Downregulation of female doublesex expression by oral-mediated RNA interference reduces number and fitness of Anopheles gambiae adult females.

BACKGROUND: Mosquito-borne diseases affect millions worldwide, with malaria alone killing over 400 thousand people per year and affecting hundreds of millions. To date, the best strategy to prevent the disease remains insecticide-based mosquito control. However, insecticide resistance as well as economic and social factors reduce the effectiveness of the current methodologies. Alternative control technologies are in development, including genetic control such as the sterile insect technique (SIT). The SIT is a pivotal tool in integrated agricultural pest management and could be used to improve malaria vector control. To apply the SIT and most other newer technologies against disease transmitting mosquitoes, it is essential that releases are composed of males with minimal female contamination. The removal of females is an essential requirement because released females can themselves contribute towards nuisance biting and disease transmission. Thus, females need to be eliminated from the cohorts prior to release. Manual separation of Anopheles gambiae pupae or adult mosquitoes based on morphology is time consuming, is not feasible on a large scale and has limited the implementation of the SIT technique. The doublesex (dsx) gene is one of the effector switches of sex determination in the process of sex differentiation in insects. Both males and females have specific splicing variants that are expressed across the different life stages. Using RNA interference (RNAi) to reduce expression of the female specific (dsxF) variant of this gene has proven to have detrimental effects to the females in other mosquito species, such as Aedes aegypti. We tested oral RNAi on dsx (AgdsxF) in An. gambiae.
METHODS: We studied the expression pattern of the dsx gene in the An. gambiae G3 strain. We knocked down AgdsxF expression in larvae through oral delivery of double stranded RNA (dsRNA) produced by bacteria and observed its effects in adults.
RESULTS: Our results show that feeding of AgdsxF dsRNA can effectively reduce (> 66%) the mRNA of female dsx transcript and that there is a concomitant reduction in the number of female larvae that achieve adulthood. Control groups produced 52% (± 3.9% SE) of adult males and 48% (± 4.0% SE) females, while AgdsxF dsRNA treated groups had 72.1% (± 4.0% SE) males vs 27.8% females (± 3.3% SE). In addition, the female adults produce fewer progeny, 37.1% (± 8.2% SE) less than the controls. The knockdown was sex-specific and had no impact on total numbers of viable male adults, in the male dsx transcripts or male fitness parameters such as longevity or body size.
CONCLUSIONS: These findings indicate that RNAi could be used to improve novel mosquito control strategies that require efficient sex separation and male-only release of An. gambiae by targeting sex determination genes such as AgdsxF. The advantages of using RNAi in a controlled setting for mosquito rearing are numerous, as the dose and time of exposure are controlled, and the possibility of off-target effects and the waste of female production would be significantly reduced.

Background

Despite recent progress in malaria control, the disease remains a global threat. Nearly half of the world’s population is at risk and the disease kills more than 400 thousand people a year [1]. Anopheles gambiae (s.l.) is the main malaria vector in sub-Saharan Africa, the region with the highest mortality rates, due to the combination of this mosquito’s high anthropophily and its susceptibility to Plasmodium falciparum infection [2-4]. Only female mosquitoes transmit the disease, as they ingest the blood meal that is necessary for the completion of their reproductive cycle [5]. Thus, female sex dimorphisms in An. gambiae include sex-specific morphological and physiological adaptations for host-seeking, blood-feeding and digestion [6].

Mosquito control to prevent malaria transmission is currently being undermined by the rapid spread of resistance to common insecticide classes [7-9]. Alternative tools and innovative strategies are needed to improve integrated management programmes [10], which rely on complementary strategies to minimize resistance. The sterile insect technique (SIT) is a system based on the mass rearing and release of adult males sterilized using gamma or X-rays to compete with wild males for matings with wild females [11-13]. Because An. gambiae females generally are monogamous, after mating with a sterile male, most females would only produce non-viable eggs. As a result, the mosquito population would be significantly reduced after a few generations if sufficient mating with sterile males occurred [14]. The SIT is, however, a self-limiting strategy that depends on the continuous and periodic release of sterile male mosquitoes. The technical difficulties in separating males from females in the mass production required for SIT, as well as the fitness costs of sterilization to males, have limited the implementation of this technique [12-15]. Importantly, the contamination with females should be avoided, because, even if they are sterile, they still may be able to transmit the disease. Furthermore, sex separation techniques are also needed for other methods currently under development, such as the release of Wolbachia infected incompatible males as a means of population reduction or the release of genetically modified mosquitoes [16].

