Background
Chloroquine (CQ) was first synthesized in 1934 [
1], and has been widely used in Yunnan Province since 1960 [
2]. In the late 1950s, CQ-resistant
Plasmodium falciparum was found in Colombia and along the Thai-Cambodian border [
3], and also in 1973 [
4,
5] and 1974 [
6], the emergence of CQ-resistant
P. falciparum was reported in the Yunnan and Hainan provinces of China. Increasing prevalence of CQ-resistant
P. falciparum has been continuously observed thereafter [
5], and negative effects of drug resistance has also been noted.
Initially, CQ-resistant
P. falciparum was tested in Yunnan Province using the in vivo test method recommended by the World Health Organization (WHO) [
7]. Between 1981 and 1983, the widespread distribution of CQ-resistant
P. falciparum in the major malaria-endemic areas of Yunnan Province was found by using the in vivo 4-week method [
7]. In view of this, artemisinin, pyronaridine, and the compound of piperaquine with sulfadoxine have been used to replace CQ in the treatment of falciparum malaria patients in Yunnan since 1983 [
8]. However, the "4-week method" with its low compliance and the test results easily affected by the patients immunity are encountering difficulties in practice when being applied to the large-scale and longitudinal monitoring of malaria CQ sensitivity [
8]. Therefore, Liu et al
. [
9] successfully introduced the in vitro microscopic method for testing CQ resistance in
P. falciparum in China [
10], and used WHO standardized CQ applicator plates [
10,
11], self-developed CQ applicator plates, and accompanying reagents [
8] to investigate the susceptibility of
P. falciparum to CQ, amodiaquine, piperaquine, and various other anti-malarial drugs in Yunnan and Hainan provinces of China from 1984 to 2002. The monitoring results showed that
P. falciparum distributed in Yunnan and Hainan provinces were highly resistant to CQ, amodiaquine, and piperaquine, but the susceptibility of
P. falciparum to CQ was restored after discontinuing or reducing the use of CQ [
12].
CQ has been the preferred treatment for clinical episodes of vivax malaria in Yunnan Province since 1958 [
2,
13‐
15], with over 300,000 patients (treated with a total dose of 1200 mg orally over 3 days) in the last four decades alone, according to incomplete statistics[
16‐
21]. However, while the challenge of CQ resistance in
P. falciparum has been a subject of great concern, evaluation of CQ's efficacy in treating vivax malaria patients has been rarely conducted. From April 2016, when the last indigenous vivax malaria case in China was reported in Yunnan Province [
22], to the end of 2022, a cumulative total of 1371 imported cases infected with
Plasmodium vivax parasites abroad were identified, including cases introduced abroad in Myanmar, Nigeria, the Democratic Republic of the Congo (DRC), Angola, and Cameroon, but predominantly in Southeast Asian countries, particularly Myanmar, which had the highest number of introduced cases [
16,
22,
23].
Data show that CQ resistance in
P. vivax was first identified in Papua New Guinea in 1989 [
24], followed by reports of cases infected with CQ resistance
P. vivax in Indonesia [
25], northern Myanmar [
26], India [
27], and Vietnam [
28]. Myanmar, which is located between South and Southeast Asia, is considered a high-risk transmission area for drug-resistant parasites [
29]. Zeng et al
. [
30], using an in vitro microscopic method, observed that in the border area between Myanmar and China 4.4% (2/46) of all clinical isolates of
P. vivax had CQ 50% inhibition concentration (IC50) values of above 220 nM, which exceed the susceptibility threshold by a factor of 1.5. A study conducted on amodiaquine (AQ) and CQ, each in combination with sulfadoxine-pyrimethamine (SP) in Papua New Guinea had a failure rate of more than 10% in the treatment of vivax malaria patients [
31]. Ratcliff et al
. [
32] used CQ alone to treat vivax malaria patients in Indonesia and had a failure rate of 15% in the early stages and up to 65% by 28 days, but the possibility that this data was confounded by relapse events has not been excluded, which causes difficulty in assessing the efficacy of anti-malarials for
P. vivax in vivo [
33,
34]. New rounds of infections and recurrent intraerythrocytic infections caused by the activation of
P. vivax hypnozoite parasites are confounding factors that must be guarded against in the in vivo assessment of anti-malarial drug efficacy in highly endemic areas [
35]. Secondly, the lack of in vitro culture methods for
P. vivax makes it difficult to directly transfer
P. vivax, and these in vitro testing methods, such as isotopic methods [
36] and microfluorimetric methods [
37], have always been used for the drug susceptibility assays of
P. falciparum. Animal model methods, which can compensate for the inability to obtain batches of
P. vivax for in vitro testing are also impractical due to the difficulty in establishing animal models and the unsustainability in supply of primates [
38]. Furthermore, with the continuous effectiveness of malaria control interventions, it has become increasingly challenging to find vivax malaria cases that meet the eligibility criteria for evaluating
P. vivax drug resistance [
39]. Therefore, the WHO proposed in 2018 that the lack of systematic evaluation of anti-malarial efficacy could be compensated by the optional use of molecular marker surveillance [
39].
