Disruption of mosGILT in Anopheles gambiae impairs ovarian development and Plasmodium infection

Malaria remains one of the greatest threats to human health, accounting for over 200 million cases and approximately half a million deaths each year (World Health Organization, 2017). Plasmodium parasites, the causative agents of malaria, require the Anopheles mosquito for their sporogonic development and transmission to vertebrates. The parasite life cycle in the mosquito begins when Plasmodium gametocytes differentiate into gametes in the midgut lumen. Fusion of male and female gametes produces zygotes, which develop into motile ookinetes within 12–20 h. After traversing the midgut epithelium, the ookinetes transform into oocysts at the basal lamina. Mature oocysts burst and release sporozoites into the hemocoel. The sporozoites that successfully invade the salivary glands (SG) are ready to be transmitted to a new host (Ghosh et al., 2000; Sinden, 2002).

Mosquitoes mount potent immune responses against Plasmodium development between the ookinete and oocyst stages, causing substantial losses in parasite numbers (Blandin and Levashina, 2004; Christophides et al., 2002; Dimopoulos et al., 1998, 2002; Sinden, 2002). Thioester-containing protein 1 (TEP1), a complement-like glycoprotein, is one of the key antiparasitic immune effectors in mosquitoes (Blandin et al., 2004; Levashina et al., 2001). The direct binding of TEP1 to the surface of ookinetes targets the parasites for destruction by lysis or melanization (Blandin et al., 2004). Silencing of TEP1 enhances Plasmodium berghei oocyst numbers and abolishes parasite melanization in both the susceptible and refractory strains of Anopheles gambiae (Blandin et al., 2004; Collins et al., 1986; Povelones et al., 2011). In addition, positive regulators of this complement-like immunity have been identified, including two members of the leucine-rich repeat immune protein (LRIM) family, LRIM1 (Fraiture et al., 2009; Osta et al., 2004; Povelones et al., 2009) and APL1C (Fraiture et al., 2009; Povelones et al., 2009; Riehle et al., 2006, 2008), and the CLIP-domain serine protease homologue SPCLIP1 (Dong et al., 2006; Povelones et al., 2013). These mosquito immune factors constitute defense mechanisms that restrict or eliminate Plasmodium development.

There are also a variety of mosquito factors that repress the immune responses, thereby promoting the development and subsequent transmission of Plasmodium (Blandin et al., 2008; Simões et al., 2018). A. gambiae (Ag) C-type lectin (CTL) 4 and CTL mannose binding 2 were shown to protect P. berghei and Plasmodium falciparum ookinetes from melanization in susceptible mosquitoes (Osta et al., 2004; Simões et al., 2017). Mosquito serpin SRPN2 is another Plasmodium agonist in A. gambiae, which facilitates ookinete invasion of the midgut by inhibiting lysis and melanization (Michel et al., 2005). Two nutrient transport proteins, lipophorin (Lp) and vitellogenin (Vg), reduce the parasite-killing efficiency of TEP1 (Rono et al., 2010). However, most studies have used the rodent parasite P. berghei as a model. Specific mosquito factors that promote the development of human malaria parasites remain to be identified.

Previously, we identified a protein, mosquito IFN-γ–inducible lysosomal thiol reductase (mosGILT), that binds to sporozoites in the saliva of infected Anopheles mosquitoes (Schleicher et al., 2018). We showed that mosGILT can partially reduce the speed and cell traversal activity of both human and rodent Plasmodium sporozoites. The inhibition of these motility components impairs the ability of the sporozoites to migrate to the liver and establish a normal hepatic infection. To determine whether mosGILT influences Plasmodium infection in mosquitoes, we generated mosaic mosGILT-mutant mosquitoes using CRISPR/CRISPR–associated protein 9 (Cas9), a powerful technique that enables the study of mosquito gene function through efficient genome editing (Dong et al., 2018). Here we show that A. gambiae mosquitoes carrying mosaic mutations in the mosGILT gene show defects in ovarian development. These mutant mosquitoes have markedly lower numbers of oocysts, following infection by either rodent or human malaria parasites (P. berghei or P. falciparum). This leads to a dramatic decrease in sporozoite load and infection of the SGs. The refractory phenotype of mosGILT mutant mosquitoes is strongly associated with their reduced expression of Vg, which leads to a more efficient TEP1-mediated killing of ookinetes. Therefore, mosGILT is a mosquito factor that plays an important role in ovarian development and indirectly protects both human and rodent Plasmodium species from mosquito immunity.

To generate mosGILT-deficient mosquitoes, we first created a transgenic line expressing three guide RNAs (gRNAs) targeting the mosGILT gene (Fig. 1 a) based on an established CRISPR/Cas9 system in A. gambiae (Dong et al., 2018). Heterozygous females of the gRNA-expressing line were crossed with homozygous males of the vasa-Cas9 strain (Gantz et al., 2015; Hammond et al., 2016). Progeny from this cross were separated by fluorescent markers into two groups: the gRNA-RFP⁺/Cas9-GFP⁺ transheterozygotes and the gRNA-RFP⁻/Cas9-GFP⁺ siblings (referred to as siblings; Fig. S1 a). In 100% of the gRNA/Cas9 transheterozygotes (n 40), Cas9 induced polymorphic mutations at the gRNA1- and gRNA3-target sites, but not the gRNA2-target site (Fig. 1 b). Our previous study showed that mosGILT is an abundant protein in A. gambiae SGs (Schleicher et al., 2018). Endogenous SG mosGILT in female gRNA/Cas9 transheterozygotes was greatly reduced compared with that of siblings (Fig. 1 c). These data suggest that CRISPR/Cas9-mediated disruption of mosGILT is highly efficient in gRNA/Cas9 transheterozygotes (referred to as mosGILT mutants), which is consistent with a previous study showing that the vasa promoter-driven Cas9 is ubiquitously expressed in both the germ line and somatic cells (Port et al., 2014).

