Z. cucurbitae were initially collected from Haikou, Hainan Province, China, and reared in the laboratory for many generations. The insect larvae and adult were fed on an artificial diet (Liu et al., 2020), and a 1:3 yeast powder and sugar diet, respectively, and maintained under 26°C, 14 h light: 10 h dark photoperiod at 70% relative humidity. Grown larvae were shifted to wet sand prior to pupation, and then pupae were transferred to cages until adult emergence.

Nucleotide sequences of Ago1 (Gene ID: 105216341) and Gawky (Gene ID: 105220406) were obtained from the Z. cucurbitae genome (Sim and Geib 2017) by BLAST in NCBI using Ago1 and Gawky genes of D. melanogaster as queries. Both genes were then verified by reverse transcription PCR (RT-PCR) and sequencing. The SMART (http://SMART.embl-heidelberg.de) and ExPASy Prosite SCAN (https://prosite.expasy.org/scanprosite/) were used to analyze the conserved domains and functional sites. The phylogenetic tree was constructed through the neighbor-joining method by taking Tephritidae and Drosophilidae as model families with 1,000 bootstrap repetitions in MEGA-X.

To examine the expression profiles of Ago1 and Gawky, samples were collected from different developmental stages: 1^st instar larvae (1L), 2^nd instar larvae (2L), 3^rd instar larvae (3L), pupae and adults, and various tissues: head, ovary, testis, fat body, and midgut from adults at 1, 2, 5, and 7 days old. Flies were dissected separately in 1 × PBS (pH 7.4) under a binocular stereoscope (Olympus SZX12, Tokyo, Japan). For each biological replicate, ovaries from females at 1–2 days old contained 20 pooled flies, whereas ovaries from females at 5 and 7 days old had 10 pooled flies. All samples included three independent biological replicates.

Total RNA was isolated from 3^rd instar larvae using a TRI Reagent^® (Sigma, United States). cDNA was reversely transcribed from 1 μg of the total RNA template using the SuperScript III First-Strand Synthesis kit (Takara, Dalian, China) and used to amplify the open reading frame (ORF) of target genes. ORF was amplified using PrimeSTAR® HS DNA Polymerase (Takara, Dalian, China) and cloned into the pMD™18-T vector (Takara, Dalian, China) for sequencing. For dsRNA preparation, T7 promoter sequences were introduced to the 5′ ends of the forward and reverse primers of Ago1 and Gawky genes (Supplementary Table S4). A green fluorescent protein (GFP) fragment was amplified from the pCAMBIA1303 expression vector. PCR program was performed as follows: 95°C for 3 min (m), 32 cycles of denaturing at 95°C for 30 sec (s), annealing at 56–58°C for 30 s, and extension at 72°C for 28 s with a final incubation at 72°C for 10 m. PCR products were electrophoresed on a 1% agarose gel, extracted, purified, and cloned into pMD™18-T vector. Positive clones were confirmed by PCR using M13 forward and gene-specific reverse primers. Plasmid DNA was extracted from positive clones using the Sangon Plasmid miniprep kit (Sangon Biotech, Shanghai, China) and sequenced at Sangon Biotech (Shanghai, China). dsRNA was synthesized using corresponding plasmids as the template with the MEGAscript RNAi kit (Promega, United States). Nuclease-free water was used for dsRNA elution. dsRNA was quantified on a NanoDrop 2,000 spectrophotometer at 260 nm, and integrity was determined using gel electrophoresis. dsRNA was diluted to 1 μg/μl in nuclease-free water.

qRT-PCR was used to evaluate the effects of dsRNA treatments on Ago1 and Gawky expression levels in various tissues and developmental stages of Z. cucurbitae. qRT-PCR was performed using gene-specific primers (Supplementary Table S4). The amplification efficiency of the primers was first confirmed by a standard curve based on a 4-fold cDNA dilution series. qRT-PCR was performed in a 10 μl reaction (5 μl of 2x SYBR Green qPCR Supermix Plus, 0.5 μl of each pair of primers, 0.5 μl of cDNA, and 3.5 μl of distilled water). The qRT-PCR program was conducted as follows: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s, 60°C for 10 s, and 72°C for 15 s in 96-well plates on the Analytik Jena qPCR system. The elongation factor 1 alpha (EF1α) and Actin were used as an internal reference to normalize the relative transcript levels of cDNA. The transcript levels were quantified using the 2^–ΔΔCT method (Livak and Schmittgen 2001).