Previous studies have demonstrated that mutations in dihydrofolate reductase gene (
pvdhfr) and dihydropteroate synthase gene (
pvdhps) are associated with the development of resistance to SP in
P. vivax [
40,
41], while mutations in multidrug resistance 1 protein gene (
pvmdr1) of
P. vivax is one of the markers indicative of those with resistance to CQ [
42]. A survey by Zeng et al
. [
30] found that the G698S substitution in the
pvmdr1 of
P. vivax population distributed in the border area between Myanmar and China was associated with reduced susceptibility to CQ, artesunate and dihydroartemisinin. In Indonesia, Suwanarusk et al
. [
43] found that the geometric mean CQ half-inhibition concentration of 283 nM in Y976F mutant isolates was significantly higher than that of 44.5 nM in wild-type strains. In Thailand, monitoring the distribution and extent of both Y976F and F1076L mutations in the
pvmdr1 has facilitated stage-specific evaluation of changes in the susceptibility of
P. vivax to anti-malarial drugs [
44]. Although there have been instances of CQ treatment failure in vivax malaria cases in Guyana, molecular marker monitoring has not further shown specific polymorphisms in the
pvmdr1. The country chose not to adjust its current anti-malarial drug policy, indicating that the susceptibility of
P. vivax to antimalarials remains stable [
45]. However, among numerous studies on the mutation polymorphism of
pvmdr1 gene, only a few ones are based on whole gene sequences [
46,
47]. The main reason for not always targeting full gene sequences is that the 3′ end of
pvmdr1 gene is not easily sequenced due to the presence of many special structures.
In order to help fully revealing of the
pvmdr1 gene polymorphism degree in
P. vivax strains infected in vivax malaria cases from Yunnan Province, China, and avoid forming a single-faceted experience of taking the research results around the China-Myanmar border as the currently actual situation in Yunnan Province, and begin carrying out the molecular surveillance plan on anti-malarial drug susceptibility of
P. vivax in Yunnan Province [
39], this study not only explored a realistic first-generation sequencing methods for
pvmdr1 whole gene, one of the molecular markers of CQ resistance, but also finished the longitudinal comparison on the
pvmdr1 characteristics in
P. vivax strains collected from vivax malaria patients diagnosed in Yunnan Province in 2014 and from 2020 to 2022 following the strategy of segmented sequencing of the
pvmdr1 full gene, and the findings are reported below.
