Association Between Pyrethrum Knockdown Time and Sodium Channel Genotypes in California Aedes Aegypti

Background: Since their detection in 2013, Aedes aegypti has become a widespread urban pest in California. The availability of cryptic larval breeding sites in residential areas and resistance to insecticides pose signicant challenges to control efforts. Resistance to pyrethroids is largely attributed to mutations in the voltage gated sodium channels (VGSC), the pyrethroid site of action. However, past studies have indicated that VGSC mutations may not be entirely predictive of the observed resistance phenotype. Methods: To investigate the frequencies of VGSC mutations and the relationship with pyrethroid insecticide resistance in California, we sampled Ae. aegypti from four locations in the Central Valley, and the Greater Los Angeles area. Mosquitoes from each location were subjected to an individual pyrethrum bottle bioassay to determine knockdown times. A subset of assayed mosquitoes from each location was then analyzed to determine the composition of 8 single nucleotide polymorphism (SNP) loci within the VGSC gene. Results: The distribution of knockdown times for each of the ve Californian populations sampled was non-parametric with bimodal distributions. One group succumbs to insecticidal effects around 35-45 minutes and the second group lasts up to and beyond the termination of the assay (120+ minutes). We detected 5 SNPs polymorphic within California populations. One is potentially new and alternatively spliced (I915K), and four are known and associated with resistance: F1534C, V1016I, V410L and S723T. The Central Valley populations (Clovis, Dinuba, Sanger and Kingsburg) are fairly homogenous with only 5% of the mosquitoes showing heterozygosity at any given position. In the Greater LA mosquitoes 55% had at least one susceptible allele at any of the ve SNPs. The known resistance allele F1534C was detected in almost all sampled mosquitoes (99.4%). We observe signicant heterogeneity in individuals with the same genotypes, conrming that the resistance SNPs alone cannot. Conclusions: Resistance associated VGSC SNPs are prevalent, particularly in the Central Valley. Interestingly, among mosquitoes with all 4 that were then genotyped. The Ae. aegypti samples were derived from multiple locations in California to explore population-level differences in response to insecticide exposure. We hypothesize that the resistance phenotype observed in Ae. aegypti in California is due to mutations in the VGSC, and individuals with more alternate alleles in the VGSC will have longer knockdown times. This observation contrasts with median knockdown times observed in mosquitoes of the identical VGSC genotype from the Clovis and Greater LA populations. Phenotypic heterogeneity existing within and between groups sharing a common genotype were observed because of conducting bottle bioassays on individual mosquitoes. All bottle bioassays were performed coating each bottle with the same amount of chemical from the same pyrethrum batch and stock solution to ensure standardized and comparable assay conditions.


