A Morpholino oligo can modify splicing of a pre-mRNA - www.gene-tools.com



We have demonstrated that Notch genes are expressed in developing mammalian ovarian follicles. Lunatic fringe is an important regulator of Notch signaling. In this study, data are presented that demonstrate that radical fringe and lunatic fringe are expressed in the granulosa cells of developing follicles. Lunatic fringe null female mice were found to be infertile. Histological analysis of the lunatic fringe-deficient ovary demonstrated aberrant folliculogenesis. Furthermore, oocytes from these mutants did not complete meiotic maturation. This is a novel observation because this is the first report describing a meiotic defect that results from mutations in genes that are expressed in the somatic granulosa cells and not the oocytes. This represents a new role for the Notch signaling pathway and lunatic fringe in mammalian folliculogenesis.


The Notch gene family encodes transmembrane receptors that are highly conserved evolutionarily (Kimble and Simpson, 1997; Lewis, 1998; Artavanis-Tsakonis et al., 1999). In mammals, there are four Notch receptors (Notch1-4) (Weinmaster et al., 1991; Weinmaster et al., 1992; Franco del Amo et al., 1992; Lardelli and Lendahl, 1993; Lardelli et al., 1994; Uyttendaele et al., 1996), and two families of ligands, Deltalike1 (Dll1) (Bettenhausen et al., 1996), Dll3 (Dunwoodie et al., 1997) and Dll4 (Shutter et al., 2000), and Jagged1 (Lindsell et al., 1995) and Jagged2 (Shawber et al., 1997). The receptors and ligands have overlapping expression patterns in many tissues (Lardelli and Lendahl, 1993; Williams et al., 1995; Lindsell et al., 1996; Johnson et al., 2001). Activation of Notch by ligand binding triggers cleavage of the receptor in a process known as regulated intramembrane proteolysis (RIP), generating the Notch intracellular domain (NotchIC) that then translocates to the nucleus (Schroeter et al., 1996; Kopan et al., 1996; Blaumueller et al., 1997; Logeat et al., 1998). In the nucleus, Notch forms transcriptional complexes with CSL transcription factors and activates the expression of downstream target genes (Tamura et al., 1995; Hsieh et al., 1996; Struhl and Adachi, 1998). These target genes are two families of basic helix-loop-helix (bHLH) proteins referred to as hairy enhancer of split (Hes), and a related family variously referred to as Hesr, HRT, Hey, CHF, and Gridlock (Jarriault et al., 1995; Jarriault et al., 1998; Ohtsuka et al., 1999; Nakagowa, 1999; Nakagowa, 2000; Kokubo et al., 1999; Chin et al., 2000; Maier and Gessler, 2000; Zhong et al., 2000). These proteins have been implicated in the repression of tissue-specific gene transcription (Jarriault et al., 1995; Jarriault et al., 1998; Ohtsuka et al., 1999; Hsieh et al., 1999; Nakagawa et al., 2000; Chin et al., 2000).

The interaction between Notch and its ligands is modulated by O-linked fucose moieties that are added to the EGF repeats of the extracellular domain. Usually fucose is unaltered when it is added to proteins; however, on the Notch receptors fucose is modified with N-acetylglucosamine (GlcNac) added by the Fringe proteins (Moloney et al., 2000; Bruckner et al., 2000). The Fringe proteins are Golgi-localized and belong to a large family ofβ 1,3-N-acetylglucosaminyl transferases (Moloney et al., 2000; Bruckner et al., 2000; Schwientek et al., 2002). Enzymes in this family have strong substrate and target specificity and diverse functions. The only known targets of the mammalian fringe proteins are the Notch receptors (Schwientek et al., 2000). In mammals, there are three fringe proteins, radical (Rfng), manic (Mfng) and lunatic fringe (Lfng) (Johnston et al., 1997). Modification of the extracellular domain of Notch by Lfng can potentiate or inhibit the interaction between a particular Notch receptor-ligand pair. For example, Lfng potentiates the interaction between Notch1 and Dll1, but inhibits Notch1-Jagged1 interactions. However, Lfng potentiates both Dll1 and Jagged1 mediated activation of Notch2 (Hicks et al., 2000). Lfng and Mfng reportedly modify different sites in the extracellular domain of Notch2 (Shimizu et al., 2001), indicating they may have different roles to play in regulating Notch signaling.