Methods
Study subjects and blood samples
A cohort study was conducted using all vivax malaria cases diagnosed in Yunnan Province, that were available in the China Disease Surveillance Information Reporting System (CDSIRS). However, this study only included the cases in 2014, 2020, 2021, and 2022 based on the principle of cluster sampling due to the limited workload of whole gene sequencing for
pvmdr1. Each case was initially diagnosed by local county-level laboratories in Yunnan province, and then confirmed by the Yunnan Province Malaria Diagnosis Referent Laboratory (YPMDRL) to be mono-
P. vivax infection through both light microscopy and genetic testing. The genetic test for
P. vivax was conducted by the YPMDRL. The primer sequences, reaction conditions, and reaction system of the genetic test for
P. vivax are shown in (Additional file
1). The information on sex, age, initial diagnosis, place of detection, and place of introduction of the vivax malaria cases was obtained from the registration files in CDSIRS. Peripheral venous blood was collected from all vivax malaria cases during acute episodes and dried blood dots (DBD) on filter paper were gathered for in-depth analysis of
Plasmodium genetic sequencing.
Confirmation of the malaria infection source
The indigenous infection malaria cases included those who had no history of travel to epidemic areas outside Yunnan Province within 30 days before the onset of malaria; The introduced malaria cases included those who had a history of migration from malaria endemic areas, such as Myanmar and Africa, within 30 days of malaria onset.
Three 5 mm diameter DBDs were used to extract Plasmodium genomic DNA according to the instructions given in the QIAgen Mini Kit (DNA Mini Kit, QIAamp, Germany), which were stored at − 20 °C.
A strategy of segmented PCR amplification and sequencing was adopted in order to obtain the whole gene sequence of
pvmdr1. The chromosomal reference sequence (ID: NC_009915.1) of
P. vivax Sal I strain was used as a template to design the amplification of fragment 1 (F1, 365277-366271 nt), fragment 2 (F2, 364360-365400 nt), fragment 3 (F3, 363489-364447 nt), fragment 4 (F4, 362559-363620), and fragment 5 (F5, 361387-362957 nt) of the
pvmdr1 gene (Additional file
2). The primers names and thieves’ sequences, target fragment lengths, and reaction conditions for the nested PCR amplification of the five regions are detailed in Table
1. The amplification products of the second round of the five regions were expected to be 995 bp, 1041 bp, 959 bp, 1062 bp, and 1571 bp long, respectively.
Table 1
The primers names and thieves sequences, target fragment lengths for the nested PCR amplification the pvmdr1 gene in P. vivax strains
F1 | First round | DonMD-1-1F | TAACTCCTCACCGTTTGGGAAT | 1245 | 365,143–366,387 |
DonMD-1-1R | TCATTGTTTGGTTGCTGGTTGC |
Second round | DonMD-1F | GGTGTGTATATCTTGAGTTTGCAT | 995 | 365,277–366,271 |
DonMD-1R | CGTGTACTTACTGTACAGCTTT |
F2 | First round | DonMD-2-2F | TTTATTACCATATTTACGTACGCAAG | 1385 | 364,189–365,573 |
DonMD-2-2R | ATGATGATCGTAATTCTGTTTTCG |
Second round | DonMD-2F | TAACAACACCATGTCCATCATCG | 1041 | 364,360–365,400 |
DonMD-2R | TTAGATGCATTAGAACCCACCAG |
F3 | First round | DonMD-3-3F | CAACATCAAGTATAGTTTGTACAGC | 1368 | 363,295–364,662 |
DonMD-3-3R | TGAACATCTCTGTTAATATGTGCTG |
Second round | DonMD-3F | TTAGTGTTTCGAAGAAGGTGCA | 959 | 363,489–364,447 |
DonMD-3R | GTAGAGGGAGTACTTATTCGAGT |
F4 | First round | DonMD-4-4F | GCAGCATTTATAAGGACTCCG | 1387 | 362,467–363,853 |
DonMD-4-4R | CTCATCACGGTAGATTTGCC |
Second round | DonMD-4F | GCCATTATAGCCCTGAGCATTAT | 1062 | 362,559–363,620 |
DonMD-4R | GACGTTTGGTCTGGACAAGATATC |
F5 | First round/ Second round | DonMD-5-5F | GAGAAGGCTATTGATTATTCGAAT | 1571 | 361,387–362,957 |
DonMD-5-5R | TTAACTATGTTTACTACGGTTAAGGG |
The system used for all 10 PCR reactions was 1.5 μl of DNA template, 14.0 μl of PCR mix for 2 × Taq, and 0.5 μl each of upstream and downstream primers (10 μmol/l), and the volume was made to be 25.0 μl with the addition of ddH2O. The conditions for the 10 PCR reactions were the following: pre-denaturation at 95 °C for 4 min, followed by 32 cycles of amplification, with steps including denaturation at 95 °C for 45 s, annealing at 55–61 °C for 45 s, and extension at 72 °C for 90–150 s. The PCR amplification was terminated at 72 °C for 10 min. The end products of the second round of PCR amplification of the five regions were observed by 1.2% agarose gel electrophoresis, and the positive samples were sent to Guangzhou Tianyi Huiyuan Gene Technology Co for Sanger bidirectional sequencing.