Background
The yellow fever mosquito, Aedes aegypti (Linnaeus 1762), is a major vector of arboviruses such as dengue, Zika, chikungunya, and yellow fever viruses. This medically important vector was detected in California in 2013 in response to a residential service request (1). It has now been detected in 17 counties throughout the state despite aggressive surveillance and treatment efforts, presenting a signi cant challenge to local control agencies (1)(2)(3)(4). The peridomestic habits of this mosquito make reduction and eradication di cult, but effective adulticides remain essential for control in the face of threats to public health.
Pyrethroid based compounds are used as adulticides, favored for their e cacy and low mammalian toxicity (5). However, the invasive populations of Ae. aegypti in California possess genetic mutations conferring resistance to the pyrethroid class of insecticides typically applied for vector control (6,7). Pyrethroids act on insect voltage gated sodium channels (VGSC) by binding to open channels and blocking the channel in the open conformation. This results in prolonged depolarization of the membrane and failure of neuronal function (8). This class of insecticides is partitioned into two types (I and II) based on the presence or absence of a cyano moiety at the alpha carbon (9). Point mutations within the channel domain of the protein can confer type-speci c resistance by causing structural changes that reduce or eliminate the insecticides ability to bind the channel (10). These mutations change the amino acid composition of the protein at speci c locations that result in changes in charge and steric hindrance of the ion channel. These changes allow the VGSC to maintain normal function in the presence of pyrethroids.
To test for genetic markers of resistance in California, public health agencies and mosquito control districts screen for known pyrethroid resistance mutations in the VGSC gene: F1534C, V1016I, and V410L (7). These mutations are annotated based on their orthologous position in the Musca domestica VGSC protein (Genbank accession number: ANW06229) (11,12). V410L is located in the sixth transmembrane region of the rst domain, V1016I is located in the sixth transmembrane region of the second domain, and F1534C is located in the intracellular space in the rst loop of the sixth domain ( Fig. 1) (10,13). Recently, another mutation (S723T) was linked to Deltamethrin resistance, though its impact on the resistance phenotype remains unknown (14). This SNP is localized in the intracellular region of the second transmembrane repeat domain. The F1534C mutation confers a low level of resistance on its own to type I pyrethroids.
The V410L SNP, rst described in 2017 (13), confers resistance to both type I and type II pyrethroids. The V1016I mutation on its own does not confer resistance, however in conjunction with F1534C it provides elevated insensitivity to type I and type II pyrethroids (8,15,16). Mosquitoes homozygous for V1016I, F1534C, and V410L mutations exhibit a high level of resistance to both type I and type II pyrethroids (17). As each SNP provides differing levels of protection against different classes of pyrethroids, testing for multiple SNPs in the eld is relevant to screening for pyrethroid resistance (13).
Detecting and quantifying insecticide resistance in mosquitoes provides mosquito abatement groups with the tools to tailor their control strategies. The CDC Bottle Bioassay provides a convenient way to detect phenotypic resistance in adults to various adulticides (18). The CDC Bottle bioassay involves placing 10-25 mosquitoes in an insecticide coated Wheaton bottle followed by observation of the time and proportion of knockdown for two hours after exposure. This procedure allows districts to investigate population level resistance. However, a direct analysis of the relationship between individual VGSC genotype and bottle bioassay phenotype (knockdown time) has not been completed with California Ae. aegypti populations. To study this relationship, we conducted phenotype assays on individual mosquitoes that were then genotyped. The Ae. aegypti samples were derived from multiple locations in California to explore population-level differences in response to insecticide exposure. We hypothesize that the resistance phenotype observed in Ae. aegypti in California is due to mutations in the VGSC, and individuals with more alternate alleles in the VGSC will have longer knockdown times.

Materials And Methods
Adult mosquitoes were collected from four towns in the Central Valley of California: Dinuba, Clovis, Sanger and Kingsburg (Fig. 2) as well as from the Greater LA area. Eggs from these adults were collected, allowed to develop for at least 5 days, then ooded in trays of 1 L of water and reared according to standard protocols (19). Mosquitoes were reared on a diet of ground rodent chow at 27 °C under 14:10 hour (light:dark) photoperiod and adults were held at 70% relative humidity. Adults 1-3 days post-eclosion were then collected and individually exposed to a bottle bioassay to record individual knockdown time. Adult mosquitoes were fed on 10% sucrose solution ad libitum and did not receive a blood meal at all prior to insecticide exposure.

Individual Adult Bottle Bioassay
To determine time to knockdown for individual mosquitoes, a modi ed bottle bioassay was developed based on the CDC protocol (20). 250 ml Wheaton bottles (Fisher #06-404B) were coated with technical grade pyrethrum purchased from Chem Service (West Chester, PA). The pyrethrum was diluted in acetone to a concentration of 15.6 µg/ml (Lot #7581300) and bottles were coated with the insecticide following the procedure described in (21). Individual female mosquitoes were then aspirated into each bottle and observed for knockdown for up to 2 hours. Individuals were determined as knocked down when the bottle was rotated, and the mosquito could not reorient itself upright. Bottles were monitored continuously and exact time to knockdown was recorded. A susceptible lab colony, Rockefeller, was assayed as a reference for knockdown behavior in the bottle assay. For each population 80-95 adult female mosquitoes were assayed. Following knockdown, individuals were placed in the Lysis Buffer provided with the Zymo Quick DNA/RNA Miniprep kit (Cat #: D7001) and homogenized using Axygen mortar and pestles (Product PES-15-B-S1). Samples were stored at 4℃.

