WHO adult bioassay
The diagnostic dosage for Aedes mosquito was used for permethrin and the tentative dosage for Anopheles was used for pirimiphos-methyl. WHO bioassays on the field strains revealed phenotypic resistance in populations from Selangor, Penang and Kelantan with percentage mortality less than 90% after exposure towards permethrin and pirimiphos-methyl (Fig. 1). Aedes aegypti from all six populations were highly resistant towards both insecticides with percentage mortality, after 24 hours exposure, varying from 0% to only 18% and 3% to 58% for 0.25% permethrin and 0.25% pirimiphos-methyl, respectively. Full susceptibility was observed in Ae. aegypti from the VCRU laboratory strain, after the exposure towards both insecticides (Fig. 1).
High resistance towards permethrin was observed in all six populations. The lowest percentage of mortality was observed in the populations from Pauh, Panji (PNJ) and Flat Buluh Kubu (FLT), Kelantan with 0% and 1% mortality, respectively, indicating that these two populations developed very strong resistance and the need of additional assessment regarding the genetic mechanisms and distribution of such resistance. Aedes aegypti from TUDM, Subang (TDM) shows the highest mortality compared to other populations with 18%, although such percentage of mortality is still quite low. Widespread resistant towards 0.25% pirimiphos-methyl was also observed in all populations with the highest resistance level recorded in Alam Budiman, Selangor (AB) with 3% mortality. Meanwhile, the population from Balik Pulau, Penang (BP) exhibited the highest mortality against pirimiphos-methyl with 58% mortality.
Synergism assay with piperonyl butoxide (PBO)
Percentage mortality of all populations except Pauh, Panji, (PNJ) after subjected to one-hour pre-exposure to 4% PBO before being exposed to permethrin, displays an increase in the mortality ranging from 2% to 69% (Fig. 1). Pre-exposure to synergist showed an increase in mortality for populations from TUDM, Subang (TDM) and Balik Pulau (BP) with percentage mortality of 29% and 69% respectively after being exposed to permethrinThese two populations showed a major increase of percentage mortality compared to the other populations, in addition, Ae. aegypti from Balik Pulau (BP) showed a significant increase in susceptibility towards permethrin (X2 = 7.244, df= 1, P = 0.007). No impact of pre-exposure to synergist PBO was observed in the population from Pauh, Panji (Kelantan) with 0% mortality before and after exposure to synergist.
Detection of kdr mutation in the vgsc gene of Malaysian population Aedes aegypti
167 samples including dead and alive mosquitoes from all populations were genotyped in order to determine the presence of S989P, V1016G and F1534C in DII and DIII and the allelic frequencies. Results of the genotyping are shown in Tables 1 (S989P), 2 (V1016G) and 3 (F1534C).
To detect other possible point mutations in the vgsc gene of the Malaysian population Ae. aegypti, the partial vgsc gene was sequenced. The length of the first fragment is 491 bp, spanning codons 989, 1011 and 1016; the same synonymous mutation was detected at the nucleotide positions 33 and 132: A to G. Sequence comparison with the pyrethroid susceptible Musca domestica (GenBank accession number U38813.1) and Ae. aegypti China strain (GenBank accession number MF794972.1) revealed six nucleotide substitutions, four of them non-synonymous resulting in four amino acid changes in the domains II and III of the vgsc gene in the permethrin resistant and susceptible samples from the six populations in Malaysia.
Several non-synonymous mutations in domain II, have been detected at codon 989 and 1016. Samples from Sungai Dua (SD), Balik Pulau (BP), Alam Budiman (AB) and TUDM, Subang (TDM) were heterozygous and homozygous for a double amino acid change. This amino acid change occurs at codon position 989, a change of wildtype amino acid serine (TCC) to proline (CCC), due to T/C substitution at nucleotide position 52. This mutation was specifically detected in populations from Penang, Selangor and in one sample from Flat Buluh Kubu, Kelantan (FLT) with the allelic frequency ranging from 0.1 to 1.0 (Table 1).
At codon 1016, a polymorphism was detected in populations from Penang and Selangor (but not in those of Kelantan). A mutation from valine to glycine at codon position 1016 was identified within these populations. A mutation of wildtype GTA to mutant allele GGA was detected from those populations with allelic frequency ranging from 5% to 55% (Table 2). Among 167 samples genotyped, a total of 46 samples were partially sequenced in domain II to valid the results obtained from AS-PCR for point mutation V1016G. Eighteen samples showed discrepancy where heterozygous allele (V/G) (from AS-PCR) turned out to be homozygous wild type allele (V/V) after Sanger sequencing. This might be due to the presence of two consecutive alternative mutations in domain II segment 6 leading to genotyping error, hence resulting in the false positive results [28]. In addition, the amplification of the non-specific band happens due to the mismatch of the single base in the gene, hence it is unable to prevent the non-specific amplification during the PCR extension [51].
