Supporting Online Material for

Transcription

1 Supporting Online Material for Oocyte-Specific Deletion of Pten Causes Premature Activation of the Primordial Follicle Pool Pradeep Reddy, Lian Liu, Deepak Adhikari, Krishna Jagarlamudi, Singareddy Rajareddy, Yan Shen, Chun Du, Wenli Tang, Tuula Hämäläinen, Stanford L. Peng, Zi-Jian Lan, Austin J. Cooney, Ilpo Huhtaniemi, Kui Liu* *To whom correspondence should be addressed. Published 2 February 2008, Science 319, 611 (2008) DOI: /science This PDF file includes: Materials and Methods SOM Text Figs. S1 to S6 References

2 Materials and Methods Mice The mice (S1) in a BALB/c; 129S4 genomic background were obtained from the Jackson Laboratory (Bar Harbor, MN), and were backcrossed to C57BL/6J mice for 10 generations. Transgenic mice carrying growth differentiation factor 9 (Gdf-9) promoter-mediated Cre recombinase, which is specifically expressed in oocytes in primordial and further developed follicles (the GCre mice) (S2), were backcrossed to C57BL/6J mice for 6 generations. After multiple rounds of crossing, we had generated mutant female mice with a ;GCre+ genotype and control female mice with a genotype. Generation of Foxo3a -/- mice has been described previously (S3). In the current study, the Foxo3a -/- mice were backcrossed to C57BL/6J for 8 generations. To obtain ;GCre+; Foxo3a -/- double-knockout mice, an initial breeding with ;GCre+ males Foxo3a -/- females was set up to obtain pups for further breeding. After several rounds of crossing, breeding pairs of ; Foxo3a +/- females ;GCre+; Foxo3a -/- males were used to obtain ;GCre+; Foxo3a -/- female pups for experiments. The mice were housed under controlled environmental conditions with free access to water and food. Illumination was on between 0600 and 1800 h. Experimental protocols were approved by the regional ethical committee of Umeå University, Sweden. Reagents, antibodies, and immunological detection methods The rabbit polyclonal antibodies to PTEN, Akt, phospho-akt (serine 473), phospho-s6 ribosomal protein (rps6) (serine 235/236), rps6, mammalian target of rapamycin (mtor), phospho-mtor (serine 2448), phospho-glycogen synthase kinase-3 (GSK-3) α/β (serine 21/9), GSK-3α, tuberin/tsc2, phospho-p44/42 mitogen-activated protein kinase (MAPK) (threonine 202/tyrosine 204), p44/42 MAPK, and rabbit monoclonal antibody to phospho-tuberin/tsc2 (threonine 1462), were obtained from Cell Signaling Technologies (Beverly, MA). The rabbit polyclonal antibodies to Foxo3a, phospho-foxo3a (threonine 32), p70 S6 kinase (S6K), and phospho-s6k (threonine 389) were from Upstate Biotechnology (New York, NY). Pregnant mare serum gonadotropin (PMSG), human chorionic gonadotropin (hcg), and mouse monoclonal antibody to β-actin were purchased from Sigma-Aldrich Sweden AB (Stockholm, Sweden). The phosphatidylinositol 3-kinase (PI3K)-specific inhibitor LY , mtorspecific inhibitor rapamycin, MAPK kinase 1 (MEK1)-specific inhibitor PD98059, and recombinant mouse Kit ligand (KL) were obtained from EMD Biosciences (San Diego, CA). Western blots were carried out according to the instructions of the suppliers for the different antibodies, and visualized using the ECL Plus Western Blotting Detection System (Amersham Biosciences, Uppsala, Sweden). Quantification of ovarian follicles and histological analysis Quantification of ovarian follicles was performed as previously described (S4). Briefly, ovaries were fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin. To count the numbers of follicles, paraffin-embedded ovaries were serially sectioned at 8-μm thickness and stained with hematoxylin for morphological observation. Ovarian follicles at different developmental stages, including primordial, transient, type 3b, type 4, type 5, and type 6 were counted in all sections of an ovary, based on the well-accepted standards established by Pedersen and Peters (S5). Growing transient follicles were defined as follicles that have obviously enlarged oocytes but which are still enclosed in flattened pre-granulosa cells. In each section, 2

