The Dennd1a mutant strain used in this study was reported previously [7]. Wild-type Kunming mice were purchased from the Laboratory Animal Center of Shandong University. The sex was determined by PCR analysis using Sry primers (Table 1) [20]. Care and use of experimental animals described in this work comply with the guidelines and policies of the University Committee on Animal Resources at Shandong University.

Mouse tissues were dissected and fixed in 4% paraformaldehyde at 4 ${}^{\circ }$C overnight. The fixed and paraffin-embedded tissues were sectioned in a thickness of 5 $\mu$m. For immunofluorescence double-labeling, the sections were incubated with mixture of primary antibodies overnight, and then incubated with fluorescently labeled secondary antibody. The sections were mounted in medium with DAPI (abcam, ab104139). Primary antibodies used include Rabbit anti-Aldh1a1 (cell signaling technology, 12035), Rabbit anti-Aldh1a2 (cell signaling technology, 83805S), Rabbit anti-Dennd1a (abcam, ab125347). Secondary Antibodies include Goat anti-Chicken IgY (H + L) Alexa Fluor 488 (Invitrogen, A-11039), Goat anti-Rabbit IgG (H + L) Cross-Adsorbed Alexa Fluor 546 (Invitrogen, A-11010), Goat anti-Mouse IgG (H + L) Highly Cross-Adsorbed Alexa Fluor 488 (Invitrogen, A11029). Images were obtained by confocal fluorescence microscope (Dragonfly, Andro, England) and analyzed by Imaris 9 software (Oxford Instruments, England).

For immunohistochemistry, the sections of E11.5 embryos were incubated with Dennd1a antibodies which were detected with horseradish peroxidase (HRP)-conjugated secondary antibodies (ZSGB-BIO, PV-9001), followed by enzymatic color reaction (Vector Laboratories, SK-4100). Images were taken using OLYMPUS BX53F microscope and analyzed by OLYMPUS cellSens Imaging 1.15 software (OLYMPUS, Japan).

The fetal gonads were obtained from E12.5 embryos, and divided into 2 pieces and cultured on 24-well plates (corning) with 1 mL of growth medium in each well at 37 ${}^{\circ }$C with 5% CO${}_2$ for 24–96 h. The growth medium was composed of $\alpha$-minimal essential medium ($\alpha$-MEM) (Gibco, 12571063), supplemented with 10% heat-inactivated fetal bovine serum (BI, 04-001-1ACS), 0.23 mM pyruvic acid (Gibco, 11360070), 10 mIU/mL FSH (Merckserono), 100 mIU/mL penicillin G, and 100 mg/mL streptomycin sulfate (Gibco, 10378016). The gonad tissues were treated with Retinoic acid (R&D, 0695/50) at a concentration of 1 $\mu$M, Recombinant human/mouse Wnt5a Protein (R&D, 645-WN-010) at a concentration of 350 ng/mL and combination respectively.

The somatic cells were obtained from E12.5 wild-type female mesonephros, which were cultured in each well of a 6-well plate with 2 mL of Dulbecco’s modified Eagle’s medium (Gibco, 11995065) containing 10% (v/v) fetal bovine serum (FBS) (BI, 04-001-1ACS) and 1% Penicillin/Streptomycin (Gibco, 10378016) in a humidified incubator at 37 ${}^{\circ }$C with 5% CO${}_2$ for 24–96 h. The somatic cells were treated with LY294002 (Sigma, 10 $\mu$M) for 24 h to inhibit PI3K activity, or treated with Wnt3a (R&D, 5036WN, 100 ng/mL) for 24 h to induce Wnt/$\beta$-catenin signaling.

The adenoviral Dennd1a shRNA (Vigene Biosciences) containing GFP labeling was added into medium of cultured tissues. The adenovirus containing no inserted gene was used as control. The gonad tissues were successfully transfected with adenovirus at the dose of 1 $\times$ 10${}^{11}$ PFU/mL, while the somatic cells were successfully transfected at the dose of 5 $\times$ 10${}^7$ PFU/mL. The efficiency of Dennd1a knockdown was measured using Western blot and real-time RT-PCR analysis.

