Complete gene sequence diversity, amplification, and natural selection of the Plasmodium vivax Duffy binding protein locus Authors: Chaturong Putaporntip, Napaporn Kuamsab, Taweesak Tia, Yoshiaki Takashima, Somchai Jongwutiwes Published: 2025 ABSTRACT AbstractBackgroundThe Duffy binding ligand domain of Plasmodium vivax (PvDBPII) interacts with Duffy blood-group antigen on the surface of reticulocytes during host cell invasion process. Although diversity of this domain has been analysed among worldwide isolates, the evolution of the complete PvDBP gene remains to be explored.MethodsIn total, 232 P. vivax isolates from northwestern, northeastern, eastern, and southern Thailand were included for analyses of the complete gene sequence, copy number variation and duplication genotypes of Pvdbp. The Duffy blood-group genotypes of the patient were also determined.ResultsPvdbp duplication, all of the Cambodian type, was detected in 22 isolates (9.48%), characterized by identical copies (n = 17) and microheterogeneity between copies (n = 5). Across 254 sequences including duplicated copies, nucleotide diversity was highest in PvdbpII. Positive selection, evidenced by excess nonsynonymous over synonymous substitutions, occurred in both the 5′ region and PvdbpII. Codon-based analysis revealed 26 positively selected codons distributed across the protein; 22 of these overlapped predicted B cell and/or T cell epitopes. Insertions and deletions including frameshift mutations were found outside PvdbpII apart from an indel in this domain, while recombination breakpoints were confined to PvdbpII and the 3′ region. Parasite isolates from each endemic region displayed population structure. No significant difference in Duffy phenotype frequencies was found between malaria patients in this study and healthy blood donors from reported data.ConclusionsWhile PvdbpII remains the most variable domain, sequence diversity occurred outside this domain. Cambodian-type Pvdbp duplications, often with identical sequences, may enhance immune evasion via increased gene dosage. Positive selection is broadly distributed and likely driven by host immune pressure. Whether positive selection outside PvdbpII is relevant for vaccine design requires further investigation.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12936-025-05756-1.Keywords: Malaria, Plasmodium vivax Duffy binding protein, Frameshift mutation, Genetic differentiation, Natural selection, Copy number variation, Duffy blood group antigen Background Plasmodium vivax causes chronic relapsing malaria and represents a major global health burden [1]. Its resilience to control and elimination stems from the presence of hypnozoites and, more recently, the recognition of an endosplenic life cycle, both of which contribute to persistent low-density infections and sustained transmission [2–4]. Consequently, vaccine development is considered a critical complement to conventional control strategies. During the multistep process of erythrocyte invasion, the interaction between the P. vivax Duffy binding protein (PvDBP) on the merozoite surface and the Duffy antigen receptor for chemokines (DARC or Duffy blood-group antigen) on reticulocytes is essential for establishing blood-stage infection [5, 6]. Blocking this interaction prevents merozoite entry into erythrocytes, thereby halting the intraerythrocytic cycle and abrogating subsequent malaria pathogenesis. Therefore, PvDBP has been an attractive vaccine target [5, 6]. PvDBP is a 140-kDa transmembrane protein encoded by a five-exon gene. Exons 1–3 encode an N-terminal signal peptide, six extracellular regions, and a transmembrane domain, while exons 4–5 and part of exon 3 encode the cytoplasmic tail [6–9]. Considerable sequence diversity has been observed in the Duffy binding ligand domain (PvDBPII) among isolates, both within and across geographic origins [10, 11]. Antibodies against PvDBP can elicit either variant-specific or strain-transcending reactivity [12, 13]. The sequence diversity in the functional receptor-binding site of PvDBP is therefore a major concern for effective vaccine design. The Pvdbp gene occurs in single or multiple copies per genome, with copy number variation (CNV) showing marked geographic heterogeneity [14, 15]. Two distinct duplication types have been identified: the Malagasy type, characterized by a longer duplication, and the Cambodian type [15]. Increased Pvdbp copy number has been shown in vitro to confer host immune evasion and protect the parasite against invasion-inhibitory antibodies [16]. Most investigations into Pvdbp diversity have focused on the Duffy binding ligand domain (PvdbpII), while relatively little is known about polymorphisms in other regions with potential structural or functional significance [10, 17]. To address this gap, complete Pvdbp gene sequences were analysed using Sanger sequencing and assessed the distribution of CNVs among P. vivax populations in Thailand. In addition, the Duffy phenotypes of sampled populations were determined. Results revealed point mutations distributed in all regions except the transmembrane-encoding domain, while insertion/deletion of codons, frameshift mutations and intragenic recombination also contributed to diversity at