Characterization of oogonia stem cells in mice by Fragilis

Identification of oogonia stem cells would have great potentials in infertility treatment and fertility preservation. Here we tested whether Fragilis/Ifitm3 can identify oogonia stem cells in fetal mouse ovaries and their molecular features, if it can, and whether the oogonia stem cells marked by Fragilis can be found in the postnatal ovaries. It is well established that the primordial germ cells (PGCs) in embryonic gonads undergo multiple cell divisions during migration to the genital ridge to produce oogonia that can differentiate into meiotic cells (Saitou and Miyauchi, 2016), and thus are definitive oogonia stem cells. Moreover, primordial germ cells-like cells (PGCLCs) induced from ESCs or iPSCs that resemble PGCs in vivo also are able to form functional oocytes that produce live pups (Hayashi et al., 2012; Hikabe et al., 2016). We essentially investigated the utility of Fragilis as a germ cell marker during different developmental stages. Our analysis suggests that Fragilis expression on cell surface might be a useful PGC marker at fetal E10.5 and E12.5, but its expression decreases thereafter and notably Fragilis is predominantly expressed in somatic stroma or theca and epithelial cells postnatally. Moreover, the Fragilis-expressing cells in fetal gonads are competent to undergo meiosis and generate functional oocytes in a reconstituted ovary assay, but those from postnatal gonads are not similarly developmentally competent. Comparison of the fetal and postnatal Fragilis cells in molecular signatures and function reveals that Fragilis expression at cell surface can specifically identify oogonia stem cells in fetal gonads, but its expression does not detect oogonia stem cells in postnatal ovaries. PGCs highly express specific germ cell marker genes, notably Stella/Dppa3, Blimp1/Prdm1, Ddx4/Vasa/MVH, Oct4 and Fragilis/Ifitm3 (Noce et al., 2001; Saitou et al., 2002; Tanaka et al., 2004; Ohinata et al., 2005; Okamura et al., 2008; Sabour et al., 2011). These germ cell specifiers also are conserved in humans (Kobayashi and Surani, 2018). Fragilis, as transmembrane protein, could be potentially useful for identification and sorting of PGCs or oogonia stem cells. Hence, we systematically examined the molecular signatures of Fragilis-sorted cells from fetal ovaries, and also compared with those of postnatal ovaries. We explored the expression pattern of Fragilis by co-immunostaining with known germ cell markers Vasa (Ddx4) or Dazl, in mouse fetal ovaries from E10.5, E12.5, E13.5 to E16.5, and compared with the postnatal ovaries from one and six-week old mice by immunofluorescence microscopy. Fragilis was specifically expressed at the cell surface, and Vasa and Dazl were mainly localized to the cytoplasm of germ cells as reported (Figs. 1A and S1). During germ cell development in the fetal ovary, while Fragilis fluorescence signal intensity decreased from E12.5 to E16.5, the fluorescence intensity and number of Vasa and Dazl gradually increased (Figs. 1A and S1). Proportion of germ cells expressing Fragilis increased at E12.5 compared to E10.5 and remarkably decreased at E13.5 when germ cells enter meiosis, and very rare cells expressed Fragilis at E16.5 compared to other periods (Figs. 1A and S1B). Thus, PGCs at E10.5 and E12.5 are featured with co-expression of membrane Fragilis and cytoplasmic Vasa or Dazl. In addition, percentage of cells expressing only Fragilis but without Vasa also decreased during fetal development. Fragilis cells also can be found in the postnatal ovaries, and Fragilis/Vasa or Dazl cells mostly were visible in the cortex and ovarian surface epithelium (OSE) (white arrows) (Figs. 1A and S1). Careful counting of the fluorescent cells showed that PGCs simultaneously positive for both membrane Fragilis and cytoplasmic Vasa were mostly seen at E12.5 and could not be found in postnatal ovaries (Fig. 1B). In contrast, Fragilis but Vasa cells were few in fetal E12.5, and became significantly increased in postnatal ovaries. Vasa but Fragilis cells can be found in low frequency in both fetal E12.5 and postnatal ovaries (Fig. 1B). However, expression of Vasa and Fragilis can also be found in the cytoplasm of developing oocytes (Fig. S1A). Thus, the expression pattern of Fragilis, Vasa and Dazl in the postnatal ovary differs remarkably from that of the fetal ovary. This prompted us to further examine the properties of the Fragilis cells. Expression of SSEA1 in membrane and nuclear expression of Oct4 serve as important markers to identify PGCs and pluripotent stem cells (Hayashi et al., 2012; Guo et al., 2017). Expression and cellular localization of SSEA1 and Fragilis were similar in fetal ovaries. SSEA1 and Fragilis were co-expressed on the cell surface (Fig. 1C), consistent

