Vector control has been a mainstay for preventing diseases caused by arboviruses transmitted by Aedes aegypti, and for mitigating the impact of these infections on human populations. During an outbreak, insecticide-based vector control interventions are usually intensified to reduce mosquito abundance and interrupt human-vector contact. However, increased use of insecticides may result in the selection of mosquitoes carrying genetic traits associated with insecticide resistance. The emergence of insecticide resistance in natural vector populations can ultimately affect the efficacy of insecticide-based vector control.

Target site insensitivity is a major mechanism of insecticide resistance that results from point mutations in genes encoding proteins at the specific site where an insecticide binds, typically in the nervous system [1]. These mutations cause structural modifications that reduce or even completely block the binding of the insecticide. The voltage-gated sodium channel (VGSC) is the target site of pyrethroid and organochlorine (notably DDT) insecticides. Mutations in the gene encoding the VGSC have been implicated in insecticide resistance in several insect species, a phenomenon referred to as knockdown resistance (kdr) [2]. In Ae. aegypti, a total of 10 kdr mutations have been reported [1]. Among these mutations, the V1016I and F1534C mutations have been extensively investigated in pyrethroid-resistant Ae. aegypti populations from Asia, South America and, to a lesser extent, Africa [2].

In 2017, a novel kdr mutation, V410L, located in domain I of segment 6 of the VGSC, was described for the first time in a pyrethroid-resistant Ae. aegypti laboratory strain originating from Rio de Janeiro, Brazil [3]. Functional analysis of mutant sodium channels expressed in Xenopus oocytes revealed that V410L significantly reduced sensitivity to both type I and type II pyrethroids. Genotyping of natural populations from Northeast Brazil did not reveal the presence of this mutation [3]. However, in a longitudinal study carried out in Mexico, the V410L mutation was detected in a single heterozygous mosquito in 2002 and by 2012 its frequency had increased to 90% in one of the localities surveyed [4]. The same mutation was described in natural populations of Ae. aegypti from Colombia, with frequencies ranging from 0.06 to 0.36 [5].

In recent years, Africa has been experiencing major outbreaks of Aedes-borne arboviral infections. Such was the case for a dengue outbreak that occurred in Cape Verde, in 2009, which affected eight of the 10 islands of this archipelago with more than 21,000 cases reported and four deaths [6]. In 2015–2016, Angola (South-Central Africa) experienced the worst epidemic of yellow fever in the last 60 years, with 4,306 notified cases and 376 deaths occurring in the province of Luanda [7]. The country had previously been afflicted by a Dengue outbreak in 2013 (https://www.ncbi.nlm.nih.gov/pubmed/23784016). In addition to the above mentioned outbreaks, Angola and Cape Verde were also affected by Zika outbreaks in 2015 and 2016, and cases of congenital Zika syndrome causing microcephaly were reported in both countries [8–10].

On Madeira Island, a Dengue outbreak involving 2,168 notified cases occurred in 2012–2013 [11]. Although this island is located ca. 680 km east off the coast of Morocco, it belongs administratively to Portugal and the European Union, raising concern about the risk of the introduction of both vectors and viraemic cases to mainland Europe.

As a response to these outbreaks, vector control programs tend to intensify insecticide-based interventions as a means of rapidly halting arbovirus transmission. This may lead to emergence of insecticide resistance in the local vector populations. Insecticide susceptibility tests carried out in 2009 on Ae. aegypti from the city of Praia, the capital of Cape Verde, revealed resistance to DDT but susceptibility to all other insecticides tested [12]. Subsequently, a survey carried out in 2012 and 2014 in the same urban area revealed resistance to deltamethrin and cypermethrin, but no kdr alleles were detected [13]. The local Ae. aegypti population of Madeira Island was found to be resistant to all major insecticide classes currently used in public health practices. This population displays multiple mechanisms of resistance, including the presence of V1016I and F1534C kdr mutations at a high frequency [14].

In Africa, information on mechanisms of insecticide resistance in Ae. aegypti populations remains scarce [2]. The presence of both F1534C and V1016I has been detected in Ghana [15] and in Burkina Faso [16], although it has also been investigated in Cameroon [17], Cape Verde and the Central African Republic [18]. None of the previous reports have analysed the new V410L kdr mutation. Considering that this mutation has shown potential to significantly reduce sodium channel sensitivity to both type I and II pyrethroid and to increase resistance in combination with F1534C [3], it is imperative to determine the current distribution of this mutation in natural Ae. aegypti populations, as its presence may greatly undermine the utility of a wide variety of pyrethroid insecticides, which are currently the major insecticide class used in vector control.