In Drosophila, two Golgi localizedβ 1,3-N-acetylglucosamine transferases, Fringe and Brainiac, play important roles in oogenesis and folliculogenesis (Goode et al., 1996a; Goode et al., 1996b; Hicks et al., 2000; Bruckner et al., 2000; Munro and Freeman, 2000; Schwientek, 2002). Brainiac activity is needed in the germ line for proper organization of the follicle (Goode, 1996). Fringe, the homolog of Lfng, is necessary for specification of the polar cells (Grammont and Irvine, 2001). Brainiac has been demonstrated to modify glycosphingolipids by adding GlcNac residues to mannose and galactose moieties on ceramide (Schwientek et al., 2002). In mice, a null mutation of the murine homolog of brainiac demonstrated that this protein is important for very early development, as braniac–/– embryos die prior to implantation (Vollrath et al., 2001). No role for this family of proteins in mammalian folliculogenesis has been described.

It has been demonstrated that Lfng is an important regulator of Notch signaling. For example, Lfng null mutants have segmentation defects that are similar to those seen in null mutations of Notch1 and Dll1 (Evrard et al., 1997; Zhang and Gridley, 1997). In somites, where Lfng is the only family member expressed, Notch receptors and ligands are expressed normally in Lfng–/– mutants, but the Notch downstream target gene Hes5 was not detected, indicating a lack of Notch activation in the presence of ligand. However, Hes5 was expressed normally in the neural tube and developing brain of Lfng null mutants, probably due to expression of Mfng and Rfng in these tissues (Evrard et al., 1997). Interestingly, Rfng-deficient mice had no phenotype and Rfng/Lfng double null mutants had only defects associated with a lack of Lfng (Zhang et al., 2000).

Folliculogenesis is the process by which oocytes develop in response to hormonal cues. This requires the coordination of the proliferation and differentiation of granulosa cells and the growth and maturation of the oocyte. Primordial follicles consist of a small oocyte surrounded by squamous somatic cells. When recruited to develop, the granulosa cells proliferate and become cuboidal. As these cells continue to proliferate, layers develop around the growing oocyte. Once a follicle has several layers of cells a fluid filled space, the antrum, will begin to form. The antrum spatially separates the two functionally distinct granulosa cell populations, cumulus and mural. During this time the oocyte has grown and at the time of antrum formation it becomes competent to resume meiosis in response to luteinizing hormone (LH). Resumption of meiosis is marked by the breakdown of the germinal vesicle (GVB). Meiosis continues to metaphase II (MII), and oocytes are blocked at this stage until fertilization. Studies done in mice have demonstrated that reciprocal signaling between the oocyte and the granulosa cells is necessary for the differentiation of the cumulus granulosa cells and meiotic maturation of the oocyte (Rodgers et al., 1999; Erickson and Shimasaki, 2000; Eppig, 2001; Matzuk et al., 2002).

We have previously shown that Notch2, Notch3 and Jagged2 are expressed in granulosa cells, and Jagged1 is expressed in the oocytes of developing mammalian follicles (Johnson et al., 2001). Furthermore, transcripts of the Notch downstream target genes, Hes1, Hes5, Hesr1, Hesr2 and Hesr3 also were detected in the granulosa cells of follicle types 3b-8 (Johnson et al., 2001), indicating that Notch signaling was active. As all three mammalian fringe proteins can modify the Notch receptors when expressed in the same cell (Bruckner et al., 2000; Moloney et al., 2000; Hicks et al., 2000), we hypothesized that the fringe genes would also be expressed in the granulosa cells, and furthermore, that they would have a role in regulating folliculogenesis through Notch2 and Notch3. In this study, we demonstrate that Lfng is expressed in the granulosa cells and theca of developing follicles from primary to preovulatory in size. Rfng is expressed briefly in granulosa cells of early antral follicles. Mfng is only detected in the vasculature. Null mutations of the Notch receptors and ligands result in embryonic lethal phenotypes (Swiatek et al., 1994; Conlon et al., 1995; Hrabé de Angelis et al., 1997; Jiang et al., 1998; Hamada et al., 1999; Xue et al., 1999; McCright et al., 2001). Some Lfng–/– mice survive to adulthood, therefore we examined folliculogenesis in these mutants. Female Lfng null mutant mice have many aberrant follicles. When induced to ovulate they released oocytes into the oviduct, but only a small percentage could be fertilized in vitro. Examination of these oocytes demonstrated that cumulus expansion occurred in response to exogenous hormones, but the oocytes were not at metaphase of meiosis II, and had not completed meiotic maturation. Mutations that block the progression of meiosis have been described, but they are all germ-cell-specific genes. These are novel observations because the disregulation of meiosis is caused by a change in the somatic cells. This represents a new regulatory pathway in folliculogenesis and a new role for Notch signaling in mammals.

