The L1014F Kdr mutation is associated with a higher prevalence and load of the Plasmodium-blocking symbiont Microsporidia MB In Anopheles Gambiae s.l. In Benin

Abstract

Introduction

The Plasmodium transmission-blocking endosymbiont Microsporidia MB was previously identified in Anopheles gambiae s.l., but its association with the carriage of the genotypes of the L1014F kdr mutation, as well as the ecological factors driving its geographical distribution remain understudied.

Methods

Adult mosquitoes were field-collected using human landing catches (HLCs) across 60 villages in the Covè, Ouinhi, and Zangnanado communes of southern Benin. After morphological identification, a sub-sample of An. gambiae s.l. were molecularly speciated, and genotypied for the L1014F kdr mutation by Polymerase Chain Reaction (PCR). Enzyme-Linked Immunosorbent Assay (ELISA) and qPCR were also used to assess infection with Plasmodium falciparum sporozoites and Microsporidia MB, respectively. The environmental variables that drive the habitat suitability for Microsporidia MB were also assessed using Maximum Entropy (MaxEnt) modelling.

Results

The An. gambiae complex (N = 1040) was composed of 93.7% An. coluzzii, 4.4% An. gambiae s.s., 0.2% An. gambiae s.s./coluzzii hybrids, while the rest failed to amplify. Infection prevalence with Microsporidia MB was 1.6% (95% CI: 0.7–3.3) in An. coluzzii and 2.2% (95% CI: 0.1–13.2) in An. gambiae s.s. The P. falciparum sporozoite rate was 2% (95% CI: 1.2–3.1, N = 974) in An. coluzzii, and null in An. gambiae s.s. (N = 46). None of the mosquitoes infected with Microsporidia MB were infected with P. falciparum. The frequency of the L1014F kdr mutation was 75.1% (95% CI: 73.1–76.9) in An. coluzzii and 91.3% (95% CI: 83.1–95.9) in An. gambiae s.s. Microsporidia MB was absent in kdr-SS mosquitoes but was present in low proportions in both kdr-RS and kdr-RR mosquitoes (1.9%, 95% CI: 0.6–5.1). The mean load of Microsporidia MB DNA was higher in kdr-RR (23.23 ng/µl, 95% CI: 18.77–28.48) compared to kdr-RS (10.13 ng/µl, 95% CI: 7.36–13.91) mosquitoes. The elevation and soil contributed to explain, at 78% and 20% respectively, the habitat suitability for Microsporidia MB.

Conclusion

In this study, we demonstrated that An. gambiae s.l. mosquitoes bearing the L1014F kdr resistant allele was associated with a higher prevalence and load of Microsporidia MB than their susceptible counterparts. Moreover, the geographical distribution of Microsporidia MB was found to be influenced by certain environmental conditions, which warrant further large-scale investigations.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-026-04810-5.

Background

Malaria, transmitted by Anopheles (An.) mosquitoes, remains a global public health issue with approximately 263 million annual cases, mostly occurring in Africa [1]. Globally, the malaria case incidence declined between 2000 and 2015, stalled between 2016 and 2022, and increased slightly in 2023 compared to 2022 [1]. The stagnation of control progress between 2000 and 2015 is partially attributable to vector insecticide resistance jeopardizing the efficacy of insecticide-based vector control tools [2–4]. This stresses the vital need for complementary control tools, including novel biotechnological ones.

Although some microbes were found to substantially impair pathogen transmission in mosquitoes [5]; the exact pathogen blocking molecular mechanisms have not been elucidated so far. Recently, it has been discovered that Microsporidia MB—a naturally-occurring bacterial endosymbiont—can block the infection of An. arabiensis with Plasmodium (P.) falciparum [6]. Moreover, it has been shown that Microsporidia MB does not induce any major negative impacts on host longevity, development, fecundity, and fertility [6, 7]. These characteristics make Microsporidia MB a promising biological tool for blocking the Plasmodium transmission cycle in Anopheles mosquitoes. Thus, the endosymbiont-based control tools, which prevent pathogen transmission, could be deployed alone or together with traditional insecticide-based control tools to reduce malaria transmission. The potential of endosymbiont-based control strategies was recently shown by the positive effect of the deployment of Wolbachia-based strategies in controlling dengue fever in Brazil and Indonesia [8, 9]. This success was due to the ability of the Wolbachia endosymbiont to impair viral replication and dissemination, and to reduce pathogen transmission by Aedes mosquitoes [10].