Next, we sought to obtain homozygous mosGILT knockout mosquitoes by crossing the mosGILT mutants with the X1 line (Volohonsky et al., 2015). However, all progeny (51 out of 51) from mosGILT mutant males crossed with X1 females only carried WT mosGILT alleles, and none of the mosGILT mutant females (0 out of 48) were able to lay eggs after mating with X1 males (Fig. S1 a). These results indicate that the disruption of mosGILT interferes with both male and female mosquito reproductive biology. In addition, crossing of heterozygous mosGILT-gRNA₃–expressing males with homozygous vasa-Cas9 females produced fewer eggs compared with the reciprocal cross, and all the hatched progeny were gRNA-RFP⁻/Cas9-GFP⁺ (Fig. S1 b). We hypothesize that maternal deposition of Cas9 in the embryo allows an earlier disruption of mosGILT, causing embryonic lethality. Together, these data suggest that homozygous mosGILT knockout mosquitoes cannot be obtained as the mutations are not heritable. Therefore, we used the F₀ mosaic mosGILT mutants as a model for further study.

MosGILT mutants survived from larvae to adulthood with a normal lifespan, wing length, feeding propensity, and sex ratio (female:male, 1.15:1), with only a slightly delayed pupation time compared with siblings (Fig. S2 a). Their inability, however, to pass down the mutated alleles of mosGILT prompted us to examine the gonads. Morphological analysis showed that 100% of the female mutants (n 90) had underdeveloped ovaries that were completely devoid of follicles, whereas their SGs appeared normal (Fig. 2 a). Yolk accumulation and oocyte development were not observed in any of the abnormal ovaries after blood feeding, and their median size was significantly smaller than that of siblings (Fig. 2, b and c). These data suggest that mosGILT plays a critical role in ovarian development. In line with this, we found that there were detectable levels of mosGILT expression in the ovaries of WT females, in addition to the fat body and SGs (Fig. S2 b). Furthermore, we detected mosGILT expression in WT larvae and pupae (Fig. S2 c). Additionally, we found that 100% of the mosGILT mutant males (n 70) displayed developmental defects of the testes (Te), although their male accessory glands appeared to be normal (Fig. S2 d). These data indicate that mosGILT is expressed in different tissues and developmental stages, and therefore could have multiple in vivo functions. As only the female mosquitoes take blood meals and are responsible for malaria transmission, we further analyzed the female mosGILT mutants.

To assess the effect of mosGILT disruption on Plasmodium infection in the vector, we first fed the female mosGILT mutants on mice infected with the rodent malaria parasite P. berghei, using siblings as the control. MosGILT mutants displayed a significant reduction in the oocyst infection intensity at 8 d post-infection (dpi), compared with siblings (Fig. 3 a; siblings, n = 50, median = 81; and mosGILT mutants, n = 50, median = 0; 100% reduction in median; P 0.0001, Mann–Whitney test). Midgut infection prevalence (the percentage of infected midguts with at least one oocyst) in mosGILT mutants was 34.1%, a significant 2.3-fold decrease from 77.8% in siblings (Fig. 3 b; P 0.0001, Fisher’s exact test). Consistently, mosGILT mutants had a significantly lower number of SG sporozoites at 21 dpi compared with siblings (Fig. 3 c; siblings, n = 10, median = 16,263; and mosGILT mutants, n = 10, median = 1,009; 93.8% reduction in median; P = 0.0001, Mann–Whitney test). The SG infection prevalence (the percentage of infected SG) in mosGILT mutants was 12.6%, fivefold lower than 62.9% in siblings (Fig. 3 d; P 0.0001, Fisher’s exact test).

Next, we examined the permissiveness of mosGILT mutants to the human malaria parasite, P. falciparum. The oocyst infection intensity of mosGILT mutants was significantly lower than that of siblings (Fig. 3 e; siblings, n = 104, median = 9; and mosGILT mutants, n = 88, median = 4.5; 50% reduction in median; P 0.0001, Mann–Whitney test). In addition, the midgut infection prevalence of mosGILT mutants was 72.3%, compared with 93.2% in siblings (Fig. 3 f; P 0.0001, Fisher’s exact test). The number of SG sporozoites was greatly reduced in mosGILT mutants compared with siblings (Fig. 3 g; siblings, n = 63, median = 5,813; and mosGILT mutants, n = 63, median = 1,688; 71% reduction in median; P 0.0001, Mann–Whitney test). The SG infection prevalences in mosGILT mutants and siblings were 63.9% and 90.5%, respectively (Fig. 3 h; P = 0.0005, Fisher’s exact test). Together, these results demonstrate that mosGILT disruption compromises Plasmodium infection in A. gambiae mosquitoes.