this locus. Notably, most positively selected codons corresponded to predicted immunogenic epitopes, suggesting that host immune pressure plays a significant role in shaping the evolution of Pvdbp. Methods Plasmodium vivax population Blood samples were collected from symptomatic malaria patients who attended malaria clinics or district hospitals in northwestern (Tak province), northeastern (Ubon Ratchathani province), eastern (Chanthaburi province), and southern (Yala and Narathiwat provinces) Thailand. Samples were collected between 2008 and 2010, and additional samples from Tak and Ubon Ratchathani were obtained in 2007 and 2014, respectively. The male-to-female ratio was 2.8:1, and the age range was 4–66 years (mean, 29.7 years). The majority of the patient population was Thai (73.7%), and the rest were cross-border migratory foreigners. Based on available data, most patients (82.8%, n = 122) experienced their first episode of malaria, and the geometric average of parasite density was 6451 parasites/μL (range, 440 – 44,120, n = 60)(Additional file 1). The local annual parasite incidence (API) fluctuated over time during the sample collection periods (Additional file 2). Diagnosis of P. vivax infection was confirmed by microscopy and species-specific polymerase chain reaction (PCR) as previously described [18]. DNA was extracted and stored at -40 °C until use. All isolates were genotyped using allele-specific primers targeting the polymorphic block 6 of the merozoite surface protein 1 (Pvmsp1) and the merozoite surface protein 3γ (Pvmsp3γ) genes [19, 20]. Only isolates containing single alleles at both loci were included in subsequent analyses. Amplification of the complete Pvdbp gene The Pvdbp gene was amplified by PCR using the forward primer (PvDBPF0: 5′- AAGCTTTTAAAAATAGCAACAAAATTTCG -3′) and the reverse primer (PvDBPR0: 5′- GACTTGTATGCACTTGTTACAAATG-3′) derived from the upstream and downstream sequences, respectively, of the Salvador 1 strain (GenBank accession no. M61095). Each 25-µL PCR reaction contained template DNA, 2.5 mM MgCl₂, 300 µM of each dNTP, 2.5 µL of 10 × ExTaq PCR buffer, 0.3 µM of each primer, and 0.5 units of ExTaq DNA polymerase (Takara, Seta, Japan). Cycling conditions consisted of an initial denaturation at 94 °C for 1 min, followed by 35 cycles of 96 °C for 20 s, 58 °C for 30 s, and 72 °C for 4 min, with a final extension at 72 °C for 10 min. For reamplification of selected isolates to confirm singleton substitutions and rare indels, Phusion™ High-Fidelity DNA polymerase (Thermo Fisher Scientific) was used. Estimation of PvDBP copy number Pvdbp copy number was determined by quantitative real-time PCR (qPCR) using the StepOnePlus™ Real-Time PCR System (Thermo Fisher, Foster City, USA), with Pvmsp1 as the single-copy reference locus [19]. The absence of naturally occurring null mutants in a large whole genome sequencing dataset of P. vivax across endemic areas has encouraged the use of this locus as the reference gene [21]. Each 20-µL reaction contained 5 µL of genomic DNA, primers PvDBPFG (5′-AGGTGGCTTTTGAGAATGAA-3′) and PvDBPRG (5′-GAATCTCCTGGAACCTTCTC-3′) for Pvdbp, and primers VMSP19BF (5′-TACAAAGGATCCGAGAAAAAAAATGAAGTG-3′) and VMSP19HR (5′-TTTGTGAAGCTTTTAAAGCTCCATGCACAGG-3′) for Pvmsp1. The cycling profile included denaturation at 95 °C for 3 min, followed by 40 cycles of 95 °C for 30 s and 55 °C for 30 s, with a final extension at 68 °C for 1 min. Cycle threshold values were set between 15 and 35 cycles. Each sample was analysed in triplicate across two independent assays. Copy number was calculated as n-fold differences relative to Pvmsp1 in the calibrator genome, using the 2−ΔΔCt method [22, 23]. Pvdbp duplication-specific PCR Duplication-specific PCR assays were performed following established protocols [14, 15]. Three primer pairs were used: (i) AF (5′-CCATAAAAGGTAGGAAATTGGAAA-3′) and AR (5′-GCATTTTATGAAAACGGTGCT-3′), (ii) BF (5′-TCATCGAGCATGTTCCTTTG-3′) and BR (5′-TTGCACGTACTCGAAACTCAG-3′), and (iii) AF2 (5′-ACGCGATGTATCTTCTTTTCA-3′) and AR2 (5′-TAGAACGCACAGTTATTGGC-3′). These yielded amplicons of 613, 643, and 657 bp, respectively. Malagasy-type duplications produced a 584-bp product with BF and AR, whereas Cambodian-type duplications generated a 736-bp product with BF and AR2. DNA sequencing Purified PCR products of the complete Pvdbp amplicons were sequenced bidirectionally using forward and reverse primers to generate overlapping reads (Qiagen PCR Purification Kit, Qiagen, Hilden, Germany). Singleton substitutions were verified by sequencing independent PCR products derived from the same genomic DNA. Cloning into plasmid vector To determine whether duplicated Pvdbp sequences were identical or distinct, PCR products from isolates with relative quantification values > 1.5 by qPCR were cloned into the pGEM-T vector and transformed into Escherichia coli JM109. DNA templates from 5—10 randomly selected recombinant colonies were sequenced for analysis. Duffy genotyping Duffy (FY) blood group genotyping was performed using allele-specific PCR (ASP-PCR) as described by Olsson [24]. For the Fya genotype, primers FYAB2 (5′-CTCATTAGTCCTTGGCTCTTAT-3′) and FYAREV (5′-AGCTGCTTCCAGGTTGGCAC-3′) yielded a 711-bp amplicon, while the Fyb genotype produced