Materials and methods

Mating study

Eleven-week-old heterozygous male, and 8-week old null and heterozygous female, mice were paired. Each morning females were examined for the presence of a copulatory plug. If a plug was present, the female was removed and a new female introduced to the male cage. If after 6 days no copulatory plug was detected, the female was placed with a new male. Copulatory plugs and litter numbers were recorded and the genotype of the offspring was determined.

Whole-mount thick section in situ hybridization (ISH)

Whole-mount ISH was done on thick sections according to Johnson et al. (Johnson et al., 2001). Briefly, ovaries were fixed in 10% neutral-buffered formalin (NBF) (Richard-Allen Scientific, Kalamazoo, MI), and embedded in paraffin wax after stepwise dehydration in ethanol. Thirty micron (μm) sections were cut perpendicular to the axis of entry of ovarian blood vessels. Sections were dewaxed, rehydrated and antisense digoxigenin-labeled gene-specific RNA probes were hybridized. Transcripts were identified using anti-digoxigenin antibody (Roche, Indianapolis, IN) conjugated to alkaline phosphatase and the BM purple substrate (Roche, Indianapolis, IN). Replicates were performed on sections from at least 3 ovaries/genotype and probes were checked for specificity by ISH on embryos.


Tissues were fixed as above and sectioned to 10 μm. Sections were prepared by standard procedures and stained with Hematoxylin and Eosin.

Bone and cartilage preparation

The mice were skinned and eviscerated. The carcasses were placed in Alcian Blue (Sigma A3157) for 48 hours to stain the cartilage, followed by 2% KOH for 48 hours. Skeletons were then placed in Alizarin Red (Sigma A5533) to stain the bone for 72 hours.

Hormone treatment and isolation of OCC and oocytes

Mice were injected intraperitoneally (ip) with 5 international units (IU) of pregnant mare's serum gonadotropin (PMSG) (Calbiochem, Carlsbad, CA) and 48 hours later, were injected ip with 5 IU of human chorionic gonadotropin (hCG) (Calbiochem, Carlsbad, CA). Oocyte cumulus complexes (OCC) were harvested from the oviduct 16 hours later. OCC and ovaries were collected in KSOM (Specialty Media, La Jolla, CA) with 10% FBS. The OCC were incubated in KSOM containing hyaluronidase (300 μg/ml) for 30 seconds, then washed in KSOM, and fixed as described in LeMaire-Adkins et al. (LeMaire-Adkins et al., 1997). Briefly, oocytes were fixed in 2% formaldehyde, 1% Triton X-100, 0.1 mM PIPES, 5 mM MgCl2, 1 mM DTT and 2.5 mM EGTA in D2O (Sigma, St Louis, MO) containing aprotinin (Sigma, St Louis, MO) and taxol (Sigma, St Louis, MO) at 37°C. Oocytes were washed in 0.1% normal goat serum (NGS) in PBS (GIBCO/BRL, Gaithersburg MD) and blocked at 37°C in PBS containing 10% NGS and 0.1% Triton X-100. Oocytes were stored in this at 4°C until staining was performed. For staining, oocytes were transferred to 1% Triton X-100 in PBS at room temperature then incubated with a monoclonal anti-α-tubulin (clone DM 1A, Sigma, St Louis, MO) conjugated to FITC at a 1:50 dilution. The oocytes were washed in PBS, incubated in PBS containing 1μ g/ml Hoechst 33258 (Molecular Probes, Eugene, OR). Oocytes were washed in PBS and mounted in glycerol with p-phenylenediamine and visualized by confocal microscopy. Confocal analysis was done using a Leica TCS NT, final magnification of 800 ×. The FITC was visualized using an Ar laser, and Hoechst 33258 was visualized using an UV laser.