A recent study conducted in Nigeria and Niger Republic reported that insecticide phenotypic-resistant mosquitoes displayed a significantly higher prevalence of infection with Microsporidia MB than susceptible ones [11]. In the south of Benin, a recent cross-sectional study reported the natural presence of Microsporidia MB in both An. coluzzii and An. gambiae s.s. with respective infection prevalences of 57.0% (95% CI: 45.4–67.9) and 41.0% (95% CI 25.9–57.8) at the mosquito pools level [12]. Moreover, the L1014F knockdown resistance (kdr) mutation – a well-validated pyrethroid resistance marker involving a structural change in the voltage-gated sodium channel – was found close to fixation with the presence of three genotypes (RR, RS, and SS) nationwide, particularly in the Covè, Ouinhi, and Zangnanando districts [13, 14]. The present study aimed to assess the association between L1014F kdr genotypes and the Microsporidia MB infection in An. gambiae s.l. In addition, ecological niche models designed based on certain environmental variables (minimum, maximum, and average temperature; elevation; soil; aspect; slope; normalized differential vegetation index) were used to map the occurrence of Microsporidia MB, and predict the most suitable areas for its development.

Methods

Study area

Mosquitoes used for the present study were collected between March and May 2023 (onset of the rainy season), during the last round of entomological collection for a three year long randomized controlled trial (RCT). The RCT aimed to evaluate how pyrethroid-pyriproxyfen and pyrethroid-chlorfenapyr long-lasting insecticidal nets (LLINs) perform on malaria transmission, compared to pyrethroid-only LLINs [15]. The current study occurred during the final trial round to ensure that mosquito infection with P. falciparum sporozoites was not influenced by insecticide incorporated into nets, which could be a major confounding factor that would obfuscate Microsporidia MB detection. The majority of insecticide in the trial LLINs was lost only two years post-deployment [16, 17].

Of note, the RCT was conducted in 60 villages, in the Covè (07°13′08.0400′′N, 02°20′21.8400′′E), Ouinhi (07°05′00′′N, 02°29′00′′E), and Zagnanando (07°16′00′′N, 02°21′00′′E) communes of the Zou department, situated 154 km away from Cotonou, the economic capital of Benin (Fig. 1). Agriculture, livestock, hunting, fishing, hospitality industry and trade are the main economic activities in the area. In the study districts, the climate is subequatorial, with two rainy seasons (April to July and September to October), and two dry seasons (November to March and July to August). The malaria transmission, principally by An. coluzzii and An. gambiae s.s. in the area was intense [13].

Fig. 1.

Map of the study area by Adoha et al. [18]

Mosquito sampling

In each of the 60 surveyed villages, 4 houses (1 chosen at random, and 3 others selected 15–20 m around the first one) were sampled from 07:00 p.m. to 06:00 a.m. using human landing catches (HLCs). All collections were performed over 12 nights, with 5 villages surveyed simultaneously during a single night. Each village was surveyed once. In each house, there were 2 volunteers (one positioned indoors, and the second outdoors) who used haemolysis tubes and flashlights to collect all mosquitoes which landed on their legs. In each village, the first group of 8 volunteers who collected from 07:00 p.m. to 01:00 a.m. was replaced by a second group 8 other volunteers from 01:00 a.m. to 06:00 a.m.

The same sampling effort (4 collections × 60 villages = 240 collections) was performed both indoors and outdoors.

Collected mosquitoes were morphologically identified to species-level using a stereomicroscope and the taxonomic keys of Gillies and De Meillon [19] and Coetzee [20]. Collections were then stored on silica gel at room temperature for further molecular analyses.

Molecular analyses

Molecular identification of species, L1014F kdr mutation and P. falciparum sporozoite infection

The head-thoraces of 16–18 individuals of An. gambiae s.l. were randomly selected and screened for P. falciparum sporozoite infection using circumsporozoite protein (CSP) ELISA [21]. Their legs, wings, and abdomens were also used for DNA extraction using the DNeasy Blood & Tissue Kit (QIAGEN, GERMANY). The extracted DNA served to identify molecular species [22] and to assess the frequency of the L1014F kdr mutation [23] by PCR.