Gene polymorphism analysis
The sequencing results were collated using DNAStar 11.0 and BioEdit 7.2.5 software. Sequences from primer pairs DonMD-1F/DonMD-1R, DonMD-2F/DonMD-2R, DonMD-3F/DonMD-3R, DonMD-4F/DonMD-4R, and DonMD-5F/DonMD-5R amplification products were retrieved in order to obtain the coding DNA sequency (CDS) of 1 aa-250 aa, 251 aa-567 aa, 565 aa-847 aa, 848 aa-1154 aa, and 1153 aa-1464 aa of the pvmdr1 gene. The overlap region of each CDS sequence was then removed and assembled along the 5′ → 3′ direction in order to obtain the CDS sequence of the pvmdr1 whole gene (4392 bp), which was compared with the mRNA reference sequence of the pvmdr1 gene (ID: XM_001613678.1), and when both Query Cover and Identifiers were > 98%, it indicated that the collated pvmdr1 whole gene CDS sequence was correct.
The CDS sequence alignment file and the deduced amino acid sequence were created using MEGA 5.04 software, and then DnaSP 6.11.01 software was used to identify the haplotypes and single nucleotide polymorphism (SNP) loci of the CDS strand of the pvmdr1 gene and their mutation types (synonymous/missense). Nucleotide diversity (π), expected heterozygosity (He) and Ka/Ks ratio were calculated, and sequence multiplicity mutations were identified. All base substitutions were confirmed by checking the sequence file ".seq" against the corresponding ".ab1" file; Ka/Ks ratios > 1, = 1, and < 1 indicated positive, neutral and negative selection in the population, respectively. Network 10.0 software was used to create an evolutionary mediated network map of the various haplotypes, where the same locus in the CDS strand of the pvmdr1 gene was repeated across haplotypes, showing a locus mutation that was defined as a 'reverse mutation'.
Statistical analysis
The database for polymorphic analysis of the pvmdr1 gene was created using Excel software, and the differences in SNP and haplotype detection rates between years were evaluated by Chi-square test (χ2) at the level of 0.05 in the "Data and descriptive statistics" module of IBM SPSS Statistics 21 software.
Discussion
The
pvmdr1 gene is located on chromosome 10, far from the telomere, in approximately 1/4 (366095/1419739) of
P. vivax, which starts at the base of the gene (5'-ATG), consists of a complete open reading frame that can encode the 1464 amino acids of
PvMDR1, which belong to the membrane structural protein with 12 transmembrane domains located at the
P. vivax digestive vesicles [
48,
49]. The sequences length of the
pvmdr1 gene obtained in this study all showed highly conservative with the fixed a 1464 aa amino acid chain, which may be related to the fact that the
pvmdr1 gene is located far from the telomeres of chromosome 10 and is less likely to undergo fragment deletion or insertion due to genetic recombination [
50,
51]. The success rate of 82.9% (624/753) obtained the
pvmdr1 full gene sequence also indicates that the experimental method for fragment PCR amplification and sequencing of
pvmdr1 gene has good stability in this study.