TEP1 is one of the key Anopheles immune effectors involved in the elimination of invading ookinetes, through lysis and/or melanization (Blandin et al., 2004; Dong et al., 2006; Fraiture et al., 2009; Povelones et al., 2011, 2013; Volohonsky et al., 2017). Therefore, we examined whether the suppression of Plasmodium by mosGILT disruption requires TEP1. TEP1 mRNA levels were greatly reduced (85%) in siblings and mosGILT mutants injected with TEP1 double-stranded RNA (dsRNA) compared with control mosquitoes injected with luciferase (luc) dsRNA (Fig. 4 a). Strikingly, both siblings and mosGILT mutant mosquitoes showed a similar level of significantly increased numbers of P. berghei oocysts upon TEP1 silencing (Fig. 4 b). The percentages of infected midguts in both siblings and mosGILT mutants reached 100% when TEP1 was depleted (Fig. 4 c). These results suggest that the parasite loss in mosGILT mutants was dependent on the TEP1-mediated immune response.

In addition, our immunofluorescence assay showed a significant reduction of early oocysts in the midgut epithelium of mosGILT mutants compared with siblings at 48 h post-infection (hpi; Fig. 4, d and e). Very few ookinetes were observed in both groups, probably because most of them had been cleared by the TEP1-dependent immune response by that time point. Consistent with a previous study (Blandin et al., 2004), we also observed a weak and patchy binding pattern of TEP1 to some early oocysts, whereas the dying or dead ookinetes were heavily coated by TEP1 (Fig. 4 d). Furthermore, melanized ookinetes were hardly seen in the P. berghei–infected midguts 8 dpi from both siblings and mosGILT mutants, whereas silencing of CTL4, a negative regulator of melanization (Osta et al., 2004; Simões et al., 2017), triggered ookinete melanization in siblings (Fig. S3). Together, these data imply that TEP1-mediated lysis, but not melanization, is responsible for limiting parasite survival in mosGILT mutant mosquitoes.

To understand how mosGILT disruption influences the TEP1-dependent responses, we analyzed the mRNA and protein levels of TEP1 in mosGILT mutants and siblings. At 24 hpi, transcripts of TEP1 were up-regulated at similar levels in the fat body from both groups of mosquitoes (Fig. 5 a). The full-length form of TEP1 (TEP1-FL) is secreted into the hemolymph and processed into the mature C-terminal fragment (TEP1-C; Fraiture et al., 2009; Levashina et al., 2001; Povelones et al., 2009). Hemolymph analysis by immunoblotting did not reveal any obvious difference in the amounts of TEP1-FL or TEP1-C between mosGILT mutants and siblings (Fig. 5 b). These results suggest the expression and activation of TEP1 are not altered by mosGILT disruption.

Vg is a precursor protein of egg yolk that is mainly synthesized in the female fat body and provides nutrients for oogenesis in the ovaries, a process known as vitellogenesis (Attardo et al., 2005). Vg has also been shown to reduce the parasite-killing efficiency of TEP1 (Rono et al., 2010). Considering the abnormal ovaries and increased TEP1-dependent responses in mosGILT mutants, we investigated whether Vg expression was impaired in the absence of mosGILT. To assess this, we analyzed the mRNA and protein levels of Vg in mosGILT mutants and siblings. Transcripts of Vg were significantly up-regulated in the fat body of siblings after taking an infectious blood meal; however, Vg induction was severely impaired in mosGILT mutants (Fig. 5 a). Vg protein in the hemolymph can be readily visualized by Coomassie staining of SDS-PAGE gels (Rono et al., 2010). Using this approach, we further validated that the protein level of Vg was also greatly reduced in mosGILT mutants compared with siblings at 24 hpi (Fig. 5 b).

In mosquitoes, a blood meal triggers the ovaries to secrete ecdysone, an insect steroid hormone, which is converted to 20E that in turn activates Vg production in the fat body (Gabrieli et al., 2014; Hagedorn et al., 1975; Pondeville et al., 2008; Raikhel et al., 2005). Using an enzyme immunoassay (EIA) kit, we found that the amount of 20E induced by blood feeding was significantly lower in mosGILT mutants compared with siblings (Fig. 5 c). This result further indicates that the underdeveloped ovaries fail to function properly, causing the impaired production of 20E and the downstream synthesis of Vg.

In addition to the impaired 20E production, mosGILT mosaic mutations might cause additional defects that lead to the reduction of Vg and parasites. To address this possibility, exogenous 20E was supplied to female mutants. First, we found that Vg expression in blood-fed mosGILT mutants was rescued by 20E in a dose-dependent manner (Fig. S4), indicating that Vg synthesis in the fat body, a downstream event in the 20E signaling pathway, is not affected by mosGILT disruption. The amount of exogenous 20E required for optimal Vg induction in mosGILT mutants was 0.5–1.0 µg/female, similar to the doses used for artificial stimulation of vitellogenesis in sugar-fed Aedes aegypti mosquitoes (Lea, 1982). Next, we examined the effects of 20E complementation on P. berghei infection. The Vg expression, oocyst number, and midgut infection prevalence were all significantly enhanced in the 20E-injected mosGILT mutants compared with noninjected controls (Fig. 6). In addition, these parameters were not significantly different between the 20E-injected mosGILT mutants and the 20E-injected siblings (Fig. 6). Furthermore, we found that silencing of mosGILT in WT females, which have already developed normal ovaries, did not affect Vg expression or P. berghei infection (Fig. S5). Together, these data suggest that lack of 20E due to the underdeveloped ovaries is the main factor contributing to the Vg reduction and refractoriness to P. berghei in mosGILT mutants.