the same product using FYAB2 and FYBREV2 (5′-AGCTGCTTCCAGGTTGGCAT-3′). To detect the Duffy-negative genotype, FYB2 was replaced with GATAFY2 (5′-CTCATTAGTCCTTGGCTCTTAC-3′), identifying the T → C substitution in the FY promoter [24]. Primers K-HuGroHo-left (5′-TGCCTTCCCAACCATTCCCTTA-3′) and K-HuGroHo-right (5′-CCACTCACGGATTTCTGTTGTGTTTC-3′) were used as positive controls, amplifying a 434-bp fragment of the human growth hormone (HGH) gene [25]. Data analysis Nucleotide sequences were aligned using CLUSTAL with manual adjustment to preserve reading frames. Amino acid positions were referenced to the Salvador 1 strain (GenBank accession no. M61095) [7–9]. Coding regions were defined as follows: (i) 5′ region: codons 1–197, (ii) Duffy binding ligand domain: codons 198–522, (iii) 3′ region: codons 523–1070 [5]. The signal peptide spanned residues 1—22, the predicted transmembrane domain residues 1006–1026, and the cytoplasmic tail residues 1027–1070 [9]. Tandem repeats were identified using Tandem Repeats Finder v4.0 [26]. Mutation statistics, parsimony-informative sites, haplotypes, and haplotype diversity (h) were calculated using DnaSP v5.10.01 [27]. Nucleotide diversity (π), synonymous substitution rates (dS), and nonsynonymous substitution rates (dN) were estimated with MEGA v6.0, with standard errors obtained from 1,000 bootstrap pseudoreplicates [28]. Departure from neutrality was tested by comparing dS and dN using the Z-test (p < 0.05). Intragenic recombination signals were identified using the Genetic Algorithm for Recombination Detection (GARD)[29]. Codon-specific selection was analysed using the Fast, Unconstrained Bayesian AppRoximation (FUBAR) method, excluding recombination events [30]. Population differentiation (FST) was inferred using the analysis of molecular variance approach (AMOVA), with significance assessed by permutation in Arlequin 3.5.2.2 [31]. For population assignments, isolates from Yala and Narathiwat provinces were grouped together as a single endemic area (Yala-Narathiwat) due to geographic proximity and similar transmission patterns [32]. Linear B-cell epitopes were predicted using the epitope1D method [33]. Potential HLA class II–binding peptides were analysed with the IEDB recommended 2.22 algorithm [34], using a peptide length of 12–18 residues. Analyses focused on common HLA-DR, -DQA, and -DQB alleles in the Thai population (allele frequency > 0.05)[35]. Candidate epitopes were defined as peptides with percentile rank < 10 and predicted HLA-binding affinity (IC50) ≤ 1000 nM [36]. Meaningful statistical analysis of parasite density and history of previous malaria was precluded by a lack of sufficient data for these variables (Additional file 1). Results Copy number variation and genotype of Pvdbp duplication The amplification efficiencies of the primers, at 93.4% for Pvmsp1 and 97.6% for Pvdbp, along with their respective correlation coefficients (R2 values) exceeding 99%, were confirmed to be within the acceptable range for accurate application of the 2−ΔΔCt method [22](Additional file 3). Quantitative PCR analysis of 232 P. vivax isolates revealed that 22 (9.48%) carried multiple copies of Pvdbp (Table 1). Isolates with Pvdbp duplication were detected in all endemic provinces, with the highest prevalence observed in Ubon Ratchathani. Of these, 21 isolates harboured two copies, while one isolate from Tak contained three copies. Genotyping with allele-specific primers demonstrated that all duplicated isolates belonged to the Cambodian type; the Malagasy type was not detected (Table 1). Table 1Copy number and genotype of Pvdbp duplicationProvinceTakUbon RatchathaniChanthaburiYala-NarathiwatAll (%)No. isolates64526353232 (100)No. copy*Single57445851210 (90.52)Double685221 (9.05)Triple10001 (0.43)Total# (%)7 (10.94)8 (15.38)5 (7.94)2 (3.77)22 (9.48)Genotype of Pvdbp duplicationCambodian type785222 (100)Malagasy type00000 (0)Sequences§Identical384217 (77.27)Microheterogeneity40105 (22.73)*Quantitative real-time PCR#Total isolates with Pvdbp amplification§Sequence identity of gene copies in isolates with gene amplification Copy number and genotype of Pvdbp duplication *Quantitative real-time PCR #Total isolates with Pvdbp amplification §Sequence identity of gene copies in isolates with gene amplification Sequence variation of duplicated Pvdbp copies Among the 22 isolates with Pvdbp duplication, 17 (77.27%) displayed completely identical sequences across gene copies, as indicated by clear, non-superimposed electropherogram signals and consistent sequences from recombinant plasmid clones. In contrast, four isolates from Tak and one from Chanthaburi exhibited microheterogeneity between duplicated copies, evidenced by superimposed electropherogram signals and concordant nucleotide substitutions among plasmid clone sequences (Additional file 4). Three isolates (APH88, TSY2954, and CT152) differed by a single nucleotide, while APH61 and APH62 contained two and three nucleotide substitutions, respectively. In total, six codon differences were identified between duplicated copies, most of which were nonsynonymous substitutions located in the 3′ region of the gene (Table 2). Table 2Microheterogeneity of duplicated copies in PvdbpOriginIsolate (no. copies)Condon