Both wild and farmed males of 80–120 g were used for the in vivo experiment. Wild eels were caught in the Harinxma canal (Netherlands). Farmed eels were obtained from the eel farm Palingkwekerij Koolen B.V (Bergeijk, Netherlands). Both wild and farmed males were transported to the animal experimental facilities (CARUS, Wageningen, Netherlands) and transferred to tanks connected to a recirculating system with artificial seawater (16°C, 36 ppt).

For the in vitro trials, feminized eels were used. For the in vivo trials, feminized eels and wild females were used to compare the effects of feminization on treatment.

Elvers of 10 g were transferred from the eel farm Palingkwekerij Koolen B.V. (Bergeijk, Netherlands) to the animal experimental facilities of Wageningen University & Research (CARUS, Netherlands). After arrival, they were placed in 400-L tanks kept in freshwater (FW) and 24°C under dimmed light conditions. For inducing feminization, elvers were fed with 17β-estradiol (E2) coated pellets for 6 months (Chai et al., 2010). Following the feminization procedure, females were fed a broodstock diet for an additional 6 months. Premature females of ∼400 g were then selected, transferred to a 3,600-L swim-gutter with seawater (Tropic Marine, 36 ppt) and subjected to simulated migration (slightly adjusted protocol from Mes et al., 2016) for 2 months to initiate sexual maturation. The simulated migration consisted of constant swimming against a flow with a speed of 0.5 m s⁻¹, in the dark at daily alternating temperatures between 10°C and 15°C. After the simulated migration, females were anesthetized with 2-phenoxyethanol (2 mL in 10 L water), PIT tagged (TROVAN, Aalten, Netherlands) and transferred to 373-L tanks connected to a recirculating system with artificial seawater (16°C, 36 ppt). Each female was injected in the peritoneal activity with a steroid implant containing 17-methyltestosterone (5 mg) and E2 (2 mg) for 2 months to induce the start of vitellogenesis (Palstra et al., 2022).

Wild silver eels of ∼500g were caught during their seaward migration in the Van Harinxma canal (Friesland, Netherlands). Females were transported to the animal experimental facilities (CARUS, Wageningen, Netherlands). After arrival, females were anesthetized, PIT tagged (TROVAN, Aalten, Netherlands), acclimatized in tanks (16°C, 36 ppt) for 13 weeks and injected with hCG at a dose of 3000 IU.kg⁻¹ (Chorulon, MSD, Kenilworth, New Jersey, United Stated). One month later, wild females were injected in the peritoneal cavity with a steroid implant containing E2 (2 mg) for 2 months to induce the start of vitellogenesis (Palstra et al., 2022).

Eleven feminized females were artificially matured according to the previously described routine protocol (Palstra et al., 2005). Feminized females were weekly intramuscular injected with 20 mg kg⁻¹ CPE to induce further vitellogenic growth (Catfish, Den Bosch, Netherlands). Starting in week 7, 2 days after each injection, females were weighed to determine the body weight index (BWI: body weight/body weight at the moment of first CPE injection × 100%) and body girth index (BGI: body girth/body length). When BWI > 110% and BGI > 0.24, females were anesthetized and sampled for ovarian tissue by inserting a cannula through the cloaca. Oocyte development was then graded on a scale from 1 to 7 according to Palstra et al. (2005). When oocytes were on average in stage 3 (i.e., transparent oocytes with migrating germinal vesicle), females were given an additional CPE injection at a dose of 20 mg kg⁻¹ to booster oocyte maturation. One day later, females were checked again to assess the progression of oocyte maturation. When oocytes developed to stage 4 (i.e., transparent oocytes with germinal vesicle at the periphery), females are commonly injected with DHP at 2 mg kg⁻¹ to induce oocyte maturation and ovulation (Ohta et al., 1996; Palstra et al., 2005). Therefore, ovarian tissues (∼5 g) were taken by needle (inner diameter of 2.3 mm) from a standardized location (5 cm anterior from the cloaca) and used for in vitro incubation studies.

DHP (Cayman, Michigan, United States) and P (Sigma-Aldrich, Saint Louis, MO, United States) powders were initially dissolved in absolute ethanol. Stock solution of DHP and P were further dissolved in Gibco™ Leibovitz’s L15 medium GlutaMAX supplement (Thermo Fisher Scientific, Waltham, MA, United States) and aliquots were made of the desired concentration. In the aliquots, ethanol concentration was below 0.1%.