Total RNA kit I (OMEGA, R6834-02) was used to extract total RNA from gonads and mesonephros. The total RNA was subject to the first strand cDNA synthesis using the PrimeScript RT reagent kit with gDNA Eraser (Takara, Japan, RR047A). RNeasy Mini Kit (QIAGEN,74104) was used to extract RNA from the fetal germ cells and somatic cells of the gonads of E11.5 or E13.5 embryos. The first strand cDNA synthesis was performed using the Evo M-MLV RT for PCR Kit (Accurate Biology, AG11603). The cDNAs were then amplified by real-time PCR with TB Green Premix Ex Taq (Takara, RR420A) in Light Cycler real-time PCR instrument (Roche, Basel, Switzerland, LC480) according to the manufacture’s specification. Gene expression changes were normalized to Gapdh and analyzed by the 2${}^{-\mathrm{\Delta }\mathrm{\Delta }Ct}$ method. The primer sequences are listed in Table 1.

Proteins were extracted from the fetal gonads using Minute™ Total Protein Extraction Kit for Animal Cultured Cells/Tissues (Invent, SD-001/SN-002). Proteins from the tissue or cell cultures were extracted using T-PER™ Tissue Protein Extraction Reagent (Thermo, 78510). Complete™ Protease Inhibitor Cocktail (Roche, 04693116001) were used to inhibit the degradation of a broad spectrum proteases. The extracted samples were separated, then transferred onto polyvinylidene fluoride membranes (Millipore, Billerica, MA, PFL00010). The membranes were probed with Rabbit anti-Dennd1a (abcam, ab125347), Rabbit anti-ALDH1A1(cell signaling techonogy, 12035, Rabbit anti-ALDH1A2 (cell signaling technology, 83805S), Rabbit anti-Wnt5a (cell signaling technology, 2392S), Rabbit anti-non-phospho $\beta$-catenin (cell signaling technology, 8814), Rabbit anti-phospho-AKT (cell signaling technology, 4060), Rabbit anti-AKT (cell signaling technology, 9272), Rabbit-anti-phospho-Stat3 (cell signaling technology, 9145), Rabbit anti-Stat3(cell signaling technology, 9134), Rabbit anti-Axin2 (abcam, ab109307) primary antibodies at 4 ${}^{\circ }$C overnight, followed by incubation with peroxidase-conjugated goat anti-rabbit or anti-mouse immunoglobulin G antibodies (ZSGB-BIO, ZB-2301, ZB2305) for 1 h at room temperature. The interaction was monitored with Thermo Scientific SuperSignal West Pico PLUS (Thermo, 34577). Anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Proteintech, 60004-1-Ig) or Alpha Tubulin (proteintech, 11224-1-AP) antibodies were used to monitor the loading amount. BIO RAD software was used for analysis.

Supernatant culture medium of gonads or somatic cells was collected to quantify concentrations of RA using Mouse Retinoic Acid ELISA kit (CUSBIO, CSB-EQ028019MO) according to the manufacture’s specification, using a microplate reader (Molecular devices; SPECTRA MAX 384plus) set to 450 nm. The software used was SoftMax Pro7 (Molecular Devices, USA). Each standard and sample were in duplicates, calibrated with blank well without any solution, and a standard curve was set up by using software Curve Expert. The data was linearized by plotting the log of the Retinoic acid concentrations versus the log of the O.D., and the best fit line was determined by regression analysis.

Graphpad PRISM8 analysis software was used for plotting, and SPSS (version 21.0, IBM Corp., Chicago, IL, USA) analysis software was used for statistical analysis. p 0.05 was statistically significant.

At E11.5 during mouse embryo development, Dennd1a protein was found in both the hindgut mesentery and genital ridge of mouse embryos (Fig. 1A), suggesting Dennd1a may play a role when PGCs are migrating through the mesentery and colonized in the genital ridge. In fetal ovaries, the levels of Dennd1a mRNA were gradually increased on E12.5, E13.5, and E15.5, during early stages of the oocyte differentiation and meiosis (Fig. 1B). The findings suggest that Dennd1a-mediated signaling pathway may be involved in the regulation of fetal ovary development.

Fig. 1.

Expression of Dennd1a in the fetal ovary. (A) Immunohistochemistry shows the localization of Dennd1a in the genital ridge and posterior mesentery of E11.5 mouse embryo. The red arrow indicates genital ridge. The black arrow indicates posterior mesentery. Scale bar = 20 $\mu$m (n = 3). (B) Quantitative analysis of Dennd1a mRNA expression levels in the wild-type fetal ovaries of E12.5, E13.5, and E15.5 embryos by real time RT-PCR (n = 5).