The molecular mechanisms responsible for sexual dimorphism in insects include sex-specific gene splicing [17]. The doublesex/mab-3 related (Dmrt) family of transcription factors is involved in sex-specific differentiation, working as tissue-specific developmental regulators that integrate information about sex, position and time to guide specific cell types toward male or female differentiation [18-20]. The doublesex (dsx) gene appears to be conserved as a switch at the bottom of the cascade, and in many insects the sex-specific splicing of the gene is directed by the Transformer protein (TRA) with Transformer-2 (TRA2), which form a TRA/TRA2 complex [21]. The downstream targets of insect dsx are not well elucidated, but at least 50 optimal binding sites and genes have been identified for Drosophila melanogaster doublesex (Dmdsx) [22, 23].

In An. gambiae, the sex-specific morphological, physiological and behavioral traits are determined by the differential expression (sex-bias) of genes that are present in both males and females [6]. There is evidence that this differential expression of genes could be regulated by the An. gambiae dsx gene (Agdsx), with specific splicing for each sex [24, 25], regulating the genes involved in the adaptations to facilitate hematophagy, reproduction, olfaction and immune responses to pathogen challenge [21, 26]. In other mosquito species, such as Aedes spp. and Culex spp., female-specific dsx silencing has had detrimental effects on female development and more recently CRISPR-Cas9 knock-out of the same gene in An. gambiae has been shown to masculinize or completely sterilize the females [27-30].

RNA interference (RNAi) is a natural eukaryotic mechanism to silence genes in a post-transcriptional step, without permanent modifications to the genome. The use of exogenous dsRNA allows targeting a specific gene to suppress its expression by triggering the degradation of its complementary mRNA in the cell [31-34]. Previous work by Whyard et al. [29] has successfully used RNAi in larvae to silence genes involved in sex differentiation in Ae. aegypti, including dsx. In An. gambiae, Zhang et al. [35] have used chitosan nanoparticles to deliver dsRNA for chitin synthase genes, but to our knowledge there is no report of dsx silencing by RNAi in this species. To study the viability of a knockdown of dsx in An. gambiae using a system of RNAi by oral delivery, we fed young larvae with dsRNA for AgdsxF, to silence the AgdsxF and reduce the total number of female individuals in the laboratory setting. This approach can be complemented and modified to eliminate females from the male production without the need for manual sex sorting. Targets such as AgdsxF are especially appealing for this technique due to the sequence specificity that the sex-specific splicing provides. This could make the methodology species- and target-specific and would minimize the possibility of off-target effects.

Methods

Animal rearing and sexing

Laboratory-reared mosquitoes, An. gambiae G3 strain [MRA-112, BioDefense and Emerging Infections (BEI) Malaria Research and Reference Reagent Resource Center (MR4) vector activity, Centers for Disease Control and Prevention (CDC), Atlanta, Georgia, USA] were used for all experiments. Insectary conditions were 27 °C and a relative humidity of 80% with a photoperiod of 12:12 light: dark photocycle with a 30-min dawn and dusk period. Adults were fed ad libitum on 10% sugar solution with 0.2% methylparaben dissolved in sterile water. Larvae were fed on Damiens diet [tuna meal, liver powder, and Vanderzant vitamin mix (BioServ, NJ, USA)] in a 2:2:1 ratio; 2% w/v slurry using 10 g with 500 ml water) [36] unless specified otherwise.

Laboratory stocks of other mosquito species used were obtained from the Malaria Research and Reference Reagent Resource Center (MR4). Sex was determined in the pupal stage by observation of the pupal terminalia.