The trade-off between reproduction and immunity has been documented in diverse insect species (Schwenke et al., 2016). Understanding the molecular mechanisms linking reproduction and immunity in Anopheles mosquitoes is of significant importance for developing malaria control strategies. Two nutrient transport proteins in A. gambiae, Lp, and Vg, have been shown to interfere with anti-Plasmodium responses after taking an infectious blood meal (Mendes et al., 2008; Rono et al., 2010; Vlachou et al., 2005). Using CRISPR/Cas9-generated mosaic mosGILT mutant mosquitoes, we now uncover a critical role for mosGILT in mosquito ovary development and Plasmodium survival in the midgut. In this study, female mosGILT mutant mosquitoes displayed underdeveloped ovaries after eclosion and reproductive defects including severely impaired production of 20E and Vg. In addition, the mosGILT mutant mosquitoes mounted a notably stronger TEP1-mediated immune response against developing ookinetes in the midgut, resulting in a lower susceptibility to infection by either P. berghei or P. falciparum. Our results provide new evidence that mosquito proteins involved in reproduction can benefit malaria parasite infection in the vector by suppressing immunity.

We demonstrate the application of CRISPR/Cas9 in a mosaic analysis of mosGILT functions in A. gambiae. Genome editing in embryos using CRISPR/Cas9 often results in genetic mosaicism in founder animals (F₀), which can be used as models for phenotyping especially when a homozygous gene knockout is developmentally lethal (Mianné et al., 2017; Port et al., 2014; Song et al., 2018). A previous study of CRISPR tools in Drosophila has shown that vasa-Cas9 in combination with U6-gRNA allows efficient biallelic gene disruption, which could readily reveal null mutant phenotypes in the soma (Port et al., 2014). Consistent with these reports, we found that 100% of the F₀ mosGILT mutants generated by crossing vasa-Cas9 males with U6-mosGILT-gRNA₃ females were genetic mosaics. They carried diverse mosGILT mutations (Fig. 1 b). More importantly, mosGILT mutant females had a greatly reduced mosGILT protein level in the SGs (Fig. 1 c) and displayed abnormal ovaries (Fig. 2). In addition, mosGILT is probably also essential for embryogenesis, as the reciprocal cross using homozygous vasa-Cas9 females and heterozygous U6-gRNA males produced only WT mosquitoes (Fig. S1 b). We speculate that a higher maternal supply of Cas9 in the egg allows the mutagenesis of mosGILT to occur at an earlier time point, which results in embryonic lethality.

When female Anopheles mosquitoes acquire a blood meal infected with Plasmodium, they not only initiate egg development in the ovaries but also allow the parasite to infect the midgut. The nutrients in the blood trigger the ovaries to produce ecdysone, an insect steroid hormone, which is converted to 20E and initiates vitellogenesis in the fat body (Baldini et al., 2013; Hagedorn et al., 1975; Kokoza et al., 2001). Vg, one of the yolk protein precursors, is highly induced during vitellogenesis and essential for oogenesis (Attardo et al., 2005). The abnormal ovaries in mosGILT mutant mosquitoes were unable to develop follicles (Fig. 2) or produce 20E (Fig. 5 c) in response to a blood meal, resulting in the subsequent loss of Vg production (Fig. 5) and egg development (Fig. 2, b and c; and Fig. S1 a). In addition to the severe defects in ovary function, mosGILT mutant mosquitoes showed a strong refractoriness to malaria infection. P. berghei oocyst number and infection prevalence were significantly lower in mosGILT mutants compared with siblings (Fig. 3, a and b). Of the few surviving oocysts in the mosGILT mutants, we did not observe a striking difference in their size compared with that of siblings (Fig. S3), and we showed that TEP1 silencing strongly reversed the phenotype (Fig. 4, b and c). Therefore, we believe that the low permissiveness of mosGILT mutant mosquitoes to P. berghei was mostly the result of TEP1-mediated immunity. TEP1 expression and cleavage were not altered by mosGILT disruption (Fig. 5); however, we observed a significant loss of oocysts by 48 dpi, likely due to more efficient killing of parasites by TEP1 in mosGILT mutants (Fig. 4, d and e). A previous study has shown that Vg silencing decreased P. berghei survival by increasing the binding efficiency of TEP1 to ookinetes (Rono et al., 2010). Considering the low Vg production due to underdeveloped ovaries in mosGILT mutants, here we also observed a strong positive correlation between the Vg level and P. berghei oocyst number. More importantly, the refractory phenotype of mosGILT mutants extended to the human malaria species, P. falciparum (Fig. 3, e and f), which is consistent with a previous finding that TEP1 can limit P. falciparum infection in A. gambiae (Dong et al., 2006).

A recent publication demonstrated that steroid hormone signaling influences Plasmodium development in Anopheles (Werling et al., 2019). The authors generated transgenic A. gambiae females by inducing mutagenesis in the zero population growth (zpg) gene, which has been shown to be essential for germ cell development (Tazuke et al., 2002; Thailayil et al., 2011). Similar to the phenotypes that we found in mosGILT mutants, the Δzpg mutant females also showed underdeveloped ovaries that produced low amounts of 20E, and were significantly less permissive to P. falciparum infection. Both mutant mosquito models reveal a positive link between 20E signaling and intensity of Plasmodium infection. Consistent with this idea, complementation of exogenous 20E to mosGILT mutants rescued the infection by P. berghei (Fig. 6, b and c). Therefore, we propose that mosGILT benefits Plasmodium survival by influencing ovarian development and the subsequent production of 20E and Vg.

