The amino acid sequence of Culex quinquefasciatus MRJP1 (CPIJ008700) was extracted from Vectorbase (https://vectorbase.org/vectorbase/app/) and fed into NCBI blastp search (https://blast.ncbi.nlm.nih.gov/Blast.cgi) to identify the potential evolutionary origin of this protein. Amino acid sequences from the top dipteran matches were extracted and used for comparison and determination of phylogenetic relationships among dipteran species. As Drapeau et al. (14) indicate that MRJP1 is closely-related to the yellow gene in insects, the amino acid sequences of yellow from Culex quinquefasiatus, extracted from Vectorbase, and yellow and its associated paralogs in Drosophila melanogaster, extracted from FlyBase, were used for the phylogenetic analysis. The following amino acid sequences were used to make the phylogenetic tree: MRJP1 from Cx. quinquefasciatus (XP_001850268.2), MRJP1 from Apis mellifera (NP_001011579.1), MRJP1 from Aedes aegypti (XP_021693794.1), MRJP1-like from Sabethes cyaneus (XP_053682299.1), MRJP1-like isoform X2 from Anopheles funestus (XP_049285312.1), yellow from Cx. quinquefasiatus (CQUJHB005397), yellow from D. melanogaster (FBgn0004034), yellow-like from Ae. albopictus (XP_019534039.2), yellow-like from Anopheles maculipalpis (XP_050069861.1), yellow-like from An. merus (XP_041787038.1), yellow-like from An. nili (XP_053679995.1), yellow-like isoform X1 from An. funestus (XP_049285311.1), yellow-like from An. moucheti (XP_052891110.1), yellow-like from An. marshallii (XP_053669181.1), yellow-like from An. stephensi (XP_035901855.1), yellow-like from An. cruzii (XP_052868860.1), yellow-like from An. aquasalis (XP_050087457.1). The following paralogs from D. melanogaster were also included: yellow-b (FBgn0032601), yellow-c (FBgn0041713), yellow-d (FBgn0041712), yellow-d2 (FBgn0034856), yellow-e (FBgn0041711), yellow-e2 (FBgn0038151), yellow-e3 (FBgn0038150), yellow-f (FBgn0041710), yellow-f2 (FBgn0038105), yellow-g (FBgn0041709), yellow-g2 (FBgn0035328), yellow-h (FBgn0039896).

Sequence alignment was performed using Clustal Omega (https://www.ebi.ac.uk/jdispatcher/msa/clustalo) and phylogenetic analyses were conducted using the maximum likelihood method in MEGA 11 software, with bootstrap values calculated from 500 trees (35). To characterize the protein composition and domains, we conducted an InterProScan search (https://www.ebi.ac.uk/interpro/search/sequence/).

A colony of Culex pipiens established in June 2013 from Columbus, Ohio (Buckeye strain) was used in this experiment. From the time that they were first instar larvae and until the adults were collected for analyses, mosquitoes were reared at 18°C and exposed to either long-day, diapause-averting conditions (photoperiod of Light: Dark 16:8 hr) or short-day, diapause-inducing conditions (photoperiod of L:D 8:16 hr). Larvae were reared in plastic containers filled with reverse osmosis water and were fed a diet of ground fish food (Tetramin Tropical Fish Flakes) according to the procedure described by Meuti et al. (36). To optimize our metabolomic procedure and allow us to acquire spectra from single mosquitoes, we conducted a preliminary experiment where mosquito pupae were transferred to cages that contained sugar water only. For all subsequent experiments, pupae from both the long and short-day photoperiods were randomly and equally divided into two cages, one that contained sugar water only (control) and one that contained 2 g of Royal Jelly (Starkish) that was dissolved in 1.5 mL of 10% sucrose solution (experimental treatment). Adult mosquitoes consumed their prescribed dietary treatments ad libitum (4 treatments total; n≅150 adults/treatment). One week after peak adult emergence, mosquitoes were euthanized and collected for experimental analyses.

