Several recent studies have suggested that mouse embryonic stem cells (ESCs) can differentiate into female and male germ cells in vitro. The meiotic process in germ cell-like cells derived from ESCs has not been studied in detail, but it has been reported that synaptonemal complex protein-3 (SYCP3) is expressed in these cells. Here, we have carefully evaluated the meiotic process in germ cell-like cells derived from ESCs, using a panel of meiosis-specific markers that identify distinct meiotic signatures unique to meiotic prophase I development in vivo. We find that whereas SYCP3 is expressed in germ cell-like cells, other meiotic proteins, such as SYCP1, SYCP2, STAG3 (stromal antigen 3), REC8 (meiotic protein similar to the rad21 cohesins), and SMC1 (structural maintenance of chromosomes-1)-β, are not expressed. The nuclear distribution of SYCP3 in the germ cell-like cells is highly abnormal and not associated with the chromosomes of these cells. Fluorescence in situ hybridization analysis shows that the SYCP3-positive germ cell-like cells do not contain synapsed homologous chromosomes but instead display a chromosomal organization normally found in somatic cells. The absence of expression of essential meiotic proteins and a normal meiotic chromosomal organization strongly suggests that the germ cell-like cells formed from ESCs fail to progress through meiosis.
The germ cell lineage in the mouse is first specified at approximately embryonic day 6–6.5 (E6–6.5). The primordial germ cells (PGCs) colonize the urogenital ridges at E10–11 and give rise to male and female gonads visible at E12–12.5 . The female germ cells initiate meiosis a short time afterward, whereas the male germ cells enter mitotic arrest as G0/G1 prospermatogonia. Initiation of meiosis is delayed in male germ cells and does not occur until after birth, during prepubertal development. The nature of the signal(s) that initiate meiosis in female germ cells is unknown, but once meiosis has been initiated, the process continues in a cell-autonomous manner. During meiosis, a single DNA replication step is followed by two cell divisions, generating haploid gametes [2, 3]. The first meiotic division is preceded by a lengthy prophase I stage that can be divided into several distinct substages. The pairing of the newly replicated sister chromatids is completed at leptotene, and subsequently the homologous chromosomes become closely positioned in a process called synapsis at the zygotene to pachytene stages. Synapsis is promoted by DNA crossovers between the homologous chromosomes resulting from the repair of DNA double-stranded breaks introduced by the SPO11 endonuclease . The meiotic chromosomes are associated with an evolutionarily conserved meiosis-specific protein structure called the synaptonemal complex (SC). The SC consists of a central element to which two colinear axial structures are anchored by a large number of transverse filaments [2, 3]. Several different meiosis-specific proteins, including SYCP1 (synaptonemal complex protein-1), SYCP2, SYCP3, STAG3 (stromal antigen 3), REC8 (meiotic protein similar to the rad21 cohesins), and SMC1-β (structural maintenance of chromosomes-1-β), have been shown to be associated with the SC in mammalian cells [5, , , , , –11]. SYCP1, SYCP2, and SYCP3 are structural proteins that take part in axial core compaction and synapsis, whereas the cohesin complex proteins STAG3, REC8, and SMC1-β promote sister chromatid pairing. Several of these proteins have been shown to be essential for meiotic progression and germ cell survival [12, , , , –17].
So far, it has been difficult to recapitulate the germ cell differentiation process in vitro using cell culture models. Recently, however, it was shown that both ovarian structures, containing oocyte-like cells, as well as sperm-like cells could be generated from different murine embryonic stem cell (ESC) lines in vitro [18, , , –22]. The derived PGC-like cells expressed several protein markers that normally appear during early germ cell formation in vivo, suggesting that the observed in vitro differentiation process correctly mimics the in vivo process.