Previously, we reported a negative role of SG mosGILT in the transmission of Plasmodium sporozoites to the mammalian host (Schleicher et al., 2018). In contrast, here we describe a positive role of mosGILT in ovarian development and Plasmodium survival in the midgut. These different findings could be attributed to diverse functions of mosGILT at various developmental stages and distinct tissue localizations (Fig. S2). In addition, although we initially showed that mosGILT does not have thiol reductase activity in an in vitro assay at pH 4.0 compared with mammalian GILT proteins (Schleicher et al., 2018), we cannot completely exclude the possibility that mosGILT has some reducing activity in vivo. Any putative enzymatic activity of mosGILT and its role in ovarian development still require further investigation. It is possible that mosGILT directly or indirectly influences ovarian development by regulating the local redox conditions before the adult stage. Identification of the mosGILT interactome and the implicated signaling pathways would further our understanding of mosGILT in vivo functions. Additionally, it would be interesting to examine whether mosGILT homologues in other insect species play similar roles in ovarian development and pathogen transmission.

Vector reproduction intimately interacts with immune defense pathways, as these two processes are simultaneously initiated after a blood meal. Identification of vector proteins involved in the cross-talk between reproduction and immunity could provide new targets for the development of disease-control strategies. Our study demonstrates a critical role of mosGILT in the reproduction of A. gambiae mosquitoes. Particularly, disruption of mosGILT in females impairs the development and function of ovaries, leading to increased anti-Plasmodium immunity. These findings can potentially lead to new preventive strategies for malaria, aimed at inhibiting the activity of mosGILT to induce mosquito sterility and reduce the vectorial capacity of A. gambiae.

Mice were housed by the Yale Animal Resource Center at Yale University and handled under the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The animal experimental protocol was approved by the Institutional Animal Care and Use Committee of Yale University (protocol permit no. 2017–07941). Human blood for P. falciparum cultures and mosquito infections was collected from a pool of prescreened donors under an institutional review board–approved protocol at Johns Hopkins University (protocol NA00019050).

Putative gRNA sequences targeting the first and second exons of the mosGILT gene were identified using online tools such as ZiFit and CRISPR Design. gRNA sequences were required to be 20 nucleotides in length, start with a guanine to facilitate transcription by the U6 promoter, and be followed by a protospacer adjacent motif (PAM; NGG, where N is any nucleotide). Three gRNA sequences that had the fewest predicted genomic off-targets were selected (Table S1).

Plasmids expressing mosGILT-gRNAs were constructed as previously described (Dong et al., 2018). In brief, linear double-stranded DNA fragments encoding each gRNA were produced by annealing two partially overlapping oligos (primers mosGILT-gRNA1-F and -R for gRNA1, mosGILT-gRNA2-F and -R for gRNA2, mosGILT-gRNA3-F and -R for gRNA3; Table S1). To generate pKSB-gRNA modules (pKSB-mosGILT-gRNA-1, -2, -3), the three double-stranded DNA fragments were individually cloned into a pBluescript vector backbone that contained the U6 snRNA polymerase ΙΙΙ promoter (AGAP013557), CRISPR RNA invariable sequences, and the RNA polΙΙΙ TTTTT terminator. The pKSB-gRNA modules were used to assemble the three gRNA genes into the pDSAR vector via the Golden Gate assembly system (E1600S; NEB) for embryo microinjection into the A. gambiae docking line X1 (Volohonsky et al., 2015). The backbone plasmid pDSAR with 3xP3-RFP allowed screening of gRNA-positive larvae or mosquitoes by red fluorescence in the eyes.

Embryo microinjection and screening of positive gRNAs-expressing lines were performed as previously described (Dong et al., 2018). A QIAGEN Endofree Maxi Kit was used to prepare pDSAR-mosGILT-gRNA₃ and a helper plasmid containing the vasa2::ΦC31 integrase gene (pENTR-R4R3-vasa2-integrase; Papathanos et al., 2009; Volohonsky et al., 2015). The gRNA plasmid (160 ng/µl) and the helper plasmid (200 ng/µl) in a phosphate buffer (0.1 mM NaHPO₄ buffer, and 5 mM KCl, pH 6.8) were microinjected into 580 embryos of the docking strain X1 (Meredith et al., 2013; Volohonsky et al., 2015). Injected eggs were maintained on wet filter paper for 2 d before hatching. 15% of the injected eggs hatched, and 32% of the survived G₀ larvae showed RFP signals at the second instar larval stage. RFP-positive G₀ pupae were sexed (14 females and 12 males) and then organized into two cohorts. G₀ female mosquitoes were crossed with WT X1 male mosquitoes at a ratio of 1:1, and G₀ male mosquitoes were crossed with female X1 mosquitoes at a ratio of 1:5. The G₁ progeny were examined for RFP-glowing eyes at both the larval and adult stages. Positive G₁ mosquitoes were outcrossed with X1 for two generations followed by two generations of self-crossing to enrich the homozygous mosquitoes by the screening of RFP in the larvae before crossing with the vasa-Cas9 strain to generate the mosGILT mutants.

