Background

With a sharp decline in falciparum malaria, Plasmodium vivax has become the prominent Plasmodium species in Afghanistan causing more than 95% of all malaria cases [1]. The recommended first-line treatment for vivax malaria in Afghanistan is chloroquine combined with primaquine for radical cure (CQ-PQ), whereas for uncomplicated falciparum malaria the combination of artesunate and sulfadoxine-pyrimethamine (AS-SP) has been first-line treatment from 2003 to 2016, replaced by artemether-lumefantrine since 2016. Although SP has been the recommended treatment for falciparum and not vivax malaria, exposure of the P. vivax parasite population to SP might still have been quite extensive because of community based management of malaria particularly symptom-based clinical or probable diagnosed malaria in past two decades [2, 3] and SP treatment of vivax malaria in parts of the private sector or through self-medication [3]. This might have increased antifolate resistance in P. vivax in Afghanistan.

Dihydrofolate reductase (DHFR) and dihydropteroate synthase (DHPS) are 2 essential enzymes in de novo folate synthesis pathway in Plasmodium and the drug targets of pyrimethamine, and sulfadoxine, respectively [4]. SP is considered a sub-optimal treatment for P. vivax infections, because this parasite is intrinsically less sensitive to sulfonamides and resistance to SP is rapidly acquired with extensive drug exposure [5]. Through clinical, epidemiological, molecular and biochemical studies it has been identified that SP resistance in P. vivax is conferred by specific point mutations in the P. vivax dihydrofolate reductase (pvdhfr) and dihydropteroate synthase (pvdhps) genes. In areas with extensive SP use, treatment failure of vivax malaria with SP rapidly evolved, and was associated with mutations in codons 57, 58, 61, 117, 173 of pvdhfr and in codons 382, 383, and 553 of pvdhps [6,7,8,9,10].

The frequencies of SP resistance related pvdhfr and pvdhps mutations have been extensively reported from various malaria-endemic areas, including Thailand [8, 11,12,13,14], Cambodia [15], Myanmar [16], Vietnam [17], Indonesia [18], Papua New Guinea [19], Madagascar [20] and India [21,22,23,24]. However, reports from Afghanistan and Middle East countries are scarce. Previous studies from Afghanistan [25], Iran [26, 27], Pakistan [28,29,30,31] showed a majority of P. vivax strains carried wild-type pvdhfr and pvdhps, and a low frequency in pvdhfr of the double mutation in codon S58R/ S117N and single mutation in codon S117N.

In this study, the frequency of SP resistance related point mutations in the P. vivax pvdhfr and pvdhps genes in Afghanistan over time between 2007 and 2017 were reported. In addition, in silico three-dimensional modelling and molecular docking of 2 newly identified pvdhfr mutationswere performed to predict their impact on pyrimethamine sensitivity.

Methods

Study sites and sample collection

Dried blood spots (n = 185) from patients with light microscopy confirmed P. vivax infection were collected from five malaria endemic provinces in Afghanistan. In 2007 samples were collected from Takhar (bordering Tajikistan), Faryab (bordering Turkmenistan), and Nangarhar (bordering Pakistan). In 2010, samples were collected from Kunar and in 2017 from Nangarhar and Laghman, all bordering Pakistan. Blood samples of 20–30 µl were collected onto filter paper at the moment of patient presentation before antimalarial treatment. Genomic DNA was extracted by using QIAmp® DNA Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer’s instructions and purified DNA was stored at − 20 °C until further processing. Approval for this study was obtained from the Ethics Review Committee for Research in Human Subjects, Faculty of Tropical Medicine, Mahidol University, Thailand (EC approval number MUTM 2019-065-01).

In this study, the frequency of mixed infection of falciparum and vivax malaria was investigated by nested PCR detected 18S rRNA of Plasmodium spp. [32], only one sample was P. vivax and P. falciparum mixed infection (0.54%, 1/185). The low frequency of mixed infection of P. vivax and P. falciparum was explained by P. falciparum infected samples were excluded in the beginning.