A total of 52 SNPs were detected in 624
pvmdr1 gene sequences in this study, except for the 31 SNPs reported before, including c.23C>T (P8L), c.132G>A (K44K), and c.516C>T (G172G), etc., [
43,
46,
47,
52], there were 21 other newly detected SNPs (Tables
3,
4), which may be due to the large sample size of this study and the analysis of the
pvmdr1 whole gene sequence. The concentration of newly detected SNPs and low-frequency SNPs in 2014 may be due to the fact that the study sample in that year included a richer population of
P. vivax than in subsequent years and even included all the indigenous Yunnan strains (Tables
3,
4). But, based on that fewer new SNPs and fewer low-frequency SNPs were detected in the indigenous Yunnan strains than in other populations, it may suggest that the
P. vivax population should be less intensely screened than the others populations.
Of the 18 SNPs that were detected in all years, eight loci mutations were frequently detected in Burmese strains, including c.1539T>A (S513R), c.2092G>A (G698S), c.2533C>T (L845F), c.2582C>A (A861E), c.2822A>C (M908L), c.2873C>T (T958M), c.3226T>C (F1076L), and c.4179G>C (K1393N) [
46]. Of which, c.2822A>C ( M908L) and c.2873C>T (T958M) reached 100% detection, which was consistent with previous results of nearly 100% of these isolates found in the Myanmar Laza city [
53], the China-Myanmar border [
46,
51,
54], the Thai-Myanmar border [
55,
56], and the Thai-Cambodia-Thailand-Lao border [
47], which is also consistent with previous studies that have shown that the
pvmdr1 gene sequences all came from
P. vivax populations in Southeast Asia, and
P. vivax strains of the present study were mainly composed of Burmese strains, accounting for 94.0% (708/753). Meanwhile, although the Ka/Ks ratios in this study were much greater than 1, in view of no positive selection pattern of low-frequency mutation surges in the mediator network map, it may still be attributed to the combination of neutral selection and drug pressure screening that the c.2822A>C (M908L) and c.2873C>T (T958M) mutations were detected in all samples strains [
46,
57]. Another mutation, c.3226T>C (F1076L), also produced by a combination of neutral selection and drug pressure [
58], showed an increase in detection from 2014 to 2022 (Table
3; Fig.
1), approaching the previous detection rates of 75.7% (143/189) to 85% (97/113) of the Myanmar population [
55,
59], but not reaching the level of the Ethiopian populations with 100% (55/55 and 28/28) detection [
42,
60], which suggests that the Myanmar population is stably screened from both the neutral selection and drug pressure, and that the screening for the c.3226T>C (F1076L) mutation is still incomplete.
In contrast to the gradual increase in the detection of the c.3226T>C (F1076L) mutation, the 29 SNPs represented by the c.2927A > T (Y976F) mutation were only detected in the early 2014 sequence (Tables
3,
4), and the c.2927A>T (Y976F) mutation was not detected again after 2020, which is consistent with the conclusions drawn by several authors on
P. vivax. The c.2927A>T (Y976F) mutation was detected at a rate of 30.8% (4/13) in 2008 [
45], decreasing to 3.3% [
45] to 7.1% [
46] in 2012, and was not detected again after 2015 [
45,
46]. So far, the factors that have caused the reduction and disappearance of the c.2927A>T (Y976F) mutation is still unclear. However, given that the c.2927A>T (Y976F) mutation is considered to be associated with the failure of CQ monotherapy for vivax malaria patients [
35], it may be prudent to consider that CQ is still a reasonable option for treatment of the