Doublesex sequence selection

GeneBank KM978938 sequence was used to design primers specific for the female mRNA splice form. After confirmation of sequence specificity and expression verification (Additional file 1), primers dsx1586 and dsx1589 were selected for the long dsRNA construct (Table 1). PCR amplification was performed using the AccuStart II PCR Supermix (Quantabio, MA, USA) and 10 µM primers, using the Tm’s specific for them. PCR products were visualized in 1.5% agarose gels stained with ethidium bromide and the amplicon was purified for cycle sequencing using a MultiScreen®HTS Vacuum Manifold (Millipore, MA, USA). DNA cycle sequencing reactions were performed for both strands with BigDye 3.1 sequencing kit (Thermo Fisher Scientific, Waltham, MA, USA), following manufacturer standard protocols, and sequenced in an ABI3130 (Applied Biosystems, Foster City, CA, USA) automated capillary sequencer. Sequences were manually assembled and aligned with SeqmanPro (DNASTAR Inc, Madison, WI, USA) and BioEdit 7.2.0. [37] After identity confirmation, T7 tailed primers were used to amplify the product with a T7 tail (either forward or reverse) to clone it into a pGEM-T-easy vector (Promega, WI, USA) according to the manufacturer’s instructions.

Table 1

Sequences of PCR primers

Gene	Vector base ID	Primer name	Primer sequence 5′–3′	Efficiency (%)
Rps7	AGAP010592	S7qf1	AGAACCAGCAGACCACCATC	115.02
		S7qr1	GCTGCAAACTTCGGCTATTC
Actin5C	AGAP000651	ACT-2f	TACAACTCGATCATGAAGTGCGA	110.67
		ACT-3r	CCCGGGTACATGGTGGTACCGCCGGA
Elongation Factor	AGAP005128	EFf1	GGCAAGAGGCATAACGATCAATGCG	112.60
		EFr1	GTCCATCTGCGACGCTCCGG
Female doublesex	AGAP004050	newDSX-f	AGAGGGCGGGGAAATTCTAGT	111.19
		newDSX-r	GGGCTTGTGGCAGTACGAATA
Female doublesex	AGAP004050	dsRNA_dsx-f2 (dsx1586)	CAAGCGGTGGTCAACGAATA	na
		dsRNA_dsx-r3 (dsx1589)	GCCCACTCCTTAAACACTACTT
Female doublesex	AGAP004050	T7dsRNA_dsx-f2	TAATACGACTCACTATAGGGCAAGCGGTGGTCAACGAATA	na
		T7dsRNA_dsx-r3	TAATACGACTCACTATAGGGGCCCACTCCTTAAACACTACTT

RNAi experiments

Due to the sex-specific nature of the dsx exons, we selected a female-specific 260 bp fragment within the dsxF mRNA sequence and added a T7 sequence to the 3′ end and ligated it to a pGEM-T-easy plasmid to obtain a product flanked by two T7’s for dsRNA production. Escherichia coli HT115 (DE3) was transformed with T7 plasmids as described by Timmons [31] to express the selected dsRNA for the dsx gene. An E. coli HT115 (DE3) previously transformed [38] to produce an unrelated dsRNA for the Aintegumenta gene from Arabidopsis thaliana was used as control in all the feeding experiments. For each feeding, an overnight culture of one colony was diluted 1:1000 in 2× YT media (100 µg/ml ampicillin and 12.5 µg/ml tetracycline) and induced after 2 h with 40 µM isopropyl β-d-1-thiogalactopyranoside (IPTG). After a total of 4 h of growth under induction conditions, cells were pelleted by centrifugation at 4000× g for 10 min, then washed in one volume of sodium phosphate buffer (PBS) and later re-suspended in 1/100 of the initial volume. A 1 ml aliquot of each induced culture was separated to measure dsRNA concentration. Heat inactivation was performed by heating at 70 °C for 1 h and 200 µl of this suspension was mixed with our ABD (Artificial Bacterial Diet; composition: 40% fish food (Goldfish, Tetra, Germany), 43% guar gum (Sigma-Aldrich, St Luis, MO, USA) and 17% active yeast). Larvae were fed the ABD diet (after L2 stage) for 4 h a day, based on previously published protocols [39]. After this time, the food was removed and cereal (Cerelac 4 Cereals, Nestle) was added as a source of food. Development and survival rates were recorded daily (Additional file 2). All larvae were reared in Petri dishes (25 × 100 mm). In experiments where female pupae and adults were to be counted (4 replicates) or where adult females were to be blood-fed and allowed to oviposit (3 replicates) 20 larvae in each dish for each experimental group were used in each experimental replicate. In each experimental replicate where morphological characters were to be recorded (3 replicates) 30 larvae were used in each dish per experimental group.