3 follicles that contained oocytes with clearly visible nuclei were scored, as previously reported (S6). Judged from careful morphological analysis, the incidence of counting the same follicle twice or missing a follicle was low. Isolation of oocytes from postnatal mouse ovaries Mice were sacrificed by decapitation, and the ovaries were dissected free of fat and connective tissue using a microscope. The ovaries were then minced with a pair of dissection scissors before being incubated in 0.05% collagenase dissolved in Dulbecco s modified Eagle s medium-f12 (DMEM/F12; Invitrogen) supplemented with 4 mg/ml bovine serum albumin (BSA), 100 units/ml penicillin, and 100 µg/ml streptomycin, with frequent agitation and pipetting. After the tissues had mostly been digested by collagenase, usually within min, EDTA was added to this mixture to a final concentration of 40 mm, and the mixture was incubated at 37ºC with frequent pipetting for another min until clusters of granulosa cells or other cells were completely dispersed. The mixture of cells and oocytes was then washed once and cultured in a 6-cm or 10-cm tissue culture dish with the above-mentioned serum-free DMEM/F12 medium for 12 h, to allow the granulosa cells and other ovarian cells to attach to the plastic. The unattached oocytes and red blood cells were then recovered by collection of the supernatant and centrifugation at 1,000 rpm for 5 min at room temperature. Red blood cells were subsequently removed using a hypotonic buffer containing 144 mm NH 4 Cl and 17 mm Tris HCl (ph 7.2). After several washes, oocytes were collected by centrifugation. They were then starved for 4 h in serum-free DMEM/F12 medium at 37 C in a humidified atmosphere (5% CO 2 and 95% air), which was followed by KL stimulation, or by lysis in a buffer containing 50 mm Tris HCl (ph 8.0), 120 mm NaCl, 20 mm NaF, 20 mm β-glycerophosphate, 1 mm EDTA, 6 mm EGTA (ph 8.0), 1% NP-40, 1 mm DTT, 5 mm benzamidine, 1 mm PMSF, 250 μm sodium orthovanadate, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 μg/ml pepstatin, followed by centrifugation at 14,000 rpm for 20 min at 4ºC. The supernatants were collected and protein concentrations were measured using the bicinchoninic acid (BCA) protein assay, and equal amounts of proteins were used for western blot. KL stimulation of starved oocytes For KL stimulation, equal amounts of oocytes were aliquoted into wells of a 24-well plate. Typically, each well contained oocytes obtained from 3 5 ;GCre+ mice or 6 10 mice that were days old. The oocytes were first starved by culturing them in serum-free DMEM/F12 medium for 4 h, followed by treatment with 100 ng/ml KL for 2 10 min. After KL stimulation, the 24-well plate was chilled on ice, and oocytes were lysed as described above for western blot analysis. Measurement of serum hormone levels Adult ;GCre+ female mice from weeks were sacrificed randomly due to lack of regular estrus cycles; female mice of similar ages were sacrificed at the proestrus stage (based on vaginal smears) in order to measure gonadotropin levels during the follicular growth phase, but not the ovulation phase. Serum hormone levels were determined by immunoassay as described previously in the following papers: FSH (S7), LH (S8), and testosterone (S9). Gonadotropin-induced ovulation and size measurement of ovulated oocytes 3

4 To induce synchronized follicular growth and ovulation, immature 23-day-old female mice were injected i.p. with 5 IU of PMSG to stimulate follicular development, and with 5 IU hcg 48 h later to induce ovulation. Ovulation normally takes place h after hcg treatment (S10). Cumulus-oocyte complexes were recovered from oviducts, and treated with hyluaronidase (0.1%) before oocytes were collected. For size measurement of oocytes, 5 mice of each genotype were super-ovulated, and oocytes were chosen randomly for measurement of diameters using a Zeiss AX10 microscope. Statistical analysis All experiments were repeated at least 3 times. For comparisons of follicle numbers and hormone levels in ;GCre+ and mice, differences between the two groups were calculated with Student s t-test, and a difference was considered to be significant if P <

