For the in vitro and in vivo experiments, 100 M. rosenbergii females were collected from the Aquaculture Station of the Ministry of Agriculture (Dor, Israel) (Levy et al., 2017) and then transferred to a holding facility at Ben-Gurion University of the Negev (Beer-Sheva, Israel) and acclimated for two weeks. During the acclimation and throughout the experiment, Females were maintained in tanks at 27°C, 12 h daylight, and fed ad libitum with Polychaeta and protein-rich pellets. Imitation black weeds and PVC tubes were included in the tanks as hiding substrates to minimize cannibalism.

The fluorophores labeled (TAMRA or FITC) VgP and scrambled VgP ( Table 1 ), and KH₉-VgP were purchased from Peptron (Daejeon 34054, South Korea).

For the in-vitro assay, slices of M. rosenbergii ovaries, 1-3 mm thick, containing oocytes of diameter ~150 µm [from early vitellogenic ovaries stage II, (Huang et al., 2010)], were incubated at room temperature for 24 h in a biological hood in a sterile 48-well plate having a 1-cm² growth area in each well (Biofil^®). Each slice was incubated with 250 µL of Schneider’s Drosophila Medium (Biological Industries), of a fixed osmolality of 420 mOsm/L (Huong et al., 2010), supplemented with 10% fetal bovine serum (FBS; Biological Industries), PSA (final concentration: penicillin, 100 units/mL; streptomycin, 0.1 mg/mL; amphotericin B, 0.25 µg/mL; Biological Industries), and bovine insulin (Morris and Spradling, 2011), final concentration 0.2 mg/mL (Sigma-Aldrich). In addition, to verify the role of the amino acid sequence of VgP in its internalization into the oocytes, the VgP 24-amino-acid sequence was synthesized in scrambled order (scVgP) to be used as a control. The ovary slices were incubated with 6 µM each VgP-TAMRA or scVgP-TAMRA, or 6 µM VgP-TAMRA + 6 µM scVgP-FITC ( Table 1 ).

For the in-vivo assay, a solution (2 µl/g body weight) containing VgP-TAMRA + scVgP-FITC (10 µg/g body weight for each peptide, dissolved in DMSO and mixed with crustacean saline 420 mOsm/L) was injected with a manual insulin syringe (BD^®) into the hemolymph sinus at the base of the fifth walking leg of M. rosenbergii reproductive females [each weighing approximately 15 g and bearing opaque white to yellowish ovaries stage II, (Huang et al., 2010)]. After injection, the animals were kept at 28°C for 24 h in the culture facility for M. rosenbergii of Ben-Gurion University of the Negev. Thereafter, the ovaries and gills were dissected out for confocal imaging.

Ovary pieces from the in-vitro and in-vivo experiments that were incubated with the fluorescing molecules were each transferred to a slide and incubated with Hoechst staining solution (5 µg/mL) for 7 min. Thereafter, the pieces were washed with saline buffer for 1 min ×3. The slides were inspected, and images were obtained with a confocal microscope FV1000 (Olympus) at ×60 magnification. The excitation laser wavelengths were 405 for Hoechst dye, 561 nm for the TAMRA-labeled peptides, and 488 nm for the FITC-labeled control peptide, and emission was obtained at 460-490 nm, 540-640 nm, and 490-530 nm, respectively.

MST was used to evaluate the affinity of VgP and scVgP to VgR-LBD-I and VgR-LBD-II. VgR-LBD-I and II were expressed and purified as described in supplementary methods. The peptides were diluted in Tris-buffered saline containing Tween and Ca²⁺ (TBS, Tris 20 mM pH 7.4, NaCl 150 mM, TWEEN 0.1%, CaCl₂ 5 mM) to 500 to 1000 fluorescence units. A serial dilution for each LBD protein was prepared in the same buffer. Ten microliters of each dilution of the peptide of interest were added to each LBD-containing tube. The experiments were performed with MST Monolith NT.115, NanoTemper.