changes between copiesDomainTakAPH61 (2)c.2189C > T, p.A730V; c.2873G > T, p.R958I; c.3156 T > C, p.P10523′ region; cytoplasmic tailTakAPH62 (3)c.2250A > G, p.K750; c.2255A > G, p.E752G3′ regionTakAPH88 (2)c.2873 T > G, p.I958R3′ regionTakTSY2954 (2)c.2250A > G, p.K7503′ regionChanthaburiCT152 (2)c.1373 T > A, p.I458KDuffy binding ligandNonsynonymous substitutions are underlined Microheterogeneity of duplicated copies in Pvdbp Nonsynonymous substitutions are underlined Genetic diversity To assess sequence diversity, both single- and multi-copy isolates were analysed. Among 254 sequences derived from 232 isolates, 102 Pvdbp haplotypes were identified. The number of haplotypes was highest in Tak, followed by Ubon Ratchathani, Chanthaburi, and Yala-Narathiwat. Haplotype diversity in Tak, Chanthaburi, and Ubon Ratchathani was greater than that observed in Yala-Narathiwat (Table 3). Within the gene, haplotype diversity was comparable between the Duffy binding ligand domain and the 3′ region, both higher than in the 5′ region. The overall nucleotide diversity of Pvdbp was lowest in Yala-Narathiwat. Consistently, the Duffy binding ligand domain showed greater nucleotide diversity than either the 5′ or 3′ regions across all parasite populations (Table 3). In total, 27 amino acid substitutions were detected within the Duffy binding ligand domain and 18 within the critical binding motifs of PvDBP (residues 198–522 and 246–415 of the Salvador 1 strain, respectively [5, 37], generating 32 and 20 distinct haplotypes in these regions (Additional file 5). Table 3Haplotype, nucleotide diversity and rate of nucleotide substitutions in PvdbpProvince/DomainNNt sitesMPHh ± S.Dπ ± S.EdS ± S.EdN ± S.ETak715′ Region5916580.657 ± 0.0340.0029 ± 0.00120.0000 ± 0.00000.0059 ± 0.0027#Binding Ligand975–9782723180.880 ± 0.0210.0068 ± 0.00180.0034 ± 0.00180.0091 ± 0.0020#3′ Region1638–16653729430.977 ± 0.0080.0037 ± 0.00090.0049 ± 0.00190.0040 ± 0.0010All3204–32317057500.986 ± 0.0060.0045 ± 0.00070.0029 ± 0.00100.0052 ± 0.0009Ubon Ratchathani605′ Region561–5916460.571 ± 0.0540.0024 ± 0.00120.0000 ± 0.00000.0065 ± 0.0028#Binding Ligand975–9782321190.884 ± 0.0240.0070 ± 0.00170.0025 ± 0.00230.0093 ± 0.0025#3′ Region1638–16652613310.956 ± 0.0130.0030 ± 0.00070.0040 ± 0.00220.0033 ± 0.0009All3180–32315538330.960 ± 0.0130.0041 ± 0.00070.0025 ± 0.00110.0051 ± 0.0009Chanthaburi685′ Region5914450.755 ± 0.0180.0028 ± 0.00120.0000 ± 0.00000.0055 ± 0.0026#Binding Ligand975–9782421150.881 ± 0.0200.0067 ± 0.00160.0030 ± 0.00230.0098 ± 0.0024#3′ Region1644–16651615230.899 ± 0.0230.0029 ± 0.00090.0037 ± 0.00240.0031 ± 0.0010All3210–32314440270.903 ± 0.0230.0041 ± 0.00060.0024 ± 0.00110.0049 ± 0.0039Yala-Narathiwat555′ Region5913330.521 ± 0.0260.0026 ± 0.00140.0000 ± 0.00000.0048 ± 0.0026Binding Ligand975–978181560.637 ± 0.0480.0051 ± 0.00150.0037 ± 0.00270.0097 ± 0.00273′ Region1644181170.632 ± 0.0530.0029 ± 0.00080.0044 ± 0.00180.0036 ± 0.0012All3210–3213392980.651 ± 0.0520.0035 ± 0.00070.0030 ± 0.00100.0054 ± 0.0010All Regions2545′ Region561–59185110.679 ± 0.0170.0028 ± 0.00120.0000 ± 0.00000.0067 ± 0.0024##Binding Ligand975–9783328330.881 ± 0.0110.0072 ± 0.00180.0033 ± 0.00150.0095 ± 0.0023#3′ Region1638–16653932750.959 ± 0.0050.0036 ± 0.00090.0046 ± 0.00250.0040 ± 0.0011All3180–323180651020.968 ± 0.0050.0045 ± 0.00070.0078 ± 0.00100.0052 ± 0.0010N: number of sequences; Nt sites: nucleotide sites; M: number of mutation: P: parsimony informative site; H: haplotype; h: haplotype diversity; π: nucleotide diversity; dS: rate of synonymous substitutions per synonymous site; dN: rate of nonsynonymous substitutions per nonsynonymous siteTests of the hypothesis that dS equals the corresponding dN: #p < 0.05; ##p < 0.01 Haplotype, nucleotide diversity and rate of nucleotide substitutions in Pvdbp N: number of sequences; Nt sites: nucleotide sites; M: number of mutation: P: parsimony informative site; H: haplotype; h: haplotype diversity; π: nucleotide diversity; dS: rate of synonymous substitutions per synonymous site; dN: rate of nonsynonymous substitutions per nonsynonymous site Tests of the hypothesis that dS equals the corresponding dN: #p < 0.05; ##p < 0.01 Repeats, insertion/deletion and frameshift mutations Two imperfect nucleotide repeat units comprising GGAGGG(or A)AAAAGATGGAGG(or A)AGATAAAACTGG(or A) (nucleotides 351–410), were identified in nearly all isolates. One isolate (UBN21, Ubon Ratchathani) exhibited a deletion of one repeat unit, encoding KDGGDKTGEE (residues 120–129). Interestingly, similar deletions occur in Plasmodium cynomolgi orthologues PcyDBP1 and PcyDBP2 (GenBank JQ422035, XM004220981) (Fig. 1A). An insertion of codon CTA (lysine) between residues 429 and 430 of the Salvador 1 strain occurred in 30 isolates (12.93%). A deletion of codons S680 and D681 was observed in 15 isolates (5.91%), while a nine-codon insertion encoding AEFAESTKS between residues 683 and 684 occurred in 9 isolates (3.54%) (Fig. 1B). a Deletion of one repeat unit in isolate UBN21 compared to the Salvador 1 strain and the corresponding regions (bold and underlined) in Pcydbp1 (GenBank accession nos. JQ422035) and Pcydbp2 (XM004220981), b indels in isolates AP248 and APH46, and c frameshift mutations between codons 750 and 755 (boxed) among Thai isolates. Dash represents deletion Two of the three Pvdbp copies in isolate APH62 (APH62b and APH62c) carried a frameshift mutation (c.2251delG), followed by insertion of G between nucleotides 2258 and 2264, which restored the reading frame. A similar frameshift (c.2252delA) with compensatory insertion preserving the integrity of the following coding region was detected in isolates AP284 and APH43 from Tak (Fig. 1C). Notably, these insertions, deletions, and frameshifts occurred outside the Duffy binding ligand domain. Domain-based analysis of natural selection In Thai isolates, the 5′ region and the Duffy binding ligand domain exhibited significantly higher nonsynonymous (dN) than synonymous (dS) substitution rates, whereas the 3′ region showed no significant difference. At the population level, dN exceeded dS significantly in Tak, Ubon Ratchathani, and Chanthaburi, but not in Yala-Narathiwat (Table 3). Codon-based analysis of natural selection FUBAR analysis identified 26 codons under positive selection: 4 in the N-terminal region (including residue 17 in the signal peptide), 10 in the Duffy binding ligand domain, and 12 in the C-terminal region (including residues 1062 and 1067 in the cytoplasmic tail)(Fig. 2). Four codons (N662, K750, G1003, P1052) were under negative selection in the 3′ region. Amino acid substitutions in PvDBP showing the corresponding positions for positively selected residues detected by FUBAR (red); predicted B-cell epitopes (green), known linear B-cell epitopes (blue); predicted T-cell epitope (grey); known T-cell epitopes (lavender). Amino acid substitutions in red are singletons. Boundaries of domains are marked above amino acid positions Intragenic recombination GARD analysis revealed two recombination breakpoints in Tak and Ubon Ratchathani populations, and three in Chanthaburi and Yala-Narathiwat. Approximately half of the breakpoints occurred in the Duffy binding ligand domain, with the remainder in the 3′ region (Table 4). Table 4Intragenic recombination in PvdbpPopulationNo. breakpointsLocationBetween nucleotide positions*Tak23′ region1586 and 15873′ region2262 and 2263Ubon Ratchathani2Binding domain1287 and 12883′ region2138 and 2139Chanthaburi3Binding domain781 and 782Binding domain1372 and 13733′ region2262 and 2263Yala-Narathiwat3Binding domain779 and 780Binding domain1173 and 11743′ region2255 and 2256*After the coding sequence of the Salvador 1 strain Intragenic recombination in Pvdbp *After the coding sequence of the Salvador 1 strain Intron sequences Among the four introns of Pvdbp, three substitutions were identified in intron 1 (135 nucleotides): A113T and T117G (one isolate each from Tak and Chanthaburi) and C131A (one isolate from Chanthaburi). Introns 2, 3, and 4 (105, 218, and 91 nucleotides, respectively) were fully conserved among Thai isolates. Genetic differentiation Wright’s F-statistics revealed significant genetic differentiation among parasite populations (permutation test, p < 10⁻5). The highest FST values (0.149–0.229) were observed between Yala-Narathiwat and other provinces, while the lowest value (0.025) occurred between Tak and Ubon Ratchathani (Table 5). Table 5Interpopulation variance indices of P. vivax populations in ThailandProvinceTakUbon RatchathaniChanthaburiYala-NarathiwatTak (n = 71) < 10–5 < 10–5 < 10–5Ubon Ratchathani (n = 60)0.0252 < 10–5 < 10–5Chanthaburi (n = 68)0.05320.0490 < 10–5Yala-Narathiwat (n = 55)0.16420.14860.2289Pairwise FST values (lower diagonal) and their permuted p values (upper diagonal) Interpopulation variance indices of P. vivax populations in Thailand Pairwise FST values (lower diagonal) and their permuted p values (upper diagonal) Predicted B cell epitopes Linear B-cell epitope prediction (EPITOPE1D) identified 16 potential epitopes (12–16 amino acids) distributed across PvDBP. Epitope scores were notably higher in the N-terminal region and the Duffy binding ligand domain compared with the C-terminal and cytoplasmic domains (Fig. 3). Four previously reported inhibitory epitopes—H3 (DEKAQQRRKQWWNESK), M3 (PQIYRWIREWGRDYVS), L2 (AQIWTAMMYSVK), and L4 (DWDCNTKKDVCIPD)—overlapped with predicted epitopes in this region [38]. Predicted linear B-cell epitope scores across coding region of PvDBP using EPITOPE1D algorithm [33]. Scores for 12, 14 and 16 amino acids are shown in distinct lines. Broken line indicates cutoff value for epitope. Corresponding domains are shown above. Upward and downward arrows indicate known epitopes and predicted epitopes, respectively. SP, TM and CT represent signal peptide, transmembrane domain and cytoplasmic tail Predicted T helper epitopes Analysis using the IEDB resource predicted multiple HLA class II–binding peptides across PvDBP, based on common Thai HLA-DR, -DQA, and -DQB alleles (frequency > 5%) (Additional file 6)[35]. Several positively selected substitutions within predicted epitopes altered IC50 values, thereby modulating binding affinity or abolishing epitope recognition (Table 6). For example, mutation D925N in the C-terminal peptides YRDVKATREDIILMS and YRDVKATRENIILMS (HLA-DRB107:01) increased predicted binding affinity (weak → strong). Conversely, peptides LSNKFISVKNAEKVQ and LLNKFISVKNAEKVQ from the Duffy binding ligand domain