To assess the oocyte differentiation and initiation of meiosis during embryonic development, we examined mRNA expression of Sohlh2, Figla (Factor in the germLine $\alpha$), Stra8, and Rec8 in the wild-type fetal ovaries at E12.5, E13.5, and E15.5. As transcription factors in the germ cells or oocytes, Sohlh2 mRNA level was elevated at E13.5 and then declined at E15.5, while mRNA expression of Figla was activated at E13.5 and dramatically promoted at E15.5 (Fig. 2A). The transcription of Stra8 and Rec8, two genes essential for oocyte meiosis, were both activated circa E13.5 (Fig. 2A). However, using Dennd1a gene knockout mice, we discovered that in the ovaries of homozygous Dennd1a-/- mutants at E13.5, mRNA expression of these genes were insufficiently induced. Deletion of Dennd1a affected the transcription of not only Sohlh2, but Figla, Stra8 and Rec8 as well at this developmental stage (Fig. 2B).

Fig. 2.

Dennd1a deficiency impairs the initiation of oogenesis and meiosis of the fetal germ cells. (A) Real-time RT-PCR analysis shows the mRNA expression levels of Sohlh2, Figla, Stra8 and Rec8 in wild-type fetal ovaries of E12.5, E13.5 and E15.5 mouse embryos (n = 5). (B) Real-time RT-PCR analysis of the mRNA levels of Sohlh2, Figla, Stra8, and Rec8 in the fetal ovaries of E13.5 Dennd1a+/- and Dennd1a-/- embryos (n = 3). (C) (i–iv) Morphological changes of E12.5, E13.5, E14.5, E15.5 wild-type fetal ovaries, scale bar = 50 $\mu$m (n = 3); (v–viii) Morphological changes of E12.5 wild-type fetal ovaries ex vivo cultured for 0, 24, 48, or 72 h, scale bar = 100 $\mu$m; (n = 3) (ix–xii). The ex vivo culture of E12.5 wild-type fetal ovaries were infected with adenoviral Dennd1a shRNA carrying a GFP fluorescent tag for 24, 48, 72, or 96 h; scale bar = 100 $\mu$m (n = 3). (D) The mRNA expression of Sohlh2, Figla, Stra8, and Rec8 in E12.5 wild-type fetal ovaries cultured ex vivo for 24, 48, 72 and 96 h were examined by Real-time quantitative RT-PCR analysis. (E) The mRNA expression of Sohlh2, Figla, Stra8, and Rec8 in the E12.5 gonad culture infected with Dennd1a shRNA adenovirus for 72 h was shown as representative (n = 3). The knockdown efficiency of Dennd1a was examined by real time PCR and Western Blot (n = 3).

Since homozygous Dennd1a-/- mutants die circa E14.5, they are not suitable to determine whether loss of Dennd1a eventually disrupts the expression of genes associated with oocyte differentiation and meiotic initiation. Moreover, due to limited availability of Dennd1a-/- gonads, we therefore established an ex vivo gonad culture by incubating genital ridges isolated from E12.5 wild-type female embryos. The long and narrow E12.5 gonads ex vivo became short and thick after being cultured for 24, 48, and 72 h (Fig. 2C, v–viii), recapitulating the morphological changes of developing gonads in vivo from E13.5 to E15.5 (Fig. 2C, i–iv). The mRNA expression of Sohlh2, Figla, Stra8 and Rec8 in wild-type gonads ex vivo were activated after being incubated for 48 h (Fig. 2D), showing a similar expression pattern as that in gonads in vivo at E13.5 (Fig. 2A). To knockdown the expression of Dennd1a, the ex vivo E12.5 gonads were infected with adenoviral Dennd1a shRNA carrying a GFP fluorescent tag for 24, 48, 72, or 96 h (Fig. 2C, ix–xii). After 48 h or longer of infection, the expression of Dennd1a mRNA and protein were decreased to approximately half the wild-type level (Fig. 2E). In gonads with Dennd1a knockdown, the mRNA expression of Sohlh2, Figla, Stra8 and Rec8 were not fully activated, as shown in representative samples treated for 72 h (Fig. 2E). These results suggest that Dennd1a deficiency disrupted the proper activation of genes associated with oocyte differentiation and initiation of meiosis in fetal ovaries.