To check the dsRNA stability in artificial diet and midgut, an ex vivo dsRNA degradation assay was performed. Midguts were extracted from 12 adult insects. Dissection was performed in 1x PBS under a microscope by first immobilizing the insects on ice for 5 m. All the unwanted tissues were removed, and midguts were collected in a pre-chilled Eppendorf tube placed on ice and then centrifuged at 13,000 rpm for 12 m; 2 μl supernatant was carefully transferred and mixed in 25 μl of dsRNA (1 μg/μl); 2 μl ddH₂O was used in mimic control instead of midgut juice, and both samples were incubated at 26°C. About 5 μl of the sample was taken off from both reaction tubes at 0, 15, 30, and 60 m, and placed at −80°C to stop the enzymatic reaction. dsRNA integrity was evaluated on 1% agarose gel. To check the stability of dsRNA in the artificial diet, 20 μl of dsRNA (1 μg/μl) was mixed with a 2 g artificial diet (Liu, et al., 2020). The nuclease-free water was used as a control in food instead of dsRNA. After 1, 24, and 48 h, dsRNA stability was checked by re-dissolving the food in 10 μl ddH₂O and running on agarose gel for 25 m.

The dsRNA feeding assay consisted of the treatments (dsAgo1 and dsGawky) and the control group (dsGFP). Early 3^rd instar larvae and the control group were fed with an artificial diet mixed with gene-specific dsRNA and dsGFP, respectively. Three biological replicates were performed for each group. Each replicate contained 60 larvae, 2 g of artificial diet, and 20 μl of 1 μg/μl dsRNA. The food of the control group had the same ingredients, except gene-specific dsRNA. Insects were fed with dsRNA for 24 h, and then shifted to the new diet with the same dsRNA for another 24 h. For each replicate, three larvae of 12, 24, and 36 h post-feeding were used for RNA extraction to analyze the knockdown efficiency. The silencing efficiency of the target genes was tested using qRT-PCR. The experimental setup is described in Supplementary Figure S1A.

For microinjection, 1 μg/ul dsRNA solution of each gene was injected into embryos (usually 1.5 μl is used to fill the injection needle for 300–400 eggs) using FemtoJet 4i microinjector (Eppendorf, Hamburg, Germany). Three biological replicates were performed for the control and treatment groups, where each replicate contained 300 eggs injected with gene-specific dsRNA or dsGFP. To determine gene silencing efficiency, 25 eggs of 12 and 24 h post-injection and 25 1^st instar larvae of 36 h post-injection were used for RNA extraction, and qPCR was performed for gene expression analysis. The experimental setup is described in Supplementary Figure S1B.

After RNAi treatment with dsRNA injection and feeding, ovaries and testes of 14-day-old female and male adults from each treatment and control group were dissected in 1 × PBS (pH 7.4). Ovary and testis phenotypes were photographed using a Leica M205A stereomicroscope (Leica Microsystems, Wetzlar, Germany). To explore the RNAi effects of Ago1 and Gawky genes on female and male fertility, the virgin females of 9 days old were individually crossed with two virgin wild-type (WT) males of the same age. Each male was paired with two virgin WT females of the same age in a courtship chamber for 5 days. Flies were provided with yeast and sugar to promote egg-laying, and cucumber slices were provided after 24 h for females to lay eggs on it. Oviposition was examined by counting the number of eggs laid by the female flies. Eggs were transferred to artificial food and incubated at 26°C. To assess the fecundity, the number of hatched larvae was counted and the egg hatchability was calculated. Ten females and males were used from each replicate with three repetitions for dsAgo1. For dsGawky, five females and males were used from each replicate with three repetitions.

Statistical analysis of differences in mRNA expression, mortality, and fecundity was performed using the GraphPad Prism software package (GraphPad Software Inc., San Diego, CA, United States). The statistical significance of differences between means of each group was assessed using a one-way analysis of variance followed by Tukey’s honestly significant difference (HSD) test.