showed similar intermediate binding affinity for HLA-DRB114:01 (IC50 283.67 and 250.03 nM). However, mutation I458K in variant peptide LSNKFKSVKNAEKVQ abolished binding (IC50 1594.57 nM). Table 6Changes in predicted HLA class II-binding peptides in PvDBP and positively selected codons Changes in predicted HLA class II-binding peptides in PvDBP and positively selected codons Duffy blood group phenotypes ASP-PCR analysis revealed that all samples carried the wild-type T allele at position –67 of the GATA motif. Exclusive amplification with FYAB2, but not GATAFY2, indicated that all patients were Duffy positive. Phenotypic frequencies varied by endemic area, but Fy(a + b–) was most common, followed by Fy(a + b +) and Fy(a–b +) (Table 7). Comparison with published data from healthy Thai donors (n = 3,358) [39–42] revealed no significant difference between malaria patients and donors (Yates’ χ2 = 3.95, df = 3, p = 0.27). Table 7Duffy phenotype frequency in vivax malaria patientsPopulationnPhenotype frequency, n (%)Fy(a + b-)Fy(a + b +)Fy(a-b +)Fy(a-b-)Malaria patients Tak6450 (78.12)12 (18.76)2 (3.12)0 (0) Ubon Ratchathani5246 (88.46)5 (9.62)1 (1.92)0 (0) Chanthaburi6347 (74.60)15 (23.81)1 (1.59)0 (0) Yala-Narathiwat5337 (69.81)12 (22.64)4 (7.55)0 (0) All232180 (77.59)44 (18.97)8 (3.44)0 (0)Healthy blood donors All regions [39]1000789 (78.90)197 (19.7)14 (1.40)0 (0) All regions [40]16581272 (76.72)324 (19.54)61 (3.68)1 (0.06) Central region [41]500464 (92.80)34 (6.80)2 (0.40)0 (0) Tak and Yala [42]200157 (78.50)39 (19.50)4 (2.00)0 (0) All33582682 (79.87)594 (17.69)81 (2.41)1 (0.03) Duffy phenotype frequency in vivax malaria patients Discussion The evolution of Pvdbp was contextualized by analyzing complete gene sequences from P. vivax populations across multiple malaria-endemic regions of Thailand. This analysis utilized Sanger sequencing, which provides high accuracy for validating single-gene variants. Among 232 isolates, most nucleotide substitutions were concentrated in the Duffy binding ligand domain, although sequence diversity extended across nearly all regions of the gene—except for the transmembrane domain and introns 2, 3, and 4, which remained perfectly conserved. Of the four parasite populations, isolates from Tak exhibited the greatest number of mutation sites, haplotypes, and highest nucleotide diversity across Pvdbp, consistent with our earlier findings for other malaria vaccine candidate genes, including Pvtrap, Pvama1, and Pvmsp family members [20, 32, 43–46]. Isolates from Ubon Ratchathani and Chanthaburi displayed intermediate levels of diversity, whereas parasites from Yala-Narathiwat consistently showed the lowest values. These patterns suggest that variation in genetic diversity of malaria vaccine candidate genes correlates with endemicity in different regions of Thailand [32]. In addition to point mutations, insertions/deletions (indels) contributed to Pvdbp variation. Notably, isolate UBN21 carried a 10-codon deletion (residues 119–130), likely caused by strand slippage within a 30-nucleotide repeat in the 5′ region (Fig. 1A). Alternatively, similarity to P. cynomolgi orthologues (PcyDBP1 and PcyDBP2) suggests duplication of this segment may have occurred after P. vivax–P. cynomolgi speciation [47]. Although the functional significance of the indels observed in PvDBP remains unknown, previous studies have reported that a specific 29-amino acid deletion spanning residues 672 – 700 (Salvador 1 numbering; corresponding to a 38-amino acid deletion relative to the P01 strain), and a > 40% deletion in the reticulocyte-binding protein 2 (RBP2) of Plasmodium simium may facilitate more efficient invasion of simian erythrocytes, enabling reverse zoonotic transmission from humans to platyrrhine monkeys in South America [48, 49]. Frameshift mutations were identified, notably in APH62, AP284, and APH43, where -1 frameshifts were rescued by compensatory insertions, producing alternative amino acids at codons 751–752. Frameshifts occurring near homopolymeric tracts (polyG) likely arose from strand mispairing [50]. Frameshifts accompanied by compensatory mutations produced p.D751G/V/M and p.G752E in Pvdbp among isolates/clones in this study, yet downstream reading frames were preserved across all variants. Thus, the observed frameshifts are unlikely to disrupt protein function. However, frameshifts in malarial erythrocyte-binding antigen (EBA) genes have played key roles in the adaptation and host switching of malaria parasites. In P. falciparum-related species infecting African apes (Plasmodium praefalciparum, Plasmodium reichenowi, Plasmodium lomamiensis, Plasmodium adleri, Plasmodium gaboni, Plasmodium blacklocki and Plasmodium billcollinsi), the EBA165 gene remains intact and functional. By contrast, a frameshift in the P. falciparum orthologue converts the gene into a pseudogene, preventing production of a functional EBS165 protein. Ape EBA165 binds specifically to ape erythrocytes but not to human red blood cells due to differences in surface sialic acid composition. Experimental correction of the frameshift in P. falciparum leads to downregulation of EBA165 and of an adjacent invasion ligand, implying that functional EBS165 is incompatible with efficient parasite