Branched germline cysts and female-specific cyst fragmentation facilitate oocyte determination in mice

Significance

In mammalian females, a large-scale germ cell death in fetal ovaries results in only a small proportion of germ cells becoming oocytes. In this study, we identified that mouse fetal germ cells form branched cyst structures where ~17% of the cells are connected by three or four intercellular bridges. In female cysts, these germ cells are preferentially protected from cell death to become oocytes through collecting cellular content from sister germ cells. These results suggest a mechanism on how the oocytes available for sustaining adult ovarian function are determined during fetal ovary development. Unlike female germ cells, male germ cells share the same cell fate within each branched cyst.

Abstract

During mouse gametogenesis, germ cells derived from the same progenitor are connected via intercellular bridges forming germline cysts, within which asymmetrical or symmetrical cell fate occurs in female and male germ cells, respectively. Here, we have identified branched cyst structures in mice, and investigated their formation and function in oocyte determination. In fetal female cysts, 16.8% of the germ cells are connected by three or four bridges, namely branching germ cells. These germ cells are preferentially protected from cell death and cyst fragmentation and accumulate cytoplasm and organelles from sister germ cells to become primary oocytes. Changes in cyst structure and differential cell volumes among cyst germ cells suggest that cytoplasmic transport in germline cysts is conducted in a directional manner, in which cellular content is first transported locally between peripheral germ cells and further enriched in branching germ cells, a process causing selective germ cell loss in cysts. Cyst fragmentation occurs extensively in female cysts, but not in male cysts. Male cysts in fetal and adult testes have branched cyst structures, without differential cell fates between germ cells. During fetal cyst formation, E-cadherin (E-cad) junctions between germ cells position intercellular bridges to form branched cysts. Disrupted junction formation in E-cad-depleted cysts led to an altered ratio in branched cysts. Germ cell-specific E-cad knockout resulted in reductions in primary oocyte number and oocyte size. These findings shed light on how oocyte fate is determined within mouse germline cysts.

In mammals, germline differentiation during gametogenesis produces eggs and spermatozoa that are distinct in morphology and function, yet are derived from the same type of progenitors, primordial germ cells (PGCs) in fetal gonads (1). In the fetal testis, PGCs differentiate into gonocytes, the majority of which differentiate into spermatogonia stem cells (SSCs) to sustain spermatogenesis in adult testes (2–4). In fetal ovaries, PGCs differentiate into primary oocytes, which become quiescent along with the surrounding pregranulosa cells, forming primordial follicles. The pool of dormant primordial follicles serves as the ovarian reserve that sustains egg production and ovarian function in adulthood (5, 6). An unresolved question in mammalian oocyte differentiation is the differential cell fates among germ cells during this process. In both humans and mice, only a small fraction (~15 to 20%) of fetal germ cells survive to become primary oocytes, with the majority of the germ cells undergoing cell death (7, 8). Mechanisms underlying the distinct cell fates during oocyte differentiation are key to understanding ovarian reserve formation and associated ovarian health issues.

From invertebrates to humans, germ cells form a highly conserved cellular structure, the germline cyst, during both female and male gametogenesis. Within the cyst, sister germ cells derived from a single progenitor are connected via intercellular bridges formed through incomplete cytokinesis during mitotic divisions (9–14). During mouse gametogenesis, PGCs form germline cysts in the fetal ovary and testis from embryonic day 10.5 (E10.5) to E14.5. Female germ cells enter meiosis immediately following mitotic cyst formation, while male germ cells undergo G0/G1 arrest at E14.5 (1, 11). Intercellular bridges have been observed by electron microscopy (EM) and can be detected by antibody staining of specific markers, such as TEX14 (testis-expressed 14) and RacGAP (Rac GTPase activating protein) in fetal ovaries, and fetal and adult testes (8, 14–16). In adult testes, as SSCs initiate spermatogenesis, male germ cells undergo mitosis to form germline cysts. Interconnected spermatocytes enter meiosis synchronously as they differentiate into spermatozoa (15, 17).

A conserved phenomenon of gametogenesis is asymmetric cell fate in female germline cysts. During oogenesis in Drosophila and mice, organelle and cytoplasmic transport between sister germ cells within a cyst facilitates two distinct cell fates: becoming oocytes vs. undergoing cell death (18, 19). Within each Drosophila 16-cell cyst, one of the two germ cells that are connected by four intercellular bridges progresses through meiosis and differentiates into an oocyte by collecting cellular content from the remaining 15 sister germ cells, i.e., nurse cells. The differential cell fates in the Drosophila cyst are facilitated by the fusome, a cytoplasmic organelle that asymmetrically spans through the intercellular bridges during cyst formation, with the future oocyte retaining a greater proportion compared to the nurse cells (19–21). The fusome anchors germ cell mitotic spindles causing branched cyst formation and organizes microtubule minus ends into the future oocyte, facilitating directional organelle and cytoplasmic transport from the nurse cells to the future oocyte (19–22).

During mouse oocyte differentiation that takes place from E14.5 to postnatal day 4 (P4), germline cysts fragment gradually and become individual cells. Our previous studies find that on average, 20% of the E14.5 cyst germ cells differentiate into primary oocytes through organelle and cytoplasmic collection within germline cysts. The remaining 80% of the E14.5 cyst germ cells undergo cell death. During this process, an average five-fold increase in organelle content occurs in primary oocytes (11, 18). In the primary oocytes, centrosomes, Golgi complexes, endoplasmic reticulum, and mitochondria organize into a Balbiani body (B-body), in which the Golgi complexes have a characteristic spherical arrangement (18, 23). The fusome structure is not observed in mouse cysts by EM (8). How germ cells are connected structurally within the mouse cyst and how specific germ cells are selected to receive organelles and cytoplasm to become primary oocytes remain open questions.

During spermatogenesis in Drosophila and adult mice, intercellular bridges facilitate the transfer of signaling molecules between cyst germ cells, enabling synchronized meiosis and symmetrical cell fate in the cyst (15, 24). All male germ cells in the germline cyst progress through spermatogenesis to differentiate into sperm or otherwise undergo cell death when triggered in some germ cells (25–28). The Drosophila male cyst is in the same branched structure as the female cyst, with the fusome evenly distributed in each germ cell (29, 30). The structures of male germline cysts in fetal and adult mouse testes have yet to be characterized.

In the present study, we find that both female and male mouse germline cysts are in branched structures and demonstrate that branched cyst structure is involved in differential cell fates during oocyte differentiation in fetal ovaries, but not in gonocyte differentiation in fetal testes. The formation of branched cysts may be regulated by the position of intercellular bridges influenced by the E-cadherin (E-cad) junctions between sister cyst germ cells. These findings shed light on the mechanism of how the ovarian reserve, the oocytes available to sustain adult ovarian function, is established during ovary formation at the fetal stage.

Results

Terms and Definitions.

• Intercellular/connected bridge: the TEX14 (or RacGAP)-positive ring or puncta at the size of approximately 1 µm connecting two germ cells.

• Unconnected bridge: the TEX14 (or RacGAP)-positive ring or puncta at the size of approximately 1 µm connected to one germ cell.

• Germ cell clone: germ cells derived from a single lineage-labeled progenitor germ cell, regardless of whether they remain connected by intercellular bridges.

• Germline cyst: a cluster of interconnected germ cells derived from the same progenitor germ cell.

• Branching germ cell: the germ cell that is connected by three or four intercellular (or unconnected) bridges in the germline cyst.