To quantify the dsRNA production and to calculate the dsRNA delivered daily by feeding, cell lysis was performed by resuspending the pellet from the cell culture aliquot in 50 µl of 0.1% SDS and incubating at 100 °C for 2 min. To eliminate mRNA, 2 µg of RNase A and 64 µl of RNase A buffer (300 mM NaCl, 10 mM Tris, 5 mM EDTA) were added. After incubation at 37 °C for 5 min, 500 µl of Trizol® (Thermo Fisher Scientific) were added and a protocol of extraction was performed according to the manufacturer’s instructions. The final dsRNA pellet was resuspended in 20 µl of water, incubated at 60 °C for 2 min. One µl was used for spectrophotometric measurement using a Nanodrop (Thermo Fisher Scientific), and 5 µl were used for quantifying on gels using 1.5% agarose.

Gene expression analysis

RNA was extracted from whole body or dissected tissues using RNeasy Mini Kit (Qiagen, Germantown, MD, USA) according to the manufacturer’s protocol. The complementary DNA was synthesized using the High-Capacity cDNA Reverse transcription kit (Applied Biosystems). The qPCR was performed in a QuantStudio6 Real Time PCR System (Applied Biosystems) using the Power SYBR-green PCR master MIX (Applied Biosystems). The Comparative Ct method [40, 41] was used to compare the changes in the gene expression levels. The An. gambiae ribosomal S7 (GenBank: L20837.1) [42], Actin (VectorBase: AGAP000651) and Elongation factor (VectorBase: AGAP005128) [43] genes were used as endogenous controls. The oligonucleotide sequences used in the qPCR assays are presented in Table 1. Three independent biological replicates were conducted, and all PCR reactions were performed in triplicate using extracts of pools of five whole pupae.

Analysis of morphological characters

Wing lengths, used as indicators of body size, were assessed in adult mosquitoes fed with dsRNA as larvae [44]. The wings were removed from 10-day-old individuals and analyzed using an EVOS fluorescence microscope from AMG (Thermo Fisher Scientific). Lengths were measured using Image J software [45]. Four biological replicates with internal duplicates were combined for statistical analysis.

Blood-feeding, reproductive and survival assays

Mated An. gambiae adult females, 5–10 day-old, were used for fecundity assays. For cross-mating assays of dsxF dsRNA reared larvae with untreated individuals, groups of 20 virgin females were maintained in small cages with 20 adult males for 5 days to allow mating. Blood-feeding was performed through Parafilm membranes using sheep blood in water-jacketed artificial feeders maintained at 37 °C. The insects were starved for 4 h prior to the feeding. Unfed mosquitoes were removed from the cages in all the experiments. Females were allowed to oviposit 24 h after blood-feeding. For assays of individual oviposition, each female was placed in a 25 ml tube with 0.5 mm of water and filter paper to collect the eggs. Eggs were counted and then left in water to evaluate hatching and viability rates. After allowing the blood-fed females to oviposite for 72 h, 60 females were dissected in 0.1 M PBS, pH 7.2. Presence or absence of sperm in spermathecae, used as mating indicator, were observed under a stereomicroscope at 400× and 1000× magnifications.

Statistical analyses

Unpaired Student’s t-tests were applied where comparisons were made between two treatments or conditions, as indicated in the figure legends. Long rank (Mantel-Cox) test was made to analyze survival curves. Two-way ANOVA with Tukey’s multiple comparisons post-hoc test was used for analysis of more than two conditions. All statistical analyses were performed using GraphPad 7 Prism Software (Graphpad Software, CA, USA).

Results

Dsx sequence identification and expression levels

RNAi strategies are based on the delivery of dsRNA complementary to the target gene. We limited our search to previously identified sequences for the dsx gene in D. melanogaster and other insect species, including An. gambiae [18, 24, 25, 27, 28, 46, 47]. A GenBank search (NCBI) for the An. gambiae nucleotide sequence of the male and female spliced Agdsx variants returned two different results. The original annotated Agdsx female specific sequence [24], DQ137802, differs significantly from the one annotated in 2015 (KM978938) by Liu et al. [25]. Liu and collaborators suggested that the former sequence may belong to An. stephensi, as genome searches with the former using BLAST matched more closely to this species. To select the sequence for dsRNA production and for qRT-PCR assays, we designed primers (Table 1) on the conserved region at the fifth exon and used them to amplify cDNA obtained from An. gambiae G3 (MRA-121, BEI Resources). The resulting sequence is presented in Fig. 1a and was shown to be a match for the KM978938 AgdsxF sequence. We used the same primers to amplify cDNA from An. stephensi STE2 (MRA-128, BEI Resources), and the sequence was identical to DQ137802. Finally, to determine the nature and variability of the selected region, we tested the same primers with Anopheles albimanus STECLA (MRA-126) and Anopheles minimus (MRA-729) cDNAs. Due to the phylogenetic distance between these species we were not able to amplify this region from An. albimanus but we obtained a 255 bp product from An. minimus. The Multiple Alignment using Fast Fourier Transform (MAFFT) of the amplified fragments from the three species differed from one other (Fig. 1b).

