Complete in vitro generation of fertile oocytes frommouse primordial germ cellsKanako Morohakua, Ren Tanimotoa, Keisuke Sasakia, Ryouka Kawahara-Mikib, Tomohiro Konoa, Katsuhiko Hayashic,d,Yuji Hiraoe,1, and Yayoi Obataa,1

aDepartment of Bioscience, Tokyo University of Agriculture, Setagaya-ku, Tokyo 156-8502, Japan; bNODAI Genome Research Center, Tokyo University ofAgriculture, Setagaya-ku, Tokyo 156-8502, Japan; cDepartment of Developmental Stem Cell Biology, Faculty of Medical Sciences, Kyushu University,Higashi-ku, Fukuoka 812-8582, Japan; dJapan Science and Technology Agency (JST), Precursory Research for Embryonic Science and Technology (PRESTO),Higashi-ku, Fukuoka 812-8582, Japan; and eDivision of Animal Breeding and Reproduction Research, Institute of Livestock and Grassland Science, NationalAgriculture and Food Research Organization (NARO), Tsukuba, Ibaraki 305-0901, Japan

Edited by John J. Eppig, The Jackson Laboratory, Bar Harbor, ME, and approved June 6, 2016 (received for review March 7, 2016)

Reconstituting gametogenesis in vitro is a key goal for reproductivebiology and regenerativemedicine. Successful in vitro reconstitutionof primordial germ cells and spermatogenesis has recently had asignificant effect in the field. However, recapitulation of oogenesisin vitro remains unachieved. Here we demonstrate the first re-constitution, to our knowledge, of the entire process of mammalianoogenesis in vitro from primordial germ cells, using an estrogen-receptor antagonist that promotes normal follicle formation, whichin turn is crucial for supporting oocyte growth. The fundamentalevents in oogenesis (i.e., meiosis, oocyte growth, and genomicimprinting) were reproduced in the culture system. The mostrigorous evidence of the recapitulation of oogenesis was the birthof fertile offspring, with a maximum of seven pups obtained from acultured gonad. Moreover, cryopreserved gonads yielded functionaloocytes and offspring in this culture system. Thus, our in vitrosystem will enable both innovative approaches for a deeper under-standing of oogenesis and a new avenue to create and preservefemale germ cells.

oogenesis | primordial germ cells | follicle formation | oocytes | in vitro

Oocytes contain fundamental materials for perpetuating a spe-cies; that is, the maternal genome and ooplasm filled with

maternal factors that are essential for totipotency, including mito-chondria, which are transmitted to the next generation. Therefore,the process of oogenesis is of wide interest in reproductive biologyand regenerative medicine. However, mechanisms underlying theprocess are not fully elucidated. An approach to comprehensivelyresolve the mechanisms of oogenesis is to reconstitute the entireprocess of oogenesis in vitro (1, 2).In mice, primordial germ cells (PGCs) mitotically divide until

13.5 d postcoitum (dpc) in the fetal gonads, after which they imme-diately enter meiosis. Passage through prophase of the first meioticdivision and preparation for primordial follicle formation, includingthe breakdown of oocyte cysts, occurs during the remainder of theprenatal period. Shortly after birth, a cohort of primordial folliclesenters the growth phase, and after 3 wk, oocyte growth culminates inthe acquisition of competencies to resume meiosis, complete mater-nal imprinting, undergo fertilization, and support full-term develop-ment. Applications of in vitro systems to study the events involved inoogenesis have been widely used, although none has been successfulin reconstituting the entire process. To date, most successful recon-stitutions of oogenesis (but not from the onset) with proven fertilityhave used neonatal oocytes, which were already in the prophase ofthe first meiosis and were ready to be assembled into primordialfollicles (3–6).Thus far, the ectopic oogenesis from PGCs to fertile mature

oocytes has been achieved only by means of grafting PGCs intoother mice to facilitate key steps (7, 8). Even PGC-like cells,originally produced from mouse embryonic stem and inducedpluripotent stem cells, can develop into functional oocytes afterreaggregation with gonadal somatic cells and grafting beneaththe bursa (9, 10). However, many previous studies attempting in

vitro oogenesis without the help of grafting have documented theoccurrence of aberrant follicular formation and insufficient growthand ability of the oocytes after in vitro growth (IVG) (1, 2, 11).Thus, the crucial step remaining to create a connected “chain” ofin vitro oogenesis is to eliminate the gap between what can beachieved using grafting and what has become possible by culture(i.e., IVG of oocytes from the primordial follicle stage) (3–5).Three events need to be achieved to successfully reconstitute

oogenesis in vitro: the initial phase of meiosis, follicular assembly,and appropriate conditions to support sufficient oocyte growth andcomplete maturation. Elucidating the mechanisms of oogenesis isimportant, and if the above three requirements are fulfilled, thenin vitro oogenesis can enable mass generation of oocytes.In this study, we reconstituted the entire process of oogenesis

across meiosis and complete maturation in vitro. Meiosis, follic-ular formation, oocyte growth, and maternal imprinting, all ofwhich are required for the development of a functional oocyte,were found to be accomplished without apparent abnormality.