Quantitative real time PCR (qRT-PCR) was used to assess how the abundance of CpMRJP1 was affected by supplementing the diet of adult mosquitoes with royal jelly. The procedure for qRT-PCR was based on (37). We designed primers for CpMRJP1 (CPIJ008700-RA) using Primer3 (Forward: TGAACGATCGTCTGCTGTTC; Reverse: TCCTCCCACATGGTATCGTT; (38). A standard curve verified that the primers met the MIQE guidelines (39). RNA was isolated from female mosquitoes (n=5 females/biological replicate; 5 biological replicates/rearing condition and feeding treatment) one week following adult emergence using TRIzol according to the manufacturer’s instructions. Complementary DNA (cDNA) was synthesized using the SuperScript III kit (Invitrogen), following the manufacturer’s instructions. All qRT-PCR reactions were done in triplicate on a 96-well plate using a CFX Connect qRT-PCR machine (BioRad). Each well contained a 10 μL reaction, consisting of 5 μL of iTaq Universal SybrGreen Supermix (BioRad), 400 nM of forward and reverse primers for either CpMRJP1 or our reference gene (Ribosomal Protein 19; RpL19; 39), 3.2 μL of molecular grade water, and 1 μL of cDNA. qRT-PCR reactions were completed through an initial denaturation at 94°C for 2 min, followed by 40 cycles at 94°C for 15 sec and 60°C for 1 min. A melt curve was run after each qRT-PCR reaction to ensure that only a single PCR product was produced. The abundance of CpMRJP1 transcripts was normalized to the abundance of RpL19 using the 2^-ΔCT method as previously described (40). We ran a model to ensure that the abundance of RpL19 mRNA did not differ significantly among dietary treatments (p = 0.778) and therefore was a suitable reference gene for this study.

To assess the diapause status of long and short day-reared females that had consumed sugar water (control) or diets that included royal jelly, we used two common markers of diapause: egg follicle length and fat content. One-week after adult emergence, we randomly euthanized 20 female mosquitoes from each dietary and rearing treatment and dissected their ovaries and egg follicles in a 0.9% saline solution (NaCl). The length of 10 egg follicles/female were measured at 200 times magnification using an inverted microscope (Nikon; n = 20 females/treatment). We also randomly selected eight, one-week-old females from the same cohorts and dietary treatments and measured the fat content in each female mosquito using a Vanillin lipid assay (41) that was modified to allow us to measure samples using a plate reader (42). The data were normalized by dividing the measured lipid content by the lean mass of the whole-body mosquito (lean mass = μg mosquito wet mass - μg of lipid).

As royal jelly is protein-rich, we also wanted to determine whether supplementing the diet with royal jelly affected the whole-body protein content of female mosquitoes. The protein content within eight, randomly selected individual female mosquitoes from the same cohort and dietary treatments as the lipid and egg follicle treatments was measured using a Bradford Assay kit (BioRad) following the manufacturer’s instructions (43). In brief, each female mosquito was weighed and then homogenized in 200 μL of a 10% ethanol solution. Samples were added in triplicate to a 96-well plate, and 250 μL of Quick Start Bradford 1X Dye Reagent (BioRad) was added to each well. The absorbance of each sample was measured using a FLUOStar Omega Microplate Reader. Measurements were normalized by dividing the protein content by the wet mass of each female mosquito (44).

In the field, diapausing females are able to survive for 3–6 months in without access to food (45). Therefore, we wanted to determine how dietary conditions affected the lifespan of female mosquitoes in the absence of food. Long and short-day reared mosquito pupae were placed in cages with access to sugar water or royal jelly (4 treatments total). One week after peak adult emergence, all male mosquitoes as well as the royal jelly or sugar-water food sources were removed from each cage. The remaining females (n = 70 – 75 females/treatment) were counted and allowed constant access to water. Every week thereafter we counted and removed dead females from the cage until no female mosquitoes remained.