To date, the meiotic process in germ cell-like cells derived from ESCs has not been studied in detail, but it has been reported that SYCP3 is expressed in these cells [18, 19]. We have used a set of meiosis-specific markers to validate the meiotic cell cycle in the germ cell-like cells. In agreement with previous studies, we found SYCP3 to be expressed in the germ cell-like cells. Surprisingly, other meiotic proteins, such as SYCP1, SYCP2, STAG3, REC8, and SMC1-β, were not detected in these cells. We found no evidence for synapsis in the germ cell-like cells despite SYCP3 expression. Furthermore, the organization of the chromosomes in these cells resembled what is seen in somatic cells.
Materials and Methods
ESCs and Mouse Strains
The ESC line R1 (passage 14) was provided by Dr. Andras Nagy (Samuel Lunenfeld Research Institute, Mount Sinai Hospital, University of Toronto, Toronto, http://www.mshri.on.ca). Additional primary ESC lines were derived from E3.5 blastocysts resulting from matings of C57B6/129N mice .
ESCs were maintained on mitomycin (Mutamycin; Bristol-Myers Squibb, Princeton, NJ, http://www.bms.com) C-treated mouse embryonic fibroblasts (MEFs) in 0.1% gelatin-coated tissue culture plates in Dulbecco's modified Eagle's medium (DMEM) containing 4.5 g/l glucose, 15% fetal bovine serum (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), 2 mM l-glutamine (Invitrogen), 100 mM nonessential amino acids (Invitrogen), 1 μM β-mercaptoethanol (Sigma, St. Louis, http://www.sigmaaldrich.com), and 50 μg/ml penicillin/streptomycin (Invitrogen) supplemented with 1,000 U/ml leukemia inhibitory factor (LIF) (ESGR; Chemicon, Temecula, CA, http://www.chemicon.com). To initiate differentiation, the ESCs were trypsinized and grown in medium without LIF and MEFs. The medium was replaced after 3 days in culture. By the 4th day, many different types of cells appeared. Six days after removal of LIF, embryoid bodies appeared, followed by the formation of small colonies. The colonies continued to grow in size, and by day 14, a subpopulation of cells loosened off from the colonies, forming aggregates in suspension. The aggregates were collected from days 16 to 20, dissociated by trypsin, fixed, and analyzed. Follicle-like structures derived from the aggregates were observed from day 18 and afterward. In parallel, cell culture plates with ESCs were cocultured with a Flp-In 3T3 cell line stably expressing bone morphogenic protein 4 (BMP4), as described .
The concentration of estradiol in the media collected from ESC cultures was assayed using the 17β-Estradiol Correlate-EIA kit (Assay Designs, Ann Arbor, MI, http://www.assaydesigns.com) as described by the manufacturer. The measurements were repeated on four different occasions with similar results.
Immunofluorescence Microscopy and Fluorescence In Situ Hybridization
Cells from germ cell-like aggregates were collected from day 12 to day 20, trypsinized, and fixed in 1% paraformaldehyde, 0.15% Triton X-100. The fixed cells were stained with a set of primary antibodies including mouse-α-SSEA1 (1:100; Chemicon), mouse-α-SSEA3 (1:100; Chemicon), rabbit-α-OCT4 (1:100; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, http://www.scbt.com), rabbit-α-SYCP3 (1:100), guinea pig-α-SYCP2 (1:500), guinea pig-α-SYCP1 (1:100), guinea pig-α-STAG3 (1:100), guinea pig-α-REC8 (1:100), guinea pig-α-SMC1-β (1:100) [24, 25], or human CREST antiserum (1:3,000), as well as secondary antibodies: swine-α-rabbit-fluorescein isothiocyanate (FITC) (1:100; Dakocytomation, Glostrup, Denmark, http://www.dako.com), donkey-α-guinea pig-CY3 (1:700; Jackson Laboratory, Bar Harbor, ME, http://www.jax.org), and goat-α-human-CY5 (1:1,000; Jackson Laboratory). Fluorescence in situ hybridization (FISH) was performed on ESCs or pachytene oocytes (fixed in 1% paraformaldehyde and 0.15% Triton X-100), using