ResultsMeiotic Entry and Follicular Assembly in Vitro. In our initial experi-ments, female gonads obtained at 12.5 dpc were cultured onTranswell-COLmembranes, using the α-MEM supplemented with10% (vol/vol) FBS, as previously described (3, 11). On the basis ofthe meiotic marker synaptonemal complex protein 3 (SYCP3)(12), entry into meiosis occurred during the first 5 d of culture,irrespective of the absence of the mesonephros (Fig. S1A), whichcontrasts with previous data that suggested the involvement ofmesonephroi in the entry into meiosis (13). On day 17 of culture,corresponding to an age of 10 d in vivo (Fig. 1A), the ovary

Significance

Throughout the life of female mammals, only a small number ofviable oocytes are produced. The mechanisms underlying thecreation and selection of competent oocytes remain unclear.Here, we propose a novel approach for elucidating these un-solved questions, involving the use of an in vitro system estab-lished in the present study, which can fully reproducemammalianoogenesis from mouse fetal primordial germ cells. Reconstitutionof the entire oogenesis process has not been previously accom-plished. Our system will assist in understanding the mechanismsof oogenesis and also create a new gamete resource in mammals.

Author contributions: K.H., Y.H., and Y.O. designed research; K.M., R.T., K.S., Y.H., and Y.O.performed research; R.K.-M. and T.K. contributed new reagents/analytic tools; K.M., R.T.,and K.S. analyzed data; and K.M., K.H., Y.H., and Y.O. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.1To whom correspondence may be addressed. Email: yujih@affrc.go.jp or y1obata@nodai.ac.jp.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1603817113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1603817113 PNAS | August 9, 2016 | vol. 113 | no. 32 | 9021–9026

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

contained many growing oocytes; however, most were contained inhypoplastic follicles with abnormal and unclear external layers (Fig.S1 B and D). The oocytes in these follicles were easily denudedduring follicle isolation attempts and were no longer capable ofgrowth in culture, thereby indicating severe defects in follicular as-sembly. Despite the presence of >100 growing oocytes per ovary,only four to six secondary follicles were successfully isolated (Fig. 1 Band E), which is consistent with previous studies (1, 11). An analysisof laminin-immunostained images (14) of the follicle basementmembrane indicated that the hypoplastic follicles had incompletelaminin envelopes, unlike normal in vivo follicles, which develop acomplete envelope (Fig. S1B). Some neighboring in vitro folliclesshared the same theca or granulosa cell layer, and multioocyte fol-licles were occasionally detected (Fig. 1B, Fig. S1B, and Movie S1).

Cause of Hypoplastic Follicle Formation in Vitro. We hypothesized thatthis abnormally structured secondary follicle has its root in the earlystage of primordial follicle assembly; if so, this could be caused by a

mechanism that involves estrogen and governs the oocyte cystbreakdown and primordial follicular formation (15, 16).To explore the cause of an abnormal follicular state and test this

hypothesis, RNA sequencing (RNA-seq) analysis was performedusing ovaries cultured in α-MEM + FBS on day 7 and ovariesfrom mice at 0 d postpartum (dpp) (Fig. S1E). In the RNA-seqanalysis, the global gene expression pattern that was observed forin vitro-derived ovaries was highly correlated (R = 0.992) (Fig.S2A) with that of neonatal ovaries, indicating that developmentalprogression in culture was similar to development in vivo. How-ever, 547 genes were differentially expressed between the in vitro-cultured ovaries and neonatal ovaries (normalized signal value,>5; P < 0.05; greater than threefold change) (Table S1). Ingenuitypathway analysis clearly demonstrated that β-estradiol, trans-actingtranscription factor 1 (SP1), and β-catenin (CTNNB1), which bindto estrogen receptors and regulate gene expression (17), consti-tuted the top three predicted upstream regulators of the 547 genes(Fig. 1C and Table S2). These results indicated that an elevated

A B

C D E

F

Fig. 1. Refinement of follicular (fol) assembly in cultured gonads. (A) Schematic representation of oogenesis and crucial events that require reconstitutionin vitro. (B) Histological section (Left) and isolated follicles (Right) from the ovary cultured for 17 d in α-MEM + FBS or the 10-dpp ovary. (Scale bars, 50 μm.)(C) Predicted upstream transcriptional regulators of 547 DEGs in ovaries cultured with α-MEM + FBS. (D) Histological section (Left) and isolated follicles (Right)from the ovary cultured for 17 d with α-MEM + FBS/10 μM ICI. (Scale bars, 50 μm.) (E) Mean number of secondary follicles successfully isolated from each ovaryunder various culture conditions. Significant differences (*P < 1.0E−3) in the number of isolated secondary follicles were determined by the Bonferroni multiple-comparison correction. Error bars indicate SDs. Numbers in parentheses indicate the number of ovaries examined. (F) A scatter plot showing the ratio of oocyte/follicle diameters from ovaries cultured with α-MEM + FBS/10 μM ICI (red) and those of in vivo origin (blue). Regression lines of in vitro-derived (solid) and in vivo-derived (dashed) ovaries are shown with the slope, intercept, and correlation coefficient. The slopes of the regressions were not significantly different.