The experimental protocol for all metabolomic analyses employed tissue extraction followed by Nuclear Magnetic Resonance (NMR) analysis and was based on the procedure described in Wu et al. (46). One mosquito was weighed and placed in a 2 mL microtube with approximately 750 mg of ceramic beads. For our initial experiment, 3 short-day reared, diapausing and 3 long-day reared, nondiapausing mosquitoes that had consumed sugar water only were used. For our second experiment, 10 independent female mosquitoes from each photoperiodic and dietary treatment, and from the same cohort as the egg follicle, fat content, and protein measurement experiments were randomly selected (n = 40 total). Mosquito samples were homogenized in 400 µL cold methanol and 85 µL water, and the homogenate was transferred to a separate tube without beads. Next, 400 µL chloroform and 200 µL water were added to the homogenate, which was then vortexed and centrifuged (2,000 rcf for 5 min at 4°C). The aqueous (methanol) layer was isolated and collected in a new 1.5 mL microtube before being dried in an evaporator. Deuterium oxide (heavy water), trimethylsilylpropanoic acid (TSP), and boric acid were added to the evaporated extracts and vortexed. The pH of the samples was manually adjusted to 7.4 and then transferred to 5 mm NMR tubes.

The metabolites in each mosquito sample were measured using NMR spectroscopy following the procedure detailed in (47). 1D ¹H NOESY spectra were obtained for the aqueous extracts. In addition, one ¹H-¹³C HSQC spectrum of a pooled sample was acquired for the four separate dietary and photoperiodic conditions according to (48). Avance III HD 800 and Ascend 850 MHz spectrometers, each with an inverse cryoprobe and z-gradients (Bruker BioSpin, Billerica, MA), were utilized to obtain NMR measurements, and resulting NMR spectra were analyzed as described in Newell et al. (47) and (49). Topspin 3.6.1 and AMIX 3.9.15 software (Bruker BioSpin, Billerica, MA) were used for preprocessing. The untargeted approach of spectral binning was chosen, where each spectrum is divided into small sections, or bins, which are subsequently analyzed. It should be noted that this approach provides broad coverage across various classes of metabolites, as compared to a targeted approach where only a predefined subset of metabolites of interest is quantified. In the initial experiment to optimize the protocol, 1D NMR spectra were binned with a bin width of 0.005 ppm. In contrast, for the second experiment that evaluated the effects of consuming diets that included royal jelly, the bin width was lowered to 0.003 ppm because these measurements were taken using a higher-frequency instrument (850MHz). Spectra were binned using the R package mrbin [Version 1.5.0; (50)}. Signals that showed large inter-spectra chemical shift differences were manually added to broader bins. Noise signals were automatically removed, and data was scaled using PQN (probabilistic quotient normalization) to correct for differences in sample mass and extraction efficacies. Each bin was then scaled to unit variance. Principal Component Analysis (PCA) models were generated to visualize metabolic data and screen for outliers regarding data quality. Signals of interest were identified using public databases and identifications were validated using the acquired HSQC spectrum and measurements of pure samples.

RNA interference (RNAi) was used to knock down CpMRJP1 mRNA to evaluate how this protein affects the diapause status of female mosquitoes. The procedure for RNA interference was based on the protocol detailed in (37). Double-stranded RNA (dsRNA) specific to CpMRJP1 and Beta-galactosidase from E. coli (β-gal; control) were synthesized using the Promega T7 RNAi Express Kit according to the manufacturer’s instructions. We selected β-gal dsRNA as a control because mosquitoes do not consume lactose and hence genes encoding β-gal proteins are absent from their genomes, and because we have previously used β-gal dsRNA as a control in RNAi experiments (37, 40). We designed primers to synthesize a 230 bp fragment of CpMRJP1 (CPIJ008700-RA) in Cx. pipiens using Primer3 (Forward: CACCGCCAAACCGAACAAAT; Reverse: TGAGCAGCCCAAAGTACAGG; 37), which served as the template to create dsRNA. On the day of adult emergence, 3 μg of either β-gal or CpMRJP1 dsRNA was injected into the thorax of long and short day-reared mosquitoes. Following injection, females were placed into small plastic containers (4.62 x 6.75 x 7.19 inches) where they consumed 10% sucrose solution ad libitium. To confirm gene knockdown, RNA was isolated using Trizol according to the manufacturer’s instructions from 3 biological replicates each containing 5 whole-body, female mosquitoes that were euthanized two days after dsRNA injection. cDNA was synthesized and qRT-PCR was conducted as described above (section 2.3), except that we used CpMRJP1 qRT-PCR primers that were previously designed by Sim et al. (18), and that after normalizing CpMRJP1 expression to the RpL19 reference gene, CpMRJP1 expression was again normalized to its expression in β-gal-dsRNA injected mosquitoes from the same photoperiodic condition (40).