Fig. 1

Doublesex sequence and profile expression in An. gambiae G3. a Sequence alignment between the two previously annotated sequences for the dsx gene in An. gambiae with the sequence of PCR products amplified from the MR4 An. gambiae G3 and An. stephensi STE2. The An. gambiae G3 sequence matched the KM978939 sequence and the An. stephensi STE2 sequence matched the DQ137802 sequence. Marked in blue are the sequences that match in at least four consecutive bases the KM978939 sequence for AgdsxF exon. Three other regions from our An. stephensi strain resulted in 100% match to DQ137802 (Additional file 3). Thus, all subsequent primer designs were based on the KM978939 sequence. b Sequence conservation among Anopheles for the AgdsxF region of interest (1823–2066) was evaluated. PCR products were amplified from An. gambiae, An. minimus, An. stephensi and An. albimanus cDNAs (Additional file 3) and the sequences analyzed. A MAFFT alignment is shown, indicating regions with significant similarity in green. c PCR amplification of a AgdsxF 260 bp fragment, in the fifth exon, showed that the expression was limited to females (using cDNAs of whole body extract) of each sex, consisting of three 1-day-old pupae or five 5-day-old adults. d Real-time PCR (RT-PCR) analyses of relative gene expression reveal low levels of AgdsxF expression between the L1 and L3 stages, and significantly higher from L4 forward. Two-way ANOVA with Tukey’s multiple comparisons test was performed. e Gene expression analysis for AgdsxF by qPCR of different tissues. In adult female mosquitoes less than 24 h post-emergence (virgin females), dsxF expression was detected only in ovarian tissue (Ov) and salivary glands (SG). In mature adult females (more than 5-days-old and mated) detectable levels of AgdsxF expression were observed in the midgut (Mg) and ovaries (Ov), and especially high levels in the salivary glands (SG). Black points indicate the samples that had no detectable expression. AgdsxF expression was not detected in either time-point in the head, including antennas and proboscis (Hd) or fat body (FB). Pools of tissue from 12–15 individuals were analyzed

The region selected was confirmed to be female-specific (Fig. 1c) and to have the expected expression profile with detectable expression across the different life stages (Fig. 1d), with significantly higher levels of the transcript in adult females when compared to pupae (used as basal condition), and other life-stages (F(5,10) = 4.966, P =  0.0152). Finally, to determine if the AgdsxF is tissue and age-specific in An. gambiae G3 adult females, we measured its expression in different individual tissues (salivary glands, midgut, ovaries, fat body, and heads including proboscis and antennae). From mosquitoes at less than 24 h post-emergence and 5 days post-emergence (Fig. 1e) we found that AgdsxF expression was tissue-specific and depended on mosquito age, with the most abundant transcript found in the salivary gland of 5-day-old females (F(4,10) = 1263, P < 0.0001).

Dsx expression profile and silencing through RNAi in larval stages

Because the AgdsxF gene is expressed across the different life stages in An. gambiae, including larvae, we used oral delivery of AgdsxF dsRNA, produced in an E. coli HT115 (DE3), a bacterial strain modified for dsRNA production. We fed the larvae with the Artificial Bacterial Diet (ABD), with approximately ≤ 2.3 µg of dsRNA per day, from L2 to L4 stages. A 66% (± 9.9% SE) gene knockdown was confirmed in the pupal stage (Fig. 2a) using the ∆∆Ct method (AgdsxF dsRNA M = 0.3417, SD = 0.1969 and Ctrl dsRNA M = 1.001, SD = 0.05223; t(5) = 6.592, P = 0.0012). As RNA-dependent RNA polymerase mediated transitive amplification is absent in An. gambiae [48], we did not expect a long-lasting effect and the phenotypes observed can be considered a result of the RNAi from the larval stages.