Immediately after extraction, ovarian tissues were placed in ice-cold culture medium supplemented with 2.5 g L⁻¹ HEPES, 0.1 g L⁻¹ streptomycin and 100,000 IU L^-1 penicillin (Pen Strep, ThermoFisher, Waltham, MA, United States). For one female, twenty oocytes were placed in 4% buffered paraformaldehyde, refrigerated at 5°C overnight and stored in 70% ethanol for later confirmation of the integrity of the oocyte follicles. For each female, 60 oocytes were randomly stocked per well in 21 wells of two 24-well plates. In each plate, oocytes were stocked in 1 mL medium supplemented with the treatment (DHP at 1, 10 and 100 ng mL⁻¹ or P at 10, 100 and 1,000 ng mL⁻¹) or with hormone free media for the control group, and tested in triplicate. The difference in applied concentrations of DHP and P is based on the higher potency of DHP vs. P to activate the receptor (Todo et al., 2000). One culture plate was incubated for 12 h and one plate was incubated for 18 h at 20°C to mimic the in vivo conditions. Just before incubating (at 0 h), the remaining oocytes were sampled for microscopy and gene expression analysis as described in the following two sections. After incubations, oocytes were again sampled for microscopy and gene expression analysis.

For each female, a total number of 20 oocytes that represented stage 2 (i.e., transparent oocyte at the start of GV migration) and further, were used for microscopy analysis according to the oocyte scale developed by Palstra et al. (2005). Oocytes were placed in Serra’s fix (ethanol:formalin:acetic at 6:4:1, diluted ×20 in PBS) for 3 min to stain the GV. When maturing oocytes lacked the GV, GVBD was considered to have occurred. Stained oocytes were photographed for later assessment of the percentage of oocytes that displayed GVBD and measurements of the lipid droplet and oocyte diameter. Lipid droplet diameters were measured according to Unuma et al. (2011) with the open source software ImageJ. On visual assessment, the ten largest lipid droplets were measured and the five maximum values averaged for each oocyte. For determination of the oocyte diameter, the maximum diameter was measured.

Oocytes were fixed in cold 4% buffered paraformaldehyde, kept refrigerated at 5°C overnight and then preserved in 70% ethanol. Ten oocytes were randomly selected, placed in 0.9% type II low gelling agarose (Sigma Aldrich, Saint Louis, MO, United States), dehydrated via an ethanol/xylene series and embedded in paraffin. Oocytes were then sectioned using a microtome into 5 µm thick sections and stained with Mayer’s hematoxylin & eosin. Oocytes were photographed with a Leica DFC450c color camera attached to a Leica DM6b microscope.

For each female, a total number of 20 oocytes including different stages from stage 3 onwards were selected under the binocular according to the oocyte scale developed by Palstra et al. (2005). Stage 0 (i.e., opaque oocytes), stage 1 (i.e., opaque oocytes with a centered nucleus becoming visible) and stage 2 (i.e., fully transparent oocyte containing numerous oil droplets) were excluded from the gene expression analysis since these oocytes are unlikely to respond to DHP and P. After selection, oocytes were photographed, placed in 1 mL RNAlater (Ambion Inc., Huntingdon, United Kingdom), kept refrigerated at 5°C overnight and then stored at −80°C until RNA extraction.

For RNA extraction, oocytes were homogenized with a tissue lyzer (Qiagen, Tissuelizer II) in 1 mL Trizol (Invitrogen, California, United States). Possible contaminant traces of DNA were digested with ISOLATE II RNA Mini Kit (Bioline, London, United Kingdom). After RNA extraction and DNase treatment, RNA purity was assessed by the 260/280 ratios (2.2 ± 0.1) on the Nanodrop (ThermoFisher, Waltham, MA, United States) and, by the absence of RNA breakdowns, on the 2100 Bioanalyzer (Agilent, Santa Clara, CA, United States). Complementary DNA was generated from RNA using random primers and dNTPs with Superscript III Reverse transcriptase (ThermoFisher, Waltham, MA, United States).