To determine how CpMRJP1 dsRNA affected seasonal phenotypes, the egg follicle lengths (n = 20) and fat content (n = 8) of randomly selected females were measured ten days following dsRNA injection as described above (section 2.4). After feeding on sugar water for one week, the lifespan of β-gal or dsMRJP1 dsRNA-injected females (n = 37 – 49 females/treatment) in the absence of food was also measured as described above (section 2.5).

All data analyses were conducted in R version 3.3.3 (51). Two-way ANOVAs and Tukey’s post-hoc tests were used to determine whether dietary treatment and/or photoperiod significantly affected CpMRJP1 mRNA abundance, egg follicle length, lipid content, and protein content in female mosquitoes. Student’s T-tests were used to determine whether injecting CpMRJP1 dsRNA effectively reduced CpMRJP1 mRNA abundance, while two-way ANOVAs were used to determine whether dsRNA injection and photoperiod significantly affected the egg follicle length or fat content. A value of alpha < 0.05 was applied to discern statistical significance.

To determine how supplementing the diet with royal jelly and knocking down CpMRJP1 with RNAi affected the longevity of female mosquitoes in the absence of food, we used Cox-proportional hazards models (survival package) that included the effects of diet and photoperiod. Hazard ratios were obtained from these models as an estimate of the ratio between the risk of dying between photoperiodic and dietary treatments, where negative hazard ratios (HR) indicate a protective effect. Kaplan Meier survival curves were also plotted and used to provide the median survival time or the time at which 50% of the population was still alive as defined by (52, 53), and log-rank tests were used to determine significant differences between treatments.

For analysis of NMR metabolomics data, for each spectral bin, a general linear model (GLM) was created to account for the effect of diet (sugar water or royal jelly), photoperiod (long or short day-rearing conditions), and the interaction term between diet and photoperiod. To correct resulting p-values multiple testing, we used a False Discovery Rate (FDR) of 5% (54). Signals that were significant in the GLM analysis were tested for pairwise group differences using Tukey’s Honest Significant Difference (HSD) test. An Artificial Neural Network (ANN) was generated using R packages keras (version 2.11.1) and tensorflow (version 2.11.0). A dense network was generated with one input neuron per NMR bin (N=1823), 200 hidden neurons with ReLU activation function, and 2 output neurons, representing “short day” and “long day” rearing conditions. The ANN was trained on the data from the “sugar water” group using leave-one-out cross validation. Samples from the “royal jelly” group were then predicted using the trained ANN.

Pathway enrichment analysis was performed by downloading KEGG pathway maps of Culex pipiens pallens with KEGG code cpii (55) using the R package KEGGREST (version 1.36.3). Fisher’s Exact Test was used to assess the number of matching metabolites per pathway. A pathway uniqueness score u was calculated as follows: For each pathway, each observed metabolite was assigned the inverse of the number of pathways in which this metabolite occurs organism-wide, then the maximum value was chosen as the pathway uniqueness score. Therefore, higher values of u indicate pathways with more metabolites that uniquely occur in that respective pathway, and we further investigated pathways with u≥0.2.

To identify the evolutionary relationships of Cx. pipiens MRJP1 with MRJP/yellow homologues in other insects, we conducted a phylogenetic analysis using amino acid sequences of MRJP/yellow proteins from dipteran species. First, we conducted a BLAST search using the protein sequence of Cx. quinquefasciatus MRJP1 (CPIJ008700) and extracted the amino acid sequences of the top hits from dipteran species. We also obtained the protein sequences of Yellow in Cx. quinquefasciatus and D. melanogaster, as well as 13 Yellow-like proteins in D. melanogaster from Vectorbase (56). Multiple sequence alignment was performed to compare sequence identity between species. The sequence identity between Cx. quinquefasciatus MRJP1 and Yellow from D. melanogaster was found to be 23.41%, while it was 21.39% with Yellow from Cx. quinquefasciatus. This result is consistent with the high identity of 20–30% shared between MRJP and Yellow paralogs in D. melanogaster, suggesting a common evolutionary origin (57). Additionally, we performed a sequence identity comparison between MRJP1 proteins from Cx. quinquefasciatus and Apis mellifera to gain insights into the evolutionary and functional aspects of this protein. Our findings unveil that MRJP1 in these species share 129 identical amino acid residues, corresponding to a percent identity of 26.88%. Through alignment of the amino acid sequences between Yellow and MRJP1 in Cx. quinquefasciatus, D. melanogaster, and Apis mellifera, we identified a consensus conserved MRJP domain in all four sequences, indicating the preservation of the MRJP domain throughout insect evolution ( Figure 1A ).