• Branched germline cyst: the germline cyst that contains one or more branching germ cells.

Female Germline Cysts, but Not Male Cysts, Fragment Extensively.

To investigate the formation and development of mouse germline cysts, we conducted single-cell lineage tracing to label individual PGCs at E10.5 by injecting a single low dose of tamoxifen into pregnant female R26-YFP mice mated with male Cag-creER; R26-YFP mice (11, 31) (Fig. 1A). We followed the development of individual germ cell clones derived from single lineage-labeled PGCs based on the expression of lineage marker yellow fluorescent protein (YFP) and germ cell marker DEAD-box helicase 4 (DDX4) (SI Appendix, Fig. S1 and Movies S1–S4). Because on average only a single PGC is labeled in each gonad, this approach allows us to study individual germ cell clones in the gonad, which are comprised of various numbers of germline cysts, with proper confidence. A consistent rate of observing lineage-labeled clones in fetal ovaries (from E11.5 to P4) in our previous study suggests that the majority of the female clones remain in the ovary without being eliminated entirely due to cell death (11). However, because an entire defective male germ cell clone could be eliminated by apoptosis in fetal testes, we should not rule out the possibility that our single-cell lineage tracing approach may bias toward healthy germ cell clone/cysts in fetal testes (27).

Fig. 1.

Both female and male mouse germline cysts are in branched structures, but female cysts, unlike male cysts, undergo extensive fragmentation. (A) Time course of oocyte and gonocyte differentiation in mouse fetal gonads. A single low dose of Tamoxifen (Tmx) was injected to Cag-creER;R26-YFP/YFP mice at E10.5 to label one primordial germ cell (PGC) per gonad on average. Fetal gonads were collected at E12.5, E14.5, E17.5, and P0 for cyst structure analysis. (B–E” ) Three-dimensional (3D) confocal images (B–E) and 3D models (B’–E’ ) of single lineage-labeled germ cell clones generated by Imaris software (see original confocal images of each clone in Movies S1–S4). (B”–E” ) Cyst structure of the clones showing the geometry of YFP-positive germ cells (green circle) and TEX14(or RacGAP)-positive bridges (red line). The cyst outlined by the blue dotted line was defined as sister germ cells connected by intercellular bridges. Each single germ cell was counted as a cyst. The clone outlined by the magenta dotted line was defined as sister germ cells derived from a single PGC labeled by lineage marker YFP. (F and F’ ) Size of germ cell clones in fetal ovaries (F ) and testes (F’ ). (G–H’ ) Z-stack confocal images (G: E14.5 ovary and H: P0 ovary) and structure model (G’ and H’ ) of lineage-labeled cysts showing examples of branching cells (blue wide-head arrows), connected bridges (arrows) and unconnected bridges (arrowheads). (I and I’ ) Number of germline cysts in a clone in fetal ovaries (I) and testes (I’ ). (J and J’ ) Size of germline cysts in fetal ovaries (J) and testes (J’ ). (K and K’ ) Number of total bridges and connected bridges in fetal ovaries (K) and testes (K’ ). (Scale bar, 10 µm). The bar in the graph represents the average value. Data are presented as mean ± SD.

In the present study, fetal ovaries were collected at E12.5, representing the cysts in the middle of cyst formation; and E14.5, E17.5 and P0, representing the cysts prior to, during, and at the end of organelle and cytoplasm transport, respectively (Fig. 1 B–E” ) (11, 18). Fetal testes at these four time points were also collected for a comparative analysis on cyst structure during gonocyte differentiation (SI Appendix, Table S1–S8). On average, each female PGC gave rise to a 27.0-cell clone at E14.5 that decreased to a 14.5-cell clone by P0 (Fig. 1F). By contrast, the clone size in fetal testes remained at ~45 cells per clone from E14.5 to P0 (Fig. 1F’   ). This observation is consistent with previous reports of large-scale germ cell loss taking place in fetal ovaries during oocyte differentiation (11, 13, 32–34).

To elucidate the structure of mouse cysts, the geometry of the cyst germ cells (identified by YFP) and bridges (identified by RacGAP in E12.5 gonads or TEX14) in each cyst was laid out based on confocal images of individual germ cell clones (Fig. 1 B”–E” ). In addition to the intercellular bridge that connects two cyst germ cells, we observed unconnected bridges, which may represent the remains of the intercellular bridges following cytokinesis (Fig. 1 G and G’ )(35); some germ cells are connected by three or four bridges (intercellular/connected and unconnected bridges), herein referred to as “branching cells” (Fig. 1 H, G’, and H’).

We found that mouse cysts undergo fragmentation. From E14.5 to P0, the number of cysts per clone increased from 4.4 to 8.4 in fetal ovaries, with cyst size decreasing from 6.1 to 1.7 cells/cyst, suggesting that cyst fragmentation and cell loss take place in parallel in female cysts (Fig. 1 I and J). In fetal testes, an E14.5 clone was comprised of 4.0 cysts, each containing 11.3 cells on average. By P0, cysts fragmented into 7.7 cysts, each containing 6.1 cells. This indicates that the male fetal cysts undergo limited fragmentation without considerable cell death (Fig. 1 I’ and J’ ). Consequently, in P0 testes, a large number of germ cells remained connected via elongated cell protrusions with TEX14-positive bridges between the germ cells (SI Appendix, Fig. S2). These data demonstrate that female cysts undergo greater fragmentation than male cysts, with germ cell loss taking place in fragmented cysts.

To elucidate the pattern of cyst fragmentation, we profiled the total number of bridges in each germ cell clone (Fig. 1 K and K’). From E14.5 to E17.5, a female germ cell clone lost, on average, 9.9 connected bridges, it only caused a slight increase in cyst number. From E17.5 to P0, each female germ cell clone lost, on average, 7.0 connected bridges, which led to cysts fragmenting into 1.7 cells/cyst by P0 (Fig. 1 I, J, and K). On average, each male clone lost 2.2 bridges, causing limited cyst fragmentation, consistent with the presence of twice as many cysts (Fig. 1 I’, J’, and K’). From E14.5 to P0, the percentage of unconnected bridges increased from 8.7% to 61.5% of the total bridges in fetal ovaries, indicating considerable levels of cyst fragmentation occurring in female cysts (SI Appendix, Fig. S3A). By contrast, in male, the frequency of unconnected bridges remained at ~10%, consistent with lower levels of cyst fragmentation (SI Appendix, Fig. S3B). Taken together, these results demonstrate that female cysts undergo extensive fragmentation caused by the loss of connected bridges and germ cells.

Female and Male Germline Cysts Are in Branched Structures.

We then asked whether all germline cysts are in branched structures by profiling the size of cysts and the percentage of them containing branching cells. Since germline cysts form through incomplete cytokinesis during germ cell mitosis, the number of bridges reflects the rounds of division the cells have undergone. By nature, an 8-cell cyst is the stage at which branching cells first appear as a 4-cell cyst undergoes one round of synchronized mitotic division. However, branching cells were observed in cysts as small as 3 cells in E12.5 ovaries (Fig. 2A); and in cysts as small as 5 cells in E12.5 testes (Fig. 2A’), reflecting cyst fragmentation during cyst formation. In E14.5 ovaries, branching germ cells were observed in a few 2-cell cysts (2.94%), many 6-cell cysts (54.5%), and all cysts larger than 8 cells (Fig. 2B). In E14.5 testes, all cysts larger than 6 cells were branched (Fig. 2B’). Both female and male cysts retained their branched structure up until P0, with female cysts having smaller sizes due to extensive germ cell loss and cyst fragmentation (Fig. 2 D and D’ ). We further found a linear association between the number of branching germ cells and cyst size (Fig. 2 E–H’ and SI Appendix, Table S9). On average, there was one branching germ cell per 4.3 germ cells in female cysts and per 4.6 germ cells in male cysts at E14.5, indicating that branched cyst formation may follow a similar pattern in fetal ovaries and testes (Fig. 2 F and F’). These results demonstrated that both female and male PGCs form branched germline cysts from E10.5 to E14.5.

