8 package [21] was used to estimate tMRCA based on the amino acid sequences. The minimum and maximum age constraints on the root age (the divergence time between P. falciparum and P. reichenowi) were set to 5 and 7 MYA, respectively. The tMRCA estimation was based on a WAG model [22] for amino acid substitutions. In the MCMC process, sampling occurred every 100 generations for 10,000 generations and the first 50,000 generations were discarded as burn-in.

To detect the signatures of natural selection, the number of nonsynonymous substitutions per nonsynonymous site (d _N) and synonymous substitutions per synonymous site (d _S) for all the pairs formed by the 32 distinct Thai P. falciparum alleles, 16 from chimpanzee Plasmodium spp., 11 from gorilla Plasmodium spp., and four from P. reichenowi sequences were estimated using the NG model with the JC correction in MEGA v.5.2 [13]. Significant difference between d _N and d _S was assessed by Wilcoxon signed-rank test. For all the 131 Thai P. falciparum isolates, the numbers of nonsynonymous substitutions per nonsynonymous site (π _N) and synonymous substitutions per synonymous site (π _S) were also calculated in the same manner as d _N and d _S. In addition, the McDonald–Kreitman (MK) test [23] was performed for detecting natural selection signal using DnaSP v5 software [24]. Tajima’s D test [25] was performed for 131 Thai P. falciparum eba-175 sequences using DnaSP v5 software [24], where the test statistic was analytically calculated. A two-sided P value of less than 0.05 was considered statistically significant.

A Wu–Kabat plot was used to estimate the level of amino acid variability for the 32 distinct Thai P. falciparum eba-175 alleles [26]. The Wu–Kabat plot estimates the level of variability for each amino acid position in the sequence alignment, measured as the number of amino acids at each site divided by the maximum frequency of amino acid for all sites.

The nucleotide sequences of regions II and III of P. falciparum eba-175 were obtained by PCR-direct sequencing. The eba-175 region III showed highly divergent dimorphic segments, the F and C segments [termed Fseg (423 bp) and Cseg (342 bp)], as reported by Ware et al. [27]. A total of 32 distinct alleles [20 Fseg alleles (ca. 2,740 bp) and 12 Cseg alleles (ca. 2,660 bp)], defined by 30 polymorphic sites including insertions/deletions (site 744–749, and 2094), were detected from 131 P. falciparum isolates from Thailand [Fig. 1; Fseg alleles: F1_1–F20 (Genbank accession numbers LC008232–LC008251) and Cseg alleles: C1–C12 (Genbank accession numbers LC008252–LC008263)]. The nucleotide sequences of the 32 alleles were translated into 31 distinct amino acid sequences. Synonymous substitutions were found at only two sites in the allele sequences. On the other hand, nonsynonymous substitutions were found at 21 sites, not including an insertion/deletion site. The F and C segments were excluded from further analyses.

A ML tree was constructed based on the nucleotide sequences of 32 eba-175 alleles from Thai P. falciparum, and two eba-175 alleles from P. reichenowi (Fig. 2). In this ML tree, Fseg alleles formed a monophyletic clade with a relatively low bootstrap value (51%). A NJ tree was generated based on synonymous substitutions in region II to further analyze the phylogenetic relations among 194 P. falciparum eba-175 alleles, including a large sequence dataset from Genbank [12] (Fig. 3). In particular, in this NJ tree, some P. falciparum alleles isolated from Asia formed a single clade rooted in the most basal node from where all P. falciparum sequences diverged. All the alleles included in the Asian clade were characterized by a mutation in site 441, as shown in Fig. 1 (Asian-specific alleles contain cytosine nucleotide at site 441, where all other sequences contain a thymine). This Asian-specific cluster was also supported by the neighbor-net method (Additional file 1) and suggests that an ancestral allele in the cluster emerged in Asia and then rapidly spread across Asia after the out-of-Africa.