In vitro fertilization (IVF)

OCC were collected after hormone administration as above. Sperm were collected from the vas deferens and cauda epididymis in human tubal fluid (HTF) and capacitated for 2 hours at 37°C. Sperm (1 ×106) were added to each OCC sample and fertilization allowed to proceed for 2 hours at 37°C. Eggs were washed three times in sperm free KSOM and incubated at 37°C. Eggs were scored as fertilized by the presence of two pronuclei and embryogenesis was scored daily.

Immunohistochemistry (IHC)

Sections were heated at 80°C for 30 minutes, cooled to room temperature, followed by xylenes, rehydrated through graded alcohols to 70% ethanol. Slides were incubated in water, then PBS, placed in 0.1 M sodium citrate (pH 6) and epitope retrieval done in the microwave. The sections were cooled to room temperature, rinsed in PBS, and incubated in 3% H2O2 in 60% methanol to destroy endogenous peroxidases. IHC was performed using the HistostainSP kit according to the manufacturer's instructions (Zymed Labs, San Francisco, CA), and primary antibodies were diluted according to this protocol except for the following: polyclonal anti-c-Kit antibody (Ab-1, Calbiochem, Carlsbad, CA) was diluted 1:25, and anti-connexin43 (Santa Cruz Biotechnology, Santa Cruz CA), 1:50. Proteins were detected with alkaline-phosphatase-conjugated anti-rabbit secondary antibody and exposed to colour reagent. No primary antibody controls were included in each experiment.

Reverse transcription polymerase chain reaction (RT-PCR)

Total ovary RNA was isolated using TRIzol (Life Technologies, Gaithersburg, MD), according to the manufacturer's directions, from 3 different animals/genotype. For oocytes, 15 oocytes per sample were denuded using hyaluronidase and total RNA extracted. cDNA was synthesized using Superscript III (Invitrogen, Carlsbad CA), according to the manufacturer's protocol. For each gene examined by semi-quantitative (sq) RT-PCR, 3 sets of samples comprising all three genotypes and no RT controls were amplified usingα -[32P]dATP (Perkin-Elmer Life and Analytical Sciences, Boston, MA). For each gene-specific primer pair the minimum number of cycles to the linear range was determined and used for all subsequent experiments. All primer sets span at least one intron. Control experiments were done using total embryo RNA. All cDNA samples were normalized using the ribosomal gene L7 (Meyuhas et al., 1990), and quantified using a Storm 860 PhosphorImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA). To detect the presence of transcripts in oocytes, a qualitative PCR protocol and amplification beyond the linear range after normalization was used (Münsterberg and Lassar, 1995).

Kinase assays

Kinase assays were carried out as described in Svoboda et al. (Svoboda et al., 2000). Single eggs were transferred in 1.5 μl of KSOM into 3.5 μl of double kinase lysis buffer (10 μg/ml aprotinin, 10 μg/ml leupeptin, 10 μM p-nitrophenyl phosphate, 20 mM β-glycerophosphate, 0.1 mM sodium orthovanadate, 5 mM EGTA) and immediately frozen in liquid nitrogen, then stored at –80°C until the assay was performed. The kinase reaction was initiated by the addition of 5 μl of double kinase buffer (24 mM p-nitrophenyl phosphate, 90 mM β-glycerophosphate, 24 mM MgCl2, 24 mM EGTA, 0.2 mM EDTA, 4.6 mM sodium orthovanadate, 4 mM NaF, 1.6 mM dithiothreitol, 60 μg/ml aprotinin, 60 μg/ml leupeptin, 2 mg/ml polyvinyl alcohol, 2.2 mM protein kinase A inhibitor peptide (Sigma), 40 mM 3-(nmorpholino) propanesulfonic acid (MOPS), pH 7.2, 0.6 mM ATP, 2 mg/ml histone (type III-S, Sigma), 0.5 mg/ml MBP with 500 mCi/mlγ -[32P]ATP (3000 Ci/mmol) (Perkin-Elmer Life and Analytical Sciences). To determine the background level of phosphorylation, 5 μl of double kinase lysis buffer was added instead of egg lysate. Reactions were incubated for 30 minutes at 30°C, and terminated by the addition of 10μ l 2 ×SDS-PAGE sample buffer and boiling for 3 minutes. Following 15% SDS-PAGE, the gel was dried and exposed to a phosphorimager screen and quantified. The mean value of the control samples was set to one and all others expressed as fold activity of control.