A novel non-synonymous substitution has been discovered at codon position 1007, a mutation from alanine (GCC) to glycine (GGC), happens due to changes of nucleotide C to G at position 1007 in the sequence. Our results show that only permethrin resistant samples from Pauh, Panji (PNJ) and Flat Buluh Kubu (FLT) have this novel amino acid substitution, with the percentage of 1007G allele were 85% and 90%, respectively. Interestingly, the populations that possess this mutation will not co-occur with other point mutations either S989P or V1016G in domain II and could only be found in samples from Kelantan state. Our results indicate this point mutation alone in domain II might be responsible in conferring the high resistance in the phenotypes since there is no other point mutation in domain II that coexist after genotyping using direct sequencing (Fig. 2). Result from the direct sequencing reveals populations from Kelantan that possess this novel mutation will coexist with another point mutation in domain III which is the F1534C mutation (Fig. 2, Fig. 3). All samples from these localities were a mixture of heterozygous and homozygous for the double substitution mutations. These populations were either heterozygous or homozygous mutant to codon 1007 and shares another amino acid change at codon 1534 and are homozygous mutant at this position (Additional file: Table S3). This might explain why the population from Kelantan were highly resistant towards permethrin.
In domain III, a change from phenylalanine (TTC) to cysteine (TTG) at codon 1534 was detected in all six populations with allelic frequency ranging from 0.028 to 0.975 and populations from Kelantan showing the highest allele frequency for the permethrin-resistant samples; more than 90%. We also recorded that the mutant allele, 1534C are common in the susceptible samples from Penang and TUDM, Subang (TDM) (Table 3, Additional file: Fig. S1).
Association between kdr mutation at domain II and III with pyrethroid resistance
To assess the correlation with the resistance phenotype, a total of 167 resistant and susceptible mosquitoes from all populations were then genotyped at domains II and III. To ascertain the impact of kdr mutations at different codons; 989, 1007, 1016 and 1534 of the vgsc gene towards pyrethroid resistance, the S989P, A1007G, V1016G and F1534C mutations were analysed separately for their associations with the permethrin resistance.
The S989P mutation in domain II is not significantly associated with the pyrethroid resistance in all populations (Fisher’s exact test, P = 0.146) and we presume the populations were not at Hardy-Weinberg equilibrium due to low samples size tested for genotyping at codon 989 (Table 1). Fisher's exact test was conducted to compare the differences in 1016G of the allelic frequency between the resistant and susceptible phenotype from each locality. At codon 1016, population from TUDM, Subang (TDM) was significantly associated with the permethrin resistance [Odds Ratio (OR): 3.696, p<0.05] and population from Sungai Dua, Penang (SD) was slightly correlated with the permethrin resistance [OR: 0.29, p<0.05] (Table 2). In most of the localities, the differences in the allelic frequencies between alive and dead mosquitoes were not significantly correlated (P = 0.429). More heterozygote mosquitoes survived after the exposure towards permethrin (Additional file: Fig. S1).
We cannot determine the correlation between novel mutation, A1007G with the permethrin resistance in population from Kelantan due to the low number of susceptible samples obtained after WHO bioassay. Despite that, we found an extremely high 1007G allele frequency ranging from 85% to 90% (Table 4). This probably explained why the mortality of the population from Kelantan were exceptionally low (Table 4).
The frequency of the 1534C allele in domain III was significantly different between alive and dead mosquitoes from all six populations (X2=51.26, df= 1, p<0.001). The 1534C mutant allele was highly significant associated in the population from Balik Pulau (BP) and TUDM, Subang (TDM) [Odds Ratio (OR): 3.318, p<0.05 and OR: 16.852, p<0.05] (Table 3). There is no significant correlation observed between 1534C genotype and permethrin resistance in Sungai Dua (SD) [OR: 1.069, p>0.05]. From the Fisher’s exacts test, we found that Alam Budiman (AB), Pauh, Panji (PNJ) and Flat Bulu Kubu (FLT) were not in the Hardy-Weinberg equilibrium (p>0.05) and we speculate that this might be due to the deficit of the heterozygote allele.
