The redox-sensing gene Nrf2 affects intestinal homeostasis, insecticide resistance, and Zika virus susceptibility in the mosquito Aedes aegypti

Production and degradation of reactive oxygen species (ROS) are extensively regulated to ensure proper cellular responses to various environmental stimuli and stresses. Moreover, physiologically generated ROS function as secondary messengers that can influence tissue homeostasis. The cap'n'collar transcription factor known as nuclear factor erythroid-derived factor 2 (Nrf2) coordinates an evolutionarily conserved transcriptional activation pathway that mediates antioxidant and detoxification responses in many animal species, including insects and mammals. Here, we show that Nrf2-mediated signaling affects embryo survival, midgut homeostasis, and redox biology in Aedes aegypti, a mosquito species vector of dengue, Zika, and other disease-causing viruses. We observed that AeNrf2 silencing increases ROS levels and stimulates intestinal stem cell proliferation. Because ROS production is a major aspect of innate immunity in mosquito gut, we found that a decrease in Nrf2 signaling results in reduced microbiota growth and Zika virus infection. Moreover, we provide evidence that AeNrf2 signaling also controls transcriptional adaptation of A. aegypti to insecticide challenge. Therefore, we conclude that Nrf2-mediated response regulates assorted gene clusters in A. aegypti that determine cellular and midgut redox balance, affecting overall xenobiotic resistance and vectorial adaptation of the mosquito.

Production and degradation of reactive oxygen species (ROS) are extensively regulated to ensure proper cellular responses to various environmental stimuli and stresses. Moreover, physiologically generated ROS function as secondary messengers that can influence tissue homeostasis. The cap'n'collar transcription factor known as nuclear factor erythroid-derived factor 2 (Nrf2) coordinates an evolutionarily conserved transcriptional activation pathway that mediates antioxidant and detoxification responses in many animal species, including insects and mammals. Here, we show that Nrf2-mediated signaling affects embryo survival, midgut homeostasis, and redox biology in Aedes aegypti, a mosquito species vector of dengue, Zika, and other disease-causing viruses. We observed that AeNrf2 silencing increases ROS levels and stimulates intestinal stem cell proliferation. Because ROS production is a major aspect of innate immunity in mosquito gut, we found that a decrease in Nrf2 signaling results in reduced microbiota growth and Zika virus infection. Moreover, we provide evidence that AeNrf2 signaling also controls transcriptional adaptation of A. aegypti to insecticide challenge. Therefore, we conclude that Nrf2-mediated response regulates assorted gene clusters in A. aegypti that determine cellular and midgut redox balance, affecting overall xenobiotic resistance and vectorial adaptation of the mosquito.
Aedes aegypti is an invasive mosquito species and represents a significant threat worldwide because of its ability to transmit dengue and, more recently, chikungunya and Zika viruses (1)(2)(3). The main strategies to eliminate the mosquito are based on vector control methods, such as removal of potential breed-ing sites and the use of insecticides, which has been intensified since the 1980s due to continuous dengue outbreaks. However, given the increasing insecticide resistance in vector populations, these control measures are becoming ineffective (4,5). Significant efforts have been devoted to developing novel strategies for mosquito control, including the use of genetically modified mosquitoes carrying genes that make them refractory to arbovirus infection (6,7). Because the midgut is the initial site of infection in the mosquito, it also constitutes the ideal site for overexpressing/activating candidate antiviral effectors (8,9).
Reactive oxygen species (ROS) 2 are implicated in direct killing of pathogens, increased tissue damage, and a variety of cell signaling processes. In mosquitoes, ROS-mediated signaling has been especially studied in association with the control of malaria parasite development in the midgut of Anopheles mosquitoes (10,11) and less intensively investigated regarding viral replication in A. aegypti (12,13). The cap'n'collar transcription factor, also named nuclear factor-erythroid 2 p45-related factor 2 (Nrf2), coordinates the regulation of drug detoxification, GSH homeostasis, and NADPH regeneration with oxidative stress. Nrf2 is structurally and functionally conserved from insects to humans (14), and it heterodimerizes with the small Maf transcription factors to bind a consensus DNA sequence (the antioxidant response element (ARE)) and regulate gene expression (15)(16)(17). Nrf2 therefore plays a pivotal role in cellular adaptation to ROS and xenobiotics (14). More recently, functions beyond stress response have been attributed to Nrf2, such as control of energy metabolism and stem cell regulation (18 -20).
Here, we show that this ancient redox-sensitive Nrf2 pathway has a pivotal role in stress responses in the adult mosquito and embryonic survival. Nrf2 depletion alters the redox balance in the midgut, which results in intestinal stem cell proliferation, microbiota growth impairment, and lowered viral infection. AeNrf2 also impacts metabolic adaptation to insecticide resistance. We discuss the importance of this pathway for the overall control of arboviral transmission.