In this study, we genotyped the kdr loci of Ae. aegypti populations from Angola, Cape Verde and Madeira Island, collected either during or after the onset of arboviral outbreaks. The objectives were: i) to determine the presence and frequency of the V410L, V1016I and F1534C mutations and ii) to investigate the association of these mutations with the pyrethroid and DDT resistance phenotype.

In total, 314 Ae. aegypti were genotyped, 91 of which were from Madeira Island, 168 from Cape Verde (80 from Praia, 39 from São Lourenço dos Órgãos (SLO) and 49 from the island of Maio) and 95 from Angola (67 from Luanda and 28 from Huambo). Of the total sample, mutation V410L was screened by direct sequencing in 91 individuals from Madeira, 67 from Cape Verde and 95 from Angola. Sequences obtained from Ae. aegypti mosquitoes from Madeira are available in GenBank with accession numbers MK656133-MK656223. Apart from the V410L, no other genetic polymorphism was detected in the 104 bp fragment analysed for Madeira population. The genotype and allele frequencies of each kdr mutation are shown in Fig 1. The 410L allele was not observed in Cape Verde. The resistance allele 1534C was detected at a low frequency (2%) in Praia, the capital, always in heterozygosity with the wild type allele 1534F. Likewise, the 1016I allele was found in two heterozygous individuals from the SLO population. In Angola, allele 410L was observed at a high frequency in Luanda (71%) and at moderate frequency in Huambo (13%). Allele 1534C was observed at high frequency in both cities of Angola, while allele 1016I was detected at much higher frequency in Luanda (91%) than in Huambo (29%) (Fig 1). On Madeira Island, the 410L allele was observed at a moderate frequency (18%). As previously reported [14,28], the 1534C mutation was found to be fixed, and 1016I was present at a moderate frequency.

Pairwise linkage disequilibrium analysis revealed significant genotypic associations between positions 1016 and 410 in all populations tested (Table 1). For positions 1016 and 1534, only the Huambo population showed significant linkage disequilibrium. Finally, positions 1534 and 410 were in linkage equilibrium in all three populations tested. This analysis was not performed in Cape Verde due to the almost complete absence of mutant alleles.

Mortality rates for the insecticides tested varied between regions, being highest in the populations of Cape Verde islands (89.9%-100.0%) and lowest in Luanda-Angola (2.6%-7.4%). These results are detailed in S1 Table. Mortalities of 10.9%-52.8% obtained in Funchal, Madeira island, were intermediate to those of the other two regions, as described in Seixas et al. [14].

For the populations of Madeira and Luanda, it was possible to analyse the association between kdr mutations and the resistant phenotype. In Luanda, there were significant associations between resistance phenotypes and genotypic frequencies for mutations V1016I and V410L, but not for F1534C (Table 2). No significant association between phenotypes and individual kdr genotypes was found in the population of Madeira. This lack of association in Madeira was also evident for tri-locus genotypes, where the most frequent VV/CC/VV was common to both susceptible and resistant individuals (Fig 3).

Conversely, in Luanda, there was a trend for individuals with a resistant phenotype to carry tri-locus genotypes with resistance-associated alleles. Here, the triple mutant homozygote II/CC/LL was present in nearly 50% of resistant individuals.

This study describes for the first time the occurrence of the recently described V410L kdr mutation in old world Ae. aegypti populations, namely from Madeira island and Angola. In both cases, this mutation coexists with mutations F1534C and V1016I, which are widely distributed across the Globe [4]. The V410L mutation was originally described in Central and South American populations of Ae. aegypti, namely from Brazil, Colombia and Mexico [4,5]. This mutation appears to increase considerably resistance to pyrethroid insecticides when in combination with other kdr mutations [3,4]. It is likely that the occurrence of this mutation will have an impact on the efficacy of pyrethroid-based control measures against Ae. aegypti currently implemented in Africa.

The detection of mutations F1534C and V1016I in two sites from Angola extends our knowledge about kdr distribution to southern African populations of Ae. aegypti. Previous reports revealed the presence of the two mutations only in West African populations of Ghana and Burkina Faso [15,16]. However, the number of kdr surveys is still scarce for the African continent (reviewed in [2]) so that absence of records may result from under-sampling rather than true absence of this resistance mechanism. Therefore, it is likely that kdr mutations are more widespread in Ae. aegypti across the African continent than previously thought. Additional kdr surveys should thus be prioritised in order to clarify the current distribution of these mutations in African Ae. aegypti populations.