Molecular detection and quantification of Microsporidia MB

Due to limited funding, DNA from only 494 An. gambiae s.l. sampled from all the 60 villages (about 8–9 mosquitoes per village) were used for the molecular identification of Microsporidia MB. These samples include DNA of all the mosquitoes carrying the kdr-SS genotype (N = 72), all those infected with P. falciparum sporozoites (N = 18), and all those detected as An. gambiae s.s. (N = 46), while the rest (N = 358) were randomly selected across all surveyed villages (Table 1).

Table 1.

Number of mosquitoes tested according to species, L1014F Kdr genotype, and status of infection with P. falciparum

	Molecular species and L1014F kdr genotypes
	An. coluzzii				An. gambiae s.s.
Infection status with Pf	RR	RS	SS	Total 1	RR	RS	SS	Total 2	Total 1 + Total 2
Negative	159	199	72	430	38	8	0	46	476
Positive	13	5	0	18	0	0	0	0	18
Grand Total	172	204	72	448	38	8	0	46	494

Pf Plasmodium falciparum, RR Homozygous resistant, RS Heterozygous resistant, SS Homozygous susceptible

Molecular detection and quantification of Microsporidia MB were performed using specific primers (MB18SF, 5’-CGCCGGCCGTGAAAAATTTA-3’; MB18SR, 5’-CCTTGGACGTGGGAGCTATC-3’) designed to detect this bacterium [6, 24] using a MIC qPCR cycler (Step One plus). A 10 µL mixture consisting of 5 µL of SYBR green, 1 µL of nuclease-free PCR water, 1 µL of each primer (forward and reverse), and 2 µL of DNA were prepared. Reactions were run in technical duplicate. The reaction conditions were: 95 °C for 15 min, 40 cycles (95 °C for 60 s, 65 °C for 90 s, 72 °C for 60 s), 72 °C for 5 min, followed by a melting curve. The efficiency of the primers was determined following linear regression, and the standard curve was generated. The data collected following the quantification of Microsporidia MB data were normalized using primers from the housekeeping gene RpS7 (S7F, TCCTGGAGCTGGAGATGAAC; S7R, GACGGGTCTGTACCTTCTGG) [25].

Ecological niche modelling for Microsporidia MB

Occurrence data for Microsporidia MB was recorded at the level of the georeferenced study villages. Fourteen environmental predictors, including climatic and topographic variables (Supplemental file, Table S1), were collected for the three months study period, from the Shuttle Radar Topography Mission (SRTM), the Normalized Difference Vegetation Index (NDVI), and the WorldClim database at 1 km spatial resolution. All raster datasets were clipped to the study area extent, resampled using bilinear interpolation, and projected to the WGS 1984 World Mercator system. Of the 14 environmental predictors initially used to run the first model, 11 showed no contribution and were removed (Supplemental file, Table S1).

The MaxEnt algorithm (version 3.4.4) was used for modelling the ecological niche for Microsporidia MB. This model, based on the principle of maximum entropy, estimates the probability of a species’ presence based on occurrence data and environmental variables.

The “cloglog” format was used to set the model output and produce a prediction map with values varying between 0 and 1, showing habitat suitability. A k-fold cross-validation (allowing cross-evaluation of model performance) with 10 replicates was used to enable the contribution of all presence records to the training and the testing of the model. The model was set up to a maximum number of 5,000 iterations, with quadratic and linear features, as well as a regularization multiplier set to 2 in order to limit overfitting due to the small number of points.

Permutation importance analysis was used to evaluate the contribution of each predictor variable. The model performance was assessed through the Area Under the receiver operating characteristic Curves (AUC). Of note, the closer the AUC value is to 1, the better performing the model is. Presence probability maps were produced in logistic format (values from 0 to 1), indicating the environmental suitability of different areas for the presence of Microsporidia.

ArcGIS software (version 10.8) was used to generate the final habitat suitability map for Microsporidia MB.

Statistical data analyses

In An. gambiae s.l., the P. falciparum sporozoite rate, and infection prevalence with Microsporidia MB was calculated by dividing the number of infected mosquitoes by the total tested. The frequency of the L1014F kdr mutation was calculated using the following formula: F (L1014F Kdr)= (2*RR + RS)/(2*(RR + RS+SS)).

The confidence intervals of the sporozoite rate, infection prevalence with Microsporidia MB, and frequency of the L1014F kdr mutation was calculated using the exact binomial test. The Chi-square test of comparison of proportions was used to compare all three outcomes.