5 Supporting Text Generation of mice with oocyte-specific deletion of Pten We deleted the Pten gene from mouse oocytes in primordial and further developed follicles by crossing mice (S1) with GCre mice (S2). A schematic representation of deletion of Pten exon 5 and creation of a Pten Δ5 allele in oocytes by Cre-mediated recombination is shown in fig. S1A. We found that the offspring of mutant ( ;GCre+) female mice that had been mated with wild-type males showed a complete deletion of Pten exon 5 in one allele of their tail-tip genomic DNA (fig. S1B). Furthermore, by western blot, we confirmed that the expression of PTEN in ;GCre+ oocytes was largely reduced relative to normal PTEN expression in oocytes (fig. S1C). The remaining low level of PTEN expression in ;GCre+ oocyte may have been from contamination of other types of ovarian cells, or from oocytes whose Pten deletion had not yet been completed (fig. S1C). When the oocyte preparation was filtered with a cell-dispersing screen with 25-µm opening, to remove the contaminating granulosa cells and other types of ovarian cells including oocytes that were smaller than 25 µm, the ;GCre+ oocytes ( > 25 µm, representing growing oocytes) showed almost no PTEN expression (fig. S1D). Thus, deletion of Pten in oocytes in ;GCre+ mice was successful. Other signaling studies in ;GCre+ oocytes We found that the phosphorylation of rps6 (serine 235/6) and S6K in ;GCre+ oocytes cultured in vitro was to a large extent sensitive to the PI3K-specific inhibitor LY (LY) and the mtor-specific inhibitor rapamycin (Rap) (fig. S4A), indicating that the activation of rps6 in ;GCre+ oocytes cultured in vitro is largely dependent on activities of PI3K and mtor. As a control, treatment with the MEK1-specific inhibitor PD98059 (PD) did not suppress the levels of p-rps6 (serine 235/6) and p-s6k (threonine 389) in oocytes (fig. S4A). In addition, the mtor inhibitor rapamycin did not suppress the level of p-akt (serine 473) in oocytes (fig. S4A), suggesting that either mtor is downstream of Akt in the signaling cascade in oocytes, or the rapamycin-sensitive mtorc1 is not involved in regulation of Akt phosphorylation, as recently suggested (S11, S12). In ;GCre+ oocytes the phosphorylation status of a common Akt substrate, GSK-3 (S13), does not appear to be affected by the loss of Pten (fig. S4B). Also, activation of p44/42 MAPK was not elevated in ;GCre+ oocytes (fig. S4B). Expression of p27, which is usually under the control of the PI3K/Akt pathway (S14), was unaltered in ;GCre+ oocytes (fig. S4B). On the other hand, in ;GCre+ oocytes, the expression and phosphorylation levels of another Akt substrate, Foxo3a, which is a transcription factor that mediates cell cycle arrest and apoptosis in other cell types (S15), were upregulated (fig. S5A). The Foxo3a in ;GCre+ oocytes, however, could not be phosphorylated further by treatment with KL, as was the case in oocytes (fig. S5B). As phosphorylation of Foxo3a by Akt indicates functional suppression (S15), our data imply that the Foxo3a molecules in ;GCre+ oocytes may be at least partially suppressed by the overactivated Akt (Fig. 4A) due to the loss of Pten. Based on our previous report that overexpression of Foxo3a in oocytes suppresses oocyte growth and follicular development (S4), and the report that conventional total knockout of Foxo3a leads to excessive activation of primordial follicles (S16), we presume that activation of the entire primordial follicle pool in ;GCre+ ovaries was partially accomplished by suppression of Foxo3a function in oocytes. This hypothesis is 5