Pvdhfr and pvdhps amplification and genotyping

Amplification of pvdhfr by semi-nested Polymerase Chain Reaction was performed following established and published methods protocols [8, 23]. For amplification, the primary and secondary reaction volumes were 25 and 100 µl, respectively, including 1 µl of template genomic DNA added in the primary amplified reaction, and 3 µl of primary amplified product added to the second round of amplification. The reaction mixture contained a final concentration of 125 nM primers forward-reverse mixture, 10 mM Tris–HCL (pH 8.3), 2 mM MgCl2, 125 µM dNTP mixture, and 0.4 U Taq polymerase (Invitrogen, Carlsbad, CA) (Additional file 1: Table S1–S7). The DNA fragments from PCR amplification or Restriction Fragment Length Polymorphism (New England BioLabs Inc., Ipswich, MA) were identified by 3% metaphor agarose gel electrophoresis (Radnor Corporate Center Radnor, PA). The amplified PCR product of pvdhfr was purified for DNA sequencing using PCR purification kit, FavorPrep™ (Favorgen, Taiwan). The purified PCR products of pvdhfr were sequenced by Macrogen, Korea. Nucleotide and amino acid sequences of this gene were aligned and compared with the P. vivax reference sequence from the original Sal1 strain (accession no. XM001615032), using BioEdit v7.2.5. For identification of established gene mutations associated with sulfonamide resistance, genotyping of pvdhps used PCR–RFLP methods, following established and published protocols [33]. The positive control of P. vivax is genomic DNA from SalI reference strain and characterized P. vivax isolates from the patient. To ensure the accuracy of the results, two positive controls and negative controls were added for quality control at every step of the procedure.

In silico modelling of mutant PvDHFR

A computer simulated three-dimensional model of the PvDHFR protein structure and its putative interactions with pyrimethamine was used to predict the impact of pvdhfr mutations on pyrimethamine binding. The wild-type PvDHFR structure (PDB ID: 2BL9) was used as a template for mutant PvDHFR modelling. The model was constructed using SWISS-MODEL (https://swissmodel.expasy.org). The constructed model was verified by PROCHECK [34]. Then, the derived structures of the active sites of mutant PvDHFR were complexed with pyrimethamine and evaluated by AutoDock Vina [35]. The structural models were visualized by Discovery Studio Visualizer–Accelrys.

Statistical analysis

The data of SNPs frequency related drug resistant genes were analysed by using MS Excel and SPSS v26.0. IBM Corp., Armonk, NY, USA. Pearson’s Chi-square test was used to compare proportions of haplotypes. A P value < 0.05 was considered statistically significant.

Results

Pvdhfr and pvdhps haplotypes

A total of 185 dry blood spots from patients presenting with vivax malaria were collected from the study sites. In the 185 genotyped samples, a total of 11 distinct pvdhfr and pvdhps haplotypes could be distinguished. Overall, the wild-type haplotype was the most frequent (64%, 119/185) (Table 1). Two other common haplotypes were a single mutant haplotype S117N (10%, 19/185) and a double mutant haplotype S58RS117N (9%, 16/185). Rare haplotypes included a mutation at codon position 383 of pvdhps in combination with pvdhfr single mutation at position 117; S117N/A383G (1%, 2/185) and a pvdhfr double mutations at 58 and 117; S58RS117N/A383G (1%, 1/185). The mutation A383G of pvdhps was detected in 8%, (3/37) of samples from Nangarhar in 2017.

Table 1 The Combination of pvdhfr and pvdhps mutations of P. vivax comparing by time and provinces
Full size table

A novel mutation at codon S93H (AGC to CAC) was identified comprising a single mutant haplotype S93H (4%, 8/185) and a double mutant haplotype S93HS117N (2%, 3/185) (Table 1). The N50I mutation (AAC to ATC) observed in this study had not been described from Afghanistan before. The mutation was part of the single N50I mutation haplotype present in 2% (3/185) of samples. While double mutant haplotype N50IS117N was found in 5% (10/185) of samples. Haplotypes with triple or quadruple mutations, which are associated with high grade antifolate resistance, were not observed in this study.