a Cy3-labeled whole mouse chromosome 1 probe according to the manufacturer's protocol (Chrombios GmbH, Raubling, Germany, http://www.chrombios.com). The FISH procedure was followed by immunofluorescent staining using a rabbit-α-SYCP3 antibody and visualized with swine-α-rabbit FITC (1:100; Dakocytomation). All slides were stained with DAPI (4,6-diamidino-2-phenylindole) (0.5 μg/ml) to control the quality of the fixation procedures, and cells with undisrupted morphology were taken for analysis. After staining, the cells were mounted in Prolong Mounting medium (Molecular Probes, Eugene, OR, http://probes.invitrogen.com). Image acquisition was done on a Leica DMRXA microscope (Heerbrugg, Switzerland, http://www.leica.com) at ×100 magnification using a Hamamatsu C4742-95 CCD (charge-coupled device) camera (Shizuoka, Japan, http://www.hamamatsu.com). Image processing was performed with Openlab 3.1.4 software (Improvision, Coventry, U.K., http://www.improvision.com) or Adobe Photoshop 7.0 (Adobe Systems, Inc., San Jose, CA, http://www.adobe.com).
Plasmid Constructs and Transfection
The Flp-In system (Invitrogen) was used to generate a stable mouse BMP4 cell line. The BMP4 coding sequence was amplified from E7.0 mouse cDNA (Clontech, Mountain View, CA, http://www.clontech.com) and cloned into the pEF5/FRT/V5-D-TOPO vector. Plasmid DNA with the correct orientation was identified, sequenced, and cotransfected into Flp-In 3T3 cells with pOG44 (a plasmid containing a recombinase gene). The transfection was performed using FuGENE (Roche, Basel, Switzerland, http://www.roche.com) following the manufacturer's instructions. Single transfected clones were obtained after selection with 200 μg/ml hygromycin B. The cells were cultivated in DMEM supplemented with 10% fetal calf serum and penicillin/streptomycin (Invitrogen). When cells reached 80%–90% confluence, protein extracts were prepared and then separated on a 13% SDS-polyacrylamide gel electrophoresis gel, followed by Western blotting. The BMP4 protein was visualized using a mouse BMP4 monoclonal antibody (1:100; Santa Cruz Biotechnology, Inc.) and a goat-α-mouse horseradish peroxide (1:3,000; Dakocytomation) as a secondary antibody, following a chemiluminescence-based detection procedure (Pierce, Rockford, IL, http://www.piercenet.com).
Differentiation of ESCs into Ovarian Follicle-Like Structures
Mouse ESCs were cultured in vitro and found to express several markers unique to undifferentiated stem cells, including stage-specific embryonic antigen-1 (SSEA1) and octamer transcription factor-4 (OCT4) (Fig. 1A–1C). The ESCs were allowed to differentiate in vitro in tissue culture plates containing ESC medium without feeder cells or LIF, as described . Large colonies that resembled early germ cell clusters formed on the cell culture plates 12 days after initiation of the differentiation experiment (Fig. 1D). Cells derived from the isolated colonies formed aggregates in the cell suspension and were found several days later to differentiate further into follicle-like structures (Fig. 1E) with a strong resemblance to primordial and primary follicles normally observed only at the dictyate stage of meiosis in ovaries . The appearance of the follicle-like structures strongly suggests that ovarian germ cell-supporting cells, such as granulosa cells, are formed during in vitro differentiation of ESCs. To investigate whether these gonadal supporting cells were functionally active, we monitored the concentration of the sex hormone estradiol in the cell culture medium. Estradiol is produced in a process that requires both granulosa cells and Theca cells within the ovarian compartment in vivo. Estradiol was first detected at day 11 to day 12. The concentration of estradiol then increased until day 20 (Fig. 2), supporting an ongoing activity of ovarian supporting cells in vitro.