6 further supported by our finding that in double-mutant mice lacking both Pten in oocytes and Foxo3a overall (the ;GCre+ ; Foxo3a -/- mice), the rate of follicle activation is similar to that in ;GCre+ ovaries, which showed no signs of synergistically enhanced follicle activation (fig. S5C). Sizes of ovulated oocytes in ;GCre+ and mice The average sizes of ovulated oocytes in ;GCre+ and mice were ± 0.54 μm (SEM, n = 40), and ± 0.36 μm (SEM, n = 32), respectively, which were not significantly different (P = 0.33). Also, based on the finding that ;GCre+ females who became pregnant and gave birth all gave litters of normal size, we believe that oocytes ovulated by the mutant mice before follicle depletion are normal. Reduced follicle death/clearance before and around the time of sexual maturity in ;GCre+ mice In mice, during the initial wave of postnatal follicular development, large numbers of follicles disappear from the non-growing follicle pool before the onset of sexual maturity, as a result of follicle atresia (S17, S18). This atresia has been proposed to be mainly initiated by the death of the oocytes (S19). In this study, we found that although the initial numbers of follicles were similar in ;GCre+ and ovaries at PD5 and PD8 (fig. S3F), the total numbers of follicles at PD23, PD35, and week 7 were significantly higher (P < 0.05) in ;GCre+ ovaries than in ovaries (fig. S3F), indicating that the follicle clearance before and around the time of sexual maturity has been reduced to some extent in ;GCre+ ovaries. At week 7, 62.6% of the follicles in ovaries were still at the primordial stage, while follicles in ;GCre+ ovaries were mostly accumulated at the transient and preantral (type 5) stages (fig. S3E and fig. S2H). The reduced follicle death in ;GCre+ ovaries is most likely caused by the activation of all primordial follicles into growing phase, as a result of loss of oocyte Pten. These accumulated follicles, however, were all depleted by weeks 12 16, causing POF in the ;GCre+ mice (fig. S2K and Fig. 2H). Moderately elevated testosterone levels in 12- to 20-week-old ;GCre+ mice In 12- to 20-week-old ;GCre+ female mice that had started to develop POF, the serum testosterone levels were found to be mildly but significantly (P = ) elevated (fig. S6). It is known, however, that in women with POF, serum testosterone levels are actually lower than in fertile women (S20). We presume that the higher levels of testosterone in ;GCre+ mice are most likely produced by the theca cells that remain in the follicle-depleted ovaries. At weeks of age, which correspond to the early stages of POF in the mice, it is possible that the elevated levels of LH (Fig. 3B) would stimulate the residual theca cells to produce higher levels of testosterone. Further work on human POF will help to clarify whether serum testosterone levels are higher in the early phase of POF. On the other hand, it is also possible that the different testosterone profiles in human and mouse POF are species-related. 6

7 Supporting Figures S1-S6 Fig. S1 A P1 loxp Pten exon 5 loxp P3 allele Gdf-9-Cre loxp loxp loxp Pten exon 5 P1 P3 Pten 5 allele in oocytes allele in other parts of the body B Pten Δ5 Pten +/Δ5 Pup #1 #2 #3 #4 #5 #6 Pten loxp/+ Wild-type 500 bp C ;GCre+ D ;GCre+ PTEN PTEN β-actin β-actin Oocytes Oocytes ( > 25 μm)

8 Fig. S1. Oocyte-specific deletion of Pten in mice. (A) Schematic representation of deletion of Pten exon 5 and creation of a Pten Δ5 allele by Gdf- 9-Cre-mediated recombination in oocytes. P1 and P3 indicate primers for genotyping of deletions of Pten exon 5. (B) PCR-genotyping of 6 offspring of ;GCre+ mothers that were mated with wild-type C57BL/6J male mice, using primers P1 and P3 as illustrated in (A), where deletion of Pten exon 5 in one allele of the genomic DNA was seen in all pups, as indicated by the PCR bands of approximately 500 bp. (C D) Western blots showing the deletion of Pten from mouse oocytes. Oocytes were isolated from ovaries of 14-day-old ;GCre+ and mice and used for western blot, as described in Materials and Methods. In (D), to remove contaminating ovarian cells in the oocyte preparation, the mixture of collagenase-digested ovarian tissues was filtered through a cell-dispersing screen with 25-µm opening, to obtain oocytes that were larger than 25 µm. The experiments were repeated 3 times. For each experiment in (C), material from 3 5 ;GCre+ mice or 6 10 mice was used. For each experiment in (D), material from ;GCre+ mice or mice was used. For each lane, 30 µg of protein was used.

9 Fig. S2 ;GCre+ A B C PD8 100 μm 100 μm 50 μm D 250 μm E F PD μm 50 μm G H I 7 weeks 250 μm 250 μm 50 μm J K L 12 weeks 250 μm 250 μm 50 μm Fig. S2. Ovarian morphologies of PD8, PD23, 7- and 12-week-old ;GCre+ and mice. Ovaries from PD8, PD23, 7- and 12-week-old ;GCre+ mice and littermates were embedded in paraffin, and sections of 8 µm in thickness were prepared and stained with hematoxylin. Note that the magnification of the inset in panel D (with a red arrow, showing primordial follicles) is the same as that of panels F. The experiments were repeated more than 3 times, and for each time and each age, ovaries from one mouse of each genotype was used.