Fig. 2.

Branching cells are preferentially protected during female cyst fragmentation. (A–D’ ) Percentage of the cysts profiled by the size of cysts (left Y axis) and percentage of the cysts that are in branched structures (right Y axis) in fetal ovaries and fetal testes. (E–H’ ) Germline cysts profiled by cyst size and the number of branching germ cells contained in fetal ovaries and fetal testes. (I and I’ ) Comparative profiling on the number of branching cells from E12.5 to P0 in ovaries (I) and testes (I’ ). (  J) Diagram of a branched cyst. The number represents the number of the bridges connected with the cell. (K and K’ ) Changes in the number of cells connected with 0 bridge, 1 bridge, 2 bridges, and 3 or 4 bridges in a germ cell clone in fetal ovaries (K ) and testes (K’ ) from E14.5 to P0. (L and M) Mathematical modeling of random loss of intercellular bridges in each germ cell clone in ovaries from E14.5 to E17.5 (L) and E17.5 to P0 (M). (N and O) Mathematical modeling of random loss of germ cells in each germ cell clone in ovaries from E14.5 to E17.5 (N) and E17.5 to P0 (O). (P) Proposed model of the change in cyst structure in a germ cell clone during oocyte differentiation from E14.5 to P0. The average numbers of germ cells per clone, cysts per clone, cells per cyst, bridges per clone, and bridges per cyst from E14.5, E17.5, and P0 clones were used for making the model. Data in the graph are presented as mean ± SD. One-way ANOVA was used for the statistical analysis between E14.5, E17.5, and P0. For modeling analyses, a two-sample t test with 0.05-significance level was used to test the null hypothesis that the data came from independent random samples from normal distributions with equal means (In silico) and equal but unknown variances (In vivo).

Branching Germ Cells Are Protected in Female Cysts.

We found that as the female cysts decreased in size from E14.5 to P0, the number of branching germ cells per cyst and the percentage of branched cysts increased (Fig. 2I and SI Appendix, Fig. S4A). By contrast, in fetal testes, the percentage of branched cysts and number of branching germ cells per cyst remained highly similar during gonocyte differentiation from E14.5 to P0 (Fig. 2I’ and SI Appendix, Fig. S4B). These results indicate that branching germ cells are preferentially protected from cyst fragmentation and germ cell death that take place in female cysts.

To determine the location of the germ cells and bridges that are lost in cysts during oocyte differentiation, we analyzed the change in cyst structure during this process by quantifying the number of germ cells connected by 0 bridge (single cell), 1 bridge, 2 bridges, and 3 or 4 bridges (branching cell) in each germ cell clone (Fig. 2 J, K, and K’). From E14.5 to E17.5, a female clone lost 37% of the1-bridge cells, 42% of the 2-bridge cells, and 25% of the 3- or 4-bridge cells per clone. From E17.5 to P0, a female clone lost 3% of the 1-bridge cells, 56% of the 2-bridge cells, and 70% of the 3- or 4-bridge cells per clone (Fig. 2K and SI Appendix, Table S10). Our analysis shows the majority of cell loss involved 1- and 2-bridge cells from E14.5 to E17.5. These results suggest that from E14.5 to E17.5, germ cell loss in cysts takes place in a selective manner with 1-and 2-bridge cells being lost and branching germ cells being protected preferentially.

In male germ cell clones, the germ cells connected by 1 or 2 bridges remained approximately the same; however, on average, a clone lost 3.6 branching germ cells from E14.5 to P0, indicating that the limited cyst fragmentation in male cysts may preferentially take place at branching germ cells (Fig. 2K’ and SI Appendix, Table S10).

To test our hypothesis that cyst fragmentation and germ cell loss in female cysts occur in a nonrandom manner with branching cells being protected, we mathematically modeled random bridge loss and germ cell loss in each female germ cell clone and compared the resultant cyst structure from in silico modeling with the observed in vivo cyst structure (Fig. 2 L–O, See SI Appendix, Supplementary File 1 and 2 for detailed mathematical methods). With mathematical modeling considering connected intercellular bridges only, unconnected bridges were excluded when counting the number of bridges connected to each in vivo germ cell in order to make a comparison between in silico and in vivo cyst structures. Based on the change in average bridge number per clone from E14.5 to E17.5 (Fig. 1K), 43.6% of the bridges were randomly removed from each E14.5 female clone (in vivo E14.5, n = 41) and the cyst structures generated by modeling (in silico E17.5) were compared with the actual cyst structure found in vivo in E17.5 clones. Similarly, 55.0% of the bridges were randomly removed from each E17.5 female clone (n = 29), and the cyst structures generated by modeling (in silico P0) were compared with actual cysts found in the in vivo P0 clones. This modeling, in which the total cell numbers do not change, demonstrated that random removal of the bridges would generate significantly more 0-bridge single cells than observed in vivo. Moreover, fewer 2-bridge and 3- or 4-bridge cells appeared in modeled clones compared with those in vivo, revealing that there were significant differences between our random bridge loss model and the experimental findings. This suggests that cyst fragmentation in vivo does not take place in a random manner, instead it occurs with bridges connected to branching germ cells being protected (Fig. 2 L and M).

We further modeled random germ cell loss (Fig. 2 N and O). On average, 33.0% of the germ cells were lost in a clone from E14.5 to E17.5 and 20.2% of the germ cells were lost in a clone from E17.5 to P0 in vivo (Fig. 1H). These experimental fractions were used for the random mathematical model as described in SI Appendix, Supplementary File 1. When germ cells (and the associated bridges) were removed randomly at such rates in each E14.5 germ cell clone (n = 41), there were significantly fewer 2-bridge and 3- or 4-bridge germ cells in in silico E17.5 clones than in vivo, suggesting that branching germ cells are preferentially protected in vivo during E14.5 to E17.5. When 20.2% of the germ cells were randomly removed from E17.5 clones (n = 29), significantly more germ cells remained connected in in silico P0 clone, suggesting extensive cyst fragmentation caused by bridge loss occurs from E17.5 to P0 in vivo.

Based on the average cell number per clone and per cyst, as well as germ cells connected with 1, 2, and 3 or 4 bridges per clone and per cyst, the average change in cyst structure was modeled in Fig. 2P. Together with mathematical modeling analyses, our results suggest that from E14.5 to E17.5, germ cells are lost in a nonrandom, selective manner, in which branching germ cells and the associated bridges are preferentially protected and periphery germ cells are lost; and from E17.5 to P0, cysts undergo extensive fragmentation, primarily due to bridge loss, breaking down into small cysts and individual germ cells (Fig. 2P).

Female Branching Germ Cells Contain Enriched Cytoplasm and Organelles.

