Putative pleiotropic effects of the knockdown resistance (L1014F) allele on the life-history traits of Anopheles gambiae

Two laboratory reference strains of An. gambiae s.s. were used. The insecticide-susceptible reference strain Kisumu, sampled from Kenya the early 1950s and was maintained at insectary [41]. The KisKdr strain, which is homozygous [kdr^RR] for the L1014F allele and resistant to both pyrethroids and organochlorines, was obtained by introgression of the kdr^R (L1014F) allele into the Kisumu genome [42]. This strain has the same genetic background as Kisumu [kdr^SS] and was free of metabolic resistance.

In order to investigate the role of kdr^R (L1014F) allele in An. gambiae s.s. blood-feeding behaviour, heterozygote [kdr^RS]-resistant mosquitoes were obtained by crossing Kisumu females [kdr^SS] with KisKdr males [kdr^RR] and Kisumu males with KisKdr females encoded F1-1 (♀Kisumu X ♂KisKdr) and F1-2 (♂Kisumu X ♀KisKdr), respectively. For routine rearing in the insectary at the Regional Institute of Public Health/ University of Abomey-Calavi (Benin), these strains were reared under soft conditions (insecticide-free laboratory environment) in a climate-controlled room at a temperature fixed at 27 °C (± 0.2), a relative humidity of 70% (± 8) and 12:12 light and dark period. Larvae were reared in plastic trays (about 30 × 20 cm) and fed with TetraMin Baby fish food. Pupae were collected and placed in small plastic cups inside a fresh cage for adult emergence. Adult mosquitoes were kept in 30 × 30x30 cm insect cages (produced locally) and continuously supplied. Mosquitoes were fed ad libitum on 10% honey solution (made with deionized water) until they were ready to be used for further assays. Female individuals were blood-fed on laboratory rabbits (used for the purpose of blood-feeding mosquitoes) twice a week. Gravid females were allowed to oviposit in plastic petri dishes containing a water-soaked cotton covered with filter paper. The eggs were collected and put in plastic trays containing dechlorinated water (1 L per tray) for hatching.

The larvae from each mosquito strain reared in insecticide-free laboratory conditions as described, were used for the survival assays. To assess larval mortality associated with kdr^R (L1014F) allele in each mosquito strain, assays were performed as described by Yahouédo et al. [43]. In total, 480 first instar larvae (L1) of each mosquito strain were used. For each replicate, 32 larvae were pipetted into a 50 mL graduated plastic beaker (9 cm diameter). The beaker was filled with dechlorinated water to the 32 mL mark and larvae were then poured into a new petri dish. The petri dishes remained covered with the lids and their positions were changed every day to compensate for any localized differences that may exist on the rack. Petri dishes were used in order to reduce variation in larval growth rate. Every day, the larvae of each petri dish were fed with 640 µg of TetraMin Baby fish food. Water was changed every two days to reduce the effect of pollution. The petri dishes containing larvae were inspected once daily and the dead pupae or larvae were recorded and removed. Daily mortality of larvae was monitored until the last one reached pupal stage. The experiments were performed three times.

Membrane feeding assays (MFAs) previously described by Kristan et al. [44] were performed to blood-feed the mosquitoes. The 3–5-days old females of Kisumu (n = 495), KisKdr (n = 200) and those from the crossings, namely F1-1 (n = 95) and F1-2 (n = 105), were used in three different experiments. Mosquitoes were glucose-starved (with access to water-soaked cotton) for 24 h and the batches of 25 individuals were separately exposed for 30 min to membrane feeders containing the blood sample pre-heated following procedures described in [45]. The fully blood-fed mosquitoes were scored 24 h later and were kept for survivorship assessment post-blood feeding.

A portion of the blood-fed mosquitoes was used to assess the blood meal size using a spectrophotometer (MULTISCAN GO, Thermo Scientific) as previously described [46]. Each experiment using at least 30 individuals per strain, was performed three times.

After the blood-feeding assays, successfully blood-fed females from Kisumu (n = 172), KisKdr (n = 168), F1-1 (n = 71) and F1-2 (n = 90) were transferred into brand-new disposable paper cups (an average 10 females per cup) and were allowed to feed on 10% honey solution. The mortality was recorded daily until the death of the last mosquito.

Data were recorded in appropriate designed forms, entered into Microsoft Excel for data cleaning and exported to R statistical software version 3.4.4 [47] and GraphPad Prism 8.0.2 software (San Diego, CA, USA) for analysis. The normality of data distribution was checked using Shapiro Wilk test [48].