For count of immunofluorescence positive or negative cells, three gonads were selected from each mouse group. Maximum sections and two views for each section were selected for each gonad under microscope by the DAPI filter to see just nuclei to avoid of possible bias. The number of cells counted in each group was combined, and at least 600 cells were counted for each group, then relative percentage of positive cells or negative cells also compared. P-value was calculated using X 2 test.

Cell size measurement
Dimensional measurements of different cells were performed using Image J software. Because Fragilis is expressed on the membrane of cell, it is counted by calculating the distance between the membranes.

Isolation and purification of Fragilis positive cells
Ovaries were freshly collected from C57BL/6-GFP mice at the age of E12.5 and 6weeks. E12.5 ovaries without attached mesonephros were mixed in a 1.5 ml centrifuge tube and dissociated in 0.25% trypsin-EDTA at 37°C for 10 min, neutralized by adding 10% fetal bovine serum (FBS) and dissociated into single cells.
Cells from ovaries of 6-week old mice were isolated using the enzymatic digestion method described previously (Zou et al., 2011;Zou et al., 2009), with minor modifications. Briefly, the ovaries were dissected, and cut into a slurry by a sterile ophthalmic scissor. The slurry was incubated at 37°C with intermittent shaking for 15 min in 0.25% trypsin-EDTA and neutralized by adding 10% fetal bovine serum (FBS).
Cell suspension from E12.5 ovaries or 6-week mouse ovaries was filtered through a cell strainer with 40-µm pore (Falcon, cat#: 352340), centrifuged, and the supernatant was carefully removed from the pellet. Cell pellet was dissociated into single cells by pipetting in MACS buffer (detailed below) and subject to MACS.
Fragilis positive cells were separated from negative cells by MACS (Woods and Tilly, 2013), according to the manufacturer's instruction (Miltenyi Biotec). Briefly, dissociated gonadal cells were incubated with Fragilis antibody at room temperature for 30 min. After being rinsed and resuspended in MACS buffer (PBS, pH 7.4, added with 0.1% BSA and 2 mM EDTA), the suspension was then centrifuged at 300 g min -1 for 5 min to collect the cell precipitate. Cell suspension was incubated with sheep anti-rabbit IgG magnetic beads (Miltenyi Biotec) at room temperature for 20 min, followed by wash in MACS buffer. The mixture of cells and magnetic beads was placed on the magnetic bead separator to obtain Fragilis positive cells. Cell suspensions dropping from the column were Fragilis negative cells. Fractions on the inner side of the MS column were Fragilis positive cells.