Our previous and present studies demonstrated that organelle and cytoplasm transport between cyst germ cells results in two germ cell populations by P0: 1) Germ cells with enriched cytoplasm and organelles, thus each containing a B-body, are preferentially protected from cell death and become primary oocytes; and 2) germ cells that are smaller in volume, without the B-body, undergo apoptosis preferentially (18).

To determine whether the branching germ cells are enriched with cytoplasm in the cyst, we measured the volume of each germ cell in the clone and plotted this against the number of bridges connected to them. We separated the germ cells into two categories: 1) single cells that have separated with the cyst and thus presumably have completed organelle and cytoplasm transport (among these cells, some still have unconnected TEX14-positive bridges, which partially reflects the history of germ cell connection); and 2) germ cells in cysts that remain connected and are in the process of organelle and cytoplasm transport. We found that the volume of germ cells increased as oocyte differentiation progressed from E12.5 to P0 (SI Appendix, Fig. S5A) (18). In E14.5 germ cell clones, single germ cells without bridges had the smallest volume (514 µm³) and branching germ cells had the largest volume (713 µm³) (Fig. 3 A and A’). By E17.5, the single germ cells with 2 or more unconnected bridges had larger volumes compared to single germ cells without bridges (Fig. 3B). Among germ cells in E17.5 cysts, bridge number and cell volume showed a positive correlation, with branching germ cells having the largest volume (817 µm³) (Fig. 3B’). By P0, single germ cells had a wide range of cell volumes, indicating that by the time of cyst fragmentation is complete, germ cells have a divergent range of cytoplasmic content due to cytoplasmic transport in germline cysts. On average, single germ cells were larger in volume compared to germ cells in cysts in P0, suggesting that germ cells in cysts have not yet completed cytoplasmic transport (Fig. 3 C and C’ and SI Appendix, Fig. S5B). These results demonstrated that branching germ cells have larger volumes, suggesting that these cells preferentially intake cytoplasm during cytoplasmic transport.

Fig. 3.

Cytoplasm and organelle content in cyst germ cells profiled based on bridge number during oocyte differentiation. (A–C’ ) The volume of single germ cells and germ cells in cysts in E14.5, E17.5, and P0 female germ cell clones. The bar in the graph represents the average value; the number above the bar represents the average volume of the germ cells; the number in the parentheses represents the number of measured germ cells. (D–D’’’’ ) A 3D confocal image showing a lineage-labeled germ cell clone with Golgi complex surface generated based on GM130 antibody staining using Imaris software (see original confocal images of the clone in Movie S5). Individual germ cells are outlined by yellow dotted lines. YFP: cyst germ cells, TEX14: bridges, GM130: Golgi complexes. (E) Diagram of the germ cell clone in D showing the geometry of the germ cells (green circles) and the bridges (red lines) connected with them. The numbers in the germ cells represent the normalized value of the content of Golgi complexes in each germ cell. (F ) Cyst germ cells profiled based on the number of bridges connected with them and the amount of Golgi complexes each germ cell contained in 4-cell cysts, 3-cell cysts, and 2-cell cysts in fetal ovaries from E17.5 to P0. (G) Comparison of the germ cell volume between the germ cells with a B-body and without a B-body in P0 clones. Data in the graph are presented as mean ± SD and a t test was used for statistical analysis. (H–H’’ ) Confocal images of an optical section showing a lineage-labeled germ cell clone in the P0 testis. YFP: cyst germ cells, TEX14: bridges, GFRα1: gonocytes. Arrowheads: germ cells with negative GFRα1 staining. (I and J ) Percentage of cysts (I) and clones (  J) containing the germ cells with only GFRα1-positive germ cells (red box), only GFRα1-negative germ cells (white box) and both GFRα1-positive and -negative germ cells (stripped box). (Scale bar, 10 µm).

To determine whether branching germ cells are also enriched with organelles, we quantified the Golgi complex volume in each germ cell of individual cysts using confocal imaged germline cysts and Imaris software (Fig. 3 D–F). We focused on 4-cell, 3-cell, and 2-cell cysts from E18.5 and P0 mouse ovaries, which are in the later stages of organelle transport in cysts. We found that the number of bridges corresponded positively with the volume of Golgi complexes in the cell (Fig. 3F and Movie S5). This suggests that branching cells preferentially accumulate organelles within germline cysts. Our previous study demonstrated that in the germ cells with enriched organelles, Golgi complexes organize into a B-body and these germ cells are preferentially protected from apoptosis in postnatal ovaries to become primary oocytes (18). To directly correlate germ cell volume, organelle content, and oocyte fate among the germ cells derived from the same PGC, we analyzed germ cell volume in P0 germ cell clones containing germ cells with a B-body. We found that germ cells with a B-body had a larger cell volume compared to germ cells without a B-body, reinforcing that germ cells with enriched cytoplasm preferentially differentiate into primary oocytes (Fig. 3G). Together, these findings suggest that branching germ cells accumulate organelles and cytoplasm from other cyst germ cells and differentiate into primary oocytes.

Uniform SSC Marker Expression in Male Fetal Branched Cysts.

To analyze whether male fetal germ cells in branched cysts show heterogeneity in cell fate, we examined the expression of SSC markers GFRα1 (GDNF family receptor alpha-1)(36, 37), PLZF (promyelocytic leukemia zinc finger) (38, 39) and UTF1 (undifferentiated embryonic cell transcription factor 1) (40, 41) in lineage-labeled germ cell clones in P0 testes. Because of germ cell connectivity, germ cells in the cyst showed a consistent expression of GFRα1 regardless of cyst structure (Fig. 3 H and I). Among the 128 cysts examined, 42% of the cysts contained germ cells that were all GFRα1 positive, 58% of the cysts contained germ cells that were all GFRα1 negative, suggesting that branched cyst structure does not influence SSC differentiation in male cysts. The germ cells in a clone did not express GFRa1 in a synchronized manner due to cyst fragmentation. Among the 26 clones examined, approximately 23% of the clones contained only GFRa1-positive cysts, 23% of the clones contained both GFRa1-positive cysts and negative cysts, and 54% of the clones contained only GFRa1-negative cysts in P0 testes (Fig. 3J). PLZF showed a similarly consistent expression in the germ cell in the cyst but not in the clone (SI Appendix, Fig. S6A). These results indicate that GFRa1 and PLZF expression may be regulated by signaling molecules that are shared among germ cells within a male cyst rather than intrinsically. UTF1 was expressed in all P0 germ cells, without detectable difference (SI Appendix, Fig. S6B).

Adult Male Germ Cells Form Branched Cysts during Spermatogenesis.

During spermatogenesis in adult testes, SSCs continually give rise to differentiating cells that form large germline cysts after a series of mitosis. We questioned whether there is a difference in cyst structure between male fetal and adult cysts. We conducted single-cell lineage tracing in the adult testis to label individual SSCs using Pax7-creER;R26-YFP mice (42). A single low dose of tamoxifen was injected into adult males (8 to 12 wk old) and testes were collected at 1 d, 10 d, and 20 d after the injection (Fig. 4A). Spermatogenesis from SSCs is divided into three major stages based on differentiation marker expression: 1) undifferentiated spermatogonia (E-cad-positive/KIT-negative); 2) differentiating spermatogonia (KIT-positive); and 3) spermatocytes (synaptonemal complex protein 3, SCP3-positive). Germ cell connectivity at stages 1 and 2 involves cell membrane connections between YFP-positive sister germ cells stained with E-cad or KIT antibodies (Fig. 4 B, D, and E). TEX14-positive bridges were observed between lineage-labeled germ cells connected by E-cad- or KIT-positive cell membranes and between SCP3-positive spermatocytes (SI Appendix, Fig. S7A and Fig. 4G).