SYCP3 Is Preferentially Expressed in Germ Cell-Like Cells Derived from ESCs but the Observed Nuclear Distribution Pattern Is Abnormal
SYCP3 is frequently used as a meiotic marker and has been shown to be expressed in the germ cell-like cells generated from ESCs in vitro [18, 19]. We analyzed the expression of this protein in cells derived from ESCs, using immunofluorescence microscopy methods, at different time points after the initiation of the differentiation experiments (day 7–day 20). Few SYCP3-positive cells were identified at day 12 to day 15. Next, we analyzed colonies and cellular aggregates derived from isolated colonies and found that although initially relatively few cells were SYCP3-positive, approximately 40% of the cells within the aggregates became SYCP3-positive between day 14 and day 16. The nuclear SYCP3-staining patterns in the cell aggregates were quite variable, exhibiting nuclear foci as well as short filamentous structures (Figs. 3 and 4). The nuclear distribution pattern of SYCP3 within colonies and cellular aggregates derived from isolated colonies was followed between day 14 and day 20; however, structures similar to the long axial cores defined by SYCP3 in pachytene oocytes were not observed. We conclude that although SYCP3 is strongly expressed in early germ cell-like aggregates derived from ESCs, the nuclear distribution pattern within the cells is abnormal.
Meiosis-Specific SC Proteins and Cohesin Complex Proteins Are Not Coexpressed with SYCP3 in Differentiating ESCs
Subsequently, we monitored the expression of a set of meiosis-specific markers in cells derived from the cellular aggregates at different time points (day 12–day 20), using immunofluorescence microscopy. These markers, including SYCP1, SYCP2, REC8, STAG3, and SMC1-β, are coexpressed in vivo together with SYCP3. We were not able to find cells derived from the aggregates that expressed the analyzed meiosis-specific proteins, except for SYCP3 (Figs. 3 and 4). We also analyzed cell samples taken from the entire cell culture plate in an unbiased manner but were not able to identify ESC-derived cells that expressed SYCP1, SYCP2, REC8, STAG3, or SMC1-β. We conclude that although a large fraction of cells derived from germ cell-like colonies in vitro express SYCP3, other essential meiosis-specific proteins are not coexpressed in these cells.
ESCs grown in the absence of feeder cells and LIF were also cocultured with a stable mouse 3T3 fibroblast cell line overexpressing BMP4. This approach has been used to improve the formation of germ-cell like cells in vitro . We found that the time required for the first appearance of aggregates that expressed SYCP3 was shortened, but BMP4 overexpression neither affected the nuclear pattern of SYCP3 nor induced the expression of other meiotic markers in the ESC-derived cells (data not shown).
Chromosomes in the SYCP3-Positive Germ Cell-Like Cells Are Not Synapsed
Meiotic prophase I chromosomes become synapsed at the beginning of the zygotene stage of prophase I and retain this chromosomal organization until the diplotene stage, a period that lasts for approximately 6 days in the embryonic ovary. The synapsed homologous chromosomes (20 bivalents at pachytene) are visualized as extended axial structures by SYCP3 during this period of meiosis. We applied FISH in order to determine the chromosomal organization within the SYCP3-positive cells. A probe against mouse chromosome 1 was used to label both cellular aggregates derived from isolated colonies and wild-type mouse embryonic ovarian meiotic cells while staining with anti-SYCP3 antibodies in parallel. We found that whereas oocytes within the ovarian wild-type structures displayed the expected single bivalent chromosome 1 signal, two separate chromosome 1 signals were observed in the SYCP3-positive ESC-derived cells (Fig. 5). Furthermore, even though the chromosome 1 signal in the oocytes clearly overlapped with the SYCP3-stained synapsed chromosomal cores, the SYCP3-positive filaments in ESC-derived cells were located separately from the two chromosomes visualized with the FISH probe. The absence of a synaptic interaction between the chromosomes in the SYCP3-positive ESC-derived cells was further exemplified by the identification of 40 individual centromeres (labeled by the CREST probe), verifying the existence of 40 individual chromosomes (Figs. 3 and 4). In contrast, 20 centromere foci revealed a synaptic organization in the ovarian meiotic cells. We conclude that the chromosomal organization in the SYCP3-positive ESC-derived cells is seemingly identical with the nuclear organization associated with somatic cell types. Furthermore, the chromosomes within these cells display no signs of synapsis or an overlap with SYCP3.