We also conducted a functional analysis of Cx. quinquefasciatus MRJP1 using InterProScan and found that it belongs to the Royal jelly/protein yellow family in the InterPro database and the MRJP family in the Pfam database. This result is consistent with previous findings indicating that all Yellow proteins have a conserved MRJP domain (58). Subsequently, we used these sequences to create a phylogenetic tree to discover evolutionary relationships ( Figure 1B ). The percentage of replicate trees in which the associated H3 clustered together in 500 bootstrap replicates is shown adjacent to the branches. The tree was drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. In the resulting phylogenetic tree, Yellow proteins in Cx. quinquefasciatus and D. melanogaster were found to be closely related and clustered together with all other Yellow paralogs in D. melanogaster. As expected, Cx. quinquefasciatus MRJP1 closely clustered with MRJP/Yellow-like proteins from other dipteran species identified in the BLAST top matches. However, MRJP1 was found to be farther away from Yellow from Cx. quinquefasciatus and D. melanogaster and other Yellow paralogs from D. melanogaster as well as MRJP1 in A. meliferra ( Figure 1B ). The distant relationship between Yellow of Cx. quinquefasciatus and D. melanogaster and MRJP1 in Cx. quinquefasciatus revealed by our phylogenetic analysis suggests that MRJP1 likely has different functions and targets than Yellow in D. melanogaster.

The relative abundance of MRJP1 mRNA did not change significantly in response to dietary treatment or photoperiodic conditions ( Figure 2A ). A two-way ANOVA revealed that there was not a statistically significant interaction between the effects of diet and photoperiod on the abundance of CpMRJP1 mRNA (F_1,16 = 0.589, p = 0.458). Simple main effects analysis showed that diet did not have a statistically significant effect on CpMRJP1 abundance (F_1,16 = 0.025, p = 0.876). Contrary to previous experiments (18), there was no significant difference in the abundance of MRJP1 transcripts between females reared in long day or short-day conditions (F_1,16 = 2.67, p = 0.122), likely due to high levels of variation in CpMRJP1 transcript abundance in the short-day reared females.

Egg follicle length can be used to determine diapause status of female mosquitoes of Cx. pipiens (5,8), such that an average egg follicle length of less than 75 μm indicates diapause, follicle lengths between 75 and 90 μm indicates an intermediate state, and follicles greater than 90 μm indicates nondiapause (42). We used a two-way ANOVA to evaluate the effects of both photoperiod and diet on egg follicle size. Our model revealed that both photoperiod (F_1,76 = 1008.2, p < 0.001) and diet (F_1,76 = 95.32, p < 0.001) independently affected egg follicle length. Moreover, there was a significant interaction between photoperiod and diet on egg follicle length (F_1,76 = 78.4; p < 0.001). As expected, one week after adult emergence all long-day reared females that consumed sugar water (controls) were in a clear nondiapause state ( Figure 2B ; average egg follicle length of 99.5 ± 1.8 μm). However, 50% of long-day reared females that consumed diets including royal jelly had egg follicles that were characteristic of being in diapause, while the remaining females were in an intermediate state, and none of the long day-reared females that had consumed diets including royal jelly had egg follicle lengths that were large enough to be considered in a nondiapause state ( Figure 2B ). Therefore, consuming royal jelly caused a significant decrease in the overall average egg follicle length in long-day reared females (average egg follicle length of 75.4 ± 1.8 μm; Tukey’s post-hoc test, p < 0.001, 95% CI = [19.3, 28.9]). Females reared in short-day conditions had significantly smaller egg follicles compared to those reared in long-day conditions (Tukey’s post-hoc test, p < 0.001, 95% CI = [-43.7, -38.5]). All females reared in short-day conditions were in a clear diapause state ( Figure 1B ), regardless of whether they consumed sugar water only (46.9 ± 0.4 μm) or diets that included royal jelly (45.7 ± 0.3 μm); therefore dietary treatment did not effect the egg follicle length of dietary treatment did not significantly impact egg follicle length of short day-reared females (Tukey’s post-hoc test, p = 0.92).