After PCR amplification and sequencing of Ago1 and Gawky genes from 3^rd instar larvae of Z. cucurbitae, the sequence analysis showed that Ago1 and Gawky contain 982 and 1,482 amino acid residues, respectively. The predicted protein structures using SMART software showed that Ago1 has a DUF domain of an unknown function besides conserved PAZ and PIWI domains (Figure 1A). RNase H-like PIWI domain is consistent with Ago proteins which give them slicer activity. Multiple sequence alignment of the PIWI domains indicated that the 5′ phosphate anchoring region, Aspartate, Aspartate, and Histidine (DDH) motif, is highly conserved in D. melanogaster, B. mori, T. castaneum, and L. migratoria (Figure 1B; Supplementary Figure S2), suggesting that Ago1 is a member of the Ago family. Z. cucurbitae has just one Gawky protein, similar to D. melanogaster (Perconti et al., 2019). Both N and C terminals and mid-region of the Z. cucurbitae Gawky protein are also rich in glycine (G) and tryptophan (W). N terminal of the Gawky protein facilitates its interaction with the PIWI domain of Ago1, and RRM (RNA recognition motif) targets mRNA and contributes to its repression (Behm-Ansmant et al., 2006). Phylogenetic analysis of Ago1 and Gawky proteins revealed that Z. cucurbitae formed a clade with other examined insect species. Ago1 of Z. cucurbitae (ZcAgo1) showed the highest similarity with B. dorsalis Ago1 protein (BdAgo1), and Z. cucurbitae Gawky protein (ZcGawky) also grouped together with its counterparts from Bactrocera genus (Figure 1C).

The expression levels of the two genes Ago1 and Gawky were measured using qRT-PCR in different developmental stages and five tissues of Z. cucurbitae (Figures 2A,B). The temporal expression showed that Ago1 and Gawky were highly expressed in embryos and adults of Z. cucurbitae (Figure 2A). In tissue-specific expression analysis, the maximum relative abundance for Ago1 was recorded in ovaries from females of 5 and 7 days old, followed by fat bodies from adults of 1–2, and 5 days old. Interestingly, the expression of Ago1 was a little higher in the fat bodies of a 1-2-day-old adult than in the ovaries of females in the same age. For the flies from day 5 to 7, the Ago1 expression reduced in the fat body but increased in the ovary. Gawky showed the highest expression in the head of 5-day-old adult flies, compared to other tissues (Figure 2B). These temporal and spatial expression profiles suggest general roles of Ago1 and Gawky in the whole body of Z. cucurbitae and in the female reproductive system.

To explore the functions of Ago1 and Gawky, we injected and fed dsRNA to embryos and larvae, respectively. Before that, we first checked dsRNA integrity in artificial food to determine how often food needed to be changed with the fresh one to ensure that larvae were exposed to stable dsRNA. 3^rd instar larvae were fed with food containing 20 µl of 1 μg/μl dsRNA. Diet samples taken at 1, 24, and 48 h were separated on 1% agarose gel to check the stability. As shown in Supplementary Figure S3A, dsRNA remains stable in the artificial diet for 24 h, and a weak level of dsRNA exists until 48 h. This indicates that enough stable dsRNA was available to larvae throughout the exposure. On the other hand, an in vitro study measuring the integrity of dsRNA in midgut juice revealed that dsRNA was stable for the first 15 m, then poorly detectable after 30 m, and finally completely degraded after 60 m at room temperature. These results indicate that no matter dsRNA degraded after 15 m in midgut juice due to the presence of nucleases, stable dsRNA was constantly available in food for more than 24 h to target the relevant genes. In contrast, dsRNA remains stable in the control (Supplementary Figure S3B). Therefore, we fed larvae with a new diet containing dsRNA every 24 h.

RNAi efficiency was determined after 12, 24, and 36 h of injection and feeding of gene-specific dsRNA and dsGFP. The embryonic injection and oral larval feeding of Ago1 and Gawky dsRNA showed significant transcript knockdown of both genes at 12 h post-delivery of dsRNA, compared to their respective injected and fed dsGFP controls (Figures 3A,B). Reduction in the relative expression of Ago1 and Gawky genes at 24 h post-injection and feeding were non-significant. However, at the longer time period of 36 h post-feeding, the expression level of Ago1 was recovered and increased, while significant suppression of the target transcript was observed by providing dsGawky compared to the controls with dsGFP (Figure 3A). Furthermore, to measure the possibility of off-target effects, the expression level of non-target genes was estimated. No significant effect on the expression of other Ago family members (Ago2, Ago3, Piwi, and Aubergine) was observed after Ago1 and Gawky knockdown (data not shown), which is suggestive of the specificity of Ago1 and Gawky knockdown.