Discussion and Conclusion
We have monitored the differentiation process that generates germ cell-like cells from ESCs and have studied the meiotic process in these cells. We found, as shown previously [18, , –21], that differentiating ESCs can give rise to both colonies and aggregates, as well as to follicle-like ovarian structures. We also found that a large fraction of the in vitro-derived germ cell-like cells express the meiotic protein SYCP3. Surprisingly, however, other meiotic markers normally coexpressed with SYCP3 in meiotic cells in vivo were not expressed in cells derived from ESCs. Several of these meiotic proteins have been shown to be essential for meiotic progression in vivo, and their absence results in male and female germ cell death [12, –14, 16, 17]. This result shows that the protein expression pattern in ESC-derived germ cell-like cells is different from the one observed within embryonic cells in vivo. In a separate set of experiments, the chromosomal organization of the SYCP3-positive germ cell-like cells was monitored. We could not identify cells that contained synapsed homologous chromosomes or chromosomes that overlapped with SYCP3 staining. Furthermore, estimation of the chromosome number in the SYCP3-positive germ cell-like cells did not reveal the expected meiotic chromosome set, as shown by FISH or by centromere staining. The absence of expression of essential meiotic proteins and of a normal meiotic chromosomal organization strongly suggests that the germ cell-like cells formed from ESCs fail to progress through meiosis.
Multiple genes associated with PGC formation have been shown to be activated in germ cell-like cells generated from ESCs in vitro, following a temporal program mimicking the expression pattern seen in the embryonic gonads [18, , –21]. The transcription factor FIGa is activated at E13 in vivo in premeiotic oocytes and regulates transcription of many genes that promote follicle formation, including ZP1 and ZP3 [27, 28]. FIGa is also activated in the germ cell-like cells, together with at least one of its downstream targets, ZP3 [18, 21]. Activation of FIGa in the germ cell-like cells likely recruits gonadal supporting cells, resulting in the formation of the follicle-like structures observed in vitro. It should be noted, however, that the physical organization of the in vitro-derived follicle-like structures is impaired as other essential targets of FIGa in vivo (e.g., ZP1) are not expressed [18, 21].
How can our results be reconciled with the observation that mature germ cell-like cells, such as oocytes and spermatids, can be generated in vitro from ESCs [18, , –21]? Both female and male germ cells enter a premeiotic stage at E12.5 in vivo, and the Sycp3 gene has been shown to be expressed in these cells [29, 30]. However, SYCP3 expression at this developmental stage in the embryonic gonad does not generate the characteristic axial cores associated with chromosomes. Furthermore, the male embryonic germ cells only transiently express SYCP3, as these cells quickly become mitotically arrested. Our results suggest that the germ cell-like cells derived from ESCs either fail to correctly initiate the specialized meiotic cell division program unique to germ cells in vivo or fail to progress correctly in the absence of complementary meiotic proteins. An important objective now is to better understand the signaling pathway that triggers meiosis in vivo but that fails to become activated in vitro. Female germ cells have been shown to contain a default program that will carry out meiosis in a cell-autonomous way once this process has been initiated . This finding is very interesting because it suggests that in vitro replication of the meiotic process should be possible if the correct cell culturing conditions are defined.
The authors indicate no potential conflicts of interest.
This work was supported by grants from the Swedish Cancer Society, the Swedish Research Council, Petrus and Augusta Hedlunds Stiftelse, and the Karolinska Institutet.