Data Analysis
Statistical analysis was performed in R Version 3.5.2 (25,26). Survival analysis was performed using the survival package (27). Pairwise comparisons of survival curves were performed through a log-rank analysis. Median knockdown times were compared using Mood's median test from RVAideMemoir (28). Knockdown distribution normality was tested using the Shapiro-Wilk test in Base R. The box plot and histograms were created with ggplot2 (29). P-value thresholds were adjusted for multiple comparison correction using the Benjamini and Hotchberg method (30).

Individual Adult Bottle Bioassay
The knockdown times of all assayed mosquitoes are displayed in Fig. 3. Mosquitoes that did not knockdown during the assay period were coded as such and considered using the Kaplan-Meier analysis which accounts for right-censored data. Aside from the susceptible laboratory strain, Rockefeller, the Sanger and Kingsburg populations had the lowest median knockdown times and are signi cantly lower than that of Clovis or Greater LA (Fig. 3, Mood's Median Test, α < 0.05). The distribution of knockdown times for each of the ve Californian populations sampled was non-parametric (Fig. 3, Fig S1) and appear to have bimodal distributions. The distribution reveals the presence of two subsets of mosquitoes in each population with one group succumbing to insecticidal effects around 35-45 minutes and the second group lasting up to and beyond the termination of the assay. However, the relative proportions of these distributions differ between populations.

SNP Genotype by Population
Samples for genetic analysis were selected primarily from the upper and lower tertiles of knockdown times for each population. We detected 5 SNPs polymorphic within California populations (Table 1). All, with the exception of I915K, have been reported previously (13,14,31). The I915K SNP is located in a region which is alternatively spliced and its presence in the resulting transcripts has not been determined conclusively ( Table 1). The V410L mutation is associated with resistance to type I and type II pyrethroids (13,14). The Central Valley populations (Clovis, Dinuba, Sanger and Kingsburg) are fairly homogenous with only 5% (12/242) of the mosquitoes tested showing heterozygosity at any given position. Alternatively, analysis of the Greater LA mosquitoes revealed that 55% had at least one susceptible allele at any of the ve SNPs (Table 1, 37/67). The F1534C mutation is nearly xed across these populations with two exceptions. One mosquito from Clovis was a heterozygote at the amino acid positions F1534C, V1016I, V410L and S723T. One mosquito from LA was heterozygous at F1534C, and homozygous susceptible at V1016, V410 and S723 (Fig. 4, Table 1). The I915K mutation was present only in the Dinuba population in heterozygotes (3/60) and these mosquitoes were also heterozygous at the V1016I, V410L and S723T positions. In the Greater LA population 15/67 were heterozygous at I915K. Of these 15, 7 were homozygous susceptible at V1016, V410 and S723. The other 8 were heterozygous at these positions (Fig. 4). The sites V1016I, V410L and S723T, appear to be inherited together frequently; an individual heterozygous or homozygous at one site was correspondingly heterozygous or homozygous across all three with exceptions in the Clovis population. In the Clovis population 8 mosquitoes were heterozygous only at sites S723T and V410L (Fig. 4).

SNP Genotype and Knockdown Time
Most of the assayed individuals were homozygous for the alternate "resistance" alleles. Still, some individuals with this genotype were knocked down in as little as 6 minutes after exposure to pyrethrum, overlapping with the susceptible Rock strain. More diversity of VGSC genotypes (6) was observed in the LA mosquitoes than Central Valley populations (4) (Fig. 4A). This is consistent with previous reports showing little to no polymorphism in VGSC fragments in central California populations (7,32). The quantitatively different genetic variations of the Greater LA population from the Central Valley population is also consistent with a previous study monitoring both Central Valley and southern California locations (7). It is noteworthy that independent studies utilizing whole genome sequence (33), SNP chip (34), and microsatellite (34) indicated different genetic makeup separating southern CA and Central CA. The similar genetic differentiation is also observed in our VGSC genotyping dataset. Only 3/60 samples from Dinuba and 9/60 samples from Clovis had "susceptible" alleles, represented by 1 and 2 genotypes respectively. Of note, only 2 of the 309 mosquitoes tested were heterozygous at position F1534C, and those mosquitoes knocked down at 5 (Greater LA) and 18 minutes (Clovis), though it is di cult to ascertain the signi cance of this result with such a low frequency of this genotype. When controlling for genotype, there were still signi cant differences between the strains (Fig. 5). All of the Sanger and Kingsburg mosquitoes were homozygous for the resistance associated alleles. Their respective median knockdown times were 42 min (N = 58) and 38 min (N = 64) in the assayed mosquitoes (Fig. 5). This observation contrasts with median knockdown times observed in mosquitoes of the identical VGSC genotype from the Clovis and Greater LA populations. Phenotypic heterogeneity existing within and between groups sharing a common genotype were observed because of conducting bottle bioassays on individual mosquitoes. All bottle bioassays were performed coating each bottle with the same amount of chemical from the same pyrethrum batch and stock solution to ensure standardized and comparable assay conditions.