The load of DNA of Microsporidia MB was estimated in positive samples from their mean Ct values by drawing a standard curve from known dilutions of the positive control. A linear regression was performed between the Ct values and the decimal logarithm of the loads according to the formula: Ct = a × log₁₀(load) + b, where a is the slope and b the y-intercept of the line. The load of DNA of Microsporidia MB was thus calculated using the following formula: DNA load = 10^((Ct − b)/a). The Poisson method was used to compare and estimate the confidence intervals of mean DNA load per L1014F kdr genotype. All analyses were performed using R statistical software, version 4.4.3.

Results

Anopheles species composition

Of the 8,855 Anopheles mosquitoes collected during the last round of the entomological data collection of the RCT, 44.8% (N = 3971) were sampled indoors, and the rest outdoors. An. gambiae s.l. accounted for 96.6% (3837/3971) of the indoor collection and 97.9% (4785/4884) of the outdoor collection, with a mean of 97.4% (8622/8855) for the total collection (indoor + outdoor). Other Anopheles mosquitoes, including An. funestus, An. pharoensis, and An. ziemanni, were also found but at lower frequencies both indoors and outdoors (< 3%) (Fig. 2).

Fig. 2.

Anopheles species composition in the study area

Of the 1040 An. gambiae s.l. molecularly speciated, 93.7% (n = 974) were An. coluzzii, 4.4% (n = 46) were An. gambiae s.s., 0.2% (2/1040) were hybrids (An. gambiae s.s./coluzzii), while 1.7% (18/1040) did not amplify. The same trend was observed both indoors (Fig. 3A) and outdoors (Fig. 3B).

Fig. 3.

Molecular species composition in the Anopheles gambiae complex indoors (A; N = 526), and outdoors (B; N = 514)

Infection prevalence with Microsporidia MB in An. gambiae s.s. and An. coluzzii

Of 493 specimens of An. gambiae s.l. tested for the presence of Microsporidia MB, 8 were positive, which equates to a mean infection prevalence of 1.6% (95% CI: 0.8–3.3).

At the molecular species level, the infection prevalence was 1.6% (95% CI: 0.7–3.3, 7/448) in An. coluzzii versus 2.2% (95% CI: 0.1–13.2, 1/46) in An. gambiae s.s. (Table 2).

Table 2.

Infection prevalence (IP) with Microsporidia MB in Anopheles Gambiae sensu stricto and Anopheles coluzzii

Mosquito Species	N identified	N tested	N Positive	IP (%)	95% CI
An. coluzzii	974	448	7	1.6	0.7–3.3
An. gambiae s.s.	46	46	1	2.2	0.1–13.2

An Anopheles, N Number of, IP Infection prevalence, CI Confidence interval

P. falciparum sporozoite rate (SR) in An. gambiae s.l. and its molecular species

Of 1,022 specimens of An. gambiae s.l. tested, 18 were positive for P. falciparum sporozoites, equating to an average SR of 1.7% (95% CI: 1.0–2.7.0.7) (Table 3).

Table 3.

Sporozoite rate in Anopheles Gambiae sensu stricto and Anopheles coluzzii

Mosquito Species	N tested	S+	SR (%)	95% CI
An. coluzzii	974	18	1.8	1.0–2.9.0.9
An. gambiae s.s.	46	0	0	0.0–9.6.0.6
An. gambiae s.s./coluzzii	2	0	0	0.0–80.2.0.2
An. gambiae s.l.	1022	18	1.7	1.0–2.7.0.7

An Anopheles, N Number of, S+ Number of sporozoite positive mosquitoes, SR Sporozoite rate, CI Confidence interval

At the molecular species level, the SR was 1.8% (95% CI: 1.0–2.9.0.9) in An. coluzzii, and null in An. gambiae s.s. and hybrids (An. gambiae s.s./coluzzii) (Table 3).

Of the eight individual An. gambiae s.l. found positive with Microsporidia MB, none harboured P. falciparum sporozoites.

Frequency of the L1014F kdr mutation in An. gambiae s.l. and its molecular species

Of the 1,022 specimens of An. gambiae s.l., 58.8% (N = 600) were kdr-RR genotype, 34.2% (N = 350) kdr-RS genotype, and 7% (N = 72) kdr-SS genotype. The mean frequency of L1014F kdr mutation for An. gambiae s.l. was 75.8% (95% CI: 73.9–77.7) (Table 4).

Table 4.