growth in human erythrocytes [51]. Thus, inactivation of this gene likely represented a key step in the parasite’s successful adaptation to humans. Although the functional significance of frameshifts in Pvdbp remains unclear, positive selection at codon 751 suggests a potential adaptive role. Importantly, all indels and frameshifts detected in this study were located outside the Duffy-binding ligand domain, indicating substantial functional flexibility in non-critical regions. Antigenic polymorphism poses a major obstacle to malaria vaccine design, particularly if immune responses target allele-specific epitopes. Domain-level analysis revealed an excess of nonsynonymous over synonymous substitutions in the Duffy binding ligand domain, consistent with positive selection. Host immune pressure is the likely driver, as polymorphisms in linear B-cell epitopes of this region influence the invasion-inhibitory efficacy of monoclonal antibodies [52]. Similarly, positive selection was detected in the N-terminal region, which, although not directly involved in receptor binding, is extracellular and potentially accessible to immune recognition. Both B- and T-cell epitopes were predicted in this region (Fig. 3, Additional file 6). Codon-specific analysis using FUBAR identified 26 codons under positive selection, 22 of which overlapped with predicted B- or T-cell epitopes (Fig. 2, Fig. 3). Experimental studies have demonstrated that amino acid substitutions within epitopes—such as R263S, D339G, and K341N—alter antibody-mediated inhibition [38, 52, 53]. Likewise, substitutions such as R263S and L288F reduce T-cell responses, reflected in diminished IFN-γ and IL-10 production [12, 54]. These findings highlight variant-specific immune responses and support the hypothesis that polymorphism in PvDBP is largely immune-driven. Recombination also contributes to diversity in malarial antigen genes. Intragenic recombination signals were detected in Pvdbp across all provinces studied, with breakpoints primarily in the Duffy binding ligand domain and 3′ region. Although recombination can generate novel haplotypes, the limited number of breakpoints in Pvdbp suggests a relatively minor contribution to overall polymorphism, in agreement with earlier observations [10]. Although most P. vivax isolates carry a single copy of Pvdbp, duplications were detected in ~ 9.5% of Thai isolates, exclusively of the Cambodian type. PCR bias during amplification prior to recombinant plasmid construction is unlikely to explain the presence of sequence variants in duplicated gene, because (i) each copy would have an equal chance of serving as a template during PCR, (ii) the primers were designed against perfectly conserved regions, and (iii) Pvdbp lacks long repeats or pronounced heterogeneity that would distort DNA structure. Most duplicated copies (77.27%) were identical, providing increased gene dosage and likely conferring immune evasion by reducing antibody efficacy [55]. A minority (22.73%) displayed 1–3 nucleotide differences, with evidence of positive selection at codons corresponding to predicted B- and T-cell epitopes (Table 2, Fig. 3). These observations suggest that both gene dosage and subtle divergence between duplicated copies may contribute to immune escape. The contrast with greater divergence among Pvdbp orthologues in P. knowlesi and P. cynomolgi suggests that duplication in P. vivax is a relatively recent event. Although identical gene duplicates are possibly produced by recent gene duplication events and the ongoing process of gene conversion that acts to homogenize the sequences. A segment of one copy can “overwrite” the corresponding segment of the other copy, making them identical in that region. If this process occurs frequently enough, it can prevent the accumulation of mutations between copies, keeping them nearly identical over long evolutionary periods. These mechanisms can confound evolutionary analyses by keeping duplicate sequences highly similar (concerted evolution), thus masking the true evolutionary divergence time of the original duplication event [56]. Intron analysis revealed substitutions exclusively in intron 1, while the remaining introns were conserved. The localized accumulation of variation may reflect a higher mutation rate or region-specific inefficiencies in DNA repair. While intron 1 of Pvdbp maintains an open reading frame, it seems to complicate the numbering of amino acid positions in this protein [8, 57]. Population genetic analysis indicated significant differentiation among Thai P. vivax populations (Table 4). The greatest divergence was observed between Yala-Narathiwat and other provinces, while Tak and Ubon Ratchathani were most closely related. These patterns mirror earlier findings based on other antigen-encoding genes (Additional file 7)[20, 32, 43–46]. Geographical isolation and limited migration likely restrict gene flow in southern provinces, whereas frequent cross-border movement in Tak, Ubon Ratchathani, and Chanthaburi enhances genetic exchange. The Duffy blood group plays a critical role in P. vivax invasion, with PvDBP binding to the N-terminal extracellular domain of DARC. Although Duffy negativity provides strong protection, polymorphisms between Fya and