The G₄ gRNA-expressing mosquitoes were crossed with the homozygous Cas9 transgenic strain (vasa2-Cas9; with green fluorescence in the eyes; Hammond et al., 2016). 100 mosGILT-gRNA female mosquitoes were crossed with 100 male Cas9 mosquitoes to generate the mosGILT mutant mosquitoes. Since the parental gRNA mosquitoes were not fully homozygous, the progeny of this cross were separated into two groups by screening the RFP signal at the larvae stage: the group of larvae with both RFP and GFP signals was gRNA/Cas9 transgenic, and the other group of larvae carrying only the Cas9-GFP transgene serves as a negative control (referred to as siblings). Pupae or adult mosquitoes were randomly selected from the gRNA/Cas9 transgenic group for genomic DNA extraction using the DNEasy Blood and Tissue Kit (69581; QIAGEN). The gRNA-targeted genomic locus was amplified by PCR with flanking primers (2.5 kb-F and -R; Table S1). PCR products were separated with an agarose gel (1%). The WT 2.5 kb bands were purified by DNA gel extraction (QIAGEN) and cloned using pCR2 TOPO-cloning system (Thermo Fisher Scientific). Individual clones were sent for DNA sequencing to confirm any Cas9-mediated mutations at the gRNA-targeted region. Due to the ubiquitous expression of Cas9 driven by the vasa promoter (Port et al., 2014), all gRNA/Cas9 transgenic mosquitoes are mosaic mutants that contain mixed genotypes with different mutations at the mosGILT gene locus throughout their tissues. The SGs of mosGILT mutant mosaic mosquitoes were dissected for Western blot analysis to confirm the reduction of mosGILT expression compared with the sibling mosquitoes.

All the A. gambiae mosquito lines were derived from the G3 strain, including the docking line X1 (Volohonsky et al., 2015), vasa2-Cas9 (Hammond et al., 2016), mosGILT-gRNA₃, mosGILT mutant and sibling mosquitoes. They were raised at 27°C, 80% humidity, under a 12/12 h light/dark cycle and maintained with 10% sucrose. Swiss Webster mice (6–8-wk-old females) were purchased from Charles River Laboratories.

After mosGILT-gRNA female mosquitoes were crossed with Cas9 male mosquitoes, the offspring larvae were maintained according to a standard procedure with 400 larvae per tray (Pike et al., 2017). For pupation time measurement, pupae were collected and screened under the fluorescent microscope to record the number of mosGILT mutant pupae (GFP⁺/RFP⁺) and their siblings (GFP⁺/RFP⁻) on a daily basis.

To examine the morphology of ovaries, SGs, and Te, at least 30 female or male mosquitoes from the mosGILT mutant and sibling group were anesthetized on ice and dissected. Representative pictures of ovaries and SGs from unfed mosquitoes were captured using an EVOS FL Auto Cell Imaging System (Thermo Fisher Scientific). Representative pictures of ovaries from blood-fed mosquitoes were taken with a Zeiss Discovery V8 stereomicroscope using a Zeiss camera AxioCam MRc equipped with AxionVision 4.8 software.

To test the ability of male mutants to pass down the mutated mosGILT gene, 50 mosGILT mutant male mosquitoes were crossed with 50 female X1 mosquitoes. 51 offspring were used for genomic DNA extraction. The gRNA-targeted genomic locus was amplified by PCR with flanking primers (2.5 kb-F and -R; Table S1), and PCR products were sent for sequencing as mentioned above.

For the fecundity assay of female mosquitoes, 50 female mosquitoes from both the mutant and the sibling groups were mated with 50 male X1 mosquitoes respectively for 4 d and then fed on naive 6–8-wk-old female Swiss Webster mice (Charles River Laboratories). Only the fed mosquitoes were selected for further analysis. 3 d after feeding, 25 female mosquitoes from each group were individually placed into Drosophila vials (AS574; Thermo Fisher Scientific) containing a conical filter paper surrounding the plastic wall and 1 cm of deionized water to lay eggs. Two days later, dead mosquitoes were excluded, and the number of mosquitoes in each group that laid eggs was recorded.

P. berghei (ANKA GFPcon 259cl2, MRA-865; ATCC) was maintained as previously described (Schleicher et al., 2018). In brief, 6–8-wk-old female Swiss Webster mice (Charles River Laboratories) were infected with parasitized mouse blood by needle injection. 5 d later, mice with parasitemia levels between 1% and 10% were used to feed mosquitoes that were deprived of sucrose for 20 h (3–5 mice/cage of 200 mosquitoes). The blood-fed mosquito cages were kept hydrated with damp paper towels at 20°C. 2 d after feeding, bottles containing a cotton wick soaked in 10% sucrose supplemented with 0.5% penicillin/streptomycin were provided to the remaining mosquitoes and subsequently changed every 3 d. 7–10 dpi, mosquitoes were anesthetized on ice, and the presence of GFP-expressing oocysts in their midguts was screened under a fluorescent microscope. 18–21 dpi, the presence of GFP-expressing sporozoites in the SGs was screened using a fluorescent microscope.