We also used a two-way ANOVA to assess whether dietary treatment and/or photoperiod affected the fat content of female mosquitoes ( Supplementary Figure S2A ), as previous studies demonstrate that diapausing females accumulate significantly higher levels of fat than nondiapausing mosquitoes (37, 42). Our analyses revealed that neither photoperiod (F_1,28 = 0.378, p = 0.54) nor diet (F_1,28 = 0.567, p = 0.458) affected fat content. However, there was a slight but non-significant interaction between photoperiod and diet (F_1,28 = 3.91, p = 0.058). Short day-reared females that consumed sugar water had higher levels of lipid relative to those that consumed royal jelly, although this result was not statistically significant (SD RJ = 22.19 ± 2.46% lipid; SD SW = 29.48 ± 4.75% lipid; Tukey’s HSD, p = 0.822).

A two-way ANOVA of the effects of photoperiod and diet on protein content revealed that photoperiod (F_1,25 = 4.97, p = 0.035) but not diet (F_1,25 = 0.03, p = 0.865) significantly affected the protein levels within female mosquitoes. Our analyses also revealed that there was no significant interaction between photoperiod and diet on protein content (F_1,25 = 0.03, p = 0.576). Females reared in long-day conditions had roughly the same amount of protein whether they consumed sugar water or diets that included royal jelly ( Supplementary Figure S2B ; average protein content LD RJ = 12.16 ± 0.76 μg/mg; LD SW = 12.60 ± 1.41 μg/mg; Tukey’s HSD, p = 0.99). Dietary treatment did not significantly affect the protein content of short-day reared females (SD RJ = 10.22 ± 0.98 μg/mg; SD SW = 9.33 ± 1.32 μg/mg; Tukey’s HSD; p = 0.95). However, females reared in short-day conditions contained significantly less protein than long-day reared females (Tukey’s HSD, p = 0.034, 95% CI = [-5.04, -0.200]).

As diapausing females can survive for prolonged periods without without food (reviewed in 11), we also assessed the starvation resistance of mosquitoes reared in different photoperiodic conditions that consumed either sugar water or diets including royal jelly for seven days prior to food removal ( Figure 2C ). Our Cox proportional hazards model revealed that photoperiod significantly impacted survival time in the absence of food, such that short-day reared females were significantly less likely to die relative to long-day reared females ( Supplementary Table S1 ; z = -8.427; p < 0.001). Although consuming diets that include royal jelly did not significantly impact survival time relative to feeding on sugar water ( Supplementary Table S1 ; z = 1.48; p = 0.138), there was a strong interaction between photoperiod and diet, such that under short-day conditions, consuming diets that included royal jelly significantly increased the risk of death (z = 12.3; p < 0.001). The median survival time of long-day reared females that consumed sugar water was eight weeks of age, while the median survival time of sugar-fed, short-day reared mosquitoes was 15 weeks ( Figure 2C ), and the risk of dying was significantly lower for short-day reared females (Tukey’s HSD, z = 8.47, p < 0.001). However, the opposite was observed with female mosquitoes that consumed diets including royal jelly; the median survival time of long-day reared mosquitoes that consumed royal jelly was 7 weeks while the median survival time of short-day reared, royal jelly-fed mosquitoes was 3 weeks, showing that under short-day conditions consuming diets that include royal jelly significantly increased the risk of dying (Tukey’s HSD, z = 10.58; p < 0.001). Under long-day conditions, consuming royal jelly did not affect the risk of dying (Tukey’s HSD, z = 1.482; p = 0.4486), whereas under short-day conditions, consuming diets that included royal jelly reduced the median survival time by 12 weeks and significantly increased the risk of death (Tukey’s HSD, z = 13.27; p < 0.001).