Isolation and purification of SSEA1 positive cells
Ovaries were freshly collected from E12.5 ovaries and dissociated into single cells as described above. MACS was performed according to the manufacturer's instructions (Miltenyi). Briefly, dissociated ovary cells were incubated with anti-SSEA1 antibody conjugated with magnetic beads at 4°C for 20 min. Cell suspensions were washed in PBS supplemented with 0.5% BSA and 2 mM EDTA and applied to an MS column (Miltenyi) to remove SSEA1 positive PGCs. SSEA1 negative cells, also namely gonadal somatic cells, were collected in the flow-through portions.
Kidney capsule transplantation was performed based on the methods described (Qing et al., 2008;Zeng et al., 2017). Briefly, one aggregate was picked up with a glass Pasteur pipette and implanted in the "pocket" which was made between the kidney capsule and kidney tissue of a bilaterally ovariectomized female recipient mouse, and 6-8 weeks old NOD-SCID females mice were used as recipients. One aggregate was picked up with a glass Pasteur pipette and implanted in the "pocket".
The transplantation procedure was completed within 5 min for each mouse.
Reconstituted ovaries (rOvaries) were obtained and dissected to examine folliculogenesis 28 days following transplantation of the aggregates.

Histological morphology of tissue sections
About 28 days after transplantation, the recipient mice were humanely sacrificed, and the aggregation-formed rOvaries were carefully retrieved from kidney capsule, fixed in 4% paraformaldehyde, dehydrated in gradient ethanol, cleared in xylene, embedded in paraffin, and sectioned for histological examination by hematoxylin and eosin (H&E) staining.

In vitro maturation (IVM) and in vitro fertilization (IVF)
The rOvaries were dissected from the recipient mouse kidney capsule, and fullygrown GV oocytes collected in IVM medium under a microscope by pricking follicles using insulin syringe. Oocytes were matured in vitro by culture in IVM medium for 17-18 h at 37 °C. IVM medium contains -MEM (Invitrogen) added with 5% FBS, 0.24 mM sodium pyruvate (Eppig et al., 2009), 1 IU/ml PMSG and 1.5 IU/ml human chorionic gonadotropin (hCG, Sigma). Oocytes at metaphase II (MII), determined by extrusion of the first polar body, were subjected to IVF. Spermatozoa were collected from the cauda epididymis of ICR males, capacitated by incubation for 2 h in HTF (Origio), and then incubated with the MII oocytes for 6 h. The zygotes were collected and transferred into human G-1 plus medium (Vitrolife). Embryos that reached the 2-cell stage after 24h culture were transferred into the oviducts of E0.5 pseudo-pregnant mice, and newborns were normally delivered on E19.5. Pups were identified initially by coat color. Contribution of Fragillis positive cells from donor mouse ovaries was confirmed by standard DNA microsatellite genotyping analysis using GFP, D8Mit94 and D12Mit136 primers (Table S2). Pups were mated with albino strain ICR mice to further examine their germline transmission competence.

Immunostaining and fluorescence microscopy of meiocyte spread
Surface spreading of meiocytes was prepared by a drying-down technique (Hodges and Hunt, 2002;Spyropoulos and Moens, 1994) and stained for synaptonemal complex proteins (Liu et al., 2004). Aggregates of Fragilis positive cells from E12.5 ovaries 5~6 days after transplantation and E17.5 ovaries or from adult ovaries were collected, minced with two forceps and dissociated by pipetting in 0.05% TE. After incubation for 7 min at 37 °C, cell suspensions were mixed with an equal volume of FBS, centrifuged for 5 min and resuspended in 100 mM sucrose. Cell suspension was spread onto glass slide by dipping onto a thin layer of fixative (1% paraformaldehyde, 0.15% Triton X-100 and 3 mM dithiothreitol, pH=9.2). The glass slides were maintained overnight in a humidified box at 4 °C. The slides were washed in water containing 0.4% Photo-flow (Kodak), and completely dried at room temperature. Dried slides were washed with 0.1% Triton X-100/PBS (PBST) for 10min, and incubated with blocking buffer (3% BSA, 2% goat serum/PBST) for 1 h at room temperature. Spreads were then incubated anti-SCP3, SCP1 or MLH1 antibody in blocking buffer at 4°C overnight, washed three times, and then incubated with appropriate secondary antibodies (goat anti-mouse IgG (H+L) FITC or goat anti-rabbit IgG (H+L) AlexaFluor® 594) added with DAPI. The slides were washed and mounted in Vectashield mounting medium (Vector Laboratories).