Results
The cap'n'collar locus in Drosophila melanogaster produces multiple alternatively spliced transcripts, giving rise to three main Cnc proteins, CncA, CncB, and CncC, of different sizes and domain compositions. All of these include the COOH-terminally located bZIP DNA-binding and dimerization domain (21). There are two proteins in A. aegypti, AAEL015467 and AAEL005077, with sequence similarity to the D. melanogaster Cnc. Both are identical at their C-terminal regions (Neh domains 1, 3, and 6), but only AAEL005077 has an N-terminal sequence that contains domains similar to Neh2, -4, and -5. These latter domains are important for transactivation and redox regulation in the D. melanogaster (CncC) and mammalian (Nrf2) homologs ( Fig. S1A and Fig. S2). We therefore designated AAEL005077 as AeNrf2/CncC and AAEL015467 as AeNrf2/CncA because the latter has higher similarity to the shorter Cnc A isoform (Fig. S2). We focused on AeNrf2/CncC (hereafter referred to as AeNrf2) for genetic targeting and characterization of the associated phenotypes because of its greater similarity to CncC, previously described as a redox-regulated transcription factor (22).
Nrf2-knockout mice are viable and fertile, although they exhibit increased sensitivity to environmental stressors (14). We attempted to generate somatic loss-of-function mutants to investigate the role of AeNrf2 in mosquito development by injecting Cas9 protein and in vitro-transcribed guide RNAs (gRNAs) targeting AeNrf2 and KMO (kynurenine monooxygenase) gene into early-stage embryos (60 -75 min after egg deposition). KMO was targeted as a positive control for Cas9 activity (23). In contrast to the results in mice, AeNrf2 knockouts in A. aegypti result in decreased embryonic survival (Table S1). Embryonic survival rates following multiplex injections of one KMO gRNA together with an increasing number (1, 2, and 5) of different Nrf2-targeted gRNAs were 19.1, 8.6, and 6.8%, respectively. DNA lesions in the KMO gene were confirmed and associated specifically with the KMO gRNA target site. However, the majority (75%) of the Nrf2 gRNA target sequences analyzed had a much lower efficiency rate (Յ30%) for Cas9-induced mutations, and Ͼ50% of the analyzed individuals had no insertions or deletions detected in their pooled DNA sequences (Table S2). Taken together, these results support the conclusion that the disruption of Nrf2 probably produces an embryonic lethal phenotype, and this results in no mutations being detected in later developmental stages.

Nrf2 connects antioxidant response and redox balance
Vertebrate Nrf2 and D. melanogaster CncC regulate the basal and inducible expression of antioxidant and detoxifying genes (22,24). We conducted a series of experiments to test whether the mosquito ortholog has a similar role. Paraquat-induced expressions of GSH S-transferase (GSTX2, AAEL010500), cytochrome P450 (CYP6M11, AAEL009127), and members of the GSH biosynthesis pathway (namely glutamate cysteine ligase (GCLC; AAEL008105) and glutathione synthetase (AAEL009154)) were impaired in the midgut by RNAi-mediated knockdown of Nrf2 (Fig. 1). Furthermore, Nrf2 homologs coordinate transcriptional activation induced by electrophilic reagents through the ARE (22,24,25). A bioinformatics de novo motif discovery (26), on GSTX2 and CYP6M11 DNA regions, identified two putative ARE motifs that have high sequence similarity to Nrf2/Maf binding sites from the publicly accessible DNA motif JASPAR database (27) (Fig. S1B). This consideration was based on the hypothesis that sequence similarity reflects functional analogy. These putative AREs, of 14 and 19 base pairs, were named AeARE1 and AeARE2, respectively, and were found in tandem repeats in the promoter regions of both GSTX2 and CYP6M11 (Table 1). Furthermore, a search using previous transcriptomic profiling data from A. aegypti cells under heme or paraquat stimulation (28) found that ϳ70% of the up-regulated transcripts have cis-located elements with significant matches to either AeARE1 or AeARE2 in their promoters (Table S3). These include some genes that are not directly involved with detoxification processes, such as energy metabolism genes. This finding indicates that a large fraction of heme and paraquat-induced transcriptional responses are associated with Nrf2 signaling.
We used HPLC quantification to evaluate the redox state of the gut by assaying fluorescent products of a ROS-sensitive probe, dihydroethidium (DHE), following Nrf2 knockdown. Nrf2 disruption resulted in significantly increased ROS levels in the midgut even after a regular blood meal ( Fig. 2A). Taken together, these findings support the conclusion that transcriptional control of key cytoprotective genes through the Nrf2/ ARE axis is an ancient mechanism conserved in evolution and