A. gambiae mosquitoes were starved for 3–5 h and then fed on NF54 P. falciparum gametocyte cultures (NF54: BEI Resources, MRA-1000; mature gametocytes were adjusted to 0.02% in 40% human red blood cells [O⁺] containing 60% heat-inactivated human serum [O⁺]; please refer to the ethics statement) through artificial membranes at 37°C at the Johns Hopkins Malaria Institute Core Facility as previously described (Angleró-Rodríguez et al., 2016; Dong et al., 2018; Trager and Jensen, 1976). Unfed mosquitoes were removed immediately after blood feeding, and the remaining mosquitoes were incubated at 27°C for 8 d to check oocysts in the midgut, or 14 d to check sporozoites in the SGs.

P. berghei–infected midguts were dissected 8 dpi, stained with 0.2% mercurochrome (M7011; Sigma-Aldrich) in PBS, and examined using a ZEISS Axio Scope A1 optical microscope equipped with a Zeiss camera AxioCam MRc 5 and ZEN 2011 (blue edition) software in bright-field under a 5× objective. Images were imported into ImageJ (v1.49), and the numbers of live oocysts and melanized ookinetes were counted manually. For P. berghei sporozoite counting, infected SGs were dissected into 100 µl PBS. Then sporozoites were released from the SGs by repeated passaging through a 28 1/2-gauge insulin syringe (10 times). The sporozoite SG mixture was filtered through a 40-µm cell strainer (352340; Falcon) and centrifuged at 17,200 ×g for 10 min. The sporozoite pellet was resuspended in 30 µl of PBS and counted using a hemocytometer.

P. falciparum–infected midguts were dissected 8 dpi, stained with 0.2% mercurochrome (M7011; Sigma-Aldrich) in PBS, and examined using a phase-contrast microscope (Leica) as previously described (Dong et al., 2018). At least two to three biological replicates were performed for each experiment. For P. falciparum sporozoite counting, SGs were dissected and individually placed in 30 µl of PBS, followed by homogenization on ice. 10 μl of this homogenate was placed in a Neubauer counting chamber and counted after 5 min using a Leica phase-contrast microscope at 400× magnification.

Mosquitoes were anesthetized on ice and kept on a cold glass slide with their ventral side up. The proboscis was clipped using dissection scissors, and the lateral aspects of the thorax were gently pressed by fine forceps to force out a visibly clear drop of hemolymph from the cut proboscis (Povelones et al., 2009). Hemolymph droplets from 15 mosquitoes were collected into 15-µl Laemmli sample buffer using a pipette tip. Proteins in the hemolymph were separated by SDS-PAGE and analyzed by Coomassie staining or Western blot. Each lane was loaded with hemolymph collected from 10 mosquitoes.

P. berghei–infected mosquitoes were dissected 48 hpi. The midgut was cut open along its length to remove the bulk of the blood meal and then fixed in 4% paraformaldehyde in PBS for 45 min at room temperature. After washing three times with PBS, midguts were blocked in PBS with 1% BSA and 0.05% Triton X-100 for 4 h at room temperature, followed by incubation with rabbit polyclonal anti-TEP1-C antibody (1:300; Fraiture et al., 2009) overnight at 4°C. They were then washed three times with PBS containing 0.05% Triton X-100 and incubated with goat anti-rabbit Alexa Fluor 555 secondary antibody (1:500; A-21428; Thermo Fisher Scientific) overnight at 4°C. Midguts were washed as before, and stained with monoclonal anti-GFP antibody (1:250; A-11120; Thermo Fisher Scientific) and a goat anti-mouse Alexa Fluor 488 secondary antibody (1:500; A-11029; Thermo Fisher Scientific), mounted with Prolong Gold Antifade containing DAPI (Thermo Fisher Scientific), and monitored using a Leica SP5 confocal microscope. Three replicates of at least 10 midguts from mosGILT mutant and sibling mosquitoes were analyzed. Images were processed using ImageJ (v1.49).

Tissues (SGs, midguts, ovaries, or fat bodies) from 5 mosquitoes, 5–10 larvae, or 5–10 pupae were dissected into an RLT buffer (a guanidine-thiocyanate–containing lysis buffer), and total RNA was extracted using an RNeasy Mini Kit (QIAGEN). cDNA was made with an iScript kit (Bio-Rad) following the manufacturer’s protocol. Quantitative PCR was performed using iQ SYBR Green Supermix (Bio-Rad) on a CFX96 real-time system (Bio-Rad) as previously described (Schleicher et al., 2018). The relative expression of different mosquito genes was normalized to A. gambiae ribosomal protein s7 (rps7; primers in Table S1) using the comparative Ct method (Livak and Schmittgen, 2001).

Proteins were resolved by SDS-PAGE using 4–20% Mini-Protean TGX gels (Bio-Rad) at 300 V for 20 min and then transferred onto a 0.2-µm nitrocellulose membrane for 60 min at 80 V in a Tris-Glycine transfer buffer with 25% methanol. Blots were blocked in 5% nonfat milk in PBS containing 0.1% Tween-20 for 30 min at room temperature. Primary antibodies were diluted in blocking buffer and incubated with the blots for 1 h at room temperature or 4°C overnight (4C10G9 mouse anti-mosGILT monoclonal 1:1,000, 0.8 µg/ml; mouse anti-β actin monoclonal, Abcam 8224, 1:1,000; rabbit anti-TEP1 polyclonal 1:200, and rabbit anti-prophenoloxidase 2 [PPO2] polyclonal 1:15,000, two kind gifts from Dr. Elena A. Levashina, Vector Biology Unit, Max Planck Institute for Infection Biology, Berlin, Germany). Infrared fluorescent secondary antibodies, goat anti-mouse IRDye 680LT (926-68020; LI-COR), or goat anti-rabbit IRDye 800CW (926-32211; LI-COR) was diluted in blocking buffer at 1:5,000 and incubated for 1 h at room temperature. Blots were washed in PBS with 0.1% Tween-20 three times for 5 min and then imaged with a LI-COR Odyssey Fc imaging system. Gel densitometry was performed using the in-built quantification tools on the LI-COR Odyssey Fc imaging system and normalized to the PPO2 signal. Data are presented as relative band signal intensity comparing the mosGILT mutants to siblings.