Frequency of the L1014F Kdr mutation in Anopheles Gambiae sensu Lato collected from villages in Covè, Ouinhi and Zangnanado communes in Zou Department, Southern Benin

Mosquito Species	Total	RR	RS	SS	F(L1014F kdr)	95% CI
An. coluzzii	974	561	341	72	75.1	73.1–76.9
An. gambiae s.s.	46	38	8	0	91.3	83.1–95.9
An. gambiae/coluzzii	2	1	1	0	75	21.9–98.7
An. gambiae s.l.	1022	600	350	72	75.8	73.9–77.7

An Anopheles, CI Confidence interval, RR Homozygous resistant, RS Heterozygous resistant, SS Homozygous susceptible

At the molecular species level, the frequency of L1014F kdr mutation was 75.1% (95% CI: 73.1–76.9) in An. coluzzii and 91.3% (95% CI: 83.1–95.9) in An. gambiae s.s. (Table 4).

Infection prevalence with Microsporidia MB per L1014F kdr genotype (SS, RS, and RS) in An. gambiae s.l.

Overall, the infection prevalence with Microsporidia MB was null in kdr-SS, and 1.9% in both kdr-RS (95% CI: 0.6–5.1, N = 212), and kdr-RR (95% CI: 0.6–5.1, N = 210) genotypes of An. gambiae s.l. (Table 5).

Table 5.

Infection prevalence with Microsporidia MB per L1014F Kdr genotype (SS, RS, and RR) in Anopheles Gambiae sensu Lato populations collected from Covè, Ouinhi and Zangnanado communes in Zou Department, Southern Benin

L1014F kdr Genotypes	N tested	N Positive	IP (%)	95% CI
SS	72	0	0	0.0–6.3.0.3
RS	212	4	1.9	0.6–5.1
RR	210	4	1.9	0.6–5.1

N Number of, IP Infection prevalence, CI Confidence interval, RR Homozygous resistant, RS Heterozygous resistant, SS Homozygous susceptible

The mean load of DNA of Microsporidia MB was significantly higher in the kdr-RR genotype (23.23 ng/µl, 95% CI: 18.77–28.48) than in the kdr-RS genotype (10.13 ng/µl, 95% CI: 7.36–13.91) (Table 6).

Table 6.

Load of Microsporidia MB in each specimen bearing the L1014F Kdr mutation

Genotypes	Load of DNA of Microsporidia MB (ng/µl)
	Mosquito 1	Mosquito 2	Mosquito 3	Mosquito 4	Mean	95% CI
RS	0.06	5.27	0.06	35.13	10.13	7.36–13.91
RR	12.20	18.40	28.54	33.77	23.23	18.77–28.48

Overall, Microsporidia MB was absent in kdr-SS genotype, and had low prevalence with lesser load in kdr-RS genotype, and low prevalence with higher load in kdr-RR genotype.

Ecological niche model for Microsporidia MB

The ecological niche model that was developed showed good predictive power of the potential distribution of Microsporidia MB, with an average AUC value of 0.825, indicating satisfactory performance.

The modelling results showed that the most suitable areas for the development of Microsporidia MB were the southwest of Covè, the south of Zangnanado, and the entire western part of Ouinhi (Fig. 4). Of the sixty investigated villages, the seven where Microsporidia MB was detected are identified on Fig. 4.

Fig. 4.

Ecological niche model for Microsporidia MB (MMB occurrences: Status of Microsporidia MB occurrences in villages, MMB- : Villages with Microsporidia MB negative samples, MMB+ : Villages with Microsporidia MB positive samples)

Contribution of environmental variables to the habitat suitability for Microsporidia MB

Of all the evaluated environmental variables (Supplemental file, Table S1), the highest contributors were elevation (78%), followed by the soil (20%), and the slope (2%) (Fig. 5). The contribution of all other environmental variables used in the model was null.

Fig. 5.

Contribution of environmental variables in the habitat suitability for Microsporidia MB

Areas with low elevation were found to be particularly conducive for the development of Microsporidia MB (Fig. 6a). Furthermore, the soil also showed a negative correlation with the habitat suitability for Microsporidia MB (Fig. 6b).

Fig. 6.