Fig. 2

Feeding larvae dsxF dsRNA resulted in mRNA inhibition and lower adult female counts with reduced lifespan and fertility. Daily feedings of AgdsxF dsRNA to larval stages (from L2 to L4) resulted in a reduction of 66% (± 9.98 SE) of the AgdsxF transcript in pupae (a). Statistical difference was calculated by unpaired Student’s t-test with P < 0.05. b The silencing of the AgdsxF gene in the larval stages resulted in a reduction of the total number of female pupae (47% ± 3.0% SE of female pupae in the control vs 36% ± 5.6 SE in the dsx dsRNA-treated group) and subsequently, of the (c) adult females (48% ± 4.0% SE of female adults in the control vs 27% ± 3.3% SE in the dsx dsRNA-treated group). Results are from four independent replicates. d Body size, measured by wing length, did not show significant changes in the resulting adult females. e Life span of the adult females was significantly reduced by over 5 days. For the males from the treated AgdsxF dsRNA groups, no significant variation in body size (f) or adult longevity were observed (g). The fecundity of the AgdsxF dsRNA females (h) was significantly lower when compared to that of the control group and that of control females mated with dsxF dsRNA males. i Sperm were present in the spermathecae of females that laid eggs and absent in those that did not. Of the control females, 2% ± 1.66% SE did not have sperm whereas 30% ± 4.7% SE of AgdsxF dsRNA females did not. Phenotypic assays were done in parallel with the qRT-PCR for the treated groups. Except when specified, results are a biological triplicates and in all experiments the control group was fed with a non-related dsRNA, for the Aintegumenta gene from Arabidopsis thaliana

Phenotype description of the dsxF silencing by larval feeding

Feeding larvae with AgdsxF dsRNA to induce gene silencing resulted in a distortion of the typical ~ 1:1 male to female pupae ratio (Fig. 2b). The percentage of the male and female adults, 52% (± 3.9% SE) and 48% (± 4.0% SE), respectively, in the control (n = 80) was significantly different in the AgdsxF dsRNA (t(10) = 3.303, P = 0.03243). Groups fed with AgdsxF dsRNA resulted in twice as many males than females reaching adulthood: 72.1% (± 4.0% SE) males vs 27.8% (± 3.3% SE) females (t(10) = 7.631, P = 0.000036) (Fig. 2c). From the mosquitoes that reached adulthood, we evaluated different parameters such as body size, lifespan and fecundity (Fig. 2d-i). The females resulting from AgdsxF dsRNA-treated larvae had shorter lifespans (Long-rank test, χ2 = 127.8, df = 1, P < 0.0001) and reduced fecundity [male Ctrl dsRNA mated with Ctrl dsRNA females vs. male AgdsxF dsRNA mated with Ctrl dsRNA females, t(58) = 1.203, P = 0.2340; male Ctrl dsRNA mated with Ctrl dsRNA females vs male AgdsxF dsRNA mated with AgdsxF dsRNA females, t(58) = 2.401, P = 0.0196; male Ctrl dsRNA mated with AgdsxF dsRNA females vs male AgdsxF dsRNA mated with AgdsxF dsRNA females, t(58) = 0.7006, P = 0.4863], 37.1% less than the controls, ± 8.2 SE. However, no difference in body-size was found (t(38) = 0.1198, P = 0.9053).

To confirm that the method was female-specific and did not affect males, the same parameters were recorded for the males (Long-rank test, χ2 = 1.061, df = 1, P = 0.3030 for survival curve and t(40) = 1.113, P = 0.2725 for wing size) (Fig. 2e-f). Our results not only confirm that the AgdsxF gene of An. gambiae can be downregulated through RNAi, but also that it is possible to reduce the total number of adult females and to impact their fitness, without affecting male fitness, using RNAi.