The open reading frame (ORF) of M. rosenbergii PAX6-like was sequenced (GenBank OP292287). Two dsRNA: dsPAX6A (230bp) and dsPAXB (198bp), distinct parts of the ORF sequence were then synthesized in our laboratory ( Supplementary Figure 1 ). The templates for the transcription were amplified by adding the following primer pairs to the PCR: the sense strand was synthesized using a gene-specific forward primer containing the T7 promoter sequence vs. a gene-specific template reverse primer. The reverse primer containing the T7 promoter sequence was used vs. the gene-specific template forward primer to synthesize the antisense strand. Primers and T7 promoter sequences for dsRNA synthesis are given in Table 2 . PCR amplicons were separated on a 1.3% agarose gel. DNA bands were visualized with SYBR Safe DNA Gel Stain (Invitrogen). Amplicons were purified with a PCR purification kit (NucleoSpin Gel and PCR Clean-up, Machery-Nagel). A TranscriptAid T7 High Yield Transcription Kit (Thermo Scientific) was used (according to the manufacturer’s instructions) to generate single-stranded RNA. Synthesized RNA molecules were purified by phenol:chloroform (1:1) and sodium acetate extraction. After purification, RNA was precipitated with ethanol. Reconstituted sense and antisense RNA strands were hybridized by incubation at 70°C for 15 min, 65°C for 15 min, and room temperature for 30 min. dsRNA quality was assessed on an agarose gel. dsRNA was then diluted to 5 mg/ml and stored at -80°C until use. FITC-labeled dsPAX6A (dsPAX6A-FITC) was synthesized as described above by incorporating UTP-FITC (Sigma-Aldrich) into the synthesized dsRNA. FITC-labeled UTP was used in an equal amount to free UTP in the dsRNA synthesis reaction. RNA molecules were purified by the RNeasy MinElute Cleanup Kit (QIAGEN), followed by hybridization and quality assessment, as described above.

In this work, we chose to bind the VgP to dsRNA by leveraging an electrostatic interaction between VgP and the negatively charged phosphoryl groups of dsRNA. To facilitate such an electrostatic interaction, a poly cationic tail composed of 9 KH repeats was synthesized on the N-terminal side of VgP ((KH)₉VgP).

Agarose gel electrophoresis was used to evaluate the migration characteristics of the peptide-dsRNA particles. A constant amount of dsPAX6 (25 pmole) solution was mixed with increasing amounts of peptide (dsRNA/peptide ratio: 1:1, 1:5, 1:10, 1:20 and 1:50) in Eppendorf vials. For each vial, the total reaction volume was made up to 15 µL with doubly distilled water, and the mixtures were incubated at room temperature for 20 min. An equal volume of loading buffer (50% glycerol and 0.5 M EDTA) was added to each mixture, and the total volume of each mixture was separated on a 1.3% agarose gel in TAE buffer for 30 min at 90 V. Fluorescence emissions of the compounds in the gels were scanned by a Typhoon™ FLA 9500 gel system (GE Healthcare Life Sciences). Thereafter, the gels were stained with SYBR Safe DNA Gel Stain (Invitrogen) and visualized with UV light.

The morphology and particle size of the (KH)₉VgP-dsRNA compounds were investigated by TEM (FEI Talos F200C). Samples with dsRNA:(KH)₉VgP molar ratios of 1:5 and 1:10 were prepared, and 2.5-µL of each were taken for TEM sample preparation on a 300-mesh copper grid. Excess liquid was blotted with filter paper after 1 min, and the grid was dried in air for 1 min. Contrast enhancement was achieved by negative staining with 5 µL of 2% uranyl acetate. Thereafter, the grid was blotted again to remove excess uranyl acetate. Finally, the grid was air dried before insertion into the microscope. The images were obtained with an FEI Ceta 16M CMOS camera.

The ability of VgP to piggyback dsRNA into oocytes by endocytosis in vitro was investigated by incubating ovary slices with dsPAX6-FITC linked to (KH)₉VgP-TAMRA (molar ratio 1:10) at room temperature for 5 h, and confocal images of oocytes were obtained. As the control, ovary slices were incubated with free dsPAX6-FITC.

For the in-vivo experiment, the silencing of PAX6 piggybacked on VgP was evaluated in developing embryos. In this experiment, 20 previtellogenic females (14.96± 0.41 g) were injected with ds-vitellogenesis-inhibiting hormone (dsVIH) to induce ovary maturation (Cohen et al., 2021). Each day, the females were inspected visually in the daylight to determine ovary maturation. As soon as the ovaries acquired a yellowish appearance, the females (n=10) were injected with a solution containing (KH)₉-VgP linked to dsPAX6A and dsPAX6B (10 µg dsPAX6/g body weight, 5 µg of each dsPAX6A and dsPAX6B, dsPAX6-peptide molar ratio 1:10). The control group (n=10) was injected with free dsRNA. Chloroquine (2 mM final concentration) was incorporated into the injection mixtures as an endosomal escape reagent. The female animals were monitored daily, and upon appearance of the reproductive molt, each female was released into a tank with a male for mating and fertilization. The development of the embryos on the pleopods of each female was monitored under a stereoscope ( Figure 1 ). The effect of PAX6 silencing on embryo eye development was followed by sampling 10 embryos from the berried females on the day after eye appearance and examining them under a stereoscope. Embryos were counted and categorized according to eye development, i.e., normally developing, retarded, remnant eyes, or no eye development. Additionally, the dimensions (length and width) of the pigmented area of 10 eyes from 10 different embryos of each female were measured under a microscope, and an eye length/width index was calculated.