Collection of oocytes and immunofluorescence microscopy
GV oocytes were collected from female 6-week old mice 46 h after injection of pregnant mare's serum gonadotrophin (PMSG, 5 IU per mice) by pricking follicles using insulin syringe. Denuded oocytes were fixed and extracted for 30 min at 37 °C in stabilizing buffer. Oocytes were washed extensively and blocked overnight at 4 °C in wash medium (phosphate-buffered saline, supplemented with 0.02% NaN 3 , 0.01% Triton X-100, 0.2% non-fat dry milk, 2% goat serum, 2% bovine serum albumin and 0.1 M glycine). Afterwards, oocytes were incubated with Fragilis antibody at 4 °C for overnight or without antibody served as negative control. After washing, samples were stained for 1:100 actin filaments with Texas Red-conjugated Phalloidin and incubated with appropriate secondary antibodies for 2 h, washed again and counterstained with 0.5 μg/ml DAPI in Vectashield mounting medium (Vector Lab).
Fluorescence was detected and imaged using Zeiss Laser scanning confocal microscope LSM710 (Zeiss).

Western blot
Cells were washed twice in PBS, collected, and lysed in cell lysis buffer on ice for 30 min and then sonicated for 1 min at 60 of amplitude at 2 sec intervals. After centrifugation at 10,000 g at 4°C for 10 min, supernatant was transferred into new tubes. Concentration of the protein sample was measured by bicinchoninic acid, and protein samples were boiled in SDS Sample Buffer at 95°C for 10 min. 10 µg total protein of each cell extract was resolved by 10% Acr-Bis SDS-PAGE and transferred to polyvinylidene difluoride membranes (PVDF, Millipore). Nonspecific binding was blocked by incubation in 5% skim milk or 5% BSA in TBST at room temperature for 2 h. Blots were then probed with primary antibodies by incubation overnight at 4°C with Oct4 (sc5279, Santa Cruz), Vasa (ab13840, Abcam), Fragilis (ab15592, Abcam), Stella (ab19878, Abcam) or -actin (P30002, Abmart). Immunoreactivity bands were then probed for 2 h at room temperature with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies, goat anti-Rabbit IgG-HRP (GE Healthcare, NA934V), or goat anti-Mouse IgG (H+L)/HRP (ZSGB-BIO, ZB-2305). Protein bands were detected by Chemiluminescent HRP substrate (Millipore, WBKLS0500).

RNA-sequencing and bioinformatics
RNA-seq libraries were prepared using Smart-seq2 technology as previously described (Picelli et al., 2014). 1000 cells per sample were resuspended in PBS added with 0.1% BSA (A3311-10g, Sigma) and transferred to the bottom of a PCR tube consisting of 3 µl lysis buffer containing oligo (dT) primer and a locked nucleic acid (LNA)-containing template-switching oligonucleotide, and cDNA was synthesized in the tube containing mRNA. Full-length cDNAs were amplified by 18 cycles of PCR using KAPA HotStart ReadyMix (KK2602, KAPA Biosystems).
The libraries were prepared by using TruePrep DNA Library Prep Kit V2 for Illumina® (TD503-02, Vazyme Biotech) according to the manual instruction.
Samples were barcoded and multiplex sequenced with a 150 bp paired-end sequencing strategy on Illumina Hiseq X10.
Clean reads were mapped to the Mus musculus mm10 reference genome using Hisat2 (version 2.1.0). Reads were assigned and counted to genes using Featurecounts (version 1.6.3). The resulting matrix of read counts was then loaded into RStudio (R version 3.5.2), and DESeq2 used to identify differentially expressed genes. Functional enrichment (GO annotation or KEGG) of gene sets with differential expression patterns were performed using clusterProfiler.
Heatmaps were drawn by the function "pheatmap" of R packages and correlation coefficients calculation by the function "cor" in R. Scatter plots were generated using the "ggplot2" package to graphically reveal genes that differ significantly between the two samples. P values were adjusted using the Benjamin &