9022 | www.pnas.org/cgi/doi/10.1073/pnas.1603817113 Morohaku et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

estrogen signaling level, probably triggered by some ligands inFBS, caused hypoplastic follicle formation in vitro.

Overcoming Hypoplastic Follicle Formation in Vitro. The potentialeffects of FBS-derived estrogen in the culture medium were mini-mized as follows (Fig. S1E): A serum protein substitute (SPS;44 mg/mL normal human serum albumin and 6 mg/mL α- andβ-globulins in PBS) was used instead of FBS (α-MEM + SPS);α-MEM + SPS was used only from day 5 to day 11, when oocyte cystbreakdown and primordial follicle assembly occurs (6) (α-MEM +FBS/SPS); and an estrogen-receptor antagonist (ICI 182,780; ICI)was added from day 5 to day 11 (α-MEM + FBS/ICI). These testconditions significantly improved the efficiency of the isolation ofsingle secondary follicles, each of which encapsulated a single primaryoocyte (Fig. 1D and Fig. S1 C and D). In particular, the addition ofICI resulted in a more than sevenfold increase in the number ofsingle secondary follicles compared with the original conditions (Fig.1E) (P < 1 × 10−3). Furthermore, ICI treatment promoted the for-mation of a complete laminin envelope by individual follicles (Fig.S1C). Although oocytes in the secondary follicles that were culturedin vitro were smaller than those growing in vivo, the ratios of oocyte/follicle diameters were comparable (Fig. 1F), suggesting the folliclesassembled normally in vitro. On the basis of an RNA-seq analysis ofovaries that were cultured for 7 d in α-MEM + FBS/10 μM ICI, 421genes were differentially expressed between ovaries cultured withα-MEM + FBS/ICI and neonatal ovaries (Table S3), although theglobal pattern of gene expression was highly correlated with that of

neonatal ovaries (Fig. S2A). Ingenuity pathway analysis demonstratedthat β-estradiol and SP1 were not on the top 10 list of predicted up-stream regulators of the 421 genes (Fig. S2B and Table S4). Potentialdownstream target genes responsible for hypoplastic follicle forma-tion were identified among 213 genes whose transcript levels differedspecifically in α-MEM + FBS-cultured ovaries (Fig. S2C). Finally,supplementing the culture medium with estradiol impaired follicleformation; therefore, the efficiency of secondary follicle isolation wasreduced (Fig. 1E and Fig. S1D). Collectively, this series of analysesclearly demonstrated that the appropriate regulation of estrogenbinding to its receptors is crucial for initial in vitro folliculogenesis.

Oocyte and Follicular Growth in Vitro. Secondary follicles were isolatedon day 17 of culture because a prolonged culture of whole ovarieseventually results in large-scale follicular degeneration (Fig. S3A).The isolated follicles were cultured in α-MEM supplemented with5% (vol/vol) FBS, 2% (wt/vol) polyvinylpyrrolidone (PVP; 360 kDa)(18), and 0.1 IU/mL follicle-stimulating hormone for IVG. The PVP-supplemented medium was used because of its effect on maintainingthe integrity of oocyte–granulosa cell complexes in a long-term cul-ture, which was found in our previous study in cattle (19). Wedemonstrated here that without PVP supplementation, even in theearly phase of IVG (day 20), the ratio of follicle/oocyte diameters wassignificantly smaller than that cultured in medium supplemented with2% (wt/vol) PVP (1.82 vs. 1.95; P = 0.0049) (Fig. S3B). Moreover, themRNA expression levels of born morphogenetic protein 6 (Bmp6),Bmp15, kit ligand (Kitl), and kit oncogene (Kit), which are involved in

A

BD

C

Fig. 2. In vitro growth of oocytes and follicles. (A) A representative follicle cultured on a Transwell-COL membrane on days 22–29 of culture. (Scale bars,100 μm.) (B) Concentration of progesterone (Left) and estradiol (Right) in the medium. Steroid hormones were measured by an enzyme-linked immunoassayduring follicular culture. Significant increases in the steroid concentration in the medium were determined by t tests. Asterisks indicate a significant increasein the steroid concentration compared with that measured 2 d before (*P < 0.05, **P < 0.01). Error bars indicate SD (n = 4). (C) Oocyte growth in IVG culture.The box plot shows the diameters of oocytes on the days that IVG started (day 17) and ended (day 29–33), the diameters of oocytes in secondary follicles at 10dpp, and the diameter of mature oocytes derived from adult mice. (D) DNA methylation imprints in the in vitro-derived GV oocytes at the Igf2r and H19 loci,which are methylated specifically in oogenesis and spermatogenesis, respectively. “#1” and “#2” represent two independent samples. Black and white circlesindicate methylated and nonmethylated cytosines at CpG sites in the analyzed imprinted regions, respectively.

