Synergistic action of the transcription factors Krüppel homolog 1 and Hairy in juvenile hormone/Methoprene-tolerant-mediated gene-repression in the mosquito Aedes aegypti

Arthropod-specific juvenile hormones control numerous essential functions in development and reproduction. In the dengue-fever mosquito Aedes aegypti, in addition to its role in immature stages, juvenile hormone III (JH) governs post-eclosion (PE) development in adult females, a phase required for competence acquisition for blood feeding and subsequent egg maturation. During PE, JH through its receptor Methoprene-tolerant (Met) regulate the expression of many genes, causing either activation or repression. Met-mediated gene repression is indirect, requiring involvement of intermediate repressors. Hairy, which functions downstream of Met in the JH gene-repression hierarchy, is one such factor. Krüppel-homolog 1, a zinc-finger transcriptional factor, is directly regulated by Met and has been implicated in both activation and repression of JH-regulated genes. However, the interaction between Hairy and Kr-h1 in the JH-repression hierarchy is not well understood. Our RNAseq-based transcriptomic analysis of the Kr-h1-depleted mosquito fat body revealed that 92% of Kr-h1 repressed genes are also repressed by Met, supporting the existence of a hierarchy between Met and Kr-h1 as previously demonstrated in various insects. Notably, 130 genes are co-repressed by both Kr-h1 and Hairy, indicating regulatory complexity of the JH-mediated PE gene repression. A mosquito Kr-h1 binding site in genes co-regulated by this factor and Hairy was identified computationally. Moreover, this was validated using electrophoretic mobility shift assays. A complete phenocopy of the effect of Met RNAi depletion on target genes could only be observed after Kr-h1 and Hairy double RNAi knockdown, suggesting a synergistic action between these two factors in target gene repression. This was confirmed using a cell-culture-based luciferase reporter assay. Taken together, our results indicate that Hairy and Kr-h1 not only function as intermediate downstream factors, but also act together in a synergistic fashion in the JH/Met gene repression hierarchy.

Introduction Arthropod-specific juvenile hormones (JHs) are key regulators of a large array of physiological processes, including growth, metamorphosis and reproduction [1,2]. Methoprene-tolerant (Met), a basic helix-loop-helix (bHLH)/-Per-Arnt-Sim (PAS) domain protein, has been characterized as the insect JH receptor [3][4][5][6][7][8]. Additionally, Taiman (Tai), an insect homolog of the vertebrate steroid receptor co-activator (SRC), which is also a bHLH-PAS protein, has been identified as the Met obligatory partner [9,10]. The heterodimer formation of Met/Tai, necessary for successful transduction of the JH signal, is induced by the ligand-receptor (JH-Met) interaction [11]. In the nucleus, the JH receptor complex interacts with JH response elements (JHREs), usually containing the core E-box-like motif 'CACGTG', in regulatory regions of target genes leading to their activation [9,[12][13][14][15][16][17]. Details of the nuclear import of Met have also been investigated, revealing the identity of Met-interacting chaperone heat shock protein 83 (Hsp83), which in turn works together with the components of the nuclear pore complex facilitating Met nuclear translocation [18,19]. Additionally, recent studies have indicated the existence of a not-yet fully characterized membrane JH receptor that is responsible for Met phosphorylation, a necessary step for the activity of the JH receptor complex [20,21]. This mode of action involving JH has been confirmed for several genes, including early trypsin (ET), regulator of ribosomal synthesis 1 (RRS1), Hairy, and Krüppel-homolog 1 (Kr-h1) [2,10,12,16,17,22,23]. Some targets of the JH signaling pathway, such as Hairy and Kr-h1, are transcription factors (TFs) that in turn can regulate the expression of downstream genes [2,22,23].
In the mosquito A. aegypti, JH III (JH) is the principal hormone controlling the physiological maturation in newly eclosed adult females. The JH-governed developmental period spans from adult eclosion to blood feeding, termed post-eclosion (PE) phase, and is critical for subsequent reproductive maturation and successful egg production. The JH titer increases after adult eclosion, reaching a peak at 48-54h PE [24]. The mosquito PE phase is characterized by sequential waves of highly expressed fat body (FB) genes, which are modulated by differential JH titers [15]. Met plays a central role in JH-mediated PE gene expression, regulating a total of 2151 transcripts, of which 1613 are upregulated and 538 downregulated [22]. A bioinformatics approach has predicted that, unlike activation, Met-mediated gene repression is indirect and requires the involvement of additional downstream factors [15]. Indeed, we have identified the bHLH-Orange domain protein Hairy as an intermediate factor in JH/Met-mediated gene repression [22]. Hairy binds to target gene promoters (the E-box sequence or its variants) and recruits the co-repressor Groucho (Gro1), resulting in transcriptional repression of JH/Met/ Hairy target genes [22]. Hairy is a late PE gene, having a low expression level in newly eclosed mosquito FB, gradually increasing to reach a peak at around 60h PE and maintaining high expression levels throughout rest of the PE phase [15,22]. Gro1, one the other hand, is constitutively expressed throughout PE indicating that it is the recruitment of the protein, and not its availability, that plays a crucial role in Hairy-mediated gene repression downstream of JH/ Met [15,22].
Another intermediate factor that has been implicated in JH/Met gene repression is the C 2 H 2 zinc-finger TF Kr-h1 [25]. Kr-h1 has been characterized as an early inducible gene in the JH signaling pathway downstream of Met in Drosophila and Tribolium castaneum [26,27]. The JH-receptor complex directly induces Kr-h1 expression by interacting with JH response elements in the upstream regulatory region [8,9,[11][12][13][14]16]. Kr-h1 has been demonstrated to repress two key genes, Broad-complex (BR-C) and E93, during JH-regulated metamorphosis in various insects [26][27][28][29][30][31]. In adult female A. aegypti mosquitoes, Kr-h1 shows an expression pattern similar to that of Hairy and is characterized as a late PE gene [15]. As in the case for Hairy, mosquito Kr-h1 also functions downstream of Met in the JH-mediated gene-repression hierarchy [23]. Whether Kr-h1 and Hairy interact in this hierarchy is, however, not understood. Our comparative analysis of RNA interference (RNAi) PE fat body transcriptomes for Kr-h1 and Hairy has revealed a significant overlap, suggesting simultaneous involvement of these two Met-regulated factors in gene repression. Only a simultaneous RNAi knockdown of Hairy and Kr-h1 could fully phenocopy the effect of Met depletion. Our results have suggested a synergistic action of Kr-h1 and Hairy in JH/Met-mediated gene repression, a hypothesis that was verified using cell transfection assays.
Our findings provide an important mechanistic insight into the JH signaling pathway. JH is a structurally unique and arthropod-specific hormone, the chemical analogs of which have been routinely utilized as potent insecticides. Furthermore, JH is a crucial component of the mosquito reproductive cycle, which in turn forms the basis for acquisition and transmission of pathogens of such devastating diseases as malaria, dengue, chikungunya, yellow fever and zika virus. Thus, these findings might lead to the identification of potential novel targets for mosquito control.