10 Fig. S3 No. of follicles per ovary % ns Pri 81.0% A. PD5 (n=5) ;GCre+ (n=6) ns Act % 0% D. PD35 (n=8) ;GCre+ (n=7) Pri Trans T3b T4 T5 T6 Act No. of follicles per ovary % 49.6% B. PD8 (n=6) ;GCre+ (n=5) Pri Trans T3b T4 T5 Act * * % 0% E. Week 7 (n=3) ;GCre+ (n=3) Pri Trans T3b T4 T5 T6 Act * No. of follicles per ovary % 0% C. PD23 (n=5) ;GCre+ (n=10) Pri Trans T3b T4 T5 T6 Act * F. Total no. of follicles per ovary ;GCre+ a a a a a c b e d d PD5 PD8 PD23 PD35 Week 7

11 Fig. S3. Quantification of ovarian follicles in ;GCre+ and mice. (A E) Ovaries from PD5 (A), PD8 (B), PD23 (C), PD35 (D), and week 7 (E) ;GCre+ and mice were embedded in paraffin, and serial sections of 8 µm in thickness were prepared and stained with hematoxylin. Numbers of different types of follicles per ovary (mean ± SEM), including numbers of primordial (Pri), transient (Trans), type 3b (T3b), type 4 (T4), type 5 (T5), and type 6 (T6) follicles, were counted. The percentages of primordial follicles for each age and genotype are superimposed on bars representing the numbers of primordial follicles. The numbers of activated follicles (Act) (mean ± SEM) are also presented. The numbers of mice used for each genotype and each age are indicated in the figures. One ovary from each mouse was used. *P < 0.01, P < 0.001, ns, not statistically significant. (F) Total numbers of follicles in ovaries of PD5, PD8, PD23, PD35, and week 7 ;GCre+ and mice (mean ± SEM). The follicle numbers at PD5 were less, but not statistically significantly so (P > 0.05), than those at PD8 in both genotypes, which is in accordance with a previous report that not all oocytes are enclosed in primordial follicles yet at this stage (S21). Different lowercase letters (a, b, c, d, and e) indicate groups that are significantly different (P < 0.05).

12 Fig. S4 A p-rps6 (S235/6) rps6 p-s6k (Thr 389) S6K p-akt (S473) Akt ;GCre+ B ; GCre+ p-gsk-3α/β (Ser 21/9) GSK-3α p-p42/44-mapk (Thr 202/Tyr 204) p-p42/44-mapk (Thr 202/Tyr 204) p42/44-mapk β-actin p42/44-mapk p27 β-actin C LY Rap PD PD12-14 oocytes PD12-14 oocytes Fig. S4. Signaling studies in ;GCre+ oocytes. Oocytes were isolated from ovaries of PD12 14 ;GCre+ and mice as described in Materials and Methods, and western blots were performed. (A) Phosphorylation of rps6 in cultured ;GCre+ oocytes was dependent on activities of PI3K and mtor. Treatment of oocytes with the PI3K-specific inhibitor LY (LY, 50 μm) and the mtorspecific inhibitor rapamycin (Rap, 50 nm) for 1 h largely suppressed levels of p-rps6 (serine 235/6) in ;GCre+ oocytes. As a control, treatment of oocytes with the MEK1-specific inhibitor PD98059 (PD, 50 μm) substantially suppressed the level of p-p42/44 MAPK (threonine 202/tyrosine 204), but did not lead to suppression of the level of p-rps6 (serine 235/6) in ;GCre+ oocytes. Levels of p-s6k (threonine 389) and p-akt (serine 473) under the treatment of the inhibitors are also shown to confirm the effectiveness of rapamycin and LY Levels of rps6, S6K, Akt, p42/44 MAPK, and β-actin were used as internal controls for loading of equal amounts of protein. (B) Western blots for p-gsk-3 α/β (serine 21/9), p-p42/44 MPAK (threonine 202/tyrosine 204), and p27. Levels of GSK-3α, p42/44 MPAK, and β-actin were used as internal controls to show loading of equal amounts of protein. All experiments were repeated at least 3 times. For each experiment, material from 3 5 ;GCre+ mice or 6 10 mice was used per lane. In each lane, μg of protein samples were loaded. Representative images are shown.