Morohaku et al. PNAS | August 9, 2016 | vol. 113 | no. 32 | 9023

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

follicle growth (20), were slightly reduced in follicles cultured in thePVP-free medium (Fig. S3C). After a full term of IVG, a significantlylarger proportion of oocyte–granulosa cell complexes were recoveredfrom the medium with PVP than from the medium without PVPsupplementation (Fig. S3D andE) (with PVP: 344/519; without PVP:98/524; P = 0.00488).Previous reports have described two possible IVG culture

methods, wherein the intact follicle structure (i.e., oocytes, gran-ulosa cells, the basal lamina, and thecal layers) is maintained or theoocyte–granulosa cell complex is directly exposed to the medium.The latter method requires collagenase treatment for digestion ofthe outer layer of follicles. In preliminary experiments in which IVGwas conducted without collagenase treatment, 83% of isolatedfollicles survived (116/140). However, only a few oocytes reachedmetaphase during the second meiosis [MII; 26/116; MII oocytes/cumulus cells-oocyte complexes (COCs)], supposedly because of theinsufficient growth of oocytes/follicles, none of which were normallyfertilized. Therefore, we treated the follicles with 0.1% collagenasefor at least 15 min and largely removed the thecal layers by pipettingon day 20 of culture. During IVG culture with collagenase treat-ment, granulosa cell proliferation, oocytes, and clearly outlinedgerminal vesicles (GV) with a characteristic nucleolus were ob-served in most of the follicular structures (Fig. 2A and Table 1).Mural granulosa cells often formed dorm-like structures. Immu-nostaining of hydroxy-delta-5-steroid dehydrogenase, 3 beta- andsteroid delta-isomerase (HSD3B), which is essential for progester-one production (21, 22), showed the presence of a population ofsteroidogenic cells on day 26 of culture (Fig. S3F). Continuous in-creases in the progesterone and estradiol concentrations in themedium suggested persistent steroidogenic activity in the follicles(Fig. 2B). Steroidogenic markers, such as Hsd3b1, cytochromeP450, family 17, subfamily a, polypeptide 1 (Cyp17a1) and Cyp19a1mRNA, were also expressed in the follicles on day 26 of culture(Fig. S3G). On days 29–33, COCs were collected from approxi-mately half of the cultured follicles (Table 1). During the IVGculture, the oocytes increased in size to a mean diameter of 80.0 μm(n = 85) (Fig. 2C). Because the average diameter before IVG was54.4 μm (n = 203) (Figs. 1F and 2C), the oocyte volume increasedby 3.18-fold. A control experiment showed that the average diameterof fully grown oocytes collected from superovulated mice was89.9 μm (n = 74) (Fig. 2C), suggesting that the volume increasedby 3.98-fold in vivo compared with oocytes in secondary follicles of10-dpp ovaries, whose average diameter was 56.6 μm (n = 175).Although the size and growth rate of oocytes were slightly atten-uated in the in vitro culture, oocyte-specific methylation imprints,which are necessary for mammalian ontogeny (23), were never-theless established in the oocytes grown in vitro (Fig. 2D).

Evidence for Recapitulation of Oogenesis in Vitro. The most rigorousevidence for recapitulation of oogenesis in vitro originates from thedemonstration of the reproductive ability of oocytes. To evaluatethis, the oocytes grown in this system were subjected to in vitromaturation, followed by in vitro fertilization. The collected COCswere induced to resume meiosis, using gonadotropins and epidermal

growth factor (24). The majority of oocytes released the first polarbody after 17 h and became MII oocytes, as confirmed by a karyo-type analysis (Fig. 3A). Approximately half (39–58%) of the MIIoocytes were normally fertilized (Fig. 3A and Table 1), and 83–97%of the eggs developed to the two-cell stage on the next day and weretransferred into 0.5-dpc pseudopregnant ICR albino mice (Fig. 3A).These transplantation experiments resulted in the birth of healthypups with comparable body weights to those derived from oocytesin vivo (Fig. 3 B and C and Table 1). All the pups had pigmentedeyes, and hence were considered to have clearly originated fromin vitro-derived oocytes. The frequency of development from two-cell-stage embryos to pups was 14–40% (Table 1). A maximum ofseven pups was obtained from a single gonad (Table S5), which iscomparable with the number seen by natural delivery in mice. Onaverage, 0.7–3.3 pups were obtained from each gonad, and theyshowed normal appearance and fertility (Fig. S4).

Cryopreservation of Ovary: Approach to Applicable System. Cryopres-ervation of ovarian tissue is an essential technology for preservinganimal fertility and gamete resources. Accordingly, the use of ourin vitro system to produce functional oocytes from cryopreservedtissues is an important criterion to evaluate the application of theculture system. We dissected each of the fetal gonads into two orthree sections, after which they were cryopreserved by vitrification(25). After placing the gonads in liquid nitrogen for at least 3 d, thegonads were rewarmed and cultured under the same conditions(α-MEM + FBS/1 μM ICI) that were used for fresh gonads. Al-though a decreased number of follicles formed in the thawed go-nads, apparently normal offspring were nevertheless derived fromthe resultant oocytes (Table 1). These results demonstrated thefeasibility of cryopreserving gonads and subsequent functional oocyteproduction that are expected to serve as an alternative source ofoocytes in the future.