Discussion

The importance of the dsx gene has been extensively described in metazoans [20, 49], and many of the key genes involved in the sex determination pathway have been elucidated [21]. In insects, it was first described in D. melanogaster by Baker et al. [18] in 1988, and subsequently in many other insect species from the orders Lepidoptera, Coleoptera, Hymenoptera and Diptera [21, 26, 49–51]. Interestingly, until very recently only two references had described and studied the expression of dsx sequences for An. gambiae [24, 25]. Because of the significant differences among the annotated sequences, we analyzed a conserved region using four different Anopheles species (An. albimanus STECLA strain, An. gambiae G3 strain, An. minimus strain and An. stephensi STE2 strain (Fig. 1a, b). Our results showed that the sequence product of our cDNA from An. stephensi STE2 presented 100% identity to the DQ137802.1 sequence published by Scali et al. in 2005 [24], who had published it as an An. gambiae sequence. The sequence we obtained from our An. gambiae G3 strain, presented 100% identity to the KM978938.1 sequence annotated by Liu et al. in 2015 [25]. In that work, Liu and collaborators proposed that the DQ137802.1 sequence belonged to An. stephensi but did not corroborate the fact experimentally. It is important to note that there are several regions in this female-specific exon that are highly conserved among these diverse Anopheles species, but that the sequence variation becomes more significant when larger areas of the sequences are analyzed. Additionally, the An. stephensi sequence, as annotated by Scali et al., was successfully used to design a transgenic An. gambiae line dsx-GFP, where the male mosquitoes selectively expressed high levels of eGFP [6].

To discern the expression pattern of the AgdsxF gene in An. gambiae G3 strain (Fig. 1b-d), we evaluated male and female individuals, different life stages, and different tissues. As expected, tissues with female-specific functions, i.e. those involved in the blood-meal acquisition and digestion like the salivary glands and the midgut, and the ovarian tissue, presented detectable expression of AgdsxF. However, we were not able to detect the transcript either in the head (including the antenna and the mouthparts) or in the fat body. This suggests that AgdsxF expression in An. gambiae G3 strain is not constitutive, and upstream regulators similar to TRA2 may induce tissue- and time-specific expression according to metabolic and reproductive needs. It is also possible however, that these tissues express the gene at very short and specific time-points, different from the ones that were selected for analysis.

In addition to sequence analysis and sex-specific splicing determination, functional studies of the gene in other mosquito species, such as Aedes aegypti, have been done in adult mosquitoes through injection of small interfering RNAs or using soaking and feeding of dsRNA in larvae [28, 33]. As a target for RNAi, the AgdsxF gene presents specific challenges for the design of long dsRNA, due to the GC content and the common regions with the male-specific spliced sequence. However, because systemic RNAi in An. gambiae has been reported to be successful using long dsRNAs [35, 52–55], we cloned a 260 bp fragment of the female-specific mRNA and selected it for the RNAi template. The use of long dsRNAs may be more stable during the heat-killing procedure of the bacteria. However, the use of short hairpin constructs (shRNA) for dsRNA production, as has been done for other mosquitoes like Ae. aegypti [28], could have different results and should be investigated.

There are interesting options of encapsulation to improve delivery of the AgdsxF dsRNA, such as chitosan nanoparticles [45] that could potentially improve silencing efficiency. More recently, Kyrou et al. [30] worked with AgdsxF to develop a CRISPR-Cas9 line where females were targeted, and performed an analysis of the dsx sequence in the female-specific region (exon 5) among members of the An. gambiae complex. The conservation of the initial portion of the exon was used as an indicator of the importance of the region on the function of the protein, which allowed for a successful design of a CRISPR-Cas9 gene drive to interfere with female development. The region selected for that construct overlaps with the region selected for RNAi in our work, reinforcing the selection of this sequence as a target for female elimination.

As evidenced by the CRISPR-Cas9 experiments performed in An. gambiae [30], the AgdsxF gene is haplosufficient. This means that in heterozygous individuals where the AgdsxF is interrupted on one chromosome, the females retain normal fecundity and physiology. The implication of this information for RNAi experiments is significant, suggesting that RNAi silencing of less than 50% would not have noticeable phenotypes. Furthermore, the RNAi efficiency depends on the combination of dsRNA size and concentration, point of entry to the organism, and tissue uptake [32]. If tissues with little or no expression of AgdsxF take and process the dsRNA available, it is possible that insufficient dsRNA reaches target tissues where more AgdsxF mRNA is present. In our experiments, feeding larvae of An. gambiae G3 with heat-killed bacteria with AgdsxF dsRNA induced a 66% reduction of the AgdsxF transcript, resulting in a male-biased sex ratio and reduced lifespan and fecundity in females. It is important to note, that the silencing from each tissue could be different, and the results were obtained from analyzing the whole body of pupae.