Kaplan-Meier Survival Analysis
Kaplan-Meier survival analysis is a non-parametric statistical test of survival that accounts for individuals that may not have experienced the "event" (in this case, knockdown) during the study period (35). Survival curve analysis of mosquitoes with equivalent genotypes reveals a range of median knockdown times that correlate with the relative proportions of susceptible versus resistant VGSC alleles. In cases where only a single individual represented a genotype, pvalues were not reported (Fig. 4C). In general, individuals containing susceptible VGSC alleles do not last as long as those with the fully resistant genotype.
However, some individuals carrying the homozygous resistant genotype at all alleles have knockdown times equivalent to individuals with homozygous susceptible genotypes (Fig. 4A-B). We reason this is unlikely due to methodological error, as these low knockdown times were observed across multiple populations.

Discussion
Using the individual bottle bioassay in this study allowed us to explore the phenotypic variability of resistance in Ae. aegypti, and pair that phenotypic information with genotype, a previously unexplored relationship in Ae. aegypti. Target site mutations were prevalent in our sample populations and play an important role in resistance. Still, signi cant variability in knockdown time was observed in insects when controlling for genotype, implicating the role of other mechanisms of resistance and warranting further study. This individual bottle bioassay method yielded a bimodal distribution of knockdown times. Insects with susceptible alleles did knockdown earlier, though some individuals with fully resistant genotypes exhibited knockdown times similar to those observed in insects with susceptible alleles. This indicates running bottle bioassays with the synergist PBO will provide control agencies with important information about the resistance pro le of local Ae. aegypti. Pyrethrum without PBO (MERUS 3.O-Clarke) is likely to see continued use, as organic farming and some residences inhibit the use of PBO. Deltamethrin has recently been approved for use in California, and there is evidence of less deltamethrin resistance across the state.
Deltamethrin correlative studies will also be needed. Further study into the resistance mechanisms at play in California Ae. aegypti populations will provide vital information to mosquito control districts across the state, as understanding the resistance mechanisms allows for careful planning of insecticide use, rotation, and spray cover.