Redox-sensing Nrf2 in A. aegypti
important for maintenance of the redox balance in the midgut of the mosquito.

Nrf2 signaling is involved with insecticide resistance in A. aegypti
The Nrf2 pathway is constitutively active in insecticide-resistant strains of D. melanogaster (29) and we hypothesized that AeNrf2 signaling could contribute to enhanced insecticide resistance of certain mosquito populations. Two Brazilian field populations with distinctive susceptibility to pyrethroid and organophosphate insecticides, Caseara (Susceptible, S) and Oiapoque (Resistant, R), and a control laboratory strain (Rockefeller) were used to investigate basal expression of selected putative Nrf2 target genes and of Nrf2 itself. Although Nrf2 had similar transcript expression levels among the tested populations, Kelch ECH-associating protein 1 (Keap1), GSTX2, and CYP6BB2 were expressed at significantly higher levels in the resistant Oiapoque strain, when compared with the susceptible Caseara strain (Fig. 3A).
dsRNA silencing of Nrf2 in 2-day-old Oiapoque (R) females resulted in a concomitant decrease in the accumulation of the transcripts of Keap1, GSTX2, and CYP6BB2 (Fig. S3). These mosquitoes were also more sensitive to the insecticide malathion exposition than the ones that had been injected with dsLacZ control (Fig. 3B). The mortality for the Nrf2-silenced cohorts was 3.7-fold higher compared with controls. These results support the conclusion that activation of the Nrf2 signaling pathway is important to confer tolerance to malathion toxicity.

Nrf2-mediated ROS controls ISC proliferation, microbiota growth, and viral replication in the midgut
Because redox state influences cellular proliferation in highturnover tissues (30,31), we investigated the mitosis rate in the intestinal stem cells (ISCs) in the midgut. Phosphohistone H3 (PH3) is detected only in mitotic cells; hence, specific antibodies can be used in assays to determine the fraction of cells undergoing division. Knockdown of Nrf2 increased the number of PH3-positive (PH3 ϩ ) cells in midguts either after a regular blood meal or when insects were fed with blood supplemented with paraquat (Figs. 2B and 3C). This supports the conclusions that increasing intracellular ROS levels promotes ISC proliferation and that AeNrf2 may limit mitotic stimulation by maintaining a reduced intracellular environment and preventing oxidative damage.
A vigorous oxidative burst response is employed as a defense against bacteria in barrier epithelia like the intestinal epithelium (32,33). We therefore evaluated whether changes in the midgut redox balance imposed by Nrf2 silencing would prevent the indigenous microbiota proliferation that takes place after a blood meal (34). Determination of bacterial load by qPCR using 16S ribosomal target sequences (35) showed that blood-fed dsNrf2-injected mosquitoes have lower microbiota levels in the midgut, similar to control dsLacZ-injected mosquitoes fed with paraquat (Fig. 4A). These results clearly indicate that Nrf2-mediated redox regulation in the midgut exerts profound effects on the growth and development of gut-associated bacteria.

Table 1 Nrf2 target genes analyzed in this study
Shown is a summary of transcription data available for each (with references) and a schematic representation of Nrf2-binding sites (AREs) identified in their promoter regions, ARE1 and ARE2, represented in blue and green, respectively. Numbers indicate the position relative to the transcriptional initiation site.