We suspected that Dennd1a deficiency disturbed somatic components which are important for inducing fetal oocyte differentiation and meiosis. It has been reported that Wnt5a was predominantly expressed in the somatic niches and necessary for gonadal differentiation [15]. Furthermore, Wnt5a was involved in oocyte maturation and meiotic resumption during oocyte development [19]. Therefore, we examined whether loss of Dennd1a disrupts Wnt5a production in the somatic cells. In E12.5 female gonad culture infected with adenoviral Dennd1a shRNA for 48, 72 or 96 h, the expression of Dennd1a mRNA and protein was disrupted (Fig. 3A,B). Knockdown of Dennd1a decreased the mRNA and protein levels of Wnt5a in the gonad culture (Fig. 3A,B).

Fig. 3.

Production of Wnt5a and RA is reduced in the somatic cells of the fetal ovary lacking Dennd1a. (A,B) Real time PCR and Western blot show the mRNA and protein levels of Dennd1a, Wnt5a, and Aldh1a1 in E12.5 wild-type fetal ovaries cultured ex vivo and infected with Dennd1a shRNA adenovirus (KD) or adenovirus containing no inserted gene as control for 72 h (n = 3). (C) ELISA shows the concentrations of RA in the culture medium of E12.5 wild-type fetal ovaries cultured ex vivo and infected with Dennd1a shRNA adenovirus (KD) or adenovirus containing no inserted gene as control for 72 h (n = 5). (D) The primordial germ cells (PGC) and the somatic cells (SC) were isolated from E11.5 female gonads. The mRNA expression of Aldh1a1 and Wnt5a in the PGC and SC isolated from E11.5 female gonads were examined by real time PCR (n = 5). (E) Total RNA was extracted from E13.5 wild-type fetal ovaries and mesonephros. The mRNA expression of Dennd1a, Aldh1a1, Aldh1a2, and Wnt5a was analyzed by real-time quantitative PCR (n = 5). (F) Immunofluorescence shows the localization of Aldh1a1 in the ovary and Aldh1a2 in the mesonephros of E13.5 embryos (Red). The section was counterstained with DAPI (blue). The Dashed line indicates the boundary between the fetal ovary and mesonephros (scale bar 40 $\mu$m, n = 3). (G,H) The mRNA and protein of Wnt5a and Aldh1a2 in E12.5 wild-type fetal mesonephros cultured in vitro and infected with Dennd1a shRNA adenovirus or adenovirus containing no inserted gene as control for 96 h were examined by real time PCR and Western blot (n = 3). (I) ELISA shows the concentrations of RA in the culture medium of E12.5 wild-type fetal mesonephros cultured in vitro and infected with Dennd1a shRNA adenovirus (KD) or adenovirus containing no inserted gene as control for 72 h (n = 5).

Fetal germ cells in the ovary initiate meiosis in response to RA [12]. Deficiency of Dennd1a also reduced the mRNA and protein expression of Aldh1a1, which encodes RA-synthesizing enzyme in the fetal ovary (Fig. 3A,B), and reduced secretion of RA into the culture medium (Fig. 3C). The main source of Wnt5a and RA in the fetal ovary is the somatic cells as confirmed by quantitative analysis of the mRNA expression of Wnt5a and Aldh1a1 in isolated PGCs and somatic cells from E11.5 female gonads (Fig. 3D). It seemed that Dennd1a was involved in the production of Wnt5a and RA from the somatic cells of the fetal ovary.

An alternative source of RA and Wnt5a is the fetal mesonephros. At E13.5, unlike the ovary which expressed mainly Aldh1a1, the mesonephros expressed RA-synthesizing enzymes Aldh1a2 (Fig. 3E,F) and Aldh1a3 [12]. The mRNA expression of Dennd1a in the mesonephros was at the same level as in the ovary, while the mesonephros expressed more Wnt5a than the ovary (Fig. 3E). The mesonephros were then isolated from E12.5 wild-type female embryos, and dissociated cells were cultured in vitro and infected with adenoviral Dennd1a shRNA for 48, 72 or 96 h (Supplementary Fig. 1). Knockdown of Dennd1a resulted in decreased mRNA and protein levels of Wnt5a and Aldh1a2 in the culture (Fig. 3G,H), and decreased production of RA in the medium (Fig. 3I). Therefore, Dennd1a may play a role in regulating the production of Wnt5a and RA from the somatic cells of the fetal ovary and mesonephros.