Gene expression analysis was performed as previously described (Jéhannet et al., 2021). Diluted complementary DNA was mixed with SensiFAST™ SYBR^® Lo-ROX Ki (Bioline, London, United Kingdom) and the primer set of the target gene (Table 1). Since the gene expression analysis data could not be obtained from a single 96-well plate due to the high amount of samples, several plates were run per gene. For comparing the data obtained from the different plates, the following were taken into consideration 1) reference and target genes were run on the same 96-well plate; 2) the same master mix (SensiFAST SYBR and primer set) was used between plates and 3) samples were randomized through the plates. The conditions in the QuantStudio Real-time PCR system (ThermoFisher, Waltham, MA, United States) used were 95°C during 2 min followed by 40 cycles of denaturation at 95°C for 5 s, annealing at 60°C–64°C for 10 s and extension at 72°C for 5 s. Primer-dimers artifacts and reaction specificity were checked by melt curve analysis. PCR products were electrophoresed on 1.5% agarose gel. Standard curves that were made by diluting sample cDNA had R² values above 0.98 and efficiencies between 92% and 109%. For the very few samples that did not amplify due to low gene expression, CT values were manually set at 35. Transcripts levels of each target were normalized over elf1 (no difference in expression between treatments; ANOVA, p = 0.341) and data were expressed as fold change by using the 2^−ΔΔCT method from Livak and Schmittgen (2001).

Twenty-two feminized and eighteen wild females were artificially matured according to the procedure described in Section 2.2.2. For ovulation induction, females were either injected with DHP or P at a dose of 2 mg kg⁻¹ (Ohta et al., 1996) Just before the DHP or P injection, ovarian tissues were extracted from feminized eels with a cannula. Twenty oocytes were selected (see Section 2.3.6) and stored in RNAlater (Ambion Inc., Huntingdon, United Kingdom) for RNAseq analysis. Females were then placed in tanks and water temperature was gradually increased from 18 to 20°C; the optimum temperature during induction of spawning in eels (Unuma et al., 2012). In the timespan of 11–18 h after receiving the injection, females were regularly checked for egg release. When eggs could be retrieved, females were stripped by applying gentle pressure on the abdomen. At the moment of stripping, eggs from feminized females were sampled again for RNAseq analyses to compare treatment effects of DHP and P on stripped eggs.

Males were injected with 1,000 IU human Chorionic Gonadotropin (hCG: Kahn et al., 1987) and placed back in the 373-L tanks kept at 36 ppt and 16°C. One day before fertilization, males were anesthetized and checked for spermiation by applying gentle pressure on the abdomen. Males that were spermiating were injected with 250 IU hCG, dissolved in 0.9% physiological salt solution, to increase sperm quality. Males were placed in the spawning tank with the DHP or P injected females. When females were ready to be stripped, males (N = 6–10) were anesthetized and 2–3 mL of sperm was taken which was directly diluted into 45 mL artificial eel plasma (Peñaranda et al., 2010).

After collection of the stripped eggs into plastic bowls, the diluted sperm was dripped onto the eggs. Gametes were mixed by stirring. Artificial seawater (36 ppt, 18°C) was added for a contact time of 5 min. After 5 min, 10 g of eggs were placed in a 100 mL cylinder filled with artificial seawater to determine the percentage of floating eggs in relation to the eggs that were sinking 1 hour later. The remaining eggs were placed in 3L-beakers filled with artificial seawater. After 1 h, the floating egg layer was collected from the 3L-beakers, rinsed gently over a sieve and transferred into beakers filled with fresh artificial seawater. Eggs were kept into suspension by gentle aeration and dead material was removed every 12 h to prevent microbial infections. After hatching (48–60 h post fertilization), the content of each beaker was gently mixed to uniformly disperse the larvae in the water column. Then, a water sample of 20 mL was taken to count the number of larvae and extrapolate this number per volume to the 3L-beakers. Larvae were gently transferred to plankton nets that were hanging in conic tanks filled with seawater and connected to a 338 L recirculating system. Larval survival was daily recorded until all larvae of the particular batch had died.