Preliminary data collected during initial optimization of the metabolomics protocol suggests that there are largescale differences in the metabolic profile of diapausing and nondiapausing Cx. pipiens ( Supplementary Figure S2A ). A Principal Component Analysis (PCA) reveals partial group separations for both the photoperiodic conditions and consuming royal jelly, indicating that both factors affect the metabolism of female mosquitoes ( Supplementary Figure S2B ). It should be noted that PCA does not search for differences between groups of interest, but it rather separates samples according to signals of largest variance. Frequently in metabolomics studies, confounding factors obscure PCA group separations by adding to the overall variance of the dataset. Therefore, additional analyses were performed to find differential signals. General linear models (GLM) revealed that 168 out of 1823 spectral signals significantly changed in response to at least one of the following: photoperiod, diet, or their interaction ( Supplementary Table S1 ). Figure 3A shows a heatmap of all signals that significantly changed. Notably, nondiapausing mosquitoes that consumed sugar water exhibit strong metabolic differences compared to diapausing mosquitoes that also consumed sugar water ( Supplementary Table S1 ; Figure 3A ). Visual inspection revealed that mosquitoes reared under long-day, diapause-averting conditions switch to a “diapause-like” metabolic profile after consuming royal jelly. In contrast, females reared under short-day, diapause-inducing conditions switched to a “nondiapause-like” metabolic state after consuming royal jelly ( Supplementary Table S2 ; Figure 3A ). To confirm these qualitative observations, an artificial neural network (ANN) was trained on the sugar water group to predict long-day or short-day rearing conditions. Using leave-one out cross validation of each SW sample ( Supplementary Table S2 ) reveals that the ANN correctly predicted most short-day (“diapause”) and long-day (“nondiapause”) samples in the sugar water group. However, when using the ANN to predict the photoperiod of mosquitoes that had consumed royal jelly ( Supplementary Table S3 ), most short-day samples were predicted as being long-day, and most long-day samples were predicted to be short-day. The ANN results thus confirm that consuming royal jelly switches the metabolic profile of long-day reared females to be “diapause-like” and short-day reared females to be “nondiapause-like.”

Among the significant signals ( Supplementary Table S2 ), six metabolites could be unambiguously identified using existing information in NMR databases ( Supplementary Table S3 ; Figure 3B ). Short day-reared, diapausing mosquitoes that consumed sugar water and long day-reared mosquitoes that consumed royal jelly had significantly higher levels of pimelic acid, asparagine, and choline; but significantly lower levels of L-alanine, histidine, and glycogen, compared to long-day mosquitoes that consumed sugar water ( Figure 3B ; see Supplementary Table S3 for GLM coefficients and Tukey’s HSD p-values). Identical trends are seen in mosquitoes reared in short-day, diapause-inducing conditions that consumed sugar water (significant in all except asparagine). In contrast, short day-reared mosquitoes that consumed royal jelly showed opposite metabolic trends when compared to short-day, sugar-fed controls; however, this difference was only significant for choline and pimelic acid.

Metabolic pathway enrichment results indicate that several pathways were affected by photoperiod and diet. When plotting the negative decadic logarithm of the p-value versus the uniqueness score u, pathways of interest are expected to be found toward the top (p ≤ 0.05) or the right (u ≥ 0.2) in the resulting scatterplot ( Figure 3C ). Pathways of high interest were found to be Alanine, Aspartate and Glutamate Metabolism (p = 0.0044, u = 0.5), Biotin Metabolism (p = 0.036, u = 0.5), Starch and Sucrose metabolism (u = 1.0), Glycerohospholipid metabolism (u = 0.33), and Histidine Metabolism (u = 0.33). Some pathways such as Aminoacyl-tRNA biosynthesis, ABC transporters, Sulfur relay system, and D-amino acid metabolism were significant. However, these significant differences were caused by repetitive metabolic reactions in these pathways and/or non-specific reactions, therefore we chose to treat these results as statistical artifacts and do not further discuss them.

A knockdown confirmation analysis was performed to determine if CpMRJP1 dsRNA significantly reduced the abundance of CpMRJP1 mRNA transcripts ( Figure 4A ). We found that MRJP1 dsRNA significantly reduced the abundance of CpMRJP1 transcripts in both long-day (71% reduction in transcript abundance relative to β-gal dsRNA-injected controls; T = 4.75; p = 0.009) and short-day reared females (74% reduction in transcript abundance relative to β-gal dsRNA-injected controls; T = 7.40, p = 0.0018).