The d _s values for regions II and III for all the pairs formed by the 32 distinct eba-175 alleles were calculated to estimate tMRCA of eba-175 alleles in the Thai population. The neutral mutation rate of the eba-175 gene was extrapolated from the calibration point derived from the divergence time of P. falciparum and P. reichenowi (6 MYA) [5, 20] by using the linearized tree of eba-175 alleles in the MEGA program [13]. Thus, the estimated neutral mutation rate was 1.0 × 10⁻⁸ per site per year. Accordingly, the estimated tMRCA for eba-175 was approximately 0.14 MYA. The number of synonymous substitutions observed in Thai P. falciparum isolates was small (maximum d _S value = 2); consequently, MCMCTree was used to estimate the tMRCA of eba-175 alleles based on amino acid substitutions within region III. In this case, the estimated tMRCA was approximately 0.13 MYA (95% highest posterior density confidential interval = 0.07–0.23 MYA). Notably, these estimations rely on the assumption that P. falciparum and P. reichenowi diverged 6 MYA.

To search a signal of natural selection on region II of eba-175 in Thai P. falciparum, the d _N and d _S values for 32 long sequences (2,310 bp) were calculated (Table 1). A comparison of the two values revealed that the d _N was significantly larger than d _S (P value <2.2 × 10⁻¹⁶, Wilcoxon signed-rank test), suggesting that diversifying selection has influenced eba-175 diversification in human P. falciparum. In addition, this result was further supported by the comparison between π _N and π _S, including the allele frequencies of 131 Thai P. falciparum isolates (π _N/π _S = 2.62; P value <2.2 × 10⁻¹⁶, Wilcoxon signed-rank test). Subsequently, it was examined whether diversifying selection has also operated within other taxa, using sequence data from 16 and 11 Plasmodium spp. infecting chimpanzees and gorillas, respectively, and four P. reichenowi isolates (Table 1). Long sequence data from African apes Plasmodium isolates were unavailable; consequently, region II short sequences (390–396 bp) were used for this analysis. Similar to the results using long sequences, d _N was significantly larger than d _S in human P. falciparum (P value <2.2 × 10⁻¹⁶, Wilcoxon signed-rank test); on the other hand, d _N was significantly smaller than d _S in African apes Plasmodium (each P value <2.2 × 10⁻¹⁶, Wilcoxon signed-rank test for chimpanzee and gorilla Plasmodium isolates). No significant differences between d _N and d _S were detected for P. reichenowi (P value = 0.32, Wilcoxon signed-rank test). These results suggest that diversifying selection has only acted on eba-175 in human P. falciparum.

The McDonald–Kreitman (MK) test [23] for sequences of eba-175 region II in 32 Thai P. falciparum alleles and one in P. reichenowi showed that the ratio of nonsynonymous to synonymous polymorphic sites within species (Pn/Ps = 19/1) was significantly higher than that of fixed sites between species (Dn/Ds = 116/48; P = 0.017, Fisher’s exact test). However, the MK test for Plasmodium isolates from African apes did not show significant differences between Pn/Ps and Dn/Ds (Pn/Ps = 59/23 and Dn/Ds = 14/6, P-value = 1.00 for 16 chimpanzee Plasmodium isolates and one P. reichenowi; Pn/Ps = 81/29 and Dn/Ds = 6/4, P = 0.46 for 11 gorilla Plasmodium isolates and one P. reichenowi). The results from the MK test suggest that either purifying selection or diversifying selection has acted on eba-175 in Thai P. falciparum population.

Tajima’s D test [25] was used to test for departure from selective neutrality in 131 eba-175 sequences of Thai P. falciparum. This test compares two population genetic parameters, one estimated from mean pairwise nucleotide differences and the other from the number of mutations. The results revealed a significant positive Tajima’s D value (Tajima’s D statistic = 3.01 and P value <0.01), suggesting the balancing selection on eba-175 in Thai P. falciparum. The possibility of a recent reduction in population size that provided the significant positive Tajima’s D statistic may be excluded, since both the d _N/d _S ratio test and the MK test also suggested the existence of positive diversifying selection.

To assess the relation between amino acid variability and the erythrocyte binding site, the level of amino acid variability at each site was examined using a Wu–Kabat plot (Additional file 2). The Wu–Kabat plot showed that the level of amino acid variability at some codon sites outside the sites of direct interaction between P. falciparum EBA-175 and human GYPA molecules [28] was higher than the variation level at their interaction sites. These variable sites may be located in regions recognized by human antibodies if the higher degree of variation is caused by diversifying selection favoring mutations to other amino acids.