In addition to mainland Africa, kdr mutations were also found in Ae. aegypti from two outer West African archipelagos, Madeira and Cape Verde. In Cape Verde, mutation V410L was not detected whereas mutations V1016I and F1534C were found at low frequency (≤ 3.0%) in Santiago, the main island of the archipelago. Previous analysis of samples collected in the same island in 2007, 2010 and 2012 did not reveal the presence of any of these mutations [13,29]. This suggests a very recent origin of kdr mutations in this island. Whether these mutations represent introductions or independent mutation events is unknown. A recent phylogeographic study suggests that Ae. aegypti from Cape Verde originated from populations of Senegal [29]. One cannot rule out the possibility of a recent introduction of kdr mutations from neighbouring regions of mainland West Africa, specifically of mutation F1534C found in the city of Praia, where an international airport and port are located. On the other hand, the detection of the V1016I mutation in the remote inland locality of São Lourenço dos Órgãos argues in favour of an independent mutation event. The recent emergence of kdr mutations in Cape Verde may also explain the low frequency of resistant-associated alleles despite increased insecticide pressures imposed by vector control since the Dengue epidemic of 2009 [30].

In contrast to Cape Verde, the three kdr mutations analysed were present in Madeira Island. One of these mutations, F1534C, was fixed while the others displayed moderate frequencies. These kdr frequencies do not seem to be fully explained by insecticide pressures, given that vector control in Madeira has been predominately based on the elimination of mosquito breeding sites. A more plausible explanation is that kdr frequencies reflect those of the source populations that colonized Madeira Island [31]. Phylogenetic analysis suggests that Ae. aegypti was recently introduced into Madeira Island from South-American source populations, notably from Venezuela [31]. This region has well-documented insecticide resistant Ae. aegypti populations displaying high frequency of the F1534C and varying levels of the V1016I mutation [32].

In Luanda, Angola, significant associations were found between resistant-associated kdr alleles and the resistant phenotype to deltamethrin. This was evident for tri-locus genotypes and individually for mutations V410L and V1016I. The lack of association of mutation F1534C with resistance is consistent with previous observations suggesting that this mutation alone confers low levels of resistance but co-evolved with V1016I yielding higher levels of resistance [27,33]. Linkage disequilibrium coefficients were also highest between these V410L and V1016I mutations, which may reflect epistatic selection through insecticide pressure. During the 2016 yellow fever outbreak, vector control based on larvicides and indoor/outdoor pyrethroid spraying was intensified in Luanda, which substantially reduced the mosquito population [34]. It should be noted, however, that this is the first kdr survey in Ae. aegypti from Angola, so that no historical data are available for comparison. In Angola, insecticides have been widely used for vector control since the DDT-based malaria eradication campaigns implemented in the 1950s [35,36].

The results obtained in Luanda overall agree with the findings of longitudinal surveys carried out in Mexico, where pyrethroid have been routinely used by operational programs to control malaria and arbovirus vectors since 2000 [4,37]. Coincidently, the frequency of V410L has substantially increased alongside with V1016L and F1534C between 2000 and 2016 and significant associations were found between resistant-associated alleles and pyrethroid-resistant phenotypes [4]. However, the estimates of linkage disequilibrium coefficients (R_ij) found in Luanda (0.07–0.33) were generally much lower than those obtained in Mexico (0.31–0.99; [4]), which may reflect an earlier stage of selection of resistant-associated alleles.

In contrast with the observations in Luanda and Mexico, there were no associations between resistance associated kdr alleles and pyrethroid resistance in Madeira island. This result was unexpected but it probably reflects the complex pattern of multiple insecticide resistance mechanisms present in the local Ae. aegypti population [14]. In addition to kdr, a recessive trait, microarray-based gene expression analysis provided evidence for metabolic and cuticular resistance mechanisms, which may disrupt the statistical association between phenotypes and kdr genotypes. Also noteworthy is that the highest linkage coefficient between V410L and V1016I was found in Madeira island, a result that agrees with the hypothesis of these mutations being introduced in the island from a few colonizing individuals already carrying multiple resistance associated alleles [14].

In summary, we report the occurrence of the V410L kdr mutation in populations of Ae. aegypti from the old world for the first time. As also observed in South America, this mutation appears to coevolve with V1016I providing substantial higher levels of resistance to pyrethroid in the population of Angola. However, this was not the case of Madeira island, where association between kdr and pyrethroid resistance is probably disrupted by the coexistence of multiple resistance mechanisms. These findings suggest that populations carrying kdr mutations may respond differently to pyrethroid. Further studies will be required to assess the real impact of kdr mechanisms in the outcome of insecticide-based control of Ae. aegypti.