Effect of environmental variables according to the Maxent prediction (Elevation (a), and Soil (b)

Discussion

Microsporidia MB is a bacterial endosymbiont that disrupts Plasmodium development cycle, and whose transmission occurs vertically in Anopheles mosquitoes [6]. The present study provides information on its infection prevalence, relationship with L1014F kdr genotypes (RR, RS, and SS), and ecological factors that drive its geographical distribution in the Covè, Ouinhi and Zagnanando communes in southern Benin. It was carried out over the last round of data collection of a three year long randomized controlled trial (RCT) assessing the efficacy of dual active-ingredient LLINs. During the third year of the trial, pyrethroid-pyriproxyfen and pyrethroid‐chlorfenapyr LLINs performed similarly on both entomological [26] and epidemiological [16] outcomes, compared to the standard-pyrethroid only LLIN.

The most frequent Anopheles species in the study area was Anopheles gambiae s.l., while all other Anopheles species were found at very low frequencies. This confirms Anopheles gambiae s.l. as the main malaria vector in the study area. Molecular species identification performed in An. gambiae s.l. revealed that the mosquito complex was mostly composed of An. coluzzii followed by An. gambiae s.s. Previously published results by Koukpo et al. [27] suggested that climatic conditions of southern Benin may be more conducive to the development of An. coluzzii. Moreover, the numerous permanent and semi-permanent larval habitats derived from the Oueme and Zou rivers in the study area could have been favorable to the emergence of An. coluzzii, as previously demonstrated by Diabate et al. [28] in a study conducted in Burkina Faso. Overall, the Anopheles species composition was similar to what was previously observed in some other southern locations of the country [12].

The mean infection prevalence of An. gambiae s.l. with Microsporidia MB was 1.6%, which is relatively low and could be due to the fact that the mosquito collection occurred at the onset of the wet season. A previous study conducted in the Koussin-Lélé village, commune of Covè, showed a higher infection prevalence of Microsporidia MB in the dry season (14%) compared to the wet season (0.7%) [29]. Similarly, Herren et al. [6] showed that the highest prevalence of infection of mosquitoes with Microsporidia MB was observed after the peak rainfalls.

The two molecular species had similar infection prevalence with Microsporidia MB (An. coluzzii: 1.6%, and An. gambiae s.s.: 2.2%), though there was a higher number of An. coluzzii tested (An. coluzzii (N = 448), and An. gambiae s.s. (N = 46)). The association of Microsporidia MB with both species in the present study corroborates previous findings from Akorli et al. [7] in Ghana, and Moustapha et al. [11] in Nigeria and Niger.

The P. falciparum sporozoite rate was 1.8% in An. coluzzii and null in An. gambiae s.s., and none of the mosquitoes found positive with Microsporidia MB contained sporozoites of P. falciparum. Although this finding does not have enough power because of the low number of mosquitoes found positive with Microsporidia MB, the absence of malaria parasites could be due to Microsporidia MB impairing transmission of P. falciparum, as previously demonstrated by Herren et al. [6].

Given, the infection prevalence with Microsporidia MB was found to be season-dependent [6, 29], it was important to investigate environmental drivers of its presence. The ecological niche model showed that the most suitable areas for the development of Microsporidia MB were the southwest of Covè, the south of Zangnanado, and the whole western part of Ouinhi. This could be due to the fact that these areas have a low elevation, which favors the accumulation of water that serves as habitat for the development of mosquito aquatic stages that could harbour Microsporidia MB. The amount of nutrients and heavy metals in the soil can also drive the contamination level of mosquito breeding sites and affect the survival and transmission of Microsporidia MB, as previously mentioned by Tchigossou et al. [29]. While a recent laboratory study revealed that temperature modulates the dissemination potential of Microsporidia MB [30], the present field study showed no contribution of temperature to the bacteria habitat suitability. This may be because the temperature data of the current study was only collected during three months, so we suggest a larger Microsporidia MB ecological niche modelling study before drawing any conclusions.

Overall, the L1014F kdr mutation was close to fixation because the combined genotypic frequency for both kdr-RR and kdr-RS carrying mosquitoes equated to 93%, and the mean allelic frequency of L1014F kdr mutation was 75.8% for An. gambiae s.l. This situation could be due to the triennial mass distribution campaigns of pyrethroid-based LLINs in the region, combined with the uncontrolled use of pyrethroid insecticides for agricultural purposes.