DiscussionHere, we demonstrated the first completely connected chainof in vitro oogenesis that permits full differentiation of PGCs intomature mouse oocytes. Breaks in the developmental “chain” havebeen bypassed previously through in vivo conditions by grafting PGCsbeneath the renal capsules (7, 8) or the ovarian bursa (9). We used adifferent kind of bypass, involving whole cytoplasm prepared fromin vivo-grown oocytes, which was combined with genomic materialsfrom in vitro-grown oocytes cultured from PGCs, using the nucleartransfer technique (11, 26). All such bypasses demand additionalanimals and complex procedures. More important, the existence of aperiod during which oocytes are not observable and controllablediminishes the potential of the experimental scheme involving oocyteculture as the model of oogenesis. Thus, the use of our in vitro systemwithout breaking the “chain” can add convenience in conductingexperiments and open up new possibilities, such as enabling obser-vation and manipulation throughout the course of oogenesis.A crucial factor underlying the successful establishment of our

in vitro system stemmed from an observation of hypoplastic follicleformation in the ovaries cultured in α-MEM + FBS (Fig. 1B). The

Table 1. Developmental ability of oocytes differentiated from PGCs after growth, maturation, and fertilization in vitro

Conditionfor organculture

No. ofculturedgonads

No. ofculturedfollicles

No. ofsurvivedfollicles

No. ofcollectedCOCs

No. ofoocytes matured

into MII

No. of eggsfertilizednormally

No of embryosdevelopedto two-cell

No. of embryosdevelopedto pup

No. of pupsfrom

a gonad

FBS/SPS 9 216 96 (44%) 60 (63%) 46 (77%) 18 (39%) 15 (83%) 6 (40%) 0.7FBS/1 μM ICI 16 641 470 (73%) 340 (72%) 321 (94%) 168 (52%) 155 (92%) 31 (20%) 1.6FBS/5 μM ICI 6 312 223 (71%) 178 (80%) 168 (94%) 94 (56%) 91 (97%) 20 (22%) 3.3FBS/10 μM ICI 6 505 337 (67%) 281 (83%) 256 (91%) 148 (58%) 138 (93%) 19 (14%) 3.2Vitrification 4 87 57 (66%) 39 (68%) 37 (95%) 19 (51%) 15 (79%) 2 (13%) 0.5GV (control) 19 155 139 (90%) 70 (45%) 67 (96%) 40 (60%) 2.1

9024 | www.pnas.org/cgi/doi/10.1073/pnas.1603817113 Morohaku et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

addition of FBS caused a delay or failure of oocyte cyst breakdownin the ovary, as suggested by the typical occurrence of multioocytefollicle-like structures. Recently, follicular assembly in newbornmouse ovaries was shown to be correlated with the loss of maternally,fetal, and/or ovary-derived estrogen (27, 28). A markedly reducedconcentration of estrogen and/or progesterone is essential to promoteoocyte cyst breakdown and follicle formation in mice and rats (15,29). This mechanism was apparently dominant in the case of ovariescultured in the present study, despite the absence of maternal andfetal estrogen after 5 d in culture. FBS used in the medium is mostlikely the primary candidate of the source of estrogen. This as-sumption is consistent with the finding that no apparent inhibition offollicle assembly was observed when SPS was used in place of FBS.Moreover, the positive effect of the use of SPS disappeared on theaddition of estradiol (Fig. 1 and Fig. S1). However, further study isnecessary to determine whether the cultured fetal ovary itself cangenerate estrogen; for example, in response to the absence of ma-ternally derived estrogen. In fact, Fortune and Eppig have shown thatneonatal ovaries can produce small amounts of estrogen (30).It is of special importance that ovaries cultured in α-MEM + FBS

showed differential expression of genes regulated by estrogen (Fig.1C and Table S1). Estrogen receptors comprise estrogen receptor 1(alpha) (ESR1), estrogen receptor 2 (beta) (ESR2), and G protein-coupled estrogen receptor 1 (GPER1) (17). It is well known thatESRs are activated by estrogen binding to consensus DNA se-quences and that they regulate the expression of various genes (17).ICI, an antagonist of ESR1 and ESR2, caused dramatic changes ofdifferentially expressed genes (DEGs), which include genes re-sponsible for hypoplastic follicle formation (Fig. S2). These previousfindings and our current results strongly suggest the involvement ofinduced ectopic activation of estrogen receptors in hypoplastic fol-licle formation after the addition of FBS. Therefore, it is a rea-sonable conclusion that the cyst breakdown problem was avoided bythe addition of ICI.There are differences among mammalian species regarding the

course of follicle formation. For example, the number of primordialfollicles reaches a maximum value in the ovaries of 141- to 210-d-old bovine fetuses, when the secretion of estradiol from the ovariesis almost undetectable (31). In contrast, estradiol promotes pri-mordial follicle formation in nonhuman primates (baboons) and

hamsters (32, 33), but not by an excess of estradiol (33). The reasonfor such differences among species should be clarified to make ourin vitro system applicable in practice to animals other than mice.Normal follicle formation is important for oocyte production in