Redox-sensing Nrf2 in A. aegypti
To test the hypothesis that changes in the redox balance in the mosquito could influence the midgut infection by a relevant human pathogen, we challenged Nrf2-silenced or paraquat-fed A. aegypti with Zika virus and measured the number of pfu per midgut (infection intensity) and the number of infected midguts (infection prevalence) 4 days after feeding of virus-contaminated blood. We observed a significant reduction in both infection intensity and prevalence (Fig. 4, B and C, respectively). The number of noninfected mosquitoes increased ϳ3-fold in the Nrf2-silenced group and 5-fold in the paraquat-fed group. We interpret that this reveals that redox alterations, either by reduction in antioxidant capacity through Nrf2 knockdown or direct feeding with a pro-oxidant molecule, can reduce arboviral infection of the midgut.

Discussion
Ranges of metazoan species cope with oxidative burden by activating the transcription factor Nrf2, a major regulator of cytoprotective responses to electrophilic stresses. We analyzed a range of phenotypes that result from ablation of this gene in the mosquito vector, A. aegypti. A role for this gene and its pathway was discovered for early development, adult midgut biology, insecticide resistance, gut microbiome, and vector competence for an arbovirus. This complexity supports a vital role of this gene in the biology of this insect.
Nrf2 is a key regulatory protein in the redox signaling pathway and, together with small Maf proteins, stimulates transcription through binding to the AREs in the 5Ј-end regulatory regions of target genes (21). This response is limited by Keap1, a repressor protein that binds to Nrf2, keeping it in the cytosol and promoting its degradation by the ubiquitin proteasome pathway (21). Nrf2 is released upon oxidation of specific Keap1 cysteine thiol residues that are sensitive to the cell redox status, and this leads to increased expression of target genes. The Neh2 domain at the N terminus of the Nrf2 protein contains motifs that bind to the Keap1 Kelch domain, which negatively regulates the transcriptional activity of Nrf2 (21,36). We found a relatively low level of conservation for the Neh2 domain in AeNrf2, even when compared with the D. melanogaster homolog, in contrast with the high identity values found for other domains. Nevertheless, the A. aegypti genome contains a Drosophila Keap1 homolog (AAEL005424) with a high degree of similarity, including the BTB domain, which is essential for Nrf2-specific negative regulation by Keap1 (37,38). In contrast, the Neh6 domain, which has high levels of identity to other Nrf2 homologs, has been shown to promote Keap1-independent negative regulation of Nrf2 in vertebrates (39). These data, together with the presence of corresponding ARE motifs in the promoter regions of key detoxification genes and a well con-
Nrf2 regulates the expression of proteins that collectively promote cell survival following exogenous stress. This protective role of Nrf2 is also demonstrated by inverse genetics. Our attempts to generate gene knockouts in A. aegypti through the previously validated CRISPR/Cas9-induced mutagenesis (23,43) led us to conclude that AeNrf2 encodes a protein vital for embryogenesis in this mosquito. In contrast to our results, Nrf2-knockout mice are viable and do not display any signs of an increased basal oxidation state, but have an impaired ability to respond upon challenge (41,42). In D. melanogaster, the longest splice variant, CncC, shows the highest conservation with functionally important domains of Nrf2. Interestingly, it has been suggested that Drosophila CncC performs functions that in mammals are attributed to Nrf2 but can also exert others that are typical of Nrf1 (knockout of which yields an embryonic lethal phenotype in mice) (21).
