Establish a placental-specific BDNF knockout mouse model (cKO mouse)
Cre-loxP technology was used to generate cKO mice (Fig. 1A). Elf5-cre mice (Elf5-Cre+ mice) were kindly given by Dr. Haibin Wang for trophoblast-specific gene manipulation[27]. LoxP BDNF mice (BDNFloxP/loxP mice) were purchased from Shanghai Model Organisms. Elf5-Cre+ mice were bred with BDNFloxP/loxP mice to generate heterozygous placental-specific BDNF knockout mice [cKO(HE) mice]. To produce homozygous knockout mice [cKO(HO) mice], cKO(HE) male mice were bred with BDNFloxP/loxP or cKO(HE) females. However, no alive cKO(HO) mice were born (Fig. 1B). Abnormal cKO(HO) embryo masses (early absorptions) were found at necropsy at E18.5 (Fig. 1C-D). Thus, cKO(HE) mice were used for this study as cKO(HO) appeared to be lethal. To determine the effectiveness and specificity of elf5-Cre to reduce BDNF expression in the placenta, we measured BDNF protein in the placenta and maternal brain. We demonstrated that BDNF protein expression was significantly reduced in cKO(HE) placentae by Western Blot (Fig. 1E) and immunohistochemical staining (Fig. 1F) compared to the controls. In maternal brains, BDNF protein expression was not changed among groups.
The ovarian function and reproduction were decreased in cKO(HE) female mice
We first noticed that four-month-old cKO(HE) female mice had abnormal litters. Specifically, the number of pups significantly decreased although there were no significant differences in pup weight in litters of four-month-old cKO(HE) mice compared to litters of control mice (Fig. 2A-B). The ovarian weight and volume were significantly lower in four-month-old cKO(HE) mice compared to the control mice (Fig. 2C-D). There are visible gaps and absorbed embryos in the gravid uterus of cKO(HE) mice at E11.5 and E18.5 (Fig. 2E). The number of follicles is a predictor of oocyte production. H&E staining was used to examine the ovarian follicle development (Fig. 2F). The numbers of primordial and preantral follicles in four-month-old cKO(HE) mice were significantly lower than those in the control mice, but the numbers of primary and secondary follicles were not different (Fig. 2G). BDNF protein expression in ovaries significantly decreased in four-month-old cKO(HE) mice compared with the control mice (Fig. 2H). FSH expression in ovaries significantly increased in four-month-old cKO(HE) mice compared to the control mice (Fig. 2I).
We then examined the ovarian function in younger cKO(HE) mice. We found that the number of follicles significantly decreased in postnatal-day-7-old (P7) cKO(HE) pups compared with the controls while the body weight of pups had no difference (Fig. 3A-C). The BDNF protein expression in ovaries was not different (Fig. 3D). In the two-month-old mice, the number of primordial and secondary follicles in cKO(HE) mice was significantly lower than those in the controls, while the numbers of primary and preantral follicles were not different comparing to the controls (Fig. 3E-F). The ovary and body weight of these mice were not different (Fig. 3G-H). One-month-old female mice are the best recipients for superovulation and harvesting eggs[37]. In one-month-old cKO(HE) mice, the number of eggs significantly decreased compared with the controls (Fig. 3I). H&E staining was used to examine the ovarian follicle after superovulation (Fig. 3J). There is no difference in primordial, primary, and secondary follicles. However, the number of preantral follicles in one-month-old cKO(HE) mice was lower than those in the controls after superovulation (Fig. 3K-L). These results indicate that ovaries in cKO(HE) mice are not sensitive to superovulation induction drugs, pregnant mare serum gonadotropin (PMSG), and human chorionic gonadotropin (hCG). Gonadotropins such as PMSG and hCG have follicle-stimulating activity and induce oocyte maturation, leading to ovulation and the production of mature eggs[38, 39]. The decrease in the number of mature eggs produced in young cKO(HE) mice after superovulation indicates that the number of available follicles for recruitment was reduced or cells in follicles including oocytes, granulosa cells, and theca cells are less responsive to gonadotrophin.
In summary, we observed premature ovarian failure (POI) phenotype in cKO(HE) mice.