vitro as well as in vivo (34, 35). However, an additional and perhapsmore dominant reason existed in the present study: normal robustfollicles that developed after ICI treatment showed the maximumnumber of isolated secondary follicles from cultured ovaries. A morethan sevenfold increase in the efficiency of follicle recovery after ICIexposure demonstrated a tremendous effect on the design and scaleof experiments for the latter part of oogenesis in vitro.Although a key to success in the present study was the ability to

generate an increased yield of intact follicles until day 17, collage-nase treatment of the isolated follicles also played an important rolein acquiring oocyte competency. The developmental competence ofoocytes was elicited by the collagenase treatment, which disruptsintegral follicle structure and may promote oocyte development bysupporting the direct transfer of materials to the exposed oocyte–granulosa cell complexes. In contrast, potentially because of theremoval of a proportion of theca cells by collagenase, there was adelay in the marked acceleration of steroidogenesis until day 25 ofculture (Fig. 2B).Another modification of the culture medium was the supple-

mentation of PVP, which increased the recovery rate of bovineoocyte-granulosa cell complexes after 2 wk of IVG (18, 19). In ad-dition to its remarkable effect on maintaining the oocyte–granulosacell complexes in an integrated form and attached to the insertmembrane (Fig. S3D), PVP may positively affect the expression ofgenes involved in granulosa cell proliferation, such as Bmp6, Bmp15,Kit, and Kitl (Fig. S3C). It is also of interest that BMP6, a paracrineand autocrine factor expressed in granulosa cells and oocytes, canmodulate steroidogenesis and follicle development (20).Our in vitro system will be useful as a model for visualizing and

manipulating PGCs/oocytes throughout oogenesis and promotingthe development of functionally mature oocytes. In a recent studyby Pfender et al. (36), IVG of mouse oocytes from secondary fol-licles was combined with a live-cell imaging, using RNA interfer-ence screening to identify genes essential for meiosis. Our in vitrooocyte production system expands the applications of such ap-proaches to as early as the PGC stage and is especially effective for

1

2

34

5

67

8 910

11

1213

14

15

16

1718

1920

0 0.4 0.8 1.2 1.6 2.0Birth weight (g)

in vitro

in vivo

A B

C

12

34

56

7910

1112 13

141516

1718 19

20

8

Fig. 3. Developmental competence of oocytes produced from PGCs in vitro. (A) Maturation, resumption ofmeiosis, fertilization, and embryonic development in vitro.Shown are COCs obtained from follicles on day 29 after the beginning of organ culture (Upper Left); denuded fully grown oocytes at the GV stage after follicularculture (Upper Middle Left); the karyotype of in vitro-derived oocytes at the first (Upper Middle Right) and second (Upper Right) meiotic stage with the number ofchiasmata and 20 pairs of homologous chromosomes, respectively; cumulus cell-enclosed in vitro-derived oocytes at the MII stage (Lower Left); normally fertilized eggswith two pronuclei and the second polar body (Lower Middle Left), indicating the completion of meiosis after fertilization; and two-cell (Lower Middle Right) andblastocyst-stage (Lower Middle Right) embryos. (Scale bars, 100 μm.) (B) Pups from in vitro-derived oocytes. (C) Birth weights of the offspring. White and black barsindicate the mean body weights of offspring (1.53 ± 0.282 g vs. 1.63 ± 0.128 g) from in vitro-derived oocytes (n = 68) and in vivo-derived oocytes (n = 23), respectively.Error bars indicate SD. No significant difference in body weights between the groups was observed (t test).

Morohaku et al. PNAS | August 9, 2016 | vol. 113 | no. 32 | 9025

DEV

ELOPM

ENTA

LBIOLO

GY

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020

elucidating mechanisms regulating the early stage of meiosis. Thesame system is also applicable to PGC-like cells derived from em-bryonic stem and induced pluripotent stem cells (9, 10). Our systemwill assist in bridging the current technical gaps in the entire in vitrooocyte production process from stem cells.Spermatogenesis has been recently reconstituted in an organ cul-

ture of neonatal testicular tissue (37). Thus, complete reconstitutionof gametogenesis is now possible in both males and females.

Materials and MethodsExpanded methods are available in SI Materials and Methods. All the animalswere purchased from CLEA Japan. BDF1 mouse fetuses were collected fromC57BL/6N female mice crossed with DBA/2J male mice at 12.5 dpc and wereused for the culture experiments. The animals were maintained in accor-dance with the guidelines of the Science Council of Japan, and all experi-ments were approved by the Institutional Animal Care and Use Committeeof the Tokyo University of Agriculture.

Female fetal gonadswithoutmesonephros were cultured in Transwell-COLmembranes (Corning) for 17 d. α-MEM (Gibco, Thermo Fisher Scientific)supplemented with 1.5 mM 2-O-α-D glucopyranosyl-L-ascorbic acid (TokyoChemical Industry), 10 units/mL penicillin, and 10 μg/mL streptomycin(Sigma-Aldrich) was used as a basal medium (referred to here simply asα-MEM). FBS (Gibco, Thermo Fisher Scientific), SPS (SAGE In-Vitro Fertiliza-tion), β-estradiol (Santa Cruz Biotechnology), and the estrogen receptorantagonist ICI 182,780 (Tocris Bioscience) were added at the indicated con-centrations for each experiment. Gonads were cultured for 17 d at 37 °Cunder 5% CO2 and 95% air. Approximately half of the medium in each wellwas replaced with fresh medium every other day (3, 11, 15, 38).