RNA-seq was performed on oocytes and eggs from three feminized females for each treatment (DHP, P). RNA was extracted using the miRNeasy Mini Kit (Qiagen, Venlo, Netherlands). Integrity and concentration of the RNA samples were determined using an Agilent TapeStation 4200 device. RNA concentrations varied from 132 to 796 ng μL⁻¹ and RIN values were between 7.0 and 10.0. Barcoded RNA-seq libraries were prepared from 0.5 μg of total RNA using the TruSeq Stranded mRNA Library Prep kit according to the manufacturer’s instructions (Illumina Inc., San Diego, CA, United States). RNA-seq libraries were sequenced on an Illumina NovaSeq 6000 System according to the manufacturer’s protocol. Image analysis and base calling were done by the Illumina pipeline (Illumina Inc., San Diego, CA, United States). RNA-seq data yield varied from ∼12 to ∼16 million paired-end 2 × 150 bp reads per sample, corresponding to ∼3.7–4.9 Gb per sample.

Quantitative analysis of the RNA-seq datasets was performed by alignment of reads against the Anguilla anguilla reference genome (https://www.ncbi.nlm.nih.gov/data-hub/genome/GCF_013347855.1/) using TopHat (version 2.0.13) (Trapnell et al., 2009). In total between ∼54% and ∼67% of the RNA-Seq reads could be mapped against the reference. The resulting files were filtered and secondary alignments were removed using SAMtools (version 1.2 using htslib 1.2.1) (Li et al., 2009). For comparison of gene expression levels between groups, aligned fragments per predicted gene were counted from SAM alignment files using the Python package HTSeq (version 0.6.1p1) (Anders et al., 2014). In order to make comparisons across samples possible, these fragment counts were corrected for the total amount of sequencing performed for each sample. As a correction scaling factor, we employed library size estimates determined using the R/Bioconductor package DESeq (Anders and Huber, 2010). Read counts were normalized by dividing the raw counts obtained from HTSeq by its scale factor. Aligned reads were processed using DESeq whereby treatment groups were each compared with the control group. Raw and processed RNA-Seq datasets have been submitted to NCBI’s GEO repository with reference GSE218444 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE218444; GSM6745397-GSM6745408). The comparisons between 1) oocytes sampled before the DHP injection and eggs stripped after DHP and 2) oocytes sampled before the P injection and eggs stripped after P were analyzed to assess differential gene expression and their functional clustering by GO analysis using UniProt (https://uniprot.org). Differences were considered significant when p < 0.01.

Fold change of each target gene, oocyte diameter, lipid diameter and the percentage of GVBD were compared between timepoints and doses of DHP by using the following linear mixed model: ln y i j k l = μ + T R i + T j + T R i × T j + A k + P l + e i j k l Where y i j k l is the variable to be explained (fold change of each target gene, oocyte diameter, lipid diameter and GVBD) from treatment i (control, 1 ng DHP, 10 ng DHP, 100 ng DHP) at incubation time j (0, 12 and 18 h) observed in eel k (k = 1–11) and kept in plate l (l = 1–22). In this model, μ is the overall mean, T R i is the fixed effect of the i ^th treatment where i is the hormonal concentration, T j is the fixed effect of the j ^th time, T R i × T j is the fixed interaction effect for the i ^th treatment and j ^th time, A k is the random effect of the k ^th eel, P l is the random effect of the l ^th plate, and e i j k l is the residual. The same linear mixed model was used to investigate the effect of P treatment on the fold change of each target gene, oocyte diameter, lipid diameter and GVBD. All random effects were assumed to be independent and normally distributed. Data were analyzed using R (R foundation for statistical computing, Vienna, Austria) and considered significant when p < 0.05.

The number of injections to reach maturation, the timespan after the DHP or P injection until egg release, the percentage of floating eggs, the number of larvae after hatching and larval survival were compared between DHP and P injected females with the non-parametrical Wilcoxon rank sum test. Data were analyzed with R and differences were considered significant when p < 0.05.

Oocyte samples that were used for the in vitro experiment were a mix of both folliculated and de-folliculated oocytes (Figure 1). While some oocytes were enveloped by theca and granulosa cells (Figure 1A), other oocytes were only surrounded by their membranes (Figure 1B).