Fig. 4.

Structures of germline cysts in spermatogenesis in the adult testis. (A) Diagram showing the strategy for lineage-labeling individual Pax7-positive spermatogonia stem cells (SSCs) in the adult testis. Testes were collected on days1, 10, and 20 after tamoxifen (Tmx) injection. (B) Confocal image of an optical section showing a lineage-labeled YFP-positive undifferentiated spermatogonia expressing E-cad (red) 1 d after Tmx injection. (C) Composition of lineage-labeled spermatogonia containing different numbers of germ cells 1 d after Tmx injection. As: single SSC; Apr: two connected SSCs; Aal-3: three aligned SSC; Aal-4: four aligned SSCs. (D–F) Optical sections of confocal images showing lineage-labeled germ cell clones that are E-cad positive (D), KIT positive (E) or SCP-3 positive (F) observed in the testes 10 or 20 d after Tmx injection. (G) Confocal image of an optical section showing meiotic germ cells labeled by a SPC3 antibody and intercellular bridges labeled by a TEX14 antibody. (H–K) Structure of germline cysts profiled by the percentage of germ cells connected by 0, 1, 2, and 3 or 4 bridges (H) in E14.5 testes (I), E-cad-positive undifferentiated spermatogonia (J), KIT-positive differentiating spermatogonia (K) and spermatocytes (L). 1,128 cells from 25 clones, 722 from 118 clones, 798 from 22 clones, and 640 cells were analyzed in (I), (J), (K), and (L) respectability. (Scale bar, 10 µm in B, F, and G. Scale bar, 50 µm in D and E ).

At 1 d after tamoxifen injection, individual linage-labeled SSCs were recognized by YFP and E-cad double-positive staining (Fig. 4B). Among these cells, 72.3% were single (As), 22.3% were paired (Apr; two connected SSCs), 0.9% were at Aal-3 (three aligned SSCs), and 4.46% were at Aal-4 (four aligned SSCs) (Fig. 4C). By 10 d after tamoxifen injection, 92.0% of the labeled cysts were E-cad-positive/KIT-negative undifferentiated spermatogonia (SI Appendix, Fig. S7B). To detail how cyst structure develops as spermatogenesis progresses, we quantified the number of germ cells harboring different numbers of membrane connections or TEX14-positive bridges in the clone (Fig. 4H). Given that lineage-labeled spermatocytes form several hundreds to thousands of cells, cyst structure was analyzed by counting the number of TEX14-positive bridges for each SCP-positive spermatocyte. We found that SSCs at the undifferentiated stage (E-cad-positive) were mostly connected via the linear structure, with only 3.7% of the germ cells having 3 bridges (Fig. 4J). The percentage of germ cells with 3 or 4 bridges increased as spermatogenesis progressed: 18.3% in differentiating spermatogonia (KIT-positive) and 26.6% in spermatocytes (SCP3-positive) (Fig. 4 K and L). The proportion of spermatocytes with different numbers of bridges was similar to those in fetal male cysts, suggesting that male germ cells are connected in a similar configuration in both fetal and adult cysts (Fig. 4 I and L).

E-cad Facilitates the Formation of Branched Cysts in Fetal Gonads.

The germline cyst arising from a single PGC has a linear structure at the 4-cell stage after two rounds of synchronized mitotic divisions (Fig. 5A). When a 4-cell cyst divide synchronously into an 8-cell cyst, the following three types of cysts can be produced: 1) a linear cyst; 2) a branched cyst with one branching germ cell (i.e.,1-branching cell cyst); and 3) a branched cyst with two branching germ cells (i.e., 2-branching cell cyst). If branched cysts are produced randomly due to mitotic spindle orientation, the expected percentage is 25%, 50%, and 25% for linear, 1-branching, and 2-branching cell cysts, respectively (SI Appendix, Fig. S8A). However, when we analyzed 8-cell cysts collected from E12.5 mouse gonads (in which female and male cysts share the same structure), we found that there were 48% linear, 19% 1-branching, and 33% 2-branching cell cysts, suggesting that branched cyst formation in mouse fetal gonads is not random (Fig. 5B).

Fig. 5.

E-cad junctions are involved in branched cyst formation by positioning intercellular bridges. (A) Diagram demonstrating three types of 8-cell cysts (linear, 1-branching cell, and 2-branching cell) that form through incomplete cytokinesis after synchronized mitotic division of a 4-cell cyst. (B) Percentage of three types of 8-cell cysts expected from random spindle orientation and observed in wild-type (Cag-creER;R26-YFP;E-cad+/+) and E-cad knockout (Cag-creER;R26-YFP;E-cad fl/fl) germ cell clones. (C and D) Optical sections from confocal images showing centrosome (γ-tub) locations and the E-cad junctions in cyst germ cells. Yellow arrowheads in C: two paired centrosomes; white arrowhead in C: a single centrosome; yellow arrowheads in D: two unpaired centrosomes. (E) Confocal image of an optical section showing that in a cyst with mitotic germ cells, E-cad distributed along the cell membrane without a junction between germ cells (see original confocal images of the cyst in Movie S6). (F) The distance between E-cad junction and centrosomes in 4-cell germline cysts. (G) The distribution of E-cad junctions in 4-cell cysts. B1: the bridge from 1st division; B2a and B2b: two bridges from the 2nd division. B2a/b: the E-cad junction observed at either one of B2; B2a&b: E-cad junction observed at both B2a and B2b. (H) A 3D confocal image showing the location of E-cad junctions and intercellular bridges stained by a RacGAP antibody in a 4-cell cyst. (I) Optical sections from confocal images showing an intercellular bridge found in the middle and at the edge of the E-cad junction. (J) Percentage of the intercellular bridges located in the middle or at the edge of the E-cad junction in 4-cell cysts. (K) Diagram showing that the location of intercellular bridges influences the structure of the branched cyst during cyst formation. (L) The percentage of fragmented and connected cysts in the wild-type (Cag-creER;R26-YFP;E-cad+/+) and E-cad knockout (Cag-creER;R26-YFP;E-cad fl/fl) 4-cell and 8-cell cysts. Chi-square test was used for statistical analysis between wild-type and knockout cysts. (M and N) Distribution of the angles measured between two adjacent intercellular bridges in the wild-type (Cag-creER;R26-YFP;E-cad+/+) (M) and E-cad knockout (Cag-creER;R26-YPF;E-cad fl/fl) (N) 4-cell cysts. (O) Number of total oocytes in the P4 wild-type (Figla-cre−/−;E-cad fl/fl) and E-cad knockout ovaries (Figla-cre+/−;E-cad fl/fl). (P) The percentage of E-cad-negative oocytes and the number of total oocytes in the E-cad knockout ovary. Magenta-lined dots: ovaries with less than 50% E-cad-negative oocytes; Blue-lined dots: ovaries with more than 50% E-cad-negative oocytes. (Q) Percentage of the primary oocytes containing a B-body in the P4 wild-type and E-cad knockout ovaries. (R) Volume of primary oocytes in the P4 wild-type and E-cad knockout ovaries. (S) A diagram demonstrating the timeline of cyst formation, fragmentation and cytoplasm and organelle transport-mediated oocyte differentiation in germline cysts. (Scale bar, 10 µm). Data in the graph are presented as mean ± SD. The bar in the graph represents the average value. t tests were used for the statistical analysis between the two groups.