To optimize the organ culture conditions, the following conditions wereevaluated (Fig. S1E): culture for the complete 17-d period in α-MEM supple-mented with 10% (vol/vol) FBS (α-MEM + FBS), culture for the complete 17-dperiod in α-MEM supplemented with 10% (vol/vol) SPS (α-MEM + SPS), culture inα-MEM + FBS with a shift to α-MEM + SPS from day 5 to day 11 (α-MEM + FBS/SPS), and culture in α-MEM + FBS with the addition of 1, 5, or 10 μM ICI fromday 5 to day 11 (α-MEM + FBS/1, 5, or 10 μM ICI).

After 17 d of organ culture, the secondary follicles were isolated from theovaries, using a Tungsten needle in L15 medium (Sigma-Aldrich). Follicles werefurther cultured in α-MEM supplemented with 2% (wt/vol) PVP (Sigma-Aldrich), 5% (vol/vol) FBS, and 0.1 IU/mL FSH (FOLLISTIM Injection 50; MSD).Follicles were cultured on a Millicell membrane (Merck Millipore) in a 35-mmculture dish (Falcon, Corning) for 12–16 d.

On day 20 of culture, follicles were treated with 0.1% collagenase type I(Worthington Biochemicals) in L15medium for 15min at 37 °C. Then, the thecalayer of the follicles was removed by pipetting in part. Follicles were culturedon Transwell-COL or Millicell membranes for another 9–13 d at 37 °C in me-dium under 5% CO2 and 95% air. Approximately half the medium in each wellwas replaced with fresh medium every other day (26).

ACKNOWLEDGMENTS. We thank Prof. Satoru Kobayashi (University ofTsukuba), Prof. Takehiko Ogawa (Yokohama City University), Ms. NaokoMochida, Dr. Akiko Hasegawa, and Prof. Hiroaki Shibahara (Hyogo Collegeof Medicine) for their comments; Prof. Ken-ichirou Morohashi (KyushuUniversity) for giving us the HSD3B antibody; and Dr. Kazuya Kobayashi(Hirosaki University) for preparing the histologic sections of the ovaries. Thiswork was supported by Grants-in-Aid for Scientific Research 26450449 (toY.O.), 25114008 (to Y.O. and Y.H.), and 25114006 (to K.H.); and by aMinistry ofEducation, Culture, Sports, Science, and Technology (MEXT)-Supported Pro-gram for the Strategic Research Foundation at Private Universities (S1311017).

1. Shen W, et al. (2007) In vitro development of mouse fetal germ cells into matureoocytes. Reproduction 134(2):223–231.

2. Zhang ZP, et al. (2012) Growth of mouse oocytes to maturity from premeiotic germcells in vitro. PLoS One 7(7):e41771.

3. Eppig JJ, O’Brien MJ (1996) Development in vitro of mouse oocytes from primordialfollicles. Biol Reprod 54(1):197–207.

4. O’Brien MJ, Pendola JK, Eppig JJ (2003) A revised protocol for in vitro development ofmouse oocytes from primordial follicles dramatically improves their developmentalcompetence. Biol Reprod 68(5):1682–1686.

5. Morohaku K, Hirao Y, Obata Y (2016) Developmental competence of oocytes grownin vitro: Has it peaked already? J Reprod Dev 62(1):1–5.

6. Pepling ME, Spradling AC (2001) Mouse ovarian germ cell cysts undergo programmedbreakdown to form primordial follicles. Dev Biol 234(2):339–351.

7. ShenW, et al. (2006) Live offspring produced by mouse oocytes derived from premeioticfetal germ cells. Biol Reprod 75(4):615–623.

8. Matoba S, Ogura A (2011) Generation of functional oocytes and spermatids from fetalprimordial germ cells after ectopic transplantation in adult mice. Biol Reprod 84(4):631–638.

9. Hayashi K, et al. (2012) Offspring from oocytes derived from in vitro primordial germcell-like cells in mice. Science 338(6109):971–975.

10. Hayashi K, Ohta H, Kurimoto K, Aramaki S, Saitou M (2011) Reconstitution of themouse germ cell specification pathway in culture by pluripotent stem cells. Cell146(4):519–532.

11. Obata Y, Kono T, Hatada I (2002) Gene silencing: Maturation of mouse fetal germcells in vitro. Nature 418(6897):497.

12. Lammers JH, et al. (1994) The gene encoding a major component of the lateral elements ofsynaptonemal complexes of the rat is related to X-linked lymphocyte-regulated genes.MolCell Biol 14(2):1137–1146.

13. Bowles J, et al. (2006) Retinoid signaling determines germ cell fate in mice. Science312(5773):596–600.

14. Berkholtz CB, Lai BE, Woodruff TK, Shea LD (2006) Distribution of extracellular matrixproteins type I collagen, type IV collagen, fibronectin, and laminin in mouse folli-culogenesis. Histochem Cell Biol 126(5):583–592.