Transcriptomic analysis of Kr-h1 depleted mosquito FB tissue
To investigate the role of Kr-h1 in JH/Met-mediated gene repression, we performed an RNAibased transcriptomic screen for this TF. Illumina RNA-seq technology was used to identify genes regulated by dsRNA-mediated knockdown of Kr-h1 (iKr-h1) in the female mosquito FB. The high-throughput experiments were performed as previously described [22]. Setting a cutoff log-fold change of more than +1 (2-fold change), a total of 231 transcripts were found to be upregulated in Kr-h1 RNAi-depleted mosquitoes when compared with the RNAi Luciferase (iLuc) control (S1 Dataset). Of the transcripts activated by Kr-h1 RNAi, 92% were also activated by the RNAi mediated knockdown of Met (Fig 1, S1 Dataset).
To verify the reproducibility of the result, two additional biological replicates of RNA-seq libraries were constructed for each RNAi treatment followed by sequencing. Results like that mentioned above were also obtained for the two repeats, with activation of 244 and 223 transcripts in iKr-h1 samples of replicates 2 and 3, respectively (S2 Dataset). Of the iKr-h1" transcripts from replicates 2 and 3, 91% and 90%, respectively, were also induced by the knockdown of Met (S2 Dataset). Comparison of transcripts activated by iKr-h1 in the three biological replicates revealed that 207 transcripts were common to all three samples (S1 Fig), indicating reproducibility of the parallel experiments.
The overlap between Met and Kr-h1 RNAi-depleted transcriptomes suggests a hierarchy in the JH-mediated repression of target genes. Kr-h1 has been implicated to play a role as an intermediate factor downstream of Met in gene repression by JH in various insects [26][27][28][29][30][31], including mosquitoes [23]. Our results agree with these data.