Fecundity of each mosquito strain was assessed as the total number of eggs over the total number of females that contributed to oviposition. A correlation between kdr^R genotype and fecundity was calculated using negative binomial model (NBM) defined as follow: log (Ov) = Genotype + ε where Ov is the number of eggs/female; Genotype is the two-level factor corresponding to the different genotypes tested; ε is the error parameter which follows a negative binomial distribution. For each mosquito strain, fertility was evaluated as percentage of hatched larvae by dividing the total number of first instar larvae over the total number of eggs. A correlation between kdr^R genotype and fertility was calculated using NBM, defined as follow: log (Ha) = Genotype + ε where Ha is the percentage of larvae/egg batch. Descriptive statistics were used to calculate pupation percentage (number of pupae/number of first instar larvae), blood-fed mosquito percentage (number of blood-fed mosquitoes/number of exposed mosquitoes). The Chi-square independence test was performed to compare proportions using the R statistical software [47]. The Mann–Whitney procedure was used to compare the means between mosquito strains. For the larval and blood-fed females survivorships, differences in the computed survival curves of Kisumu and KisKdr strains were analysed using Kaplan–Meier pair-wise comparisons [49]. The Log-rank test was performed to evaluate the difference in survival time between the mosquito strains [50]. Differences in larval survival time and in adult survival time post-blood meal between the two genotypes were tested using Cox proportional hazards regression model (Cox model) with a binomial error distribution. The models were calculated as follows: Survival = Genotype + ε, where Survival is a proportion of dead larvae or adults; Genotype is the two-level factor corresponding to the different genotypes tested; ε is the error parameter which follows a binomial distribution. The pupae were censored in the larval survivorship analysis. The significance of differences in blood-feeding rates between the genotypes was assessed with the following generalized linear models (GLM): Fed = Genotype + ε, where Fed is the blood-fed status; Genotype is a three-level factor corresponding to the different genotypes tested ([kdr^SS], [kdr^RS] and [kdr^RR]); ε is the error parameter which follows a binomial distribution. All these analyses were set at significance threshold of p < 0.05.

The mean number of eggs laid per mosquito female (fecundity) and the average larval hatching rate (fertility) were significantly different between the two strains (30.72 ± 19.92 eggs/KisKdr female vs 87.98 ± 44.51 eggs/Kisumu female, p = 1.07 × 10^–10; Fig. 1) and (72.89 ± 15.7% hatched larvae/KisKdr female vs 81.89 ± 12.4% for Kisumu female, p = 0.02 × 10^–1; Fig. 2). Moreover, the KisKdr female fecundity and fertility decreased by 1.05 (GLM.NB: F = 58.21, Δdf = 1, p = 8.71 × 10^–12) and 0.12 (GLM.NB: χ² = 1062, Δdf = 1, p = 0.01 × 10^–1), respectively, when compared to those of Kisumu females. Overall, the reproductive success of KisKdr [kdr^RR] females was significantly lower than that of Kisumu [kdr^SS] females.

The median survival times of Kisumu and KisKdr larvae were, respectively, 10 days and 11 days (Fig. 3A). However, the survival time of Kisumu larvae was significantly shorter than that of KisKdr larvae (Log-rank test: χ² = 110, Δdf = 1, p = 2.10^–16). Furthermore, more than 50% of KisKdr larvae were still alive and have reached the pupal stage at the end of the larval following-up period (Fig. 3A). The risk of death of individual larvae when bearing kdr^R allele at homozygote state [kdr^RR] is reduced by a factor of 59% compared to homozygote susceptible larvae [kdr^SS] (Cox model: likelihood ratio test (LRT): χ² = 114.7, Δdf = 1, p = 2.10^–16). Consequently, pupation rate in KisKdr females was significantly higher (85.84%, CI_95% = [84.12–87.75]) than that recorded for Kisumu strain (54.05%, CI_95% = [51.34–56.74]) (Fig. 3B).

Overall, 84% (168/200) of KisKdr females and 34.75% (172/495) of Kisumu females subjected to the assays took a blood meal, as shown in Fig. 4A. The KisKdr females showed a significantly higher blood-feeding rate than the Kisumu ones (χ² = 136.32, df = 1, p = 2.2 × 10^–16). Interestingly, the offspring heterozygote [kdr^RS] females F1-1 and F1-2 displayed also consistently higher per cent of blood-fed individuals (respectively, 74.74% (71/95) and 85.71% (90/105)) than that of Kisumu [kdr^SS] individuals (χ² = 121.89, df = 2, p = 2.2 × 10^–16) (Fig. 4A). In all cases, mosquitoes harbouring the kdr^R allele at both homozygote and heterozygote states showed higher blood-feeding ability compared to the susceptible homozygote Kisumu strain (GLM: (RLT): χ² = 215.28, Δdf = 2, p = 2.2 × 10^–16).

When using other batches of mosquito females for the same blood-feeding assays, the average blood volume ingested by KisKdr individuals, was similar to that of Kisumu specimens (p = 0.22) while the average amount of blood ingested by the heterozygous offspring (1.68 µL/mg) was significantly higher than for Kisumu mosquitoes (1.36 µL/mg) (p = 8.10^–4), as shown in Fig. 4B.

The median survival times after blood-feeding of the homozygous susceptible (Kisumu) and resistant (KisKdr) mosquitoes were, respectively, 7 days and 8 days (Fig. 5A). No significant difference in the survival time was observed between the two strains (Log-rank test: χ² = 0.6, Δdf = 1, p = 0.4).

Moreover, the offspring heterozygote [kdr^RS] displayed a longer median survival time after blood-feeding (10 days) compared to those of their parents (8 days for KisKdr; Log-rank test: χ2 = 48, Δdf = 2, p = 4.10^–11 and 7 days for Kisumu; Log-rank test: χ2 = 54.9, Δdf = 2, p = 10^–12). In addition, these offspring displayed a higher survival rate when compared to KisKdr females (hazard ratio = 0.44; Cox model: (LRT): χ2 = 38.12, Δdf = 1, p = 7.10^–10) and Kisumu specimens (hazard ratio = 0.41; Cox model: (LRT): χ2 = 44.93, Δdf = 1, p = 2.10^–11) as shown in Fig. 5A, B.