We investigated how branched cysts form from 4-cells to 8-cells. In many cell types, cadherin junctions formed between the cells orient mitotic spindles during division (43, 44). We examined the distribution of centrosomes and E-cad in 4-cell cysts. E-cad junctions were found in 4-cell female and male cysts between two adjacent cyst germ cells, without a preference in forming between certain two germ cells (Fig. 5 C, D, and G and SI Appendix, Fig. S9). By antibody staining, E-cad was not detected in somatic cells in fetal gonads (SI Appendix, Fig. S9). Germ cell centrosomes tended to localize near the E-cad junctions in interphase female and male germ cells at E12.5 (Fig. 5 C and D). The distance between the centrosome and E-cad junctions varied from less than 1 µm to over 10 µm for single centrosomes, duplicated centrosomes (paired), and two separated centrosomes (unpaired) (Fig. 5 C, D, and F). The E-cad junctions disappeared when germ cells enter the M-phase, suggesting that E-cad junctions may not be involved in spindle orientation during cyst formation (Fig. 5E and Movie S6).

We found that intercellular bridges in cysts were located at the edge of E-cad junctions at a high frequency (Fig. 5 H and I). In 4-cell cysts, 90.7% of the bridges were found at the edge of the junction (Fig. 5J). The location of the bridges influences the location of future daughter cells and branched cyst structure (Fig. 5K and SI Appendix, Fig. S8B). To elucidate whether E-cad junctions are involved in positioning bridges between germ cells and in branched cyst formation, we knocked out E-cad in lineage-labeled individual PGCs using Cag-creER;R26-YFP;E-cadherin-flox mice, where the injection of a low dose of tamoxifen induces Cre recombinase activity that drives the expression of YFP protein and depletion of E-cad simultaneously in single PGCs. To avoid the potential effect from the loss of E-cad in fetal somatic cells on bridge location, only the lineage-labeled germ cell clones without lineage-labeled somatic cells nearby were used for cyst structure analysis. We compared cyst formation and structure between mutant (Cag-creER+/−;R26-YFP/YFP;E-cadherin flox/flox) and wild-type (Cag-creER+/−;R26-YFP/YFP;E-cadherin +/+) cysts and found that E-cad knockout led to increased cyst fragmentations at the 8-cell cyst stage (Fig. 5L), indicating that E-cad junctions are involved in holding cyst germ cells together, in combination with the intercellular bridges during cyst formation.

To quantitatively demonstrate the location of intercellular bridges in 4-cell wild-type and E-cad-depleted cysts, we measured the angles generated from the center of the cell to each of the two bridges that it has in 4-cell cysts (Fig. 5 M and N). We found that the angles were distributed over a wide range of values in 4-cell wild-type cysts (Fig. 5M). By contrast, angles measured in the E-cad knockout 4-cell cysts had a distribution with the majority being ~90 degrees, showing a close coincidence with the pattern of random bridge distribution on the cell surface based on calculated circumferences (Fig. 5N). The change in bridge location at the 4-cell stage corresponded to an altered percentage in 8-cell cysts of three types of structures. The percentages of linear, 1-branching cell, and 2-branching cell cysts were 22%, 56%, and 22%, respectively, which is remarkably similar to the expected ratio when 8-cell cysts form randomly (Fig. 5B and SI Appendix, Fig. S8).

To investigate the effect of altered cyst structure and cyst fragmentation on oocyte differentiation, we examined oocyte differentiation in Figlα-cre+/−; E-cadherinflox/flox mice where E-cad knockout occurs specifically in germ cells, starting at around E12.5 (45). We found that the efficiency of E-cad knockout in germ cells (i.e., percentage of E-cad-negative oocytes) varied in the mutant ovaries and no significant difference was observed in the average number of oocytes per ovary between wild-type and mutant ovaries at P4 (Fig. 5O). However, there was a negative association between the number of oocytes in the ovary and the percentage of E-cad-negative oocytes, suggesting that the loss of E-cad causes reduced oocyte numbers (Fig. 5P). Primary oocytes with E-cad knockout were smaller in cell volume and showed a reduction in the percentage of primary oocytes containing a B-body, indicating that altered branched cyst formation and increased cyst fragmentation in E-cad mutant germ cells lead to defects in cytoplasmic and organelle enrichment during oocyte differentiation (Fig. 5 Q and R).

Discussion

Branched Cyst Structure Maximizes Intercellular Communication.

During gametogenesis in germline cysts, intercellular bridges allow the exchange of signaling molecules and small organelles between cyst germ cells. Germline cysts in many animals are in branched structures, including Drosophila ovaries and testes, Xenopus ovaries, and rat adult testes (19, 46, 47). The pattern of physical connections between germ cells likely influences intercellular communications. Here, we have found that germline cysts in mouse fetal ovaries and testes are in branched cyst structures, with an average of 16.8% female germ cells and 20.4% male germ cells being connected by three or four intercellular bridges by the time cyst formation ceases at E14.5. Similarly, during spermatogenesis in adult testes, cysts become more branched as spermatogenesis progresses such that 26.6% of the germ cells are branching cells when they enter meiosis.

The branched cyst structure may provide an advantage in facilitating germ cell communication. To analyze the difference in the efficiency of germ cell communication between different cyst structures, we compared the efficiency of information sharing in three types of 8-cell cysts using mathematical modeling, which measured the time needed to spread the information from a single cell until it reaches a steady-state distribution among the cells in the entire 8-cell cyst (SI Appendix, Supplementary File 3). The distribution time in the linear cyst, one-branching cell cyst, and two-branching cell cyst were 40 s, 30 s, and 22 s, respectively, thus branched cyst structures facilitate more efficient intercellular communications, compared with linear cysts (Movies S7–S9). This result may explain why male germ cells form branched cyst structures without significant fragmentation to ensure efficient exchange of signaling molecules needed for gonocyte differentiation and spermatogenesis.

Selective Loss of Germ Cells and Intercellular Bridges Facilitate Oocyte Determination.

The branching germ cells connected with three or four bridges are likely the initial founder cells of the cyst, since the number of bridges on the germ cell reflects the rounds of mitosis the cell has experienced. In female cysts, branching germ cells are preferentially protected from cell loss and cyst fragmentation from E14.5 to E17, increasing their chances of enriching their cellular content and becoming oocytes (Fig. 2K). Our analyses on the change in cyst structure suggest that cytoplasmic transport in germline cysts is conducted in a directional manner. The preferential loss of 1-bridge and 2-bridge germ cells from E14.5 to E17.5 indicates that cellular content is first transported locally between peripheral germ cells and further enriched in branching germ cells. The selective germ cell loss within the cyst from E14.5 to E17.5 may help facilitate the directional communication among cyst germ cells, which is needed for organelle and cytoplasmic enrichment when cysts undergo rapid fragmentation from E17.5 to P0, caused primarily by bridge loss. Mechanisms for selective bridge closure/loss between E17.5 and P0, in particular, whether the cellular machinery of cytokinesis is found at certain bridges, should be addressed after acquiring a better understanding of bridge formation, maintenance, and closure.