15. Chen Y, Breen K, Pepling ME (2009) Estrogen can signal through multiple pathways toregulate oocyte cyst breakdown and primordial follicle assembly in the neonatalmouse ovary. J Endocrinol 202(3):407–417.

16. JeffersonW, Newbold R, Padilla-Banks E, PeplingM (2006) Neonatal genistein treatmentalters ovarian differentiation in the mouse: Inhibition of oocyte nest breakdown andincreased oocyte survival. Biol Reprod 74(1):161–168.

17. Levin ER (2005) Integration of the extranuclear and nuclear actions of estrogen. MolEndocrinol 19(8):1951–1959.

18. Hirao Y (2012) Isolation of ovarian components essential for growth and developmentof mammalian oocytes in vitro. J Reprod Dev 58(2):167–174.

19. Hirao Y, et al. (2013) Production of fertile offspring from oocytes grown in vitro bynuclear transfer in cattle. Biol Reprod 89(3):57.

20. Otsuka F (2013) Multifunctional bone morphogenetic protein system in endocrinol-ogy. Acta Med Okayama 67(2):75–86.

21. Miyabayashi K, et al. (2015) Heterogeneity of ovarian theca and interstitial gland cellsin mice. PLoS One 10(6):e0128352.

22. Morohashi K, Baba T, Tanaka M (2013) Steroid hormones and the development ofreproductive organs. Sex Dev 7(1-3):61–79.

23. Obata Y, et al. (2011) Epigenetically immature oocytes lead to loss of imprintingduring embryogenesis. J Reprod Dev 57(3):327–334.

24. Cortvrindt R, Smitz J, Van Steirteghem AC (1996) In-vitro maturation, fertilization andembryo development of immature oocytes from early preantral follicles from pre-puberal mice in a simplified culture system. Hum Reprod 11(12):2656–2666.

25. Wang X, Catt S, Pangestu M, Temple-Smith P (2011) Successful in vitro culture of pre-antralfollicles derived from vitrified murine ovarian tissue: Oocyte maturation, fertilization, andlive births. Reproduction 141(2):183–191.

26. Obata Y, Maeda Y, Hatada I, Kono T (2007) Long-term effects of in vitro growth ofmouse oocytes on their maturation and development. J Reprod Dev 53(6):1183–1190.

27. Lei L, Jin S, Mayo KE, Woodruff TK (2010) The interactions between the stimulatoryeffect of follicle-stimulating hormone and the inhibitory effect of estrogen on mouseprimordial folliculogenesis. Biol Reprod 82(1):13–22.

28. Dutta S, Mark-Kappeler CJ, Hoyer PB, Pepling ME (2014) The steroid hormone envi-ronment during primordial follicle formation in perinatal mouse ovaries. Biol Reprod91(3):68.

29. Kezele P, Skinner MK (2003) Regulation of ovarian primordial follicle assembly anddevelopment by estrogen and progesterone: Endocrine model of follicle assembly.Endocrinology 144(8):3329–3337.

30. Fortune JE, Eppig JJ (1979) Effects of gonadotropins on steroid secretion by infantileand juvenile mouse ovaries in vitro. Endocrinology 105(3):760–768.

31. Yang MY, Fortune JE (2008) The capacity of primordial follicles in fetal bovine ovariesto initiate growth in vitro develops during mid-gestation and is associated withmeiotic arrest of oocytes. Biol Reprod 78(6):1153–1161.

32. Wandji SA, Srsen V, Nathanielsz PW, Eppig JJ, Fortune JE (1997) Initiation of growthof baboon primordial follicles in vitro. Hum Reprod 12(9):1993–2001.

33. Wang C, Roy SK (2007) Development of primordial follicles in the hamster: Role ofestradiol-17beta. Endocrinology 148(4):1707–1716.

34. Heller DT, Cahill DM, Schultz RM (1981) Biochemical studies of mammalian oogenesis:Metabolic cooperativity between granulosa cells and growing mouse oocytes. DevBiol 84(2):455–464.

35. Carabatsos MJ, Sellitto C, Goodenough DA, Albertini DF (2000) Oocyte-granulosa cellheterologous gap junctions are required for the coordination of nuclear and cytoplasmicmeiotic competence. Dev Biol 226(2):167–179.

36. Pfender S, et al. (2015) Live imaging RNAi screen reveals genes essential for meiosis inmammalian oocytes. Nature 524(7564):239–242.

37. Sato T, et al. (2011) In vitro production of functional sperm in cultured neonatalmouse testes. Nature 471(7339):504–507.

38. Mochida N, et al. (2013) Live births from isolated primary/early secondary folliclesfollowing a multistep culture without organ culture in mice. Reproduction 146(1):37–47.

39. Kumaki Y, Oda M, Okano M (2008) QUMA: Quantification tool for methylationanalysis. Nucleic Acids Res 36(Web Server issue):W170–W175.

9026 | www.pnas.org/cgi/doi/10.1073/pnas.1603817113 Morohaku et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 2

3, 2

020