Analysis of genes co-repressed by both Hairy and Kr-h1
It is clear from previous publications and our present results that Kr-h1 and Hairy are two TFs that act downstream of Met in the JH gene-repression hierarchy [22,23]. We identified a large overlap (137 transcripts) between the Kr-h1-and Hairy RNAi-upregulated transcriptomes (Fig 2A and S3 Dataset). Of the iHairy and iKr-h1 upregulated genes, 44.1% (137/311) and RNAidepleted (iKr-h1) female A. aegypti fat body. RNAi-depleted Luciferase (iLuc) fat body transcriptome was used as a control. Venn diagram comparing iMet and iKr-h1 transcriptomes shows that 92% of RNAi Kr-h1 (iKr-h1) upregulated genes (212 genes) were also upregulated by iMet. iMet", indicates more than two-fold upregulation by dsRNA-mediated depletion of Met; iKr-h1", more than two-fold upregulation by dsRNA-mediated depletion of Kr-h1. https://doi.org/10.1371/journal.pgen.1008443.g001 Kr-h1 and Hairy synergistically repress JH-controlled mosquito genes Comparative analysis of iKr-h1 and iHairy transcriptomes shows a significant overlap between these two gene sets. (A) Venn diagram comparing iMet, iHairy and iKr-h1 transcriptomes revealed that 130 target genes were activated by both iHairy and iKr-h1 in the iMet background, indicating that the two factors acted either co-operatively or hierarchically in repression of JH/Met target genes. iMet", more than two-fold upregulation by dsRNA depletion of Met; iKr-h1", more than two-fold upregulation by dsRNA-mediated depletion of Kr-h1; iHairy", more than two-fold 59.3% (137/231), respectively, were found to be common among the transcripts regulated by the two TFs. Of these 137 transcripts, 130 were also found to be repressed by Met (Fig 2A and  S3 Dataset). Similar results were also observed in the other two biological replicates.
qRT-PCR analysis of the expression of Met, Hairy and Kr-h1 in iMet, iHairy and iKr-h1 samples suggests that neither Hairy nor Kr-h1 regulates the transcription of each other, ruling out a hierarchy between these two factors ( Fig 2B). Therefore, the observation that a significant number of genes downstream of Met are regulated by both Hairy and Kr-h1 led us to the hypothesis that Hairy and Kr-h1 act synergistically in JH/Met-regulated gene repression. A careful look at the fold changes of the target genes indicated widespread differences among the iMet, iHairy and iKr-h1 transcriptomes (Fig 3). In most of these Met/Hairy/Kr-h1 repressed genes, the intensity of activation as a result of Met knockdown was significantly higher than that in either iHairy or iKr-h1 treatments (Fig 3). The trend was also observed in the other two biological repeats.
We conducted further experiments to establish JH repression of the Met/Hairy/Kr-h1-regulated gene AAEL005093 (Clip-Domain Serine Protease). We first examined the effect of JH by treating newly eclosed female mosquitoes with JH III. qRT-PCR analyses revealed a clear repression of the tested gene AAEL005093 by JH in comparison with the acetone-treated (Solvent) or untreated (NT) control mosquitoes ( Fig 4A). Next, we utilized cycloheximide (CHX), a potent translational blocker, in the in-vitro fat body culture (Solvent, JH, JH+CHX, CHX) according to Saha et al. (2016) to test the necessity of intermediate factors in the JH-mediated repression [22]. The tested gene transcript level was high after solvent (acetone) treatment but was significantly reduced in the JH-treated samples ( Fig 4B). Addition of CHX to the JH-containing culture medium rendered the AAEL005093 gene unresponsive to the repressive action of this hormone (Fig 4B), clearly indicating the involvement of intermediate factor(s) downstream of Met in regulation of target genes. To investigate the involvement of specific Met downstream factors, we performed experiments combining dsRNA-mediated knockdown and upregulation by dsRNA-mediated depletion of Hairy. (B) qRT-PCR analysis of Met, Kr-h1 and Hairy mRNA expression in iMet, iHairy and iKr-h1 mosquito fat body. iLuc was used as control. The knockdown of Hairy did not impact the expression of Kr-h1 mRNA or vice versa, ruling out a hierarchy between the two factors. Error bars represent ± SD. � p < 0.05; �� p < 0.01; ��� p < 0.001.
https://doi.org/10.1371/journal.pgen.1008443.g002 in-vitro fat body culture, as previously described [22]. FBs from mosquitoes treated with dsRNA for Luc (control), Met, Hairy or Kr-h1 were incubated in the culture medium in the presence or absence of JH. As for iMet, knockdown of both Kr-h1 and Hairy resulted in a compromised JH action on the repression of the target gene AAEL005093 (Fig 4C), establishing that both factors are involved in the JH-mediated regulation of the target gene.
To probe the synergistic action between Kr-h1 and Hairy, we performed simultaneous knockdown of the two factors ( Fig 4D). The expression of AAEL005093 was induced in all the test samples-iMet, iKr-h1, iHairy and iKr-h1+iHairy in comparison with iLuc controls. However, the intensity of induction of AAEL005093 in iKr-h1 and iHairy samples was significantly lower than in iMet samples ( Fig 4D). However, the level of the target gene induction was comparable between iMet and iHairy+iKr-h1 samples, suggesting a possible synergistic action of Hairy and Kr-h1 factors downstream of Met in the JH signaling pathway (Fig 4D). Specifically, Kr-h1 and Hairy synergistically repress JH-controlled mosquito genes the expression of AAEL005093 in iKr-h1+iHairy samples was approximately 110% of that of iMet, 300% of iHairy and 400% of iKr-h1 ( Fig 4D). Efficiency and specificity of RNAi knockdowns in this experiment were confirmed using Western blot analysis ( Fig 4E). We further confirmed our results by testing a second gene (AAEL006978, protein-glutamine gamma-glutamyltransferase) from this gene set. Like AAEL005093, AAEL006978 is regulated by JH and requires intermediate factor downstream of Met for its repression (S2A and S2B Fig). Most importantly, AAEL006978 expression is impacted to comparable levels in iMet and iHairy +iKr-h1 samples, suggesting a synergistic action of these two factors in this gene as well (S2C Fig). The results of RNAi experiment for the two tested genes AAEL005093 and AAEL006978 conform to our RNAseq results (Fig 3 and S3 Dataset).

Ontology analysis of genes co-repressed by Kr-h1, Hairy and Met
Gene ontology analysis based on adjusted non-supervised orthologous groups (NOGs) for the 130 genes co-repressed by Kr-h1 and Hairy in a Met background was performed as described previously [22]. The genes were mapped to two major functional categories-Cellular Processes and Signaling, and Metabolism-and none was mapped to the third major functional category-Information Storage and Processing. Significantly overrepresented OGs  Table). Additionally, two transcripts were mapped to [W] extracellular processes under this broad category. Within Metabolism, the overrepresented groups are [P] inorganic ion transport and metabolism, and [Q] secondary metabolites biosynthesis, transport and catabolism (S3 Fig and S1 Table).