Declarations
Ethics approval and consent to participate Not Applicable

Consent for publication Not Applicable
Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
AC acknowledges funding support from the Paci c Southwest Regional Center of Excellence for Vector-Borne Diseases funded by the U.S. Centers for Disease Control and Prevention (Cooperative Agreement 1U01CK000516) pacvec.us. The funding body played no role in the design of this experiment.
The establishment of Ae. aegypti in California has raised concerns that they could facilitate local transmission of arboviruses such as dengue and Zika as seen in Florida, Texas and Puerto Rico (36)(37)(38). While pyrethroids are favored as a chemical method for their control, resistance is widespread in California and globally (7,8,32,39). Despite signi cant work on pyrethroid resistance in Ae. aegypti, questions remain concerning the identi cation and relative compounding roles that VGSC mutations, metabolic detoxi cation mechanisms, other resistance mechanisms, and environmental factors have on conferring insecticide resistance. Additional resistance mechanisms such as reduced cuticular permeability and behavioral resistance have been shown to be of signi cant importance in insects (40,41).
The VGSC mutations V1016I, F1534C and V410L are prevalent in Ae. aegypti in the Americas and have been closely monitored by the California Department of Public Health (CDPH) (7,13,15,17). These surveillance efforts found that between 2015 and 2017 resistant alleles were largely xed in the Central Valley population, and less abundant (> 80% for F1534C and > 61% for V1016I from 2015-2017) but increasing in prevalence in the Southern population of Ae. aegypti (7). Studies in Mexico have identi ed similar trends (17). Our results were similar to the CDPH ndings (Table 1). In our samples F1534C was nearly xed, with only two heterozygous individuals (one from Clovis and one from Greater LA). The detection of a heterozygote from Clovis was surprising, given our relatively small sample size compared to the 1200 mosquitoes tested by Liebmen et al. between 2015 and 2017. Our sample mosquitoes were captured in late 2018, and follow the general trend found in Liebmen et al. 2019 (7). In California, Ae. aegypti appear to have dispersed along the major Interstate 5 route so it is possible that mosquitoes from the Southern Population and Central Valley population are routinely moved between regions (42). Recent whole genome sequencing data as well as SNP chip data support multiple introductions of Ae. aegypti into CA, and found evidence of distinct genetic clusters converging in the Central Valley (34,43).
Most of our samples had all 5 VGSC mutations (Fig. 4-5). Resistance alleles were likely prevalent in the founder populations in California . Previous research has found that the Central Valley populations appear genetically similar to Ae. aegypti from the South-Central US, while the Greater LA mosquitoes are more similar to southwestern Ae. aegypti (2,34). Resistance is widespread in populations in the United States, though the F1534 allele is still found in some areas (44,45). Interestingly, bottle bioassay results reported by local control agencies had not always followed the patterns that would be expected given the relative proportions of resistance associated SNPs, which was also in line with our results using the individual bottle bioassay test (Fig. 3-5).
The Xenopus oocyte expression system for Ae. aegypti sodium channels has facilitated investigation into the role of individual mutations assayed in pyrethroid resistance (13,31,46,47). The mutations V410L + F1534C and V1016I + F1534C have even been studied in combination (13,16,47). Little is known about the S723T mutation (48) though with the high frequency of this mutation in all populations tested (59% in Greater LA and > 92% in all Central Valley populations, Table 1), future functional analyses would be important to understanding the role this mutation plays in resistance. These mutations act in combination with metabolic mechanisms of pyrethroid resistance mediated by the upregulation, overexpression or duplication of cytochrome P450 enzymes encoding genes (14,(49)(50)(51). Bottle bioassays with cytochrome P450 inhibitor, piperonyl butoxide (PBO), have implicated that cytochrome P450s play a synergistic role in combination with VGSC mutations to confer resistance in mosquitoes from Clovis (32). The signi cant differences in knockdown time found in our samples with the resistant genotypes (Fig. 5) supports the hypothesis that other mechanisms are important and variable within California populations and these other mechanisms should be explored in future studies. This indicates that running bottle bioassays that include the cytochrome P450 inhibitor PBO will provide local control agencies important information when evaluating resistance. Given the functional evidence in the literature, and our assay results (Fig. 4-5), the increasing prevalence of the F1534C mutation, particularly in combination with V410L and V1016I reliably indicates an elevated level of resistance to type I pyrethroids (7,13,31,47). However, the individual bottle assays employed here prompts us to question that, assuming good conditions, the presence of resistance alleles in the VGSC gene may not guarantee adulticide failure, and susceptible alleles may not ensure control success.
For this reason, it is important to pursue detailed investigations in the biology underlying the resistance phenotype. Additional knowledge on this subject would facilitate identi cation of additional genetic and/or biochemical markers. The ideal goal would be to identify markers that provide quantitative measures of the resistance phenotype in eld caught mosquitoes which would boost the predictive power of these assays. In addition, identi cation of new targets underlying the phenotype opens the door for development of alternative strategies and new synergists to augment existing insecticidal compounds. ETK, LKM, KB, AJC, YL, and GMA conceived and designed the study. ETK, LKM, KB, KVS, AZ, and TVS collected data. ETK, LKM, and GMA analyzed the data.
ETK and LKM prepared the manuscript. ETK, LKM, AJC, YL, and GMA revised and edited the nal manuscript. All authors read and approved the nal manuscript.