Temporal changes in pvdhfr and pvdhps haplotypes between 2007 and 2017

In 2007, the wild type pvdhfr and pvdhps haplotype was most frequent (81%, 80/99). Other haplotypes were rare, including the single mutant haplotype S117N in 6%, (6/99), and the double mutant haplotype N50IS117N in 3% (3/99) of samples. Also in the year 2010 the wild type pvdhfr and pvdhps haplotype was most frequent (64%, 18/28). The single mutant haplotype S117N (18%, 5/28) and double mutant haplotype S58RS117N (11%, 3/28) were also observed. In contrast in 2017, the frequency of wild-type pvdhfr and pvdhps haplotype was compared to 2007 significantly reduced to 36% (21/58) of samples (p value ≤ 0.001). In 2017, the frequencies of the single mutant haplotype S117N (14%, 8/58, p value ≤ 0.001) and double mutant haplotype S58RS117N had increased significantly (21%, 12/58, p value ≤ 0.001) (Table 1) (Figs. 1, 2).

Fig. 1
figure 1

Frequencies of mutations (%) in pvdhfr and pvdhps in five border provinces of Afghanistan. Nine non-synonymous mutant codons of pvdhfr; I13L, P33L, N50I, F57L, S58R, T61I, S93H, S117N, I173L and 2 codon non-synonymous mutant codons of pvdhps; A383G, A553G were observed

Full size image
Fig. 2
figure 2

Comparison of pvdhfr/pvdhps haplotype frequncies (%) between year 2007 to 2017

Full size image

In silico modelling of novel pvdhfr mutations

Since the two nonsynonymous pvdhfr mutation at position N50I, AAC (Asn) to ATC (Ile), and S93H, AGC (Ser) to CAC (His) had not been observed earlier in Afghanistan, the corresponding three-dimensional structures of the mutated proteins and the impact on pyrimethamine binding were modelled and predicted. The mutation N50I is located within the binding pocket, approximately 5 Å from the pyrimethamine molecule but does not directly interact with the drug. The mutation S93H is placed on the other side of the protein molecule, far away from the binding pocket (Figs. 3, 4, 5). Replacement of Asn with Ile at position 50 disrupts the hydrogen bonds between Asn and water molecules. The N50I mutation also causes a conformation rearrangement of the helix (residues 50–63), disturbing the favorable interactions between the enzyme and inhibitor. The model predicted that the pvdhfr N50I mutation would affect only moderately pyrimethamine binding, whereas the S93H mutation would not affect binding.

Fig. 3
figure 3

Structural model of PvDHFR N50I mutation complexed with pyrimethamine. A mutation at position 50 (in yellow circle) is located within the binding pocket of pyrimrthamine (in purple circle)

Full size image
Fig. 4
figure 4

Structural model of PvDHFR S93H mutation complexed with pyrimethamine. A mutation at position 93 (in yellow circle) is located far away from the binding pocket of pyrimrthamine (in purple circle). This mutation does not have any effect on pyrimethamine binding

Full size image
Fig. 5
figure 5

Molecular interactions of pyrimethamine and PvDHFR. The binding analysis showed that mutation at position 50 (in red circle) is located approximately 5°A from pyrimethamine (purple) but not directly involved in the binding

Full size image

Discussion

The combination of AS-SP has been used as first-line treatment for uncomplicated.

P. falciparum in Afghanistan since 2003 to 2016, and SP has been widely available in private sectors for a long time. Since an episode of falciparum malaria triggers relapse infection with P. vivax a few weeks later, this provides a window of selection for resistant parasites to outgrow sensitive parasite still killed by the remaining SP drug levels. Drug pressure from SP on the P. vivax parasite population over the past decade is thus likely, which might cause gene mutations in pvdhfr and pvdhps causing reduced SP sensitivity. Indeed, in this study, the frequency of the double mutant pvdhfr haplotype at S58RS117N and single mutant haplotype at S117N had increased significantly over the course of ten years since 2007 were observed. However, the triple and quadruple haplotype mutations associated with higher level of antifolate resistance were not found, suggesting only low or moderate drug pressure from SP on the P. vivax parasite populations in the study areas.

These findings agree with previously published reports showing low prevalence (less than 5%) of triple and quadruple mutant pvdhfr and pvdhps haplotypes in P. vivax parasite populations from Iran, Pakistan and India (Table 2) [21, 22, 26,27,28,29,30,31, 36,37,38,39]. In the wider study region, it was observed that the pyrimethamine resistance marker at position 117 in pvdhfr arose first and was followed by the subsequent mutation in positions 117 and 58, conferring increasing levels of resistance as report earlier [23, 26,27,28,29,30,31].