Larvae from the VgP-dsPAX6 injected females and from the control group were fixed and dehydrated by immersion in increasing ethanol concentrations (50%, 70%, 80%, 90%, 95% and twice in 100%; 15 min each). The samples were dried in a critical point dryer, placed on stubs, and coated with gold. The specimens were examined with a scanning electron microscope (SEM; JEOL model JSM-7400F).

Graphical and statistical analyses were performed using OriginPro and SPSS. For multiple comparisons, statistical differences were determined by one-way ANOVA. Differences were considered statistically significant at P < 0.05. The most important evaluation metric for checking any classification model’s performance is the Receiver Operating Characteristics (ROC). ROC is a probability curve, and the Area Under the Curve (AUC) represents the degree or measure of separability. It tells how much the model is capable of distinguishing between classes. The higher the AUC, the better the model predicts impaired eye development. The classes are the (KH)₉VgP-dsPAX6 injected versus the control ds injected females. The classifier is the length/width index of the eye pigment. ROC analysis was performed to define the percentage of embryos possessing impaired eye development and therefore, the efficiency of the treatment.

To enable evaluation of the endocytosis ability of VgP into early vitellogenic oocytes, both VgP and the control scrambled peptide, scVgP, were labeled with the TAMRA fluorophore. When early vitellogenic oocytes were incubated with the peptides and inspected under a confocal microscope, fluorescence emission at 564 nm was detected. Both VgP and scVgP were visible in the vicinity of the oocyte membrane ( Figures 2A–D ). VgP was distributed mainly on the cytosolic side of the oocyte membrane where yolk droplets were forming ( Figures 2A, C , black arrow). Figure 2C shows that VgP accumulated mainly in the peripheral area of the oocyte, near the membrane (white arrow), being scattered on the external surface of newly formed droplets. In contrast, scVgP was dispersed mainly on the outer surface of the oocyte ( Figures 2B, D ).

A dose-dependent test using VgP-TAMRA and scVgP-TMARA was performed to determine the optimal peptide concentration for in-vitro incubation of the peptide with the ovary slices ( Supplementary Figure 2 ). Unlike the distribution patterns for scVgP, the distribution patterns of VgP in the oocyte remained similar, irrespective of the peptide concentrations ( Supplementary Figure 2 , upper row). For VgP, a clear fluorescence signal was detected at concentrations of 12, 6, and 1.5 µM, while the fluorescence signal for scVgP was weak at high concentrations (12 µM and 6 µM), and no signal was detected at 1.5 µM. No signal was obtained at 0.75 µM for both peptides ( Supplementary Figure 2 ).

To evaluate the binding affinity of VgP and scVgP to the two LBDs (LBDI and LBDII) of VgR, the kinetic dissociation constant (Kd) was calculated from kinetic curves obtained by microscale thermophoresis (MST) ( Table 3 and Supplementary Figure 3 ). For VgP, Kd values of 3.42 ± 2.16 and 50.89 ± 26.47 μM for binding to LBDII and LBDI, respectively, were obtained ( Table 3 ). In contrast, there was no apparent interaction of scVgP with the receptor LBDs.

VgP and its control counterpart (scVgP) were labeled with different fluorophores, TAMRA and FITC, respectively, to enable differentiation of the specific localization of each in the oocyte. In accordance with the results obtained from incubating ovary slices with the peptides labeled with the same TAMRA fluorophore, the fluorophore experiment showed the distributions of the two fluorescently labeled peptides in the same oocyte. Endocytosis of VgP-TAMRA into the oocytes was evident, with the labeled peptide being distributed along the inner side of the membranes ( Figures 2E, H ). In contrast, scVgP-FITC was visible only on the membrane’s outer surface, overlapping with the outer surface of the follicular cell nuclei stained with Hoechst stain ( Figures 2F–H ).

In in-vivo experiments designed to determine whether VgP-TAMRA was internalized specifically into oocytes, a mixture of the VgP-TAMRA and the control-FITC peptides was injected into the circulatory system of a female prawn. The control-FITC peptide was absent from oocytes dissected 24 h after injection ( Figure 3B ). For VgP-TAMRA, fluorescence emission was evident inside the oocytes ( Figures 3A, C ), mostly inside yolk droplets. To verify that the two peptides did indeed circulate through the female prawn body, the prawn gills were dissected out and examined by confocal microscopy. Clear dual (TAMRA and FITC) fluorescence emission was evident, indicating that most of the injected peptide had indeed been washed out through the osmoregulatory system (namely, the gills; Figures 3D–F ).