A recent study conducted in Nigeria and Niger showed a possible association between infection of An. gambiae s.l. with Microsporidia MB, and phenotypic resistance to pyrethroids [11]. Indeed, this study revealed a significantly higher infection prevalence with Microsporidia MB in resistant mosquitoes than in susceptible ones. Similarly, the assessment of the infection prevalence with Microsporidia MB per L1014 kdr genotype in the present study revealed an infection prevalence of 1.9% in both kdr-RR and kdr-RS genotypes, while no mosquito with kdr-SS genotype harboured Microsporidia MB. Moreover, the mean load of DNA of Microsporidia MB was significantly higher in mosquitoes with the kdr-RR genotype than in those with the kdr-RS one. These results that show a graded effect for the prevalence/load of Microsporidia MB (absence in kdr-SS genotype, low prevalence with lesser load in kdr-RS genotype, and low prevalence with higher load in kdr-RR genotype), which suggests a possible genotype-dependent association. Thus, the carriage of the R resistant allele of the L1014F kdr mutation might have conferred an advantage to the mosquitoes of becoming more infected with Microsporidia MB. One potential explanation could be that Microsporidia MB stimulates the mosquito’s immune system or alters its metabolism, making it more susceptible to infection, or because the resistance process weakens the mosquito’s defence against parasitic infections. These hypotheses deserve confirmation through a larger Microsporidia MB-focused study. Another study demonstrated an association between overabundance of Asaia and Serratia Bacteria and deltamethrin resistance in Côte d’Ivoire [31]. Indeed, there was increased evidence that resistant mosquitoes bear those bacteria (Asaia and Serratia) as they can degrade insecticides [31]. All these display the pivotal role played by mosquito microbiota.

Although, the method used for selecting An. gambiae s.l. mosquitoes screened (all kdr-SS, all P. falciparum positive, all An. gambiae s.s., and a random selection of the rest) to identify Microsporidia MB is a pragmatic approach given funding limits, it introduces a potential bias and may affect statistical power of the analyzes, which is the main limitation for the study. The overall low prevalence and the used sampling method mean that the observed association, while suggestive, require validation with a larger, randomly selected sample set to confirm causality and generalizability. Moreover, it would also be interesting to investigate how the physico-chemical characteristics of mosquito breeding sites (pH, turbidity, dissolved oxygen, conductivity, total dissolved solids, salinity) could influence the dissemination of Microsporidia MB.

Conclusions

Findings of the present study show that the carriage of the L1014F kdr resistance allele by An. gambiae s.l. mosquitoes was associated with greater prevalence and load of Microsporidia MB, compared to homozygous susceptible ones. Moreover, elevation and soil were the ecological drivers that influence infection with Microsporidia MB, which requires further investigation.

Authors’ contributions

AS, MJA, RA, GST, NP, JC, MCA, LAM, and CJA wrote the main study protocol and designed it. AS, MJA, AD, and CJA performed the statistical and spatial analysis. MJA, GST and LAM and conducted the molecular analyzes. BY, GST, SZH, DMZ, AKK, and CJA performed the field work. AS, RA, NP, JC, GGP, MCA, and LAM supervised the study data collection. NP, JC MCA, and GGP provided administrative support. AS, TS, MJA, AD, and CJA wrote the original draft of the manuscript that was edited by BY, UT, RCM, TS, RA, CZK, SZH, DMZ, AKK, HS, VG, RO, AAO, NP, JC, GGP, MCA, and LAM. All authors read and approved the final manuscript.

Funding

This study, which is part of a larger project, “The New Nets Project”, was financially supported by a grant to the London School of Hygiene and Tropical Medicine from UNITAID and the Global Fund via the Innovative Vector Control Consortium (IVCC).

Data availability

The datasets generated and/or analysed during the present study are available on reasonable request from the corresponding author.

Declarations

Ethics approval and consent to participate

The protocol of the present study was reviewed and approved by the Benin’s Comité National d’Éthique pour la Recherche en Santé (N°30/MS/DC/SGM/DRFMT/CNERS/SA, Approval n°6 of 04/03/2019), and the Ethics Committee of the London School of Hygiene and Tropical Medicine (16237-1). Prior to their involvement in the study, all community members and leaders provided their informed consent. At the onset of the study, collectors were trained on how to catch mosquitoes on their lower limbs while avoiding being bitten. They were also immunized against yellow fever and taken care of at the nearest health facility when they suffered from malaria or other illnesses that had similar symptoms. The study was conducted in compliance with the Helsinki Declaration, and all national health guidelines and regulations were enforced.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.