Although approximately 80% of the E14.5 germ cells undergo cell death from E14.5 to P4, apoptotic germ cells detected by TUNEL or cleaved-PARP antibody are  at a very low rate of between 0 to 4.3% in fetal ovaries (8, 16, 48). To examine whether periphery cyst germ cells undergo cell death preferentially, we stained individual isolated germline cysts with an antibody to Annexin V that detects early apoptotic cells and an antibody to TEX14 for bridges (SI Appendix, Fig. S10 and Movie S10) (18). In healthy germ cells, the intercellular bridge was found in the narrow cytoplasmic region between two connected germ cells (SI Appendix, Fig. S10, arrowhead). However, in the germ cells that were Annexin V-positive, a gap between the cytoplasm and the nucleus, and asymmetrically distributed cytoplasm was observed. The change in the distribution of cytoplasm disallowed us to analyze whether the Annexin V-positive cells were connected by bridges and the location of these cells in the cyst confidently. The gap observed here in the Annexin V-positive germ cells is consistent with a previous report on mouse fetal germ cells undergoing programmed cell death during oocyte differentiation (49). The separation between the dying nucleus and cytoplasm may prevent the spread of cell death signals in the entire female cyst.

E-cad Junctions Facilitate Branched Fetal Cyst Formation.

We observed E-cad junctions between cyst germ cells from E10.5 to E14.5, suggesting that they may be involved in germline cyst formation (SI Appendix, Fig. S9). The E-cad junctions may be essential for holding cyst germ cells together, which are otherwise connected via intercellular bridges. E-cad junctions thus may protect cysts from fragmentation caused by somatic cell invasion during cyst formation, an idea confirmed by our observation that E-cad-depleted germ cell clones had a higher rate of cyst fragmentation (Fig. 5L).

We found that E-cad junctions play a role in branched cyst formation by positioning the intercellular bridge between two sister germ cells (Fig. 5K). E-cad is involved in spindle orientation in Drosophila germline stem cells, as well as sensory organ precursor cells and epithelial cells in Drosophila and mammals (22, 50–56). Here, we found that, although germ cell centrosomes showed a strong association with the E-cad junctions during interphase, the junctions became dispersed during mitotic division, indicating that E-cad junctions in the germline cyst are not involved in orienting mitotic spindles during germ cell division. It is intriguing that when E-cad was depleted in individual 4-cell cysts, both the angles between two adjacent bridges and the percentages of linear and branched 8-cell cysts resembled the patterns caused by random positioning of intercellular bridges by E-cad junctions (Fig. 5 B and N and SI Appendix, Fig. S8B). This result indicates that branched cyst formation in wild-type ovaries is not caused by randomly positioned intercellular bridges. The dynamics of E-cad junction formation and maintenance between cyst germ cells may provide insight into how bridge location and resultant branched cysts formation are regulated.

We observed that in mice with germ cell-specific depletion of E-cad, mutant primary oocytes had a decreased cell volume and the percentage of primary oocytes containing a B-body, suggesting that cytoplasmic transport in these germ cells is compromised. However, given the role of E-cad in PGC survival and cell polarity, how altered cyst structure changes oocyte determination in germline cysts needs to be addressed in future studies with the appropriate genetic mutant mouse models.

Despite the onset of sex differentiation around E11.5, female and male PGCs form branched cysts in a similar structure (1). Consistent with limited cyst fragmentation observed in male cysts, branched cyst structure may promote intercellular communications between male germ cells and facilitate symmetric germ cell fate in the cyst, including the loss of whole germ cell clones containing defective germ cells in fetal testes (27). The phenomenon of asymmetrical cell fate in females and symmetrical cell fate in males in the similar branched cyst structure, reported in the present study, is conserved in Drosophila gametogenesis (25). Whether protein composition of the intercellular bridges is different between female and male cysts, as observed in Drosophila, will be investigated in our future study (57, 58). Addressing this question will further help us understand selective loss of bridges in female cysts and limited cyst fragmentation in male cysts.

In summary, our present study identifies branched germline cyst structures during mouse gametogenesis and differential patterns of cyst fragmentation between female and male cysts in fetal gonads. The patterns of cyst fragmentation, germ cell loss, and distribution of organelle/cytoplasmic content in female cysts indicate that branching germ cells differentiate into primary oocytes within the germline cysts. E-cad is involved in branched cyst formation, which may influence primary oocyte number and organelle content in the primary oocytes (Fig. 5S). Our study sheds light on how the number of primary oocytes is determined during mammalian ovary formation.

Materials and Methods

Animals.

Cag-creER (004682) (59), R26-YFP (006148)(60), Figlα-GFP-icre (037373) (45) and Cdh1/E-cadherin-flox (005319) (61) mouse strains were acquired from the Jackson Laboratory. All mice were maintained at C57BL/6 background and housed and bred according to the protocol approved by the Institutional Animal Care & Use Committee (IACUC) at the University of Michigan (PRO00008693), the Buck Institute for Research on Aging (A10207) and the University of Missouri (36647). Mice euthanasia in this study was performed by strictly following the protocol from the IACUC.

Single PGC Lineage Labeling and E-cad Knockout.

A single dose (0.2 mg per 40 g body weight) of tamoxifen was administered to female (R26-YFP/YFP) mice that were plugged by male (Cag-creER+/−; R26-YFP/YFP) mice at E10.5 by intraperitoneal injection (11, 18, 31). Fetal gonads were collected and stained with a green fluorescent protein (GFP) antibody to reveal germ cell clones. For single SSC lineage labeling, adult (8 to 12 wk and older) male Pax7-creER+/−; R26-YFP/YFP mice were given with a single dose of 5.0-mg tamoxifen by intraperitoneal injection. Testes were collected at days 1, 10, and 20 after the injection and stained with a GFP antibody. To knockout E-cad in the single germ cell clone, a single dose (0.4 mg per 40 g body weight) of tamoxifen was administered to female (R26R-YFP/YFP; E-cad-flox/flox) mice that were plugged by male (Cag-creER+/−; R26-YFP/YFP; E-cad flox/flox) mice at E10.5 by intraperitoneal injection. Germ cell clones with Cre recombinase activity and E-cad depletion were identified by YFP expression in the germ cells.

Whole-Mount Immunostaining.

Fetal gonads and adult testes were dissected in phosphate-buffered saline (PBS), and seminiferous tubules were untangled (11, 62). Tissues were fixed in cold 4% paraformaldehyde (PFA) for 1 h (E12.5 and E14.5 gonads and seminiferous tubules) or 2 h (E17.5 and P0 gonads). Tissues were washed in PBST₂ (PBS with 0.1% Tween-20 and 0.5% Triton X-100) and incubated with the primary antibodies (SI Appendix, Table S11) at 4 °C overnight in PBST₂ with 10% donkey serum, 10% bovine serum albumin (BSA), and 100 mM Glycine. Tissues were washed with PBST₂, incubated with the secondary antibodies in PBST₂ at 4 °C overnight (SI Appendix, Table S11), and stained with DAPI to visualize nuclei. The stained tissues were placed on the coated glass slides and mounted on slides with mounting medium for confocal imaging.

3D Models of Germ Cell Clones and Germline Cyst Structure Analysis.

Germ cell clones were imaged by confocal microscopy at 0.8 to 1.0 µm per scan. Imaris software was used to generate the 3D model of a series of confocal images from a germ cell clone. To create a 3D model, spheres were placed on each germ cell (YFP-positive) using the Spots tool. TEX14 (or RacGAP)-positive intercellular bridges of the germ cell clone were selected using the Mask tool and marked using the Spots tool. To lay out the 2D structure of each germline cyst, the location of the bridges and germ cells were laid out manually by following the location of germ cells and intercellular bridges using the Image-J software. To analyze the number of bridges per spermatocyte in the adult testis, bridges attached to each lineage labeled SCP3-positive cell were counted.