Knocking down CpMRJP1 dsRNA significantly affected egg follicle length in short-day reared females ( Figure 4B ). A two-way ANOVA was used to determine how photoperiod and dsRNA injection affected egg follicle length, and revealed that both photoperiod (F_{1, 76} = 394; p < 0.001) and dsRNA treatment (F_{1, 76} = 15.4; p = 0.0019) had significant effects. Moreover, there was a significant interaction between photoperiod and dsRNA treatment (F_{1, 76} = 28.3; p <0.001). All females reared in long-day conditions that were injected β-gal and CpMRJP1 dsRNA were in a clear nondiapause state, such that the average egg follicle length of β-gal dsRNA-injected (100.7 ± 0.5 μm) and CpMRJPI dsRNA-injected mosquitoes (98.1 ± 2.5 μm) were not significantly different ( Figure 3B ; Tukey’s HSD; p = 0.76). Females reared in short-day conditions that were injected with β-gal dsRNA were in a clear diapause state, with an average egg follicle length of 53.3 ± 0.4 μm. However, the average egg follicle length of females reared in short-day conditions that were injected with CpMRJP1 dsRNA was 70.7 ± 2.7 μm, and females were found to be in a mixture of diapause (70%), intermediate (25%), and nondiapause (5%) states. Overall, knocking down CpMRJP1 significantly increased egg follicle length in short-day reared females (Tukey’s HSD; p < 0.001; 95% [CI = 10.4, 24.4]).

We also used a two-way ANOVA to determine whether photoperiod and knocking down CpMRJP1 affected fat content ( Supplementary Figure S3B ). Our analyses revealed that dsRNA injection did not affect fat content (F_{1, 26} = 2.28, p = 0.143). However, photoperiod had a significant impact on fat content (F_{1, 26} = 4.47; p = 0.045), such that β-gal and MRJP1 dsRNA-injected, short-day reared females had higher levels of fat than dsRNA-injected long-day reared females. This was largely driven by a non-significant trend where females reared in long-day conditions and injected with β-gal dsRNA had slightly less fat (6.45 ± 0.78%) than long day-reared mosquitoes injected with CpMRJP1 dsRNA (10.97 ± 0.40%; Tukey’s HSD, p = 0.181), and short-day reared mosquitoes injected with β-gal dsRNA (Tukey’s HSD, p = 0.092) or CpMRJP1 dsRNA (Tukey’s HSD, p = 0.072). Our 2-way ANOVA also confirmed that that there was not a significant interaction between photoperiod and dsRNA injection (F_{1, 26} = 2.12, p = 0.157).

We also investigated how injection with MRJP1 dsRNA affected the lifespan of long and short day-reared females relative to β-gal dsRNA-injected controls in the absence of food ( Figure 4C ). A Cox proportional hazards model revealed that short photoperiods had a protective effect on female lifespan in the absence of food ( Supplementary Table S1 ; HR = -1.31 ± 0.209; z = -6.28, p < 0.001), while knocking down CpMRJP1 had a slight, but non-significant protective effect (HR = -0.307 ± 0.167; z = -1.83; p = 0.0668). The median survival time of long-day reared, β-gal dsRNA-injected females was six weeks while the median survival time of short-day reared, β-gal dsRNA-injected females was eleven weeks (Tukey’s HSD, z = -6.28, p < 0.001). The median survival time of long-day reared, CpMRJP1 dsRNA-injected females was four weeks, while the median survival time of short-day reared, CpMRJP1 dsRNA-injected females was nine weeks (Tukey’s HSD, z = -1.314, p < 0.001). While knocking down CpMRJP1 did not significantly affect the likelihood of dying in females reared in long-day conditions relative to β-gal injected controls (Tukey’s HSD, z = -1.833, p = 0.258), dsRNA against CpMRJP1 significantly extended the total lifespan in short-day conditions ( Figure 4C ; all β-gal-injected females were dead at 11 weeks; all MRJP1-dsRNA injected females were dead at 15 weeks; Log Rank Test of survival differences: X² ₁ = 5.9, p = 0.02).