Characterization of the A. aegypti Kr-h1 binding site
C 2 H 2 domain containing Kr-h1 is expected to bind to specific DNA elements. Kayukawa et al. (2016) have demonstrated that Kr-h1 binds directly to a Kr-h1 binding site (KBS) in the 5 0 regulatory region of the Broad complex (BR-C) gene in silkworm Bombyx mori [30]. In that study, the authors defined a 30-bp sequence-5 0 GACCTACGCTAACGCTAAATAGAGTT CCGA 3 0 -as the KBS core region. A more detailed EMSA study revealed that this 30-mer KBS core region binds two Bombyx Kr-h1 protein molecules, indicating that it might harbor two closely placed Kr-h1 binding sites [30]. It was also suggested that each zinc finger domain of BmKr-h1 recognizes a different sequence in the KBS core region, due to the diversity of the amino acid residues of the α helix of each zinc finger domain in BmKr-h1 [30]. In a subsequent study, the same group characterized KBS consensus using chromatin immunoprecipitation sequencing analysis (ChIP-seq) [32]. When aligned with KBS sequences in the promoters of BmBr-C and BmE93A, the identified KBS consensus demonstrated that 13 nucleotides of the 15-bp-long consensus sequence were highly conserved. Also, the first 9 conserved bases were separated from the last 4 by a more degenerate base. With the knowledge that the 30-mer KBS core region binds two Bombyx Kr-h1 protein molecules in the promoter of BmBR-C and that each zinc finger domain of BmKr-h1 might recognize a different sequence in the KBS core region [30], we wanted to identify the Kr-h1 binding site in A. aegypti (AaKBS). In particular, we were interested in detecting the AaKBS within the upstream regions of the Met/Kr-h1/ Hairy regulated genes.
First, we analyzed 5-kb 5´upstream regions of the Met-and Kr-h1-repressed genes. We used an 8-base sliding window to search 5-kb upstream regions of the upregulated transcripts that were activated after the depletion of both Met and Kr-h1. These regions were scanned for 23 different 8-mers from within the 30-mer KBS and their reverse complementary sequences, and the frequency of their occurrences (hits) was plotted ( Fig 5A). The decision to divide the 30-mer KBS into 8-mers was based on the conservation demonstrated in the alignment between the BmKr-h1 ChIP-seq consensus and KBS sequences on the promoters of BmE93A and BmBR-C [32]. It was observed that for 12 (KBS10 -KBS21) out of the 23 different 8-mers, the number of putative occurrences within the dataset of 5´upstream regions was statistically significant, that is the observed values were significantly higher than the values expected by random chance. Motif "TAAATAGA" (KBS16) and its reverse complementary sequence "TCTATTTA" scored the maximum number (61) of hits, suggesting that this sequence may be enough for binding Kr-h1 ( Fig 5A).
Next, we checked regions 5-kb upstream of the Met/Kr-h1/Hairy downregulated genes in a similar way to that described above. It was observed that the number of hits for the same 8-mers (KBS10-KBS21) were statistically significant (S4 Fig). This suggested that AaKr-h1 may bind to any of these 12 KBSs or a combination of these putative sites depending on the actual length of the binding site. Kr-h1 and Hairy synergistically repress JH-controlled mosquito genes To test A. aegypti Kr-h1 binding to the characterized KBS site, we used oligonucleotide sequences from the promoter region of a Met/Kr-h1/Hairy-repressed gene, AAEL005093, which carries the reverse complementary sequence 'TCTATTTA' of KBS16 that was shown to record the maximum number of hits ( Fig 5A) in our bioinformatics analysis. We selected this promoter for EMSA analysis because most of the sequences on the right and left flanking regions of this core putative KBS did not match the 30-mer reported by Kayukawa et al. (2016) [30], other than a guanine "G" which is to the right of the last adenine "A", essentially making a 9-mer core KBS that is a combination of KBS16 and KBS15. This presumed core AaKBS 'TCTATTTAG' along with 10-bp flanks at both 3´and 5´ends of the KBS were then used for EMSA. We utilized A. aegypti fat body nuclear extracts, collected at 72 h PE, to perform these studies. Using the above-mentioned oligonucleotide harboring AaKBS, a strong binding was observed, as evident from the distinct band in the EMSA assay ( Fig 5B). Mutation of the AaKBS 'TCTATTTAG' to 'GACTGGGTT', a sequence not predicted to bind Kr-h1 and very different from the consensus AaKBS, resulted in complete disappearance of the band, indicating the absolute necessity of the KBS for the DNA-protein interaction. Competition assays with 50-fold excess of unlabeled specific probe, but not the nonspecific competitor SP1, resulted in the elimination of the observed band, thus indicating the specificity of the DNA-protein interaction ( Fig 5B). Next, we investigated the presence of A. aegypti Kr-h1 protein in the DNA-protein complex by utilizing our customgenerated polyclonal anti-Kr-h1 antibody. Inclusion of this antibody in the EMSA reaction resulted in a super-shift of the band, indicating the presence of Kr-h1 protein in the DNA binding complex (Fig 5B).
Thus, we have identified a 9-mer sequence from the upstream regulatory region of an A. aegypti Met/Kr-h1/Hairy target gene that interacts with Kr-h1 protein. The core binding sequence was predicted as a KBS in our bioinformatics search, and it is expected that other variations of this KBS would also interact with Kr-h1. Validations of all these putative KBS sites are beyond the scope of this investigation.

Distance correlation between Kr-h1 and Hairy binding sites in target gene promoters
We performed a bioinformatics analysis for the KBS and the Hairy Binding Sites (HBS, characterized in Saha et al. 2016 [22]) within the upstream regions of the Met/Hairy/Kr-h1-repressed genes to check whether we could detect any distance correlation between these two sites. In 92% of the genes analyzed, the distance between the two sites was less than 1kb. Interestingly, the two sites were separated by fewer than 300 bp in 70% of the genes (S2 Table). There was no clear orientation bias for the sites, as HBS was downstream of KBS in 53% of the genes and KBS was downstream of HBS in 47% (S2 Table). These results further support our hypothesis regarding synergistic action of Kr-h1 and Hairy in JH-mediated repression of target genes.