13 Fig. S5 A GCre+ B GCre+ Foxo3a p-foxo3a (Thr 32) p-foxo3a (Thr 32) β-actin Foxo3a β-actin KL C PD13 50 μm 50 μm 50 μm ;GCre+ ;GCre+; Foxo3a -/- Foxo3a -/- Fig. S5. Foxo3a expression and function in ;GCre+ oocytes. Oocytes were isolated from ovaries of 12- to 14-day-old and ;GCre+ mice, starved for 4 h in serum-free medium, and lysed directly for western blots, or treated with KL (100 ng/ml) for 10 min as described in Materials and Methods. (A) In oocytes of ;GCre+ mice, Foxo3a expression was elevated, which may be due to counterbalancing of the overactivated PI3K/Akt signaling. The level of phosphorylated Foxo3a (p-foxo3a, threonine 32) was also elevated in the ;GCre+ oocytes. The level of β-actin served as an internal control for loading of equal amounts of protein. (B) KL treatment did not lead to further phosphorylation of Foxo3a (threonine 32) in starved ;GCre+ oocytes, as was the case in oocytes. Levels of Foxo3a and β-actin served as internal controls. For (A) and (B), all experiments were repeated at least 3 times. For each experiment, material from 3 5 ;GCre+ mice or 6 10 mice was used per lane. In each lane, μg of protein samples were loaded. Representative images are shown. (C) To investigate whether suppression of Foxo3a in ;GCre+ oocytes was one of the causes of the excessive follicular activation, doublemutant mice carrying both oocyte-specific loss of Pten and overall loss of Foxo3a ( ;GCre+; Foxo3a -/- ) were generated. As compared to the rate of follicular activation in PD13 ;GCre+ ovaries, concurrent loss of Pten and Foxo3a in oocytes did not lead to synergistically enhanced follicular activation.

14 Fig. S6 Serum testosterone levels (ng/ml) P= Mutant Control Fig. S6. Elevated levels of testosterone in adult ;GCre+ mice. Sera from 12- to 20-week-old ;GCre+ (Mutant) and (Control) mice were collected for measurement of testosterone levels. Serum samples from 19 ;GCre+ mice and 18 mice were used. P =

15 Supporting References S1. M. Groszer et al., Science 294, 2186 (2001). S2. Z. J. Lan, X. Xu, A. J. Cooney, Biol. Reprod. 71, 1469 (2004). S3. L. Lin, J. D. Hron, S. L. Peng, Immunity. 21, 203 (2004). S4. L. Liu et al., Development 134, 199 (2007). S5. T. Pedersen, H. Peters, J. Reprod. Fertil. 17, 555 (1968). S6. J. Johnson, J. Canning, T. Kaneko, J. K. Pru, J. L. Tilly, Nature 428, 145 (2004). S7. J. I. van Casteren, W. G. Schoonen, H. J. Kloosterboer, Biol. Reprod. 62, 886 (2000). S8. A. M. Haavisto et al., Endocrinology 132, 1687 (1993). S9. I. Huhtaniemi, H. Nikula, S. Rannikko, J. Clin. Endocrinol. Metab 61, 698 (1985). S10. K. Liu et al., Dev. Biol. 295, 615 (2006). S11. D. D. Sarbassov, D. A. Guertin, S. M. Ali, D. M. Sabatini, Science 307, 1098 (2005). S12. D. A. Guertin et al., Dev. Cell 11, 859 (2006). S13. D. A. Cross, D. R. Alessi, P. Cohen, M. Andjelkovich, B. A. Hemmings, Nature 378, 785 (1995). S14. V. Chandramohan, S. Jeay, S. Pianetti, G. E. Sonenshein, J. Immunol. 172, 5522 (2004). S15. D. Accili, K. C. Arden, Cell 117, 421 (2004). S16. D. H. Castrillon, L. Miao, R. Kollipara, J. W. Horner, R. A. DePinho, Science 301, 215 (2003). S17. M. J. Faddy, E. Telfer, R. G. Gosden, Cell Tissue Kinet. 20, 551 (1987). S18. S. K. Bristol-Gould et al., Dev. Biol. 298, 149 (2006). S19. Y. Morita, J. L. Tilly, Dev. Biol. 213, 1 (1999). S20. S. N. Kalantaridou et al., Fertil. Steril. 86, 1475 (2006). S21. M. E. Pepling, A. C. Spradling, Dev. Biol. 234, 339 (2001).