RNA interference was performed as previously described (Povelones et al., 2011; Schleicher et al., 2018). dsRNA targeting mosGILT, TEP1, CTL4, or an irrelevant luc gene from Renilla reniformis (Schleicher et al., 2018), were transcribed using the MEGAScript RNAi kit (Thermo Fisher Scientific, Ambion; primers in Table S1).

2-d-old A. gambiae mosquitoes were injected with dsRNA using a Nanoject II Auto-Nanoliter Injector (Drummond) and allowed to recover in paper cups with 10% sucrose for 2 d. For silencing of CTL4 or TEP1, 0.35 µg of total dsRNA was used for each mosquito. For silencing of mosGILT, 1.0 µg of total dsRNA was injected into each mosquito. On day 4 after dsRNA injection, these mosquitoes were deprived of sucrose for 20 h before allowing them to feed on P. berghei–infected mice with 3–5% parasitemia. Unfed mosquitoes were immediately discarded, and the fed remaining mosquitoes were maintained with 10% sucrose containing 0.05% penicillin/streptomycin for 8 d at 20°C. On day 8 after infection, oocysts and melanized ookinetes in the midgut were counted as mentioned above.

Ecdysteroid titers were determined by a 20E EIA kit (Cayman Chemical) according to the manufacturer’s protocol. At 24 h after infectious blood meal, individual female mosquitoes were stored in 100 µl of 100% methanol at −80°C. The mosquito whole body was homogenized and kept at 4°C for 30 min before centrifugation (10,000 ×g, 10 min). The supernatant was transferred to a new 1.5-ml Eppendorf tube, and the solvent was removed by centrifugation at reduced pressure. The extracted total ecdysteroids were redissolved in 50 µl of EIA buffer, and loaded on the 96-well strip plate precoated with mouse anti-rabbit IgG (Cayman Chemical). Samples in duplicate were incubated with 50 µl tracer and 50 µl 20E antiserum overnight at 4°C and then developed with 200 µl Ellman’s Reagent for 90–120 min. Absorbance was measured by a BioTek Synergy Mx microplate reader at 412 nm.

2 h after mosquitoes took a blood meal, 20E (CAS 5289–74-7, Santa Cruz Biotechnology, sc-202407) was microinjected into each mosquito using a Nanoject II Auto-Nanoliter Injector (Drummond). At 24 h after blood feeding, the fat bodies from five mosquitoes were dissected into RLT buffer, and total RNA was extracted to analyze the Vg expression by quantitative RT-PCR (RT-qPCR). P. berghei oocyst number and infection prevalence in 20E-injected mosquitoes were examined at 8 dpi as mentioned above.

All primers used in this study are listed in Table S1.

All data analysis, graphing, and statistics were performed in Prism (v7.0, GraphPad Software).

Fig. S1 shows a schematic describing the experimental crosses. Fig. S2 reveals the influence of mosGILT disruption on pupation time and male reproduction system, in addition to mosGILT expression in WT A. gambiae at larval, pupal, and adult stages. Fig. S3 depicts representative images of P. berghei parasites in the midguts of siblings and mosGILT mutants. Fig. S4 demonstrates the restoration of Vg expression in mosGILT mutants by 20E injection. Fig. S5 shows that mosGILT silencing does not alter Vg expression and P. berghei infection in WT mosquitoes that have normal ovaries. Table S1 lists primers used in this study.

We thank Kathleen DePonte, Mingjie Wu, Alfred Jiang, and Dr. Akash Gupta for their assistance and laboratory support. We thank the Johns Hopkins Malaria Research Institute insectary and parasitology core facilities for providing mosquito rearing resources and P. falciparum gametocyte cultures. We thank Dr. Elena A. Levashina for rabbit anti-TEP1 and rabbit anti-PPO2 antibodies. We thank Dr. Eric Marois for fruitful discussion on the CRISPR/Cas9 technology and Dr. Yue Li for helpful discussions on the manuscript.

This work was supported by funding from the Howard Hughes Medical Institute and the National Institutes of Health (138949).

The authors declare no competing financial interests.

Author contributions: J. Yang, T.R. Schleicher, Y. Dong, H.B. Park, J. Lan, P. Cresswell, J. Crawford, G. Dimopoulos, and E. Fikrig designed, performed, and analyzed experiments. Y. Dong performed the embryo microinjections, generated the mosGILT-gRNA–expressing mosquito line, and performed the P. falciparum infection studies. H.B. Park helped with the detection of mosquito hormones. J. Lan helped with the maintenance of mosquito colonies, mosquito crossings, and genotyping. The paper was written by J. Yang, T.R. Schleicher, and E. Fikrig. All authors discussed the results and commented on the manuscript.