Table 2 Published molecular surveys of pyrimethamine and sulphadoxine markers in south-western Asia
Full size table

Antifolate drugs inhibit two essential enzymes in the folate biosynthesis pathway, necessary for DNA precursor synthesis. Sulfadoxine inhibits DHPS while pyrimethamine inhibits DHFR, which results in parasite death [4]. In P. falciparum, the intensive use of SP as first line treatment in south-east Asia since 1970s resulted in rapid acquisition of mutations in pfdhfr and pfdhps, causing amino acid changes in the enzymes at positions in the binding pockets for pyrimethamine and sulfadoxine, respectively, causing SP resistance [7, 8]. The antifolate resistant mechanism in P. vivax, with incremental resistance acquired by the accumulation of single point mutations in pvdhfr and pvdhps, is thought to be similar to the antifolate resistance mechanism in P. falciparum. In P. vivax, the sequence of non-synonymous mutations in pvdhfr is in codons 13, 58, 117, 173, equivalent to the orthologous pfdhfr positions 16, 59, 108 and 164 in P. falciparum. In P. vivax, biochemical and protein functional studies showed that pyrimethamine effectively inhibits wild-type PvDHFR, but were approximately 60 to  > 4000 times less active against mutant enzymes. Double mutant S58R and S117N PvDHFR was 10–25 fold less inhibited by pyrimethamine than the S117N mutant [10]. This relates to the steric hindering of the pyrimethamine binding pocket caused by the S117N mutation in the PvDHFR enzyme [9, 10].The studies also showed that a change in the effect of pyrimethamine on the enzyme kinetic properties of mutant PvDHFR is closely related with a change in sensitivity to pyrimethamine [9, 10].

The current study identified a novel nonsynonymous mutation S93H in pvdhfr and for Afghanistan newly observed N50I mutation in pvdhfr. The N50I mutation corresponds to the orthologous mutation N51I in P. falciparum. The pvdhfr N50I mutation was previously observed in low frequency in Pakistan [29]. Previous study shown that mutant N50I PvDHFR expressed in yeast confers an increase in pyrimethamine IC50 compared to wild type PvDHFR, whereas, the combination of mutation N50I and S117N conferred a-57 fold increase in IC50 compared to wild type [40]. To assess the interaction of these mutations on pyrimethamine binding, homology modelling and molecular docking in-silico were performed. This showed that the pvdhfr N50I mutation interrupted moderately the binding of pyrimethamine. In contrast, the pvdhfr S93H mutation, changing amino acids from AGC (Ser) to CAC (His) located away from the binding pocket of pyrimethamine, was predicted not to interrupt pyrimethamine binding. However, these in silico results will need confirmation in further in vitro or in vivo drug sensitivity studies.

An additional marker for SP resistance in P. falciparum is increased copies number of GTP cyclohydrolase I gene (pfgch1). GTP cyclohydrolase is another important enzyme in folate biosynthesis, and an increased copy number might compensate for the fitness loss associated with triple and quadruple mutations in pfdhfr and pfdhps [41, 42]. So far, there is limited number of reports on gene amplification of pvgch1, therefore still no finding of the association of mutations in pvdhfr/pvdhps and pvgch1 copy number variation. Whether copy number variation (CNV) in the P. vivax orthologous gene pvgch1 is important in antifolate resistance needs further study. Pvgch1 CNV was not assessed in the current study, but less likely to be present because of the absence of triple- or quadruple mutations in pvdhfr in our samples.

Conclusions

This study shows that although wild-type pvdhfr and pvdhps haplotypes have become less frequent in the Afghanistan P. vivax parasite population between 2007 and 2017, but only double and single nonsynonymous mutations in pvdhfr were observed, which would confer only moderate SP resistance. This suggests relatively low drug pressure from SP on the P. vivax parasite population in the study areas. Of the two novel pvdhfr mutations identified in Afghanistan, only the N50I was predicted to moderately affect pyrimethamine binding.