Synergistic action of Kr-h1 and Hairy in the repression of target genes
In order to test the hypothesis about the synergistic action of Hairy and Kr-h1 in the repression of JH/Met target genes, we performed cell culture-based luciferase reporter assays. We used the regulatory region 1-kb 5´upstream of the Met/Kr-h1/Hairy-repressed gene AAEL005093 that harbors both KBSs and HBSs, within 255 bps of each other (Fig 5C). This upstream region was cloned into the luciferase reporter vector pGL3basic (AAEL005093 1kb -Luc). The fulllength A. aegypti Kr-h1 was amplified by PCR and cloned into pAc5.1 vector with a C-terminal Myc tag (Kr-h1-Myc). Previously cloned Hairy-Flag and Gro1-V5 [22] were also utilized in this study. Gro1-V5 was included because it has been demonstrated to be a necessary co-repressor for the successful functioning of Hairy in A. aegypti mosquitoes [22]. A 15-to 20-fold basal induction in the reporter activity was observed when AAEL005093 1kb -Luc was transfected into Drosophila melanogaster S2 cells, rendering this experimental system feasible to address gene repression (Fig 6A). Co-transfection of either Kr-h1-Myc or Hairy-Flag and Gro1-V5 along with AAEL005093 1kb -Luc resulted in significant repression in the promoter activity ( Fig 6A). However, the level of repression was dramatically affected when 100 ng of Hairy-Flag, Gro1-V5 and Kr-h1-Myc was transfected together with AAEL005093 1kb -Luc, reducing the luciferase signal to almost background levels (Fig 6A).
To further verify our finding about the synergistic action of the two factors, we mutated Kr-h1 (KBS-5´TCTATTTAG3´) and Hairy (HBS-5´CACGTG3´) binding motifs in the promoter of the AAEL005093 gene. The binding sites were mutated by inserting restriction sites in place of HBS (AAEL005093 1kb ΔHBS-Luc) or KBS (AAEL005093 1kb ΔKB-S-Luc), or both (AAEL005093 1kb ΔKBSΔHBS-Luc). Hairy-Flag, Gro1-V5 and Kr-h1-Myc were co-transfected with either AAEL005093 1kb -Luc (as control) or its mutant versions (AAEL005093 1kb ΔHBS-Luc, AAEL005093 1kb ΔKBS-Luc and AAEL005093 1kb ΔKBSΔHB-S-Luc). In the control AAEL005093 1kb -Luc transfection, a more than tenfold repression was observed in the presence of Hairy, Gro1 and Kr-h1 (Fig 6B). A partial loss of repression in the promoter activity was noted when either AAEL005093 1kb ΔHBS-Luc or AAEL005093 1kb ΔKBS-Luc reporter plasmids were co-transfected along with Hairy-Flag, Gro1-V5 and Kr-h1-Myc ( Fig 6B). However, transfection of S2 cells with the reporter plasmid AAEL005093 1kb ΔKBSΔHBS-Luc harboring mutations for both HBSs and KBSs in the presence of Hairy-Flag, Gro1-V5 and Kr-h1-Myc resulted in a complete loss of repression of the promoter activity (Fig 6B). The observed luciferase signal under the above experimental condition is comparable to the control samples transfected with the non-mutated reported plasmid AAEL005093 1kb -Luc only, and lacked expressed Hairy, Gro1 or Kr-h1. The schematic representation of the reporter constructs with the HBSs and KBSs and the various mutations used for the luciferase transfection assays are provided in Fig 6C. The expression of tagged proteins Hairy-Flag, Gro1-V5 and Kr-h1-Myc in the cell culture system was confirmed by means of immunoblots using anti-Flag, anti-V5 and anti-Myc antibodies (S5A and S5B Fig). Using the same experimental approach, a synergistic action of Hairy and Kr-h1 on the promoter of the Met/Hairy/Kr-h1 target gene AAEL006978 has also been demonstrated (S6A, S6B and S6C Fig), providing further support for our hypothesis.

Hairy and Kr-h1 binds to specific sites in the promoter region of the target gene
To confirm the binding of the two factors Hairy and Kr-h1 to specific binding sites in the promoter of Met/Hairy/Kr-h1 target gene, we performed chromatin immunoprecipitation (ChIP) assays for the gene AAEL005093. The experiments were conducted in the cell culture system, using Drosophila S2 cells transected with AAEL005093 1kb -Luc, Kr-h1-Myc and Hairy-Flag. ChIP assays were performed using Aedes anti-Hairy and anti-Kr-h1 antibody following a modified protocol specifically designed for S2 cells [33]. IP with an IgG antibody was used as control. Quantifications were performed by qRT-PCR using specific primers targeting the HBS and KBS locations in the cloned promoter of gene AAEL005093. An enrichment of HBS but not the KBS region was observed in ChIP assays using anti-Hairy antibody (Fig 7A). Similarly, use of anti-Kr-h1 antibody resulted in a substantial enrichment of the KBS site but not the HBS site (Fig 7B). These results clearly confirmed the physical binding of Hairy-HBS and Kr-h1-KBS sites in the target gene promoter. A primer targeting the plasmid backbone was utilized as control for both the experiments (Fig 7A and 7B).

Effect of Hairy and Kr-h1 double knockdown on ovarian follicle growth
Kr-h1 is a downstream factor of Met in the JH signaling pathway for gene repression. Thus, we checked the ovarian primary follicle length at 72h PE in iKr-h1 mosquitoes. A reduction (approximately 20%) in the length of primary follicles was observed in iKr-h1 mosquitoes when compared with iLuc, indicating the important role of Kr-h1 in PE follicle maturation (Fig 8). However, RNAi knockdown of Met had a more dramatic effect with an observed 52% reduction of follicle length (Fig 8). The iKr-h1 had a phenotype very similar to that observed for iHairy mosquitoes, where a difference was observed between follicles of iHairy and iMet mosquitoes [22]. We then checked the follicle length of iKr-h1+iHairy double knockdown mosquitoes. A greater reduction in follicle length was observed in iKr-h1+iHairy samples than in single knockdowns of iHairy or iKr-h1 samples, suggesting synergistic action of these factors (Fig 8). However, in iMet mosquitoes, a much greater reduction was observed. This could be due to the central role of Met in the JH signaling pathway. In addition to acting through Hairy and Kr-h1, Met regulates multiple genes directly [15,17].