P. vivax population of Myanmar, as the c.2927A>T (Y976F) mutation is not detected in this specific population.
In this study, the He obtained from 624
pvmdr1 gene sequences was 0.9515, within the interval of (0.869–0.983 vs. 0.879–0.974) [
46] calculated by other scholars for
P. vivax strains in the China–Myanmar border region. However, a total of 105 haplotypes were more than the 75 [
46] species that had been previously identified in the China–Myanmar border region by other scholars in 2015, and the 10 [
61] to 27 [
33] species found in the South-North Amazon Basin, the North Coast of Peru, and India. In terms of haplotype species, except for the seven haplotypes including threefold mutations (G698S/M908L/T958M) [
51,
62], fourfold mutations (G698S/M908L/T958M/F1076L) [
30,
51], fivefold mutations (G698S/L845F/M908L/T958M/F1076L) [
30,
51] and (G698S/M908L/T958M/Y976F/F1076L) [
51], sixfold mutations (K44K/G698S/L845F/M908L/T958M/F1076L) and (T529T/G698S/L845F/M908L/T958M/F1076L) [
63], and eightfold mutation (L493L/T529T/G698S/L845F/M908L/T958M/F1076L/E1233E) [
63]. the remaining 98 haplotypes had been not previously reported; furthermore, unprecedented ninefold and tenfold mutant haplotypes were also detected. These differences may be due to the fact that more and longer
pvmdr1 sequences were analysed in this study.
Although 105 haplotypes were identified in this study, there was a trend of gradual decrease in the number of haplotypes detected annually, from 88 in 2014 to 15 in 2020, then 21 in 2021, and most recently only 13 in 2022, with low frequency haplotypes being more common only in the early years, particularly 2014 (Table
5; Fig.
3; Additional file
7). In contrast, the mutability of haplotypes shifted towards increasingly complex types, with threefold mutant haplotypes and fourfold mutant haplotypes mostly detected in 2014 and, in recent years, an increase in haplotypes with 10–13-fold loci mutations, such as Hap_40, Hap_91, Hap_2 (Fig.
3). This concentration of joint mutations in a small number of types has facilitated the generation of relatively uniform management strategies, but new adaptive strategies are necessary in order to deal with the problem of multiple mutations.
It is important to mention that in this study, an unexpected occurrence was observed in the description of the
pvmdr1 whole gene sequence, such that the fold of multiple loci mutations obtained from the sequence alignment was inconsistent with those identified by the network evolutionary analysis, where only 15 haplotypes (14.3%, 15/105) remained consistent with the multi-locus joint mutations identified by both methods, whereas the remaining haplotypes (85.7%, 90/105) always experienced more multiple mutations in the network evolutionary analysis (Additional file
7). The reason is that reverse mutations occurring 2–4 times at nine loci, including c.132, c.1477, c.1539, c.1559, c.1587, c.2092, c.3226, c.4074 and c.4179, could not be identified in sequence alignment, which makes it more difficult to accurately describe the sequence diversity of the
pvmdr1 gene. However, this study can provide readers with a better understanding of the polymorphism of the
pvmdr1 gene, and thus is valuable to the public.
In this study, 12 SNPs were found during the TMD of
PvMDR1, with only one deleterious mutation c.2533C > T (L845F), and 37 SNPs were found within the digestive vesicle (Additional file
10), with one deleterious mutation c.4179G>C (K1393N) identified in 2.7% (1/37) of samples. PROVEAN and SIFT analyses suggest that these deleterious mutations may lead to altered amino acid charge and hydrophobicity, resulting in a lack of protein structural integrity [
46]. Additionally, high-frequency mutations were observed, including c.1539T>A (S513R) (18.9%, 118/624) and c.4349C>T (S1450L) (15.4%, 96/624) within the digestive vesicle and c.3226T>C (F1076L) (75.6%, 472/624) at TMD domain (Additional file
10), which could also affect protein function based on previous research [
47,
64].
In this study, the prevalent trend of the molecular markers associated with drug resistance in P. vivax strains infected with vivax malaria cases in Yunnan Province are systematically revealed, and a set of pvmdr1 full gene sequencing of P. vivax strains from the Myanmar population were obtained in batch, which will provide valuable information and enrich the GenBank data. However, there are some limitations to this study. Firstly, the copy number of pvmdr1 genes was not assessed. Secondly, due to limitation on the length of the article, the others molecular markers for drug resistance monitoring other than pvmdr1 were not included in the analyses. Future research should conduct a study on the association between resistance molecular markers and anti-malarial drug susceptibility phenotypes, as well as improving the monitoring data on molecular markers for drug resistance of P. vivax.