Silencing Ago1 and Gawky in Z. cucurbitae via feeding larvae and microinjecting embryos caused developmental defects. In contrast to the dsGFP control group that had normal yellow ventral segments, adult flies with Ago1 silencing showed color transition from yellow to white or white ventral patches (Figure 4A). The treatments of dsAgo1 injection and feeding resulted in adult flies with 38.33 and 46.6% white ventral patches, respectively (Supplementary Tables S2, S3). Severe developmental defects were also observed in the flies with Gawky silencing, such as a reduction in abdominal segmentation in relation to the dsGFP-treated control (Figure 4A). About 33.33 and 35.29% abnormal phenotypes were observed with dsGawky injection and feeding, respectively (Supplementary Tables S2, S3). Additionally, compared to dsGFP, Gawky silencing presented the highest mortality rate of about 80–85% in both dsRNA delivery groups (Figure 4B). The most efficient mortality occurred in the pupal stage after ingestion and injection of ds Gawky. Taken together, these results indicate that Gawky is an essential gene for both growth and development of Z. cucurbitae, and dsAgo1 does not affect its viability.

The role of Ago1 and Gawky in reproduction of Z. cucurbitae was investigated by genetic crosses. The females and males (9 days old) from both treated groups (feeding and microinjection) were individually crossed with two virgin males and virgin females, respectively. In the control group, one WT male was crossed to two WT females, and one WT female was crossed to two WT males. For the treated groups, one treated male was crossed to two WT females, and one treated female was crossed to two WT males in courtship cages for five consecutive days. Similar crosses were performed for the dsGFP group. The eggs laid by each group were counted daily. No effect on egg-laying was observed when the treated males of dsAgo1 and dsGawky were crossed with WT females compared to the control group in both dsRNA delivery groups. Egg-laying was significantly reduced (about 60%) in both dsRNA delivery methods when the dsAgo1- and dsGawky-treated female flies were crossed with WT males for five consecutive days (Figures 5A,B).

To investigate the gene silencing effects of Ago1 and Gawky on reproductive capacity of Z. cucurbitae, we further carried out hatching assays of laid eggs. There were no significant differences in egg hatching in both treated methods (feeding and injection) when the treated males of dsAgo1 were crossed with WT females (Figures 6A,B) or the treated females of dsAgo1 were crossed with WT males (Figures 6C,D). In both the dsRNA delivery methods, egg hatching was significantly reduced when the dsGawky-treated males and dsGawky-treated females were crossed with WT females and males, respectively. For injection of dsGawky, 60–70% egg hatching reduction was observed in both the treated male and female groups compared to the control group (Figure 6). In view of these results, the reduction in egg hatching was probably due to the developmental defects observed in the physiology of the dsGawky treated flies, where there was no effect of dsGawky on reproductive tissues of Z. cucurbitae.

After the effects on reproductive capacity with Ago1 and Gawky silencing were observed, we speculated that these genes are probably required for reproductive tissue development. To test this assumption, reproductive tissues of 14-day-old females and males after genetic crosses were dissected to check the effects of target genes on ovary and testis morphology. Ago1 silencing caused ovarian abnormalities, such as undifferentiated and undeveloped ovaries, decreased number of ovarioles, and abnormal oviposition in female adults (Figure 7). Different phenotypes in ovaries were observed after feeding and injection of dsAgo1. Feeding of dsAgo1 affected the ovary morphology. One ovary from each pair has an undifferentiated and tiny round ovariole, while the other has no reproductive tissue, and its development was completely arrested (Figure 7B). Egg count was also significantly reduced with dsAgo1 compared to dsGFP. After injection of dsAgo1, one spheroid of the ovary was fully developed with long ovarioles, while the other one had few long ovarioles compared to the normal ovary in the control group (Figure 7C). No effect was observed on the morphology of testes of Z. cucurbitae after the Ago1 knockdown. These results suggest that Ago1 plays a vital role in the reproduction of Z. cucurbitae females. No testis and ovary malformation were observed, and each ovary contained normal ovarioles in the dsGawky treated flies, similar to the dsGFP control group. Our findings are partially coherent with our speculation that decrease in egg number is due to ovarian defects in the dsAgo1 treated females. However, reduced number of egg hatchability are not associated with these reproductive tissues in the dsGawky-treated flies.