Discussion
In female mosquitoes, JH is responsible for developmental events after adult eclosion during the PE phase. This JH-controlled phase is critical for a female mosquito to become competent for blood feeding and subsequent egg maturation. PE is characterized by sequential waves of gene expression, the most important of which are the early (EPE) and late (LPE) gene cohorts, exhibiting their respective peaks at 6-12 h and 72 h PE, which correlate with low and high JH titers, respectively [15,24]. At its high titer, JH downregulates the expression of the EPE genes and upregulates the LPE genes. Zou et al. (2013) [15] have shown that the JH receptor Met mediates induction of the LPE genes via direct binding to their promoters, and later work confirmed the direct upregulation of selected LPE genes by Met [34]. Zou et al. (2013) also postulated that Met downregulates EPE genes indirectly requiring the involvement of intermediate TFs [15]. Indeed, our recent study demonstrated that the Met-regulated TF Hairy acts as a DNA-interacting transcriptional repressor that recruits a co-repressor Groucho and downregulates EPE gene expression [22]. Here, we have demonstrated the role of the zinc-finger domain protein Kr-h1 as a transcriptional repressor of a larger cohort of EPE genes downstream of Met. In a temporal microarray analysis, Kr-h1 has been classified as the LPE gene [15]. The molecular analysis utilizing in-vivo and in-vitro experiments has established Kr-h1 as an intermediate factor of the JH/Met gene-repression hierarchy. Furthermore, we were able to characterize a mosquito Kr-h1 binding site from our Kr-h1 repressed gene set, which was validated by EMSA analysis. It is worth mentioning that the same 8-mers (KBS10-KBS21) are overrepresented in both Met/Kr-h1-and Met/Hairy/Kr-h1-repressed gene sets. Also, the binding site we identified and subsequently validated matches 9 bp of the 30-bp KBS core, identified in the promoters of B. mori genes [30,32]. Recently, using a chromatin immunoprecipitation (ChIP)-based approach, Ojani et al. (2018) identified a 12-bp-long DNA binding Kr-h1 and Hairy synergistically repress JH-controlled mosquito genes site for A. aegypti Kr-h1 [23]. The sequence shares similarities with the 30-mer KBS core from promoters of B. mori genes [30,32] on which our bioinformatics analysis is based. The predicted sequence sits upstream of the 9-mer KBS that we identified and validated through the EMSA analysis. In their analysis, Ojani et al. (2018) did not rule out the possibility of the 9-mer sequence "TAAATAGA" working as a Kr-h1 binding site [23].
As a component of the JH signaling pathway, the role of Kr-h1 in reproduction has been investigated in several insects [35]. JH plays various roles in insect reproduction, and unlike mosquitoes where it guides a developmental phase (PE) in preparation for vitellogenesis, it induces vitellogenesis and oogenesis in other insects. In the orthopteran Locusta migratoria, JH-Met-Kr-h1 controls vitellogenesis and oocyte maturation in females [36]. In this insect, Kr-h1 is involved in regulation of vitellogenin gene expression in the fat body, lipid accumulation in primary oocyte, development of follicular epithelium, and ovarian growth [36]. However, in the hemipteran linden bug Pyrrhocoris apterus and the coleopteran beetle T. castaneum, no clear involvement of Kr-h1 in reproduction could be ascertained [37,38]. Ojani et al. (2018) have implicated Kr-h1 in gene activation and repression in A. aegypti female mosquitoes [23]. Our data corroborates their observations by showing that Kr-h1 plays a crucial role in We report hereby that the two Met downstream factors, Hairy and Kr-h1, act synergistically coordinating the repression of a subset of JH/Met-regulated genes. Several lines of evidence suggest this interaction. Gene transcripts were induced at different levels by the dsRNA-mediated depletion of Met, Hairy or Kr-h1. Simultaneous knockdown of Kr-h1 and Hairy resulted in levels of upregulation of common target genes that are comparable to those for iMet. This led us to hypothesize for a possible synergistic action of these two factors in co-repression of a subset of JH/Met-regulated genes. Observation from bioinformatics analysis showed that within the promoters of about two-thirds of the Met/Hairy/Kr-h1-repressed genes, with both KBSs and HBSs, the two sites were very close to each other. The proximity of the two sites was not observed in the promoters of the Met/Hairy and the Met/Kr-h1 repressed gene sets, further strengthening our hypothesis of synergistic action. Cell culture-based luciferase reporter assays confirmed the synergistic action of these two factors. Furthermore, the necessity of these factors for the synergistic repression was confirmed when the complete loss of a target gene repression could only be observed with the mutation of both HBSs and KBSs in the regulatory region of the said gene. Taken together, our findings have revealed a novel aspect of the JH gene-repression hierarchy by demonstrating that Hairy and Kr-h1 synergistically repress a subset of JH/Met downregulated target genes. It appears that the synergistic action of Kh-h1 and Hairy is dictated by a target gene structure and requires the presence of corresponding binding sites for these two Met downstream TFs in regulatory regions of common target genes. It should be pointed out that Kr-h1 and Hairy together regulate about 20% (329 of 1613 total) of Met repressed genes (Fig 2A), indicating that there are additional factors involved in the JH/Met gene repression pathway, the identity of which remains unknown. This observation further confirms that gene regulation by JH is a complex phenomenon requiring a network of downstream factors.
Synergistic or cooperative regulation of gene expression is a widely occurring phenomenon [39][40][41][42][43]. In most cases, TFs responsible for synergistic or cooperative regulation of gene expression co-occupy regulatory regions of common target genes. Furthermore, synergistic TF action takes place in both gene activation [39,41,44,45] and gene repression [43,44,46,47]. In Drosophila, members of the ETS family of TFs have been shown to collaboratively repress the even skipped gene [47]. Interestingly, the co-repressor Groucho, which is a component of the Hairy/Kr-h1-mediated synergistic suppression of JH/Met target genes, is also required for even skipped repression [47]. Additionally, the Krüppel-like family of TFs has been implicated in the synergistic regulation of gene expression in conjunction with other TFs [45].
The TFs that are involved in synergistic or cooperative gene regulation are often observed interacting directly with each other [39,48]. During Drosophila development, many TF-binding pairs are located in relatively short distances from each other on the DNA. In-vitro protein-protein binding experiments have shown that more than 65% of these TF pairs directly bind to each other, with some of them implicated in cooperative gene regulation [48]. Whether Kr-h1 and Hairy directly interact after their binding to a common target gene promoter remains to be determined.