To explore the mechanism by which Dennd1a regulates RA and Wnt5a in the somatic cells, we used the culture of fetal mesonephros which produce RA and Wnt5a, and also contribute to the somatic niche for germ cells, in order to avoid the complexity associated with the mixture of germ cells and somatic cells in gonadal culture. As a major GEF for Rab35 to regulate endosomal membrane trafficking, Dennd1a may be involved in Rab35-mediated PI3K/Akt pathway [21]. To determine whether Dennd1a deficiency may compromise PI3K/Akt pathway, the E12.5 fetal mesonephros infected with or without adenoviral Dennd1a shRNA for 48 or 72 h were assessed. Dennd1a knockdown led to decreased phosphorylation of Akt in the somatic cells (Fig. 4A,D). Dennd1a deficiency also caused reduced accumulation of non-phosphorylated $\beta$-catenin and expression of Axin2, a downstream target of $\beta$-catenin-dependent transcription (Fig. 4B,D). In addition, Dennd1a knockdown resulted in decreased expression and phosphorylation of Stat3 (Fig. 4C,D).

Fig. 4.

Knockdown of Dennd1a decreases the phosphorylation level of Akt and accumulation of active $\beta$-catenin in the mesonephros. Proteins extracted from E12.5 wild-type fetal mesonephros, which was cultured in vitro and infected with Dennd1a shRNA adenovirus (KD) or adenovirus containing no inserted gene as control for 96 h, were examined by Western blot. The expression level of Tublin or Gapdh was analyzed as a loading control (n = 5). (A) Phosphorylation levels of Akt. (B) Protein levels of non-phosphorylated active $\beta$-catenin and Axin2. (C) Phosphorylation levels of Stat3 were analyzed by Western blot. (D) Quantitative analysis of protein levels.

To provide evidence of a direct causal link between PI3K/Akt axis and Wnt5a or RA production in the mesonephros culture, we treated the somatic cells with LY294002 to inhibit PI3K activity, the efficacy of which was tested by measuring the phosphorylation levels of the PI3K target protein Akt. Treatment with LY294002 (10 $\mu$M) inhibited the phosphorylation of Akt, which was accompanied with reduced protein levels of both Wnt5a and Aldh1a2 in the somatic cells (Fig. 5A,C), as well as less RA production in the culture medium (Fig. 5D). Moreover, in the mesonephros cell culture treated with LY294002, accumulation of active non-phosphorylated $\beta$-catenin was reduced (Fig. 5B,C), suggesting that $\beta$-catenin signaling is downstream of PI3K/Akt pathway. Inhibition of PI3K/Akt pathway also reduced the phosphorylation level of Stat3 (Fig. 5B,C), which has been linked to the regulation of Wnt5a expression.

Fig. 5.

Inhibition of PI3K/Akt pathway downregulates $\beta$-catenin signaling and reduces the production of Wnt5a and RA in the mesonephros. (A) The mesonephros culture was treated with LY294002 to inhibit PI3K activity, the efficacy of which was tested by measuring the phosphorylation levels of the PI3K target protein Akt. Western blots also show the protein levels of Wnt5a and Aldh1a2 (n = 5). (B) Immunoblot analysis shows the levels of non-phosphorylated $\beta$-catenin (active), phospho-Stat3, and Stat3 (n = 5). (C) Quantitative analysis of protein levels shown in A and B. (D) ELISA shows the concentrations of RA in the culture medium of E12.5 wild-type fetal mesonephros cultured in vitro and treated with LY294002 for 24 h (n = 5).

The role of Wnt/$\beta$-catenin signaling in regulating the production of Wnt5a or RA was then examined. Treatment with Wnt3a (100 ng/mL) induced accumulation of non-phosphorylated $\beta$-catenin in the mesonephros culture (Fig. 6A,C). Interestingly, the protein level of Wnt5a and phosphorylation level of Stat3 was increased by Wnt3a treatment (Fig. 6A–C), indicating that Wnt/$\beta$-catenin signaling may induce Wnt5a expression in the mesonephros. Treatment with Wnt3a also increased the levels of Aldh1a2 in the mesonephros and RA in the culture medium (Fig. 6A,C,D). The results suggest that Akt/$\beta$-catenin pathways may be involved in Dennd1a-mediated Wnt5a and RA production in the somatic cells during fetal germ cell development.