Distribution of triple-loci and quadruple-loci of the genotypic combination in the domain II and III of the vgsc gene in Ae. aegypti
In a total of 46 samples genotyped in domain II and III, we found 13 different combination patterns of substitutions in Malaysian populations of Ae. aegypti with nine types of triple-locus and four type of quadruple-locus combination observed within those populations. The combination of the triple-locus genotype detected in each population (Selangor and Penang) ranging from one single type to four different types. Meanwhile, two to three types of combination for quadruple-locus genotyped was observed in the population from Kelantan (Additional file: Table S2). Triple-locus wild type homozygote, S989+V1016+F1534 (Type 1) and quadruple-locus wild type homozygote, S989+A1007+V1016+F1534 (Type 10) was found in four susceptible samples from Selangor and one susceptible sample from Kelantan respectively. Most of the locus genotyped were a combination of two to three amino acid substitution. We noticed triple-loci kdr genotypes (Type 3,4,5,6) resulted in the phenotypic resistant in the samples from Selangor and Penang. The presence of single mutation, S989+1016G+F1534 (Type 2) were seen in the population from Selangor only. Interestingly, the combination of the triple mutant, 989P+1016G+1534C happens to be present in one susceptible sample from Penang. This individual which possesses this genotype combination survived after 24 hours exposure period, suggesting the heterozygous allele at codon 1534 might confer lack resistance phenotypically in this sample. Four combination of quadruple-locus genotype was particularly present in population from Kelantan only with three combinations of locus; 989P+A1007+1016G+F1534; S989+1007G+V1016+1534C; S989+1007A/G+1534C (Type 11, 12 and 13) were detected in permethrin-resistant samples and one locus, S989+A1007+V1016+F1534 (Type 10) found only in permethrin-susceptible sample. Those loci consist of the new mutation with a high number resulting in 15 individuals (Kelantan population) surviving after the exposure of the permethrin.
Haplotype distribution and polymorphism analysis of the vgsc gene fragment in Malaysian population of Ae. aegypti
Seven haplotypes were identified with five haplotypes present in domain II and two haplotypes was observed in domain III. These haplotype variations produced four amino acid substitution. In the coding region of domain II, we found five polymorphic sites resulting in three non-synonymous changes (S989P, A1007G and V1016G mutations) and two synonymous changes. Meanwhile in the coding region of domain III, one polymorphic site at codon 1534 could be observed leading to the two haplotypes created in this vgsc fragment. In general, the vgsc gene exhibits low polymorphism level for all six populations in Malaysia with a low number of the mutational steps between the haplotypes in domain II and III as shown in the TCS network (Fig. 4). From the TCS network analysis in domain II, four resistant haplotypes were observed with one singleton haplotype 4 (H4) and haplotype 5 (H5) which is a combination of the new mutation and synonymous change were detected in Kelantan samples only.
There are five different haplotypes with two major haplotype that are established from all six Ae. aegypti populations in Malaysia (Fig. 4A). Interestingly, in haplotype 4 (H4) and haplotype 5 (H5), we found a novel regional mutation in domain II segment 6 which comprised of resistant samples from Kelantan state only and consisted of both homozygous and heterozygous A1007G mutation. These samples from Pauh, Panji and Flat Bulu Kubu were homozygous susceptible for S989P and V1016G mutations.
There are only two haplotypes that were found to be associated with the F1534C mutation in domain III of the vgsc gene. In domain III, haplotype 1 (H1) was associated with F1534C mutation revealing that all resistant samples from six populations were homozygous resistant for F1534C mutation (Fig. 4B). Haplotype 2 (H2) has homozygous susceptible allele, F1534.
There is no significant difference in the Tajima's D estimation in the vgsc fragment for both domain II and III demonstrating a low number of the polymorphisms in the vgsc within those populations (Table 5). The presence of the predominant haplotypes in both domain II and III gave us an idea that there is selection pressure in the vgsc gene fragment in both domain II and III which is in agreement with the existence of the kdr mutation in this Malaysian population of Ae. aegypti. Maximum parsimony phylogenetic tree analysis of the vgsc gene display an association between the pyrethroid resistance and the single nucleotide polymorphism of domain II and III in the vgsc gene respectively (Additional file: Fig. S2A, Fig. S2B). The existence of polymorphism in exon 15 to 16 domain II and exon 23 to 25 of domain III of vgsc gene potentially correlated with the permethrin resistance. Reconstruction of the maximum parsimony tree revealed the haplotype pattern within those domains clustered according to the phenotype of the mosquito samples.