Both DHP and P treatments (p < 0.001), but not time (p > 0.05), significantly affected GVBD (Figures 2A, B). Also, a significant eel effect (p < 0.05) was detected. Oocytes incubated at a dose of 100 ng DHP underwent GVBD at a higher rate than oocytes treated at a dose of 10 ng DHP (Figure 2B). After 12 h of incubation, GVBD was observed in 27% of the oocytes at doses of 10 ng DHP and in 67% of the oocytes at 100 ng DHP (Figure 2B). After 18 h of incubation, GVBD was observed in 35% of the oocytes at doses of 10 ng DHP and in 71% of the oocytes at 100 ng DHP (Figure 2B). When incubating ovarian tissues with P, maturing oocytes were lacking a GV at doses of 100 ng P and 1,000 ng P (Figure 2B). After 12 h of incubation, GVBD was observed in 63% of the oocytes at doses of 100 ng P and in 65% of the oocytes at 1,000 ng P (Figure 2B). After 18 h of incubation, GVBD was observed in 65% of the oocytes at doses of 100 ng P and in 75% of the oocytes at 1,000 ng P (Figure 2B). No statistical differences were detected between 100 ng P and 1,000 ng P (Figure 2B).

For both DHP and P, lipid diameter was significantly affected by time (p < 0.001) and treatment (p < 0.05). Moreover, a significant eel (p < 0.001) effect was observed. Lipid diameter increased from 101 ± 18 μm at the start of the incubation to 137 ± 34 µm after 12 h of incubation and 146 ± 38 µm after 18 h of incubation in the controls (Figure 3). At a dose of 100 ng DHP after 18 h of incubation, the lipid diameter increased from 146 ± 38 μm to 163 ± 33 µm which was not statistically different from the 18 h control, probably due to the high variation in the data (Figure 3A). At a dose of 1,000 ng P after 18 h of incubation, the lipid diameter significantly increased from 146 ± 38 μm to 176 ± 23 µm when compared to the 18 h control (Figure 3B). Oocyte diameter did not change with time nor treatment (Supplementary Figure S1).

For both DHP and P, expression of the nuclear progestin receptor pgr1 was significantly affected by time (p < 0.01), treatment (p < 0.001) and time × treatment interaction (p < 0.05), and a significant eel effect (p < 0.001) was observed. Pgr1 expression decreased over 4-fold with time and over 15-fold after 18 h of incubation, for both treatments (Figure 4A). At a dose of 100 ng DHP, pgr1 expression decreased after 18 h of incubation when compared to the controls (Figure 4A1). Similarly, at the highest doses of P (100, 1,000 ng), pgr1 expression decreased after 18 h of incubation when compared to the controls (Figure 4A2). No statistical differences were detected between the doses of 100 ng and 1,000 ng P after 18 h of incubation (Figure 4A2).

For both DHP and P, expression of the nuclear progestin receptor pgr2 was significantly affected by time (p < 0.05) and treatment (p < 0.001), and a significant eel effect (p < 0.01) was observed. Expression of pgr2 decreased over 2-fold with time and over 20-fold after 12 and 18 h of incubation for both treatments (Figure 4B). At doses of 10 ng DHP, 100 ng DHP, 100 ng P and 1,000 ng P, the expression of pgr2 decreased after 12 and 18 h of incubation when compared to the controls (Figure 4B). A dose of 100 ng DHP decreased the expression of pgr2 further than a dose of 10 ng DHP (Figure 4B1). No statistical differences were detected between doses of 100 ng and 1,000 ng P (Figure 4B2).

For both DHP and P, expression of the luteinizing hormone receptor lhcgr1 was significantly affected by time (p < 0.001) and treatment (p < 0.05) and a significant eel effect (p < 0.05) was observed. Expression of lhcgr1 decreased over 20-fold with time and remained low with exception of the lower doses for both treatments after 12 and 18 h of incubation (Figure 4C). When incubating ovarian tissues with DHP for 12 and 18 h, at the lower doses of DHP (1 and 10 ng), but not DHP at 100 ng, the expression of lhcgr1 increased when compared to the controls (Figure 4C1). Similarly, at the lowest dose of P (10 ng) after 12 and 18 h of incubation, but not the highest ones (P 100 ng, P 1,000 ng), the expression of lhcgr1 increased when compared to the controls (Figure 4C2). The expression of the luteinizing hormone receptor lhcgr2 remained low and stable with time and treatment; significant differences were detected but differences were not correlated with time and treatment (Supplementary Figure S2).