In contrast, we achieved successful Nrf2 knockdown in adult insects, showing that this transcription factor regulates the basal and stress-inducible expression of a battery of genes encoding key components of the GSH-based antioxidant systems and drug-metabolizing isoenzymes. This signaling pathway is important also for thioredoxin-based antioxidant response, as well as multidrug resistance-associated efflux pumps (44). GSH S-transferase (GST) is a canonical Nrf2 transcriptional target, having a prototypic ARE in its promoter (16,22). Similarly, the Nrf2 pathway is associated with cytochrome P450 -mediated metabolic resistance to insecticides in D. melanogaster (29,45) and more recently in other insects, such as beetles (46,47) and aphids (48). It was shown in A. aegypti that detoxification enzymes, namely members of the GST and cytochrome P450 families, contribute to increased levels of resistance to insecticides (49 -51). Specifically for the CYP genes studied here, the CYP6M subfamily has been functionally characterized for their role in insecticide resistance in anophelines; Anopheles gambiae CYP6M2, whose best hit in the A. aegypti genome is CYP6M11, has been shown to metabolize permethrin and deltamethrin (52,53). Furthermore, heterologously expressed A. aegypti CYP6BB2 exhibited strong metabolic activity for permethrin (51). This same gene was also pointed out as a solid candidate for imidacloprid metabolism based on gene expression data and substrate binding predictions (54). Constitutively activated Nrf2 controls insecticide resistance in D. melanogaster through up-regulation of cytochrome P450 and GSH transferase genes (29,45).
Regarding insecticide resistance, microarray analyses (55, 56) and RNA-seq studies (57)(58)(59) in A. aegypti-resistant strains revealed genomic changes (including polymorphism, copy number variation, and gene amplification events) in detoxification enzymes, such as P450s, esterases, and GSTs, that validated the use of these transcripts as genetic markers for resistance. Taken together, these findings suggest that insecticide resistance is not only conferred via multiple resistance genes, but also that regulation of transcription is a key factor in resistance gene amplification (reviewed in Ref. 5). We investigated the transcriptomic database of resistant A. aegypti populations present in Faucon et al. (59), the most recent and geographic diverse thus far, in a search for ARE-containing genes (Table  S4). Surprisingly, we found that ϳ70% of the transcripts up-regulated in these populations presented AeARE1 and/or AeARE2 in their promoters, implicating this cis-regulatory element as a relevant genomic marker of the resistance phenotype in natural populations. Our results further suggest that a field population naturally resistant to insecticides (Oiapoque) presents an inherently elevated transcriptional regulation of the targets, considering that the large control of the pathway is at the level of the transcription factor protein stability. Hence, one cannot exclude additional components that could allow this constitutive activation in resistant mosquitoes (e.g. proteins involved in proteasomal degradation, co-factors, histones, and chromatin remodeling proteins). However, the contribution of Nrf2 itself is demonstrated by the elevated basal (unstressed, sugar-fed) expression of previously identified Nrf2 targets. Additionally, the increased mortality of dsNrf2-injected resistant mosquitoes in response to malathion exposition is probably due to reduced expression of several genes that are downstream to this transcription factor, including the ones herein investigated.