Transcriptome analysis identified differentially expressed genes (DEGs) in ovaries
As shown in the volcano plot, 15 genes were significantly upregulated and 21 genes were significantly downregulated in ovaries harvested from P7 cKO(HE) mice compared to the controls. (Fig. 4A). Three DEGs with known functions were selected for RT-qPCR validation (Supplemental Table 2). RT-qPCR results confirmed that the gene expression of fibroblast growth factor 2 (fgf2) was significantly downregulated, the gene expression of protocadherin beta 21 (pcdhb21, p<0.05), and high mobility group box 1 (hmgb1) was up-regulated (p=0.05) but not statistically significant in the cKO(HE) ovaries compared with the controls (Fig. 4B). These genes have been shown to mediate mitochondrial function and apoptosis[40]. There is increasing evidence suggesting that mitochondrial dysfunction plays a key role in ovarian aging[41, 42]. Thus, we further examined mitochondria in the ovaries of P7 pups using a Scanning Electron Microscope (SEM). SEM images revealed that the number of mitochondria was reduced and the average volume of mitochondria (swollen mitochondria) was increased in P7 cKO(HE) oocytes compared with the controls. Apoptotic bodies were also observed in cKO(HE) oocytes (Fig. 4C). The combination of the transcriptome and SEM results indicate that mitochondria-mediated cell death or aging might occur in cKO(HE) ovaries.
Transcriptome analysis of oocytes harvested from one-month-old mice showed an aged gene profile in cKO(HE) oocytes compared to controls.
The transcriptome profile undergoes dramatic changes during oocyte aging. The oocytes are large and rich in RNA, which provides an advantage in RNA-seq (scRNAseq) studies. As shown in the volcano plot and heatmap, 646 genes were significantly upregulated and 230 genes were significantly downregulated in cKO(HE) oocytes compared to the controls (Fig. 5A-B). GO enrichment analysis indicated that mitochondrial inner membrane, embryonic organ development, and positive regulation of protein kinase activity were significantly enriched (Fig. 5C). KEGG pathway classification indicated that DEGs associated with chemical carcinogenesis − reactive oxygen species, thermogenesis, and oxidative phosphorylation signal pathways (Fig. 5D). The top 10 DEGs ranked according to p-values include rpl37rt, gm4350, pcna-ps2, pakap, gm17535, pla2g4e, gm10800, duox2, gm10722, lrrc14, ugcg, gm37915, uqcr11, uba52, rpl41, ndufa11, gm1673, gm10076, samhd1 and gm6166 (Supplementary Table 3). KEGG pathway enrichment analyses demonstrated that up-regulated DEGs were enriched in arginine and proline metabolism signals (Fig. 5E). And down-regulated DEGs were enriched in oxidative phosphorylation (Fig. 5F). The top 5 genes in arginine and proline metabolism pathway are azin2, arg1, aoc1, prodh, and smox. The top 5 genes in the oxidative phosphorylation pathway are uqcr11, ndufs8, cox11, ndufa7, and ndufa11 (Supplementary Table 4). Increased arginine metabolism may result in the accumulation of methylarginines which has been shown to negatively impact IVF outcomes[43, 44]. Reduced mitochondrial metabolism and oxidative phosphorylation indicate that the oocytes are aging[45]. When oxidative phosphorylation is downregulated, it can lead to decreased ATP production, increased oxidative stress, and impaired cellular function[46, 47]. Increased oxidative stress can lead to cell damage, inflammation, and cellular aging. A study found that some genes associated with oxidative stress were significantly upregulated in the oocytes of older women, while they were downregulated in the oocytes of younger women[48, 49].
The number and cell proliferation biomarkers of PGCs decreased in cKO(HE) embryos
The ovary BDNF levels of postnatal-day-7-old pups were not different between cKO(HE) mice and the controls. Thus, we suspect the POI phenotypes in cKO(HE) offspring originated from embryo development. We did not observe noticeable differences in the morphology of embryos at E11.5 (Fig. 6A). Intriguingly, alkaline phosphatase staining demonstrated that the number of PGCs significantly decreased in cKO(HE) embryos compared with the controls (Fig. 6B). As PGCs are highly proliferative between E7.5 and E13.5 to establish the ovarian reserve, we suspected that PGCs cell proliferation was inhibited in cKO(HE) mice. Previous studies reported that BDNF regulates cell proliferation through Cyclin D1[50, 51]. Here, we report that the expression of Cyclin D1 significantly decreased in cKO(HE) embryos' genital ridges compared with the controls (Fig. 6C). Overall, we demonstrated that the PGCs proliferation and PGCs pool were negatively affected by BDNF knockdown in the placenta in cKO(HE) embryos. The diminished ovarian reserve during embryonic development may underlie the observed POI in cKO(HE) offspring.
Placenta BDNF-PGCs-POI- axis
Reduction of placenta BDNF leads to a reduced Cyclin D1 expression in embryos' genital ridges and a reduced number of PGCs. which ultimately leads to POI in adulthood. The increased hmgb1 expression and decreased fgf2 expression and dysregulated mitochondrial function in P7 cKO(HE) ovaries indicate that ovarian aging in these mice may be mediated by mitochondria (Fig. 7).