For both DHP and P, expression of the follicle stimulating hormone receptor fshr was significantly affected by time (p < 0.05) and time × treatment interaction (p < 0.05) and a significant eel (p < 0.01) effect was detected. Fshr expression remained low and stable with exception of a 5-fold increase after 12 h of incubation for both treatments (Figure 4D). When ovarian tissues were incubated with DHP, treatment did not affect fshr expression (p > 0.05) with exception of one dose after 12 h of incubation. A dose of 100 ng DHP significantly increased fshr expression after 12 h of incubation when compared to the controls and the lowest DHP concentrations (1, 10 ng) (Figure 4D1). For P, treatment significantly affected fshr expression (p < 0.05). Doses of 100 ng and 1,000 ng P increased the expression of fshr after 12 h of incubation when compared to the controls and lowest P concentration (10 ng) (Figure 4D2). No statistical differences were detected between the doses of 100 and 1,000 ng P after 12 h of incubation (Figure 4D2).

For both DHP and P, expression of the prostaglandin receptor ptger4b was significantly affected by time × treatment interaction (p < 0.001) and a significant eel (p < 0.01) effect was detected. Ptger4b expression remained low and stable with exception of a 13-fold increase after 12 h of incubation for both treatments (Figure 4E). When ovarian tissues were incubated with DHP, treatment did not affect ptger4b expression (p > 0.05) with exception of 100 ng DHP which increased the expression of ptger4b after 12 h of incubation when compared to the controls and doses of 1 ng DHP and 10 ng DHP (Figure 4E1). P treatment significantly affected ptger4b expression (p < 0.05). Ovarian tissues incubated with P increased ptger4b expression in a similar manner as DHP treatment (Figure 4E2). At doses of 100 ng and 1,000 ng P, the expression of ptger4b increased after 12 h of incubation when compared to the controls and dose of 10 ng P (Figure 4E2). No statistical differences were detected between 100 and 1,000 ng P after 12 h of incubation (Figure 4E2).

For both DHP and P, treatment (p < 0.05) and time × treatment interaction (p < 0.05) significantly affected the androgen receptor ara expression. A significant eel effect (p < 0.05) was detected. Ara expression remained low and stable with exception of a 4-fold decrease after 18 h of incubation for both treatments (Figure 4F). When incubating ovarian tissues with DHP, at a dose of 100 ng DHP, ara expression decreased after 18 h of incubation when compared to the controls (Figure 4F1). Similarly, at the highest dose of P (1,000 ng), the expression of ara decreased after 18 h of incubation when compared to the controls (Figure 4F2). No statistical differences were detected between doses of 100 and 1,000 ng P after 18 h of incubation (Figure 4F2).

An overview of the reproductive success of feminized females injected with DHP or P is shown in Table 2. The number of weekly CPE injections to reach maturation was similar between females injected with DHP (8.2 ± 1.2 injections) and P (8.0 ± 1.4 injections) (p = 0.51). Also the timespan between treatment and egg release was similar between DHP (13.8 ± 1.7 h) and P (14.0 ± 0.7 h) injected females (p = 0.42). Of the twelve eels that were injected with DHP, five died, one was stripped but did not give embryos and six eels gave larvae that survived until 8 dph. Of the ten eels that were injected with P, five died, one was stripped but did not give embryos and four eels gave larvae that survived until 5 dph. The percentage of floating eggs was similar between DHP (average: 36%) and P (average: 27%) injected females (p = 0.87). Expected differences in larvae number and survival between the DHP and P injected females were not significant (DHP: 6 females; P: 4 females).

An overview of the reproductive success of wild females injected with DHP and P is shown in Table 3. The number of weekly CPE injections to reach maturation was similar for the eels used in both treatment groups. Eels that required 9 ± 1.5 CPE injections were injected with DHP, eels that required 9 ± 1.0 CPE injections were injected with P (p = 0.26). The timespan between treatment and egg release was similar between DHP (11.9 ± 1.1 h) and P (12.5 ± 2.3 h) injected females (p = 0.75). When comparing wild and feminized eels, wild females released eggs sooner after treatment (12.2 ± 1.7 h) than the feminized ones (14.1 ± 1.4 h) (p = 0.004). Of the nine females that were injected with DHP, two died and seven eels were stripped and gave larvae that survived until 20 dph. Of the nine wild females that were injected with P, two died, one released eggs but did not give embryos and six were stripped and gave larvae that survived until 21 dph. No significant differences were detected between the DHP and P injected females regarding reproductive success. The percentage of floating eggs ranged from 18% to 92% for the eels that received DHP treatment (average 54%) and from 1% to 92% for eels that received P treatment (average: 63%). The number of larvae ranged from 50 to 20,000 for DHP treatment (average: 2,993 larvae) and from 20 to 45,000 for P treatment (average: 12,753 larvae). Larvae survived on average for 7 dph for each of the two treatments.