. Silencing of AeNrf2 decreases gut bacterial microbiota and Zika virus midgut infection levels.
A, culture-independent evaluation of midgut natural microbiota in dsLacZ or dsNrf2-injected mosquitoes, 24 h after feeding with blood or blood supplemented with 1 mM paraquat, through qPCR for bacterial ribosomal 16S RNA. Results are for pools of at least three independent experiments. Error bars, S.E. Statistical analyses were made by one-way analysis of variance. B, dsRNA-injected females were fed on blood contaminated with 10 7 pfu/ml Zika virus, and 4 days after feeding, the number of pfu was determined in the midgut. C, the percentage of infected midguts (infection prevalence) was scored from the same set of data as in B. Mann-Whitney U tests were used for infection intensity (B), and 2 tests were performed to determine the significance of infection prevalence analysis (C). *, p Ͻ 0.05; ***, p Ͻ 0.001; ****, p Ͻ 0.0001.

Redox-sensing Nrf2 in A. aegypti
Recently, it has been shown that in the malaria vector A. gambiae, Nrf2 partner Maf protein expression correlates with the expression of a limited transcript set, detoxificationrelated, such as cytochromes P450 and GSTs, indeed suggesting a regulatory role for this transcription factor in insecticide resistance (60). Here, we propose a more comprehensive role for Nrf2 signaling in mosquito physiology. The use of insecticide by humankind is a very recent event in the evolutionary history of insects, and xenobiotic resistance comes from recruitment of ancestral highly conserved detoxification pathways that evolved as detoxification pathways directed to plant and microbe-borne allelochemicals. In addition, the same mechanisms contribute to protection against redox insult promoted by endogenous sources of free radicals, which generate a wide variety of electrophilic compounds with biological activity. Among the classes of genes induced in a previously described A. aegypti cellular transcriptomic response toward heme and paraquat, a large fraction is accounted for by stress response genes (28). However, transcripts from assorted functional groups in this database presented AeAREs in their promoters; these included antioxidant and phase II detoxifying enzymes, which are classically regulated by the Nrf2 pathway and have been associated with protective responses in the midgut of blood-feeding arthropods (61)(62)(63)(64)(65) as well as sequences that regulate glucose metabolism and transmembrane transporters, indicating a role for the Nrf2 signaling in the energy metabolic adaptation of A. aegypti to environmental stress, as has been suggested for other models (20,66,67). Moreover, in the midgut of A. aegypti, ROS levels are dramatically reduced after blood ingestion through a heme-driven mechanism (34). Collectively, our results show that Nrf2 signaling regulates batteries of genes involved in various aspects of cytoprotective and metabolic functions through associated AREs, and this is probably an important component of the heme-mediated protective response to avoid oxidative stress.
Modulation of the redox balance in the intestinal epithelium has been shown to induce proliferation of stem cells (68,69), being a major regulator of tissue homeostasis and an essential feature of midgut physiology, characterized by very high rates of cell turnover. We observed an increased mitosis in response to feeding with the ROS generator paraquat and in Nrf2-knockdown mosquitoes, similarly to the pattern for redox-modulated ISC proliferation found in flies, in which loss of CncC results in increased ROS accumulation, accompanied by enhanced proliferation (70), suggesting that the regulation of stem cell function by the intracellular redox milieu is an evolutionarily conserved phenomenon. ROS levels in the gut epithelia are also shown to play an important role in controlling bacterial growth (32,33,71). In mosquitoes, changes in ROS production in the midgut not only impact innate immunity responses against bacteria, but can also affect their ability to transmit human pathogens (11,34,72,73). The Nrf2 pathway has been shown to participate as a signaling conduit between the resident microbiota and the eukaryotic host, mediating beneficial effects in the gut (74). Here we present evidence for an Nrf2-mediated redox control on the mosquito's midgut microbiota. Furthermore, Nrf2 signaling depletion, rather than a collateral bacterial alter-ation, seems to be the primary event causing change in the redox state (Fig. S4).
In mammals, Nrf2-dependent gene expression has been associated with a number of pathologies that are caused or exacerbated by oxidative stress. Specifically, for DENV-mediated response, it has been recently reported that Nrf2 is directly activated by DENV in mononuclear phagocytes, transcriptionally signaling for inflammatory activation (75). Regarding data discussed here, Nrf2 has been implicated in the control of Zika virus infection (76), suggesting an important role for the induction of these intracellular mediators in retarding flavivirus replication. As shown here, Nrf2 silencing can limit Zika virus infection, an effect that can be emulated by paraquat, suggesting that the permissive role of the Nrf2 pathway could be accounted by its homeostatic function in the redox balance of the midgut. In contrast, in a recent study, Oliveira et al. (77) demonstrated that a blood-induced antioxidant response mediated by the enzyme catalase facilitates the establishment of Dengue virus (but not Zika virus) in the midgut of A. aegypti. Mosquitoes evolved a redundant antioxidant protection strategy to prevent oxidative stress following blood intake to adapt to the ingestion of large amounts of heme, an unavoidable consequence of blood digestion (34,63). Here we present evidence that the antioxidant mechanisms under Nrf2 control, while contributing to redox balance and tissue homeostasis of the midgut, also have an infection-permissive effect, affecting the so-called midgut infection barrier for Zika virus. This also indicates that the redox milieu of the mosquito can diversely impact the distinct virus this vector can transmit, and these differences deserve further investigation. Because the A. aegypti midgut is an initial site of contact between virions and mosquito cells, genetic alterations capable of improving the midgut infection barrier might be considered as potential tools for preventing the establishment of the infection and can be used as targets for disease transmission control strategies (9,78).
Overall, our results reveal that AeNrf2-mediated signaling is a major pleiotropic regulator for midgut homeostasis, which affects environmental and stress-related responses in mosquitoes, and can transcriptionally control several genes that directly or indirectly affect their vectorial competence (Fig. 5).

Ethics statement
All animal care and experimental protocols were conducted in accordance with the guidelines of the Committee for Evaluation of Animal Use for Research (Federal University of Rio de Janeiro, CAUAP-UFRJ) and the National Institutes of Health Guide for the Care and Use of Laboratory Animals. The protocols were approved by CAUAP-UFRJ under registration number IBQM155/13. Dedicated technicians in the animal facility at the Instituto de Bioquímica Médica Leopoldo de Meis (UFRJ) carried out all protocols related to rabbit husbandry under strict guidelines to ensure careful and consistent animal handling.