Experimental animals
A. aegypti mosquito larvae were cultured at 27˚C in water supplemented with a mixture of yeast and rat chow (1:1 ratio). Adult mosquitoes were maintained in specifically designed cages at 27˚C, 86% humidity, and supplied with unlimited access to 10% (wt/vol) sucrose solution and water. All dissections were performed in Aedes physiological solution at room temperature [49]. Four-day-old adult females were blood fed on white Leghorn chicken. All procedures for the use of vertebrate animals were approved by the University of California, Riverside, Institutional Animal Care and Use Committee.

dsRNA-mediated gene silencing
dsRNA-mediated knockdown of specific target genes was done following methods described by Zou et al. (2013) [15]. Briefly, dsRNA was synthesized using the MEGAscript kit (Ambion). The bacterial luciferase gene was used to generate control iLuc dsRNA. 0.6-0.8 μg of the desired dsRNA was injected into the thorax of cold-anesthetized female mosquitoes 1-day PE using the Picospritzer II (General Valve). For simultaneous knockdown of Hairy and Kr-h1, equal amounts (in μg) of both dsRNAs were mixed and the concentration of the dsHairy +dsKr-h1 was adjusted to 0.6-0.8 μg. Samples of the mosquito fat body (abdominal wall with adhered fat body) were collected 4 days post-injection. qRT-PCR was performed to verify the knockdown efficiency and specificity. All primers used are listed in S3 Table. Illumina

RNA-seq library preparation, sequencing and bioinformatics analysis
The dsRNA-mediated knockdown, fat body sample collections and subsequent library preparation and sequencing for iKr-h1 samples were performed as previously described [22]. RNAi for luciferase (iLuc) served as a control. Briefly, ten mosquito fat bodies were dissected at 4 days post-injection followed by RNA extraction using TRizol (Gibco/BRL). RNA-seq libraries were prepared using TruSeq RNA Library Preparation Kit (Illumina). Sequencing reactions were performed at the University of California Riverside Genomics Core Facility. Three independent biological replicates of the experiment were performed, resulting in three RNA-seq libraries per treatment. The obtained reads were aligned with Bowtie2 against A. aegypti transcript sequence database (AaegL1.3 geneset; Vectorbase). The resulting count table was transformed into values representing fragments per kilobase pair of transcripts per million fragments mapped (FPKM). Relative transcript abundance of iLuc and iKr-h1, defined as the sum of FPKM values of the two treatment samples, was sorted from highest to lowest. From the first set of RNA-seq experiments, 10,000 transcripts with the highest abundance were selected for further analysis. The differentially expressed transcripts were defined by a twofold increase or decrease of FPKM values.

Ontology-based functional analysis
In order to examine the ontology of the genes, Evolutionary Genealogy of Genes: Non-supervised Orthologous Groups (eggNOG) database, version 3.0, was used, which is constructed through identification of reciprocal best BLAST matches and triangular linkage clustering [50]. The OGs are annotated with functional descriptions along with functional categories, which were derived from the original Clusters of Orthologous Groups/Eukaryotic Orthologous Groups categories. Adjustments of non-supervised orthologous groups (NOGs) were performed as described in Saha et al. (2016) [22].
MiniElute cleanup kit (Qiagen) for further processing. RNA was treated with DNase I (Invitrogen), after which cDNAs were synthesized from 2 μg of this total RNA using the Omniscript Reverse Transcriptase kit (Qiagen). PCR was performed using the Platinum High Fidelity Supermix (Invitrogen). qRT-PCR was performed using the iCycler iQ system (Bio-Rad) and an IQ SYBR Green Supermix (Bio-Rad). Quantitative measurements were performed in triplicate and normalized to the internal control of actin mRNA for each sample. Real-time data were collected and exported to Excel (Microsoft) for analysis. RNA extraction, cDNA synthesis and subsequent qRT-PCR analysis were done as previously described [15]. All the primers used are listed in S3 Table.

In-vivo JH treatment
Abdominal applications of JH to newly eclosed adult female mosquitoes (3 h PE) were performed as previously described [22]. Specifically, a 0.3-μL aliquot of 1 μg/mL JH III (Sigma) or solvent (acetone) was topically applied to the abdomen of newly emerged female mosquitoes (3 h PE). Sample collections were performed at 8h after hormonal treatment. The efficiency of hormonal treatment was tested by verifying the induction of the established JH-activated genes Hairy and Kr-h1 using qRT-PCR.

In-vitro fat body culture
Fat body tissue cultures with JH III and CHX supplementations were performed as previously described [22]. For RNAi-fat body tissue culture tandem experiments, dsRNA-mediated gene silencing was performed as described above, followed by fat body collections at 4 days postinjection. The dissected fat bodies were used for in-vitro fat body culture, with and without JH.

Western blotting
cDNA encoding full-length A. aegypti Kr-h1 (AAEL002390) was cloned into pcDNA3.1 (Thermo Fisher, USA) and was sent out for antibody production. Sub-cloning, bacterial expression and the subsequent antibody production in mouse were performed by Genscript (USA). Antiserum against Kr-h1 was purified by means of affinity chromatography using the ImmunoPure IgG purification kit (Pierce). In addition to Kr-h1, previously generated antiserum against A. aegypti Hairy and Met were utilized in this study [22]. Fat-body protein sample collections and western blots were performed as previously described [51]. A specific mouse monoclonal antibody against β-actin (Sigma, USA) was used as loading control.

Electrophoretic mobility shift assay (EMSA)
The DNA oligonucleotides containing the core motif and flanking region were annealed, purified and end-labeled with [γ-32 P] ATP, as described previously [15]. Nuclear extracts from the mosquito fat body 48h PE were prepared using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific, Chino, CA) and were subsequently utilized for the EMSAs. Binding reactions were performed using the Gel Shift Assay System (Promega) and the DNAprotein complex was resolved on 5% TBE Criterion Precast Gel (Bio-Rad). Following electrophoresis, the gel was dried, exposed to phosphor imaging screens and visualized by means of autoradiography using the Personal Molecular Imager (Bio-Rad). For competition assays, 50-fold unlabeled specific motif or unlabeled SP1 (non-specific competitor oligonucleotides; Promega) motif was incubated with nuclear extract for 10 min and then further incubated with labeled motif for 20 min. Generated polyclonal A. aegypti Kr-h1 antibodies were used to test the presence of the TF in the observed DNA-protein complex.