Fig. 6.

Wnt3a/$\beta$-catenin signaling induces Wnt5a and RA synthesis in the mesonephros. (A) Immunoblot analysis shows the protein levels of non-phosphorylated $\beta$-catenin, Wnt5a, and Aldh1a2 in the mesonephros with or without treatment of Wnt3a (100 ng/mL) for 24 h (n = 5). (B) Phosphorylation levels of Stat3 in the mesonephros were examined by Western blot (n = 5). (C) Quantitative analysis of protein levels shown in A and B. (D) ELISA shows the concentrations of RA in the culture medium of E12.5 wild-type fetal mesonephros cultured in vitro and treated with or without Wnt3a (100 ng/mL) for 24 h (n = 5).

To determine the effect of Wnt5a and RA on the regulation of genes associated with oocyte differentiation and meiotic initiation, we treated the fetal gonad culture from E12.5 female wild-type mice with recombinant Wnt5a (350 ng/mL) or RA (1 $\mu$M) or both for 48 h. Exogenous Wnt5a alone induced mRNA expression of Sohlh2 in the fetal ovary, but didn’t affect gene expression of Figla, Stra8, or Rec8 (Fig. 7A). RA alone did not affect Sohlh2 expression while it decreased the expression of Figla, and as expected, RA induced the expression of Stra8 and Rec8 in the fetal ovary (Fig. 7A). Addition of both Wnt5a and RA did not significantly alter the expression of Sohlh2, Figla, and Rec8, although they increased the expression of Stra8 (Fig. 7A). These findings suggested that Wnt5a could induce oocyte differentiation, whereas it had no effect on meiotic initiation of the fetal germ cells. Furthermore, RA did not work synergistically with Wnt5a to affect oocyte differentiation while it stimulated initiation of meiosis in the fetal ovary.

Fig. 7.

Exogenous WNT5A and RA stimulate the expression of genes associated with initiation of oogenesis and meiosis in the fetal germ cells. (A) The fetal gonad culture of E12.5 female wild-type mouse embryos was treated with recombinant Wnt5a (350 ng/mL) or RA (1 $\mu$M) or both for 48 h. The mRNA expression of Sohlh2, Figla, Stra8, Rec8 were examined by real time PCR (n = 5). (B) The control and Dennd1a knockdown fetal ovaries were cultured ex vivo with or without Wnt5a or RA for 48 h. The mRNA expression of Sohlh2, Figla, Stra8, and Rec8 were examined by real time PCR (n = 5). (C) The schematic shows that Dennd1a could be involved in multiple signal pathways in the somatic cells that are critical for various processes and stages of oocyte differentiation and meiosis in the fetal ovary.

Next, we investigated whether insufficient Wnt5a and RA was responsible for dysregulation of genes associated with oocyte differentiation and meiosis initiation in the fetal ovary lacking Dennd1a. In the E12.5 fetal ovary culture infected with adenoviral Dennd1a shRNA for 48 h or 72 h, Dennd1a knockdown impaired the expression of Sohlh2. Treatment with exogenous Wnt5a (350 ng/mL) could stimulate the expression of Sohlh2 in Dennd1a knockdown fetal ovaries to the level comparable to that in wild-type control (Fig. 7B showing representative samples treated for 48 h). However, exogenous Wnt5a could not rescue the decreased mRNA expression of Figla, Stra8, or Rec8 in the fetal ovaries affected by Dennd1a knockdown (Fig. 7B). On the other hand, RA treatment (1 $\mu$M) stimulated the expression of Stra8 and Rec8 in both wild-type and adenoviral Dennd1a shRNA infected fetal ovaries, although it had no effects on the expression of Sohlh2 and Figla (Fig. 7B). The above results suggested that Dennd1a might play a role in regulating the somatic cells of fetal ovaries to produce Wnt5a and RA, which were required for the oocyte differentiation and meiotic initiation respectively (Fig. 7C).