Mosquitoes
A. aegypti (Red Eye strain) were raised in a mosquito rearing facility at the Federal University of Rio de Janeiro, Brazil,

Redox-sensing Nrf2 in A. aegypti
under a 12-h light/dark cycle at 28°C and 70 -80% relative humidity. Larvae were fed with dog chow, and adults were maintained in a cage and given a solution of 10% sucrose ad libitum. Females 4 -7 days posteclosion were used in the experiments.
For insecticide resistance experiments, mosquitoes were maintained at Laboratório de Fisiologia e Controle de Artrópodes Vetores, Instituto Oswaldo Cruz, FIOCRUZ, Rio de Janeiro, Brazil. Laboratory-reared (F6 generation) field populations of A. aegypti from Caseara (TO, Brazil) and Oiapoque (AP, Brazil) were used in the insecticide bioassays as models for susceptibility and resistance, respectively, to both pyrethroid and organophosphate insecticides. The Rockefeller reference strain was used as an additional susceptibility control.
When mentioned, mosquitoes were artificially fed with heparinized rabbit blood with or without 1 mM paraquat, a pro-oxidant compound used to induce oxidative stress. Feeding was performed using water-jacketed artificial feeders maintained at 37°C sealed with parafilm membranes. Female midguts were dissected 24 h after feeding for RNA sample preparation.

Nrf2 gene knockdown by RNAi
dsRNA was synthesized from templates amplified from cDNA of whole mosquitoes using specific primers containing a T7 tail (see supporting material). The in vitro dsRNA transcription reaction was adapted from a tRNA transcription protocol (79). Briefly, reactions were performed at 37°C for 12 h in a buffer containing 40 mM Tris⅐HCl (pH 8.0), 22 mM MgCl 2 , 5 mM DTT, 2 mM spermidine, 0.05% BSA, 15 mM guanosine monophosphate, a 7.5 mM concentration of each nucleoside triphosphate, amplified template DNA (0.1 g/l), and 5 M T7 RNA polymerase. The transcribed dsRNA was treated with DNase at 37°C for 30 min and precipitated using 1:10 (v/v) 3 M sodium acetate, pH 5.2, and 1 M isopropyl alcohol. The pellet was washed twice with 70% ethanol and then eluted in water to reach a final concentration of 3 g/l. Mosquitoes were injected intrathoracically with the dsRNA (0.4 g) with a microinjector (Nanoinjector, Drummond) and were either blood-fed or used in insecticide bioassays 48 h later. The LacZ gene was used as a nonrelated dsRNA control and was amplified from a plasmid containing a cloned LacZ fragment.

RNA isolation and quantitative real-time PCR analysis
Total RNA was isolated from insects at different developmental stages, whole bodies of adult males and females, and dissected midgut epithelia ovaries, heads, thoraces, and abdomens (carcass) of blood-fed females using TRIzol (Invitrogen). Complementary DNA was synthesized using the High-Capac- Figure 5. Schematic overview of Nrf2 signaling in mosquito physiology. Through a comprehensive transcriptional enhancement, Nrf2, bound to the ARE sites, regulates cellular redox balance. In a broader tissue context, Nrf2 signaling is able to coordinate midgut homeostasis by balancing signaling/damaging ROS and thus fine-tuning microbiota growth, viral replication, and stem cell proliferation. In consequence, Nrf2 can also systemically influence mosquitoes' xenobiotic tolerance and emerge as a central regulator of various processes important for vectorial adaptation of A. aegypti.
Quantitative gene amplification (qPCR) was performed with the StepOnePlus real-time PCR system (Applied Biosystems, Foster City, CA) using the Power SYBR Green PCR Master Mix (Applied Biosystems). The comparative Ct method (80) was used to compare RNA abundance. The A. aegypti ribosomal protein 49 gene (Rp49) was used as an endogenous control (81). The assessment of midgut bacterial growth was performed through qPCR for bacterial ribosomal 16S RNA. All oligonucleotide sequences used in qPCR assays are available in the supporting material.