Luciferase reporter assays
A. aegypti Kr-h1 with a C-terminal Myc-tag was sub-cloned into the pAc5.1 vector (Kr-h1-Myc) (Thermo Scientific, Chino CA). Previously cloned A. aegypti Hairy-Flag and Gro1-V5 were also used in this experiment [22]. A 1-kb promoter regions of the target genes AAEL005093 and AAEL006978, both harboring KBSs and HBSs, was PCR amplified and cloned into reporter vector PGL3basic (Promega), followed by sequence confirmation (AAEL005093 1kb -Luc and AAEL006978 1kb -Luc). Kr-h1 or Hairy binding motifs in both the cloned promoters were mutated by incorporating restriction sites in place of the defined KBS (AAEL005093 1kb ΔKBS-Luc and AAEL006978 1kb ΔKBS-Luc) or HBS (AAEL005093 1kb ΔHB-S-Luc and AAEL006978 1kb ΔHBS-Luc), respectively. Mutation of both KBSs and HBSs was achieved by sequential incorporation of restriction sites replacing the binding motifs (AAEL005093 1kb ΔKBSΔHBS-Luc and AAEL006978 1kb ΔKBSΔHBS-Luc). Transient transfection of cultured Drosophila S2 cells was performed using the FuGENE HD reagent (Promega) following manufacturer's instructions. 100ng of desired reporter plasmids and 10ng of the control Renilla luciferase reporter vector pCopia were co-transfected into the S2 cells along with 200ng each of the expression plasmids Kr-h1-Myc, Hairy-Flag and Gro1-V5, in various combinations. The total concentration of transfected plasmid in each well was normalized by adding the empty expression vector pAc5.1. Synthesized dsRNA samples for DmKr-h1, DmHairy and DmGroucho were supplemented in the transfection mixture as required, in order to negate the effects of endogenous factors on our experiment. Luciferase assays were performed using the Dual Luciferase Assay Kit (Promega), as previously described [22].

Chromatin immunoprecipitation (ChIP) assay
ChIP assays were performed in Drosophila S2 cells as previously described [33]. Briefly, 5-10×10 6 transfected S2 cells were fixed with 1% formaldehyde in tissue culture media for 10 min at room temperature followed by addition of glycine (0.125M) in order to stop the reaction. After washing in ice cold PBS (1X) for 2 mins, cells were re-suspended in cell lysis buffer (5 mM pH 8.0 PIPES buffer, 85 mM KCl, 0.5% Nonidet P40, and protease inhibitors cocktail) followed by the addition of nucleus lysis buffer (50 mM pH 8.1 Tris-HCl, 10 mM EDTA, 1% SDS and protease inhibitors cocktail). After a 20 min incubation at 4˚C, the nuclear extract was sheared by sonication on ice for 5 min (pulsed 10 times for 30 s with 30 s intervals) using Athena Ultrasonic Processor (Probe Sonicator) ATP-150; Athena Technology Inc. The chromatin solution was then clarified by centrifugation at 14,000 rpm for 10 mins at 4˚C. For immunoprecipitation 2μl of each Aedes anti-Hairy and anti-Krh1 were incubated with the chromatin for 2h followed by overnight binding to 50μl pre-washed protein A agarose beads at 4˚C with shaking. Anti-IgG antibody was used as mock control. The beads were washed three times with wash buffer 1 (0.1% SDS, 1% Triton, 2 mM EDTA, 20 mM pH 8.0 Tris, and 150 mM NaCl); three times with wash buffer 2 (0.1% SDS, 1% Trition, 2 mM EDTA, 20 mM pH 8.0 Tris, and 500 mM NaCl); and two times with wash buffer 3 (10 mM pH 8.1 Tris, 1 mM EDTA, 0.25 M LiCl, 1% NP40, and 1% sodium deoxycholate). The immunoprecipitated DNA was eluted from the beads in 0.1 M NaHCO 3 and 1% SDS followed by revrese cross-linking by incubating overnight at 65˚C. DNA was extracted by phenol-chloroform method using ethanol precipitation. Quantifications were performed by qRT-PCR and represented as % of input DNA. A primer pair targeting the plasmid backbone was used as control. The primers used are mentioned in the S3 Table. Supporting information Luciferase reporter assays after co-transfection of expression vectors Hairy-Flag and/or Gro1-V5 and/or Kr-h1-Myc along with reporter construct AAEL006978 1kb -Luc. Treatments with no input DNA and empty expression vector served as controls. Repression of promoter activity was observed when either Kr-h1-Myc or Hairy-Flag was overexpressed in the cell culture system. Co-repressor Gro1-V5 was required for the successful functioning of Hairy. However, overexpression of both Kr-h1-Myc and Hairy-Flag (along with Gro1-V5) dramatically enhanced the intensity of repression of luciferase activity. Error bars represent ± SD. � p < 0.05; ��� p < 0.001. (B) Mutation analysis of AAEL006978 promoter by luciferase reporter assays. The Kr-h1 and Hairy binding sites (KBS and HBS, respectively) in the AAEL006978 1kb -Luc reporter construct were mutated either separately (AAEL006978 1kb ΔHBS-Luc and AAEL006978 1kb ΔKBS-Luc) or together (AAEL006978 1kb ΔKBSΔHBS-Luc) and co-transfected along with expression vectors Hairy-Flag and Gro1-V5 and Kr-h1-Myc. The repression in the luciferase activity observed with the AAEL006978 1kb -Luc construct was partially compromised with the mutation of either KBS or HBS. A complete loss of repression in the promoter activity was observed when both KBS and HBS were mutated in the promoter upstream of the luciferase gene in reporter construct. Error bars represent ± SD. � p < 0.05; ��� p < 0.001.  Table. List of primer sequences utilized in this study. (XLSX)