The RNA-dependent RNA polymerase mediated transitive amplification that prolongs and spreads RNAi in some insects is absent in An. gambiae [48]. Therefore, we expected a short residual effect and we propose that the effects observed are probably due to the RNAi events happening during the larval and pupal stages. Injection of dsRNA into adults, the traditional delivery method of dsRNA to study mosquito biology [52-55], may have different effects but was not applied for this study since the results would not be comparable to the ones obtained by oral delivery in a larval stage. In previous studies with Rhodnius prolixus, vector of Chagas disease, we have observed that the RNAi effect for the same gene results in different phenotypes depending on the delivery method and the half-life of the protein that was silenced [38, 56]. Oral delivery of dsRNA using chitosan nanoparticles has been reported in An. gambiae. Bacteria producing dsRNA has been used for RNAi in other mosquitoes and insects [29, 32, 33, 35, 38, 39]. Building on previous experience with other insects, we adapted the bacterial delivery of dsRNA as a feasible RNAi methodology in mosquito larvae. Therefore, these methodologies can be adapted to aid in many strategies of vector control for different diseases.

Oral delivery of AgdsxF dsRNA to the An. gambiae larvae resulted in a reduction of female individuals reaching adulthood. How this reduction in AgdsxF transcript interferes with development is poorly understood. In Ae. aegypti, dsxF upregulates genes involved in the cell cycle in females [57], and in Cyclommatus metallifer the dsx gene has been linked to the juvenile hormone signaling pathway [58]. It is possible that in An. gambiae similar relationships exist, and during the silencing of the dsx gene, these pathways or others are being affected. Further work to characterize the cellular processes affected by the gene in An. gambiae and other mosquitoes could be of use to design better targets for female elimination.

One of the main findings of this work is the total reduction of the adult females in the treatment group fed with AgdsxF dsRNA. This approximately 50% reduction in females is significant, but is unacceptable in an effective SIT program. Risk perception and public assurance that the releases are safe, requires that the female elimination procedures remove females to < 1.0%. The lifespan of females produced in the AgdsxF dsRNA treatments is also significantly compromised which could mean that if there were female contaminants, they would likely die sooner in the field. However, the life-span reduction (of ~ 7 days) from the remaining viable females is not enough to indicate that it would necessarily have an impact on parasite transmission.

About 30% of the females that fed on AgdsxF dsRNA did not lay any eggs, and after examination of the spermathecae of these individuals we found that the capsule was empty (t(4) = 6.364, P = 0.003126). The lack of sperm suggested an unsuccessful mating. In D. melanogaster, silencing dsx expression in neurons makes females unreceptive to courtship or copula [59], and therefore a similar effect could have occurred in this case. More information about the sex differentiation pathway in An. gambiae is still needed to understand mating behavior.

Coupling the effect of this specific construct with short hairpin RNAs will be investigated to determine if a stronger effect on phenotypes can be achieved. In summary, feeding larvae with an appropriate target sequence of AgdsxF dsRNA is a promising method for removing females from cohorts of mass reared An. gambiae sterile males for release. This technology is especially appealing for regions where the continuous exposure to insecticides has made the vectors resistant to traditional control methods. Future studies will focus on improving silencing and dsRNA delivery to facilitate the production of male mosquitoes for release in SIT, gene drive, or other alternative vector control methods.

Conclusions

Knockdown of AgdsxF in An. gambiae females after ingestion of AgdsxF dsRNA was observed. The total number of adult females was reduced, and their lifespan and fecundity were also negatively affected. Nevertheless, the life-span reduction was not enough to suggest a significant impact on parasite transmission. Fertility was also reduced, but sterility was not achieved. More information about the sex differentiation pathway is still needed, as some parts remain poorly understood. Coupling the effect of this specific construct with different gene targets could potentially result on a stronger effect. This technology for female elimination is particularly useful in new methods in which only males are released. Such methods are promising where the continuous exposure to insecticides has selected for resistance to traditional control methods. Future studies will focus on improving silencing and dsRNA delivery to facilitate the production of male mosquitoes for release in SIT, gene drive, or other alternative vector control methods.

Additional file 1. GenBank and VectorBase sequences for dsx female splice isoform.

Additional file 2. Pupation of male and female groups of An. gambiae fed with control dsRNA or F-dsx dsRNA.

Additional file 3. Comparison of Doublesex sequence analysis in four Anopheles species (An. stephensi STE2, An. gambiae G3, An. minimus strain and An. albimanus STECLA).