36,061 transcripts that were associated with A. anguilla genes in NCBI were detected when comparing: 1) eggs stripped after DHP injection and eggs stripped after P injection; 2) oocytes sampled at the time of DHP injection and eggs stripped after DHP injection; 3) oocytes sampled at the time of P injection and eggs stripped after P injection. When comparing eggs that were stripped after either DHP or P injection, only one gene (glutathione S-transferase Mu 3-like) was significantly different and upregulated in eggs stripped after P injection (p < 0.01; Supplementary Table S1). The comparison between oocytes sampled at the time of DHP injection and eggs stripped after DHP injection yielded 1,710 differentially expressed genes (DEGs) (p < 0.01; Supplementary Table S2). While 1,599 genes of these 1,710 DEGs were shared with the comparison between oocytes sampled at the time of P injection and eggs stripped after P injection, 111 genes were not (p < 0.01). The comparison between oocytes sampled at the time of P injection with eggs stripped after P injection yielded 5,074 DEGs (p < 0.01; Supplementary Table S3). As stated, 1,599 genes of these 5,074 DEGs were shared with the comparison between oocytes sampled at the time of DHP injection and eggs stripped after DHP injection, 3,475 genes were not (p < 0.01).

Among the 1,710 DEGs that were differentially expressed between oocytes sampled at the time of DHP injection and eggs stripped after DHP injection (Supplementary Table S2), most genes that were targeted in vitro were found (Table 1). Alike the in vitro results (Figure 4), the luteinizing hormone/choriogonadotropin receptor, progesterone receptor and androgen receptor-like were downregulated between oocytes sampled at the time of DHP injection and eggs stripped after DHP injection (Supplementary Table S2). Alike the in vitro results of ptger4b after 18 h of incubation (Figure 4), the prostaglandin E2 receptor EP4 subtype-like was not differentially expressed in vivo (Supplementary Table S2). In contrast to the in vitro results (Figure 4), the follicle-stimulating hormone receptor was downregulated in vivo (Supplementary Table S2); probably due to its transient expression (Figure 4). The same results were observed when comparing the in vitro (Figure 4) and in vivo (Supplementary Table S3) results following P treatment.

On basis of fold change magnitude, the top 200 genes of the 1) 1,710 DEGs that were yielded from the comparison between oocytes sampled at the time of DHP injection and eggs stripped after DHP injection and 2) 5,074 DEGs that were yielded from the comparison between oocytes sampled at the time of P injection and eggs stripped after P injection were categorized into groups according to their biological functions to understand what happens between the injection moment and egg release (Table 4). Among these DEGs were several genes that were involved in apoptosis, cell adhesion, cell cycle, cytoskeleton organization, extracellular matrix organization, transport, lipid metabolism, muscle contraction, steroid signalling pathway and steroidogenesis. Other genes, presented in Supplementary Table S4, were involved in respiration, mRNA processing/splicing, transcription, translation and ubiquitination.

The 111 DEGs that were differentially expressed between oocytes at the moment of DHP injection and eggs stripped after DHP, and different from comparing oocytes at the moment of P injection and eggs stripped after P, were categorized into groups according to their biological function to understand what DHP induced more than P between the injection and egg release moment (Supplementary Table S5). Among these DEGs were several genes involved in apoptosis and inflammation, cell adhesion, cell cycle, transport, transcription and ubiquitination.

On basis of fold change magnitude, the top 200 genes of the 3,475 DEGs that were differentially expressed between oocytes at the moment of P injection and eggs stripped after P, and different from comparing oocytes at the moment of DHP injection and eggs stripped after DHP, were categorized into groups according to their biological function to understand what P induced more than DHP between the injection and egg release moment (Supplementary Table S6). Among these DEGs were several genes that were involved in cell adhesion, cell cycle, extracellular matrix organization, inflammation, ion transport, mRNA processing/splicing, muscle contraction, mitochondrial respiration, steroidogenesis, transcription and ubiquitination.