HPLC analysis of DHE products
To provide specific quantitative assessment of ROS levels, we performed HPLC fractionation of DHE oxidation products (82). Briefly, after incubation with 100 M DHE, the dissected midguts were opened with forceps and washed in PBS to remove intestinal contents. Pools of 15 gut epithelia each were frozen in liquid N 2 , homogenized in 100% acetonitrile (500 l), sonicated, and centrifuged at 13,000 ϫ g for 10 min. The resulting supernatant was dried under vacuum (SpeedVac SVC100 Savant), and the resulting pellet was stored at Ϫ70°C until use. The dried samples were resuspended in PBS containing 100 M diethylenetriaminepentaacetic acid (Sigma) and injected into an HPLC LC-10AT device (Shimadzu, Tokyo) equipped with a diode array (SPD-M10A) and fluorescence detectors (RF-20A). Chromatographic separation of DHE oxidation products was performed using a NovaPak C18 column (3.9 ϫ 150 mm, 5-m particle size) equilibrated in solution A (10% acetonitrile and 0.1% TFA) at a flow rate of 0.4 ml/min. After sample injection, a 0 -40% linear gradient of solution B (100% acetonitrile) was applied for 10 min, followed by 10 min of 40% solution B, 5 min of 100% solution B, and 10 min of 100% solution A. The amount of DHE was measured by light absorption at 245 nm, and the DHE oxidation products, hydroxyethidium and ethidium, were monitored by fluorescence detection.
Quantification of whole-midgut mitosis was performed by counts of individual nuclei marked by the PH3, using a Zeiss Observer Z1 fluorescence microscope equipped with a Zeiss Axio Cam MrM using a Zeiss-15 filter set (excitation BP 546/ 12; beam splitter FT 580; emission LP 590) under a ϫ20 objective. Representative images were acquired in an Olympus IX81 microscope and a CellR MT20E imaging station equipped with an IX2-UCB controller and an ORCAR2 C10600 CCD camera (Hamamatsu). Image processing was performed with the Xcellence RT version 1.2 software. Optical slices (0.1 m) were generated.

Viral infections
A Brazilian strain of Zika virus from Pernambuco was kindly provided by Dr. Laura Helena Vega Gonzales Gil (Centro de Pesquisas Aggeu Magalhães, Fundação Oswaldo Cruz, Brazil) and characterized by Coelho et al. (84). Infection procedures were performed as described previously (74). Briefly, viral stocks were propagated in C6/36 cells maintained in Leibovitz-15 medium supplemented with 5% fetal bovine serum, 1% nonessential amino acids, 1% penicillin/streptomycin and tryptose (2.9 g/liter). Culture supernatants containing viral particles were harvested, centrifuged, aliquoted, and stored at Ϫ70°C until use. Viral titers were determined by plaque assay as 2 ϫ 10 7 pfu/ml. Females were infected in an artificial blood meal containing a 1:1 mix of rabbit red blood cells and L-15 medium containing Zika virus. Midguts were dissected at 4 days postblood meal and stored individually in DMEM at Ϫ70°C until assayed.

Plaque assays
Zika plaque assays were performed (as described (74)) in Vero cells maintained in DMEM supplemented with sodium bicarbonate, 1% L-glutamine, 10% fetal bovine serum, and 1% penicillin/streptomycin. Midgut tissue was disrupted to release viral particles by vortexing the tubes for 10 min at room temperature. The samples then were centrifuged at 10,000 ϫ g at 4°C, and 10-fold serial dilutions were performed and each one inoculated in a single well. The plates were gently rocked for 15 min at room temperature and then incubated for 45 min at 37°C and 5% CO 2 . Finally, an overlay of DMEM, containing 0.8% methylcellulose and 2% FBS, was added to each well, and the plates were incubated for 5 days (at 37°C and 5% CO 2 ). Culture media were then discarded, and a solution of 1:1 (v/v) methanol and acetone and 1% crystal violet was used to fix and stain the plates. pfu were counted and corrected by the dilution factor.

Bioassays
Previously dsRNA-injected mosquitoes from different strains (Rockefeller, Caseara, or Oiapoque) were submitted to a dose discriminator bioassay (48 h after the injection) to evaluate their profile of susceptibility to the organophosphate, malathion. The test was adapted from the World Health Organization recommended protocol and consisted in confining mosquitoes in acrylic chamber tubes lined internally with Whatman grade No. 1 papers that had been previously impregnated with malathion (840 l of a 2 g/liter solution, resulting in 0.1 g/m 2 in the paper) (85). Approximately 20 females, ϳ4 days posteclosion were exposed to the insecticide for 90 min and then transferred to insecticide-free rescue tubes. Mortality was evaluated 24 h later. Two to three replicates were used, and each experiment was repeated three times.

Statistical analysis
All analyses were performed with the GraphPad Prism statistical software package (Prism version 6.0, GraphPad Software, Inc., La Jolla, CA). Asterisks indicate significant differences (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; ns, not Redox-sensing Nrf2 in A. aegypti significant.), and the type of test used in each analysis is described in its respective figure legend.