Fyb alleles also influence susceptibility: erythrocytes with Fya exhibit reduced PvDBP binding [58], and individuals with Fy(a + b–) show reduced parasitaemia and clinical risk [59]. Consistent with this, vivax malaria has a low prevalence in sub-Saharan Africa, where Duffy negativity predominates [60]. In this study, Fy(a + b–) was the most common phenotype, followed by Fy(a + b +) and Fy(a–b +), with no Fy(a–b–). The similarity in phenotype distribution between patients and healthy Thai donors suggests incomplete protection, alternative invasion pathways (e.g., via PvEBP2), or insufficient sample size to detect subtle effects. Nonetheless, associations between Duffy phenotype, parasite density, and disease severity cannot be excluded. In this study, a large dataset of Duffy blood group genotypes from the general population in Thailand was used as controls. Although extensive, this dataset may not be ideal due to the risk of population stratification—differences in allele frequencies among subpopulations arising from distinct ancestral origins, environmental factors, historical malaria exposure, or unmeasured genetic factors. Such differences may produce spurious associations or obscure true ones. A more rigorous design would include locally matched controls from the same malaria-endemic provinces. It is also likely that some minor variants in mixed-clone infections escaped detection by the Pvmsp1 and Pvmsp3γ genotyping method, and the Pvdbp sequence analysis. Previous work demonstrated that targeted amplicon deep sequencing outperformed PCR for detecting malaria parasites in Thailand, with > 80% of samples yielding concordant results [61]. While next-generation sequencing provides a more robust, high-resolution view of parasite diversity—enabling detection of rarer variants and minority clones—the fundamental population-genetic and selective processes in Plasmodium, driven by high diversity and frequent recombination in endemic settings, are broadly reflected in multiclonal infection patterns identified using traditional PCR approaches. Thus, the core biological dynamics remain consistent across methods. Conclusions Genetic diversity within Pvdbp is shaped by point mutations, indels, frameshift mutations, and intragenic recombination. All Thai isolates with duplications carried the Cambodian type; most duplicates were identical, while a minority showed minor sequence divergence. Population structure of Thai P. vivax reflects epidemiological boundaries, with southern populations showing greatest differentiation. The widespread distribution of positively selected codons across Pvdbp, many corresponding to predicted B- and T-cell epitopes, underscores immune-driven diversification extending beyond PvDBPII. These findings highlight the importance of considering broader regions of PvDBP in vaccine development to achieve effective and durable protection against P. vivax. Supplementary Information Additional file 1 Additional file 2 Additional file 3 Additional file 4 Additional file 5 Additional file 6 Additional file 7 Acknowledgements We are grateful to all patients who participated in this study. We thank the staff of the Bureau of Vector Borne Disease, Department of Disease Control, Ministry of Public Health, Thailand, for assistance in field work. Abbreviations Analysis of molecular variance approach Copy number variation Duffy antigen receptor for chemokines Nonsynonymous substitutions per nonsynonymous site Synonymous substitutions per synonymous site Fast, Unconstrained Bayesian AppRoximation Genetic Algorithm for Recombination Detection Haplotype diversity Number of haplotype Human growth hormone Number of mutation Polymerase chain reaction Nucleotide diversity Parsimony informative site Duffy binding ligand domain of Plasmodium vivax Merozoite surface protein 1 of Plasmodium vivax Merozoite surface protein 3γ of Plasmodium vivax Quantitative real-time PCR Author contributions S.J. and C.P. designed the study. N.K., Y.T., T.T. and C.P. performed the experiments. C.P., N.K., Y.T. and S.J. performed data analysis. C.P. and S.J. drafted the manuscript. S.J. and C.P. finalized the manuscript. All authors approved the manuscript. Funding This study received financial supports from Ratchadapiseksompotch Fund from Faculty of Medicine, Chulalongkorn University (Grant No. RA68/030) to S.J. and C.P. Data availability The Pvdbp gene sequences of Plasmodium vivax have been deposited in NCBI GenBank under accession numbers PX236240—PX236341. Declarations Ethical approval and consent to participate This study was reviewed and approved by the Institutional Review Board in Human Research of the Faculty of Medicine, Chulalongkorn University, Thailand. Written informed consent was obtained from all participants or, in the case of minors, from their parents or legal guardians. All procedures were conducted in accordance with relevant ethical guidelines and regulations. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Additional file 1 Additional file 2 Additional file 3 Additional file 4 Additional file 5 Additional file 6 Additional file 7 Data Availability Statement The Pvdbp gene sequences of Plasmodium vivax have been deposited in NCBI GenBank under accession numbers PX236240—PX236341.