VgP-TAMRA and scVgP-FITC were synthesized with cationic poly (KH)₉ tails to facilitate electrostatic interactions between the peptides and the negatively charged dsRNA. dsRNA-peptide complexes were prepared at different molar ratios (dsRNA/peptide ratio: 1, 1:5, 1:10, 1:20 and 1:50) and characterized by two different means, separation on agarose gel and TEM. In the agarose gel assay, gradual retardation of the dsRNA, indicative of increasing particle size or decreasing negative charge, was distinguished as the interaction ratios increased ( Figures 4A, B ). The fluorescence gel image in Figure 4A shows the migration of the cationic VgP-TAMRA toward the cathode. At a 1:50 dsRNA:peptide ratio, a sharp fluorescence band may be seen on the loading well, which probably indicates the formation of large dsRNA-peptide complex aggregates or charge-neutralized molecules ( Figure 4A ). However, at ratios of 1:5 and 1:10 a clear shift was evident in comparison to dsRNA alone ( Figure 4B ). The (KH)₉Vg-dsRNA complex was further characterized by TEM for evaluating the morphology and particle size of the dsRNA-peptide complex at molar ratios of 1:5 and 1:10. Free dsRNA appeared as long thin filaments ( Figure 4C ). (KH)₉VgP appeared as helical 50- to 450-nm filaments (oval shapes, Figure 4D ). Incubation of the peptide with dsRNA at both indicated ratios led to the formation of almost spherical particles of 100 nm diameter, surrounded by filaments ( Figures 4E, F ).

The next step was to evaluate the ability of VgP to piggyback dsRNA into oocytes. For this purpose, dsRNA-FITC, which exhibits green fluorescence, was synthesized. dsRNA-FITC was then linked to (KH)₉VgP-TAMRA and incubated in vitro with ovary pieces for 5 h. The green fluorescence of dsRNA-FITC can be seen within the oocyte boundary ( Figure 5C , arrows), and the green puncta appear to overlap with the red fluorescence of (KH)₉VgP-TAMRA, with the overlap indicating a stable complex and internalization of the two molecules ( Figure 5D , arrows). However, no green fluorescence puncta could be seen in the oocytes from ovary pieces incubated with dsRNA alone ( Figure 5A ).

To confirm that VgP can piggyback dsRNA into oocytes, which would lead to gene silencing, PAX6, which controls eye development in embryos, was selected as a candidate gene for illustrating gene silencing ( Supplementary Figure 1 ). In M. rosenbergii embryos, PAX6 gene expression begins on day 5 ( Supplementary Figure 4 ), and eyes appear on day 9. In our study, vitellogenic females were injected with (KH)₉VgP-dsPAX6 or free dsPAX6, and the development of the embryos (on the pleopods of the injected females) was monitored. Embryos were examined one day after eye appearance, and eye development was documented ( Figure 6 ). The embryos from (KH)₉VgP-dsPAX6-injected females showed either impaired eye development or a lack of eye development. A thin eye pigment was abundant in those embryos ( Figures 6A, B , black arrowhead), and maldeveloped eyes ( Figures 6A–C , red arrowhead) or a complete lack of eyes were also documented ( Figures 6A–C , white arrowhead). To identify embryos with retarded eye development, an eye pigment length/width index was calculated. The average ratio for embryos taken from a (KH)₉VgP-dsPAX6-injected female was significantly higher (4.55 ± 0.19, p<0.0001) than that for the embryos taken from a dsPAX6-injected female (3.48 ± 0.05, Figure 6G ). ROC analysis revealed that the AUC reached 0.93, which means there is a 93% chance that the model will be able to distinguish between impaired versus normally developed eyes. Indeed 87% of embryos taken from the (KH)₉VgP-dsPAX6-injected female possessed length/width eye pigment value above the AUC threshold (0.93), indicating impaired eye development ( Supplementary Figure 5 ). The eye phenotype was also examined at advanced developmental stages. A SEM investigation of the eye surface of larvae at stage 11 (11 days after hatching) from (KH)₉VgP-dsPAX6-injected females revealed irregular, elongated ommatidia, with continuous or elevated hinges between ommatidia ( Figures 6I, J ). In contrast, the control larvae demonstrated well-shaped hexagonal ommatidia ( Figure 6H ).