Mathematical Modeling of the Loss of Germ Cells and Intercellular Bridges in Germ Cell Clones.

The model was written in MATLAB using undirected graph algorithms in which each cell is considered a “node” and each connection or bridge is considered an “edge”. Adjacency matrices describing the connectivity of the cells in cysts were created using Cytoscape, all remaining adjacency matrices were created manually. Simulations of two models describing the mechanisms of cyst fragmentation were performed: random cell loss and random bridge loss. Using each clone from E14.5 (n = 41) as an initial model input sample, 250 trials were run with random cell or bridge removal in a manner that results in an average loss that corresponds with the experimental loss of cells or bridges across multiple trials (See SI Appendix, Supplemental Files 1 and 2 for additional details). Each mechanism was tested to compare the resulting population of hundreds of simulated broken-down cysts (in silico E17.5) to the E17.5 cyst structure data (in vivo E17.5, n = 29). The same approach was used to model random cell loss and random bridge loss in E17.5 clones (n = 29), the resulting population of hundreds of simulated broken-down cysts (in silico P0) was compared to the P0 cyst structure data (in vivo P0, n = 22). The connectivity of the cells is given by the degree distribution, where the degree of a cell is the number of bridges it has (or the number of other cells to which it is connected) and the degree distribution is the percentage of cells in a clone with 0, 1, 2, 3, or 4 bridges. The averages of each cell number were calculated by multiplying with total cell numbers from in vivo data. For each summary statistic, a two-sample t test with a 0.05-significance level was used to test the null hypothesis that the data came from independent random samples from normal distributions with equal means and equal but unknown variances.

Apoptosis Assay in Isolated Germline Cysts.

Germline cysts from E17.5 mouse fetal ovaries were isolated using insulin needles in a small drop of PBS. PBS was replaced with an Annexin V antibody diluted in DMEM/F12 media supplemented with 3 mg/mL BSA, and the cysts were incubated with the antibody for 30 min in a cell culture incubator. After gentle washes with PBS, the cysts were embedded in Smear Gell (Diagnocine) on a chambered slide. The cysts were fixed in 4% PFA for 10 min at room temperature and stained with the antibodies to GFP and TEX14 using the whole-mount immunostaining protocol.

Quantification of Germ Cell Volume and Golgi Complexes in the Germline Cyst.

Lineage-labeled fetal ovaries were stained with GFP, GM130 and TEX14 antibodies, germ cell clones were imaged using confocal microscopy at the Z = 1 µm per scan. To measure germ cell diameter, two vertical diameters of the biggest cross-section of the YFP+ germ cell were measured using Image J, and the average value (Rave gc) was used to represent the diameter of the germ cell. The volume of the germ cells (Vgc) was calculated using the equation: Vgc = 4/3*3.14*Rave gc^3. The volume of the Golgi complex in each germ cell was measured using the surface tool of Imaris software. To compare the volume of Golgi complexes among the germ cells of the cyst, the volume of the Golgi complex in each germ cell was normalized by dividing its Golgi volume by the smallest volume of the Golgi complex in the cyst.

Measurement of the Angles between Intercellular Bridges.

To measure the angles between two adjacent intercellular bridges in the 4-cell cysts, lineage-labeled gonads were collected at E12.5 and stained with GFP and RacGAP antibodies for confocal imaging. The angle of the triangle comprised of three points (two adjacent intercellular bridges, and the center of the germ cell nucleus) was measured on the 3D model generated from a series of confocal images using the Imaris software.

Follicle Quantification, Oocyte Volume Measurement and B-Body Quantification.

P4 ovaries were collected from the wild-type (Figla-cre−/−; E-cad flox/flox) and homozygous mutant (Figla-cre+/−; E-cad flox/flox) mice. The ovaries were fixed in cold 4% PFA for 2 h at 4 °C, washed in PBS and incubated in 30% sucrose overnight before embedding in optimal cutting temperature compound. Serial sections were cut at 10 µm of the entire ovary and stained using a DDX4 antibody to reveal oocytes. Follicles were counted in every fifth section. The number of follicles per ovary was calculated as follicles/section × total sections. To measure primary oocyte volume and the percentage of the oocytes containing a B-body, wild-type and homozygous P4 ovaries were stained with DDX4, GM130, and E-cad antibodies and imaged using confocal microscopy at the Z = 1 µm per scan. An area of 100 µm wide and 100 µm deep to the ovarian surface in the ovary was chosen for analysis, and two optical sections with at least 15-µm interval from each ovary were analyzed. Primary oocytes were recognized by DDX4-positive oocytes that are surrounded by a single layer of flattened pregranulosa cells. For wild-type ovaries, all primary oocytes in the area were analyzed. For homozygous mutant ovaries, only the primary oocytes with E-cad-negative staining were analyzed. Oocyte diameters were measured for oocyte volume calculation as described above. The B-body is defined as a circular Golgi complex in DDX4-positive primary oocytes. To quantify the percentage of primary oocytes containing a B-body, the number of total oocytes in the above defined area in the wild-type ovary and the number of the oocytes containing a B-body were quantified. For homozygous mutant ovaries, E-cad-negative oocytes and those containing a B-body in the above-defined area were quantified.

Statistics.

All data were presented as mean ± SD. Nonparametric t tests were run to analyze the difference between the two experimental groups. In addition, multiple experimental groups were analyzed using one-way ANOVA. P value level of at least P < 0.05 was considered statistically significant.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Dataset S05 (XLSX)

Dataset S06 (XLSX)

Dataset S07 (XLSX)

Dataset S08 (XLSX)

Dataset S09 (XLSX)

Dataset S10 (XLSX)

Dataset S11 (XLSX)

Supplementary File S01 (PDF)

Supplementary File S02A (PDF)

Supplementary File S02B (PDF)

Supplementary File S02C (PDF)

Supplementary File S02D (PDF)

Supplementary File S02E (PDF)

Supplementary File S02F (PDF)

Supplementary File S02G (PDF)

Supplementary File S02H (PDF)

Supplementary File S03 (PDF)

Movie S1.

A series of confocal images of a germ cell clone in the E12.5 ovary shown in Figure 1B.

Movie S2.

A series of confocal images of a germ cell clone in the E14.5 ovary shown in Figure 1C.

Movie S3.

A series of confocal images of a germ cell clone in the E17.5 ovary shown in Figure 1D.

Movie S4.

A series of confocal images of a germ cell clone in the P0 ovary shown in Figure 1E.

Movie S5.

A series of confocal images of a germ cell clone in the E18.5 ovary stained with antibodies to GFP, TEX14, and GM130.

Movie S6.

E-cadherin distributed sparsely on the germ cell membrane without forming junction during germ cell mitotic division.

Movie S7.

Efficiency of information sharing modeled in an 8-cell linear cyst.

Movie S8.

Efficiency of information sharing modeled in a 1-branching cell cyst.

Movie S9.

Efficiency of information sharing modeled in a 2-branching cell cyst.

Movie S10.

A series of confocal images of a germline cyst (E17.5 ovary) stained with antibodies to Annexin V, GFP and TEX14.

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.