Quantifying cumulus expansion

OCC were collected from the oviducts of Lfng+/– and Lfng–/– mice (n=5/genotype) post-hormone administration. OCC were photographed at 70 × and photomicrographs printed the same size. The widest diameter of each OCC was measured in mm and mean diameter±s.d. was determined.

Statistical analysis

Analysis was carried out using the SAS system and the FREQ procedure. Our data was found to be significant (P<0.0001) by Chi-square, likelihood ratio Chi-square, Continuity-adjusted Chi-square and Mantel-Haenszel Chi-square analysis.


Expression of lunatic fringe in the murine ovary

In mammals, the three Fringe proteins are often expressed in overlapping and distinct patterns within the same tissue (Johnston et al., 1997; Ishii et al., 2000). Lfng modification of the Notch extracellular domain is known to potentiate interactions between Notch2 and jagged 1 (Hicks et al., 2000). However, its effect on the interactions between Notch2 and jagged 2, or between any ligand and Notch3 is uncharacterized. Rfng has been shown to be completely redundant to Lfng (Zhang et al., 2000). The role of Mfng is not well known, but it modifies different sites on the extracellular domain of Notch1 to Lfng (Shimizu et al., 2001). Knowing the importance of the fringe proteins in modulating Notch signaling, and having demonstrated the expression of Notch2 and Notch3, the jagged ligands and the downstream target genes of Notch in developing follicles (Johnson et al., 2001), it was logical to examine whether the fringe genes were also expressed in the ovary.

In order to determine which cells in the ovary expressed Lfng, whole-mount thick section in situ hybridization (ISH) was carried out using antisense digoxigenin-labeled RNA probes, as described by Johnson et al. (Johnson et al., 2001). Using two different probes, one that included 3′ untranslated sequences and another that only included the coding region, Lfng mRNA was detected in the granulosa cells of follicles from type 3 to those preovulatory in size (Fig. 1A,B). Interestingly, in small growing follicles, Lfng transcripts clearly demarcate the outer edge of the follicle (Fig. 1B, inset). In antral follicles, Lfng expression in the theca was evident (Fig. 1A,B). Furthermore, this gene was expressed in blood vessels (Fig. 1B, red arrow), so it is possible that the expression noted in the theca is also in the vascular component. Lfng was not expressed in oocytes, primordial or primary follicles, or corpus luteum (CL). The lack of Lfng transcripts in oocytes was confirmed by RT-PCR performed using total RNA from denuded germinal vesicle (GV) stage oocytes and ovulated MII eggs (Fig. 1F).

Fig. 1.

Lunatic fringe is expressed in the granulosa cells and theca. Whole-mount thick section ISH of wild-type 42-day-old ovary was carried out using antisense digoxigenin-labeled RNA gene-specific probes. (A) Lfng-specific probes including 3′ untranslated sequences and Lfng transcripts were detected in follicles from type 3 to preovulatory in granulosa cells, but not in oocytes or interstitial cells. (B) Lfng exon only probe. Transcripts were detected in the theca and in the vasculature (red arrow indicates positive blood vessel), a band of Lfng transcripts surrounding small follicles was observed (inset). (C) Rfng was expressed transiently in the granulosa cells of antral follicles (arrow). (D) Mfng transcripts were only detected in the vasculature (red arrow). (E) RT-PCR from total ovary RNA demonstrates the presence of Mfng and Rfng transcripts. E9.5 represents total E9.5 embryo control; L+/+ represents total ovary RNA from a 42-day-old wild-type animal; mock represents no RT control. (F) RT-PCR of Lfng and Gdf9 transcripts from total RNA from GV and MII stage oocytes. E9.5, total embryo control; MW, 100 bp ladder. Scale bars: 100μ m.

Using RT-PCR of total ovary RNA, both Rfng and Mfng transcripts were detected (Fig. 1E). ISH done with an antisense Mfng probe demonstrated that transcripts for this gene were detected only in the vasculature (Fig. 1D). Rfng transcripts were detected in the vasculature and transiently in granulosa cells of early antral follicles (Fig. 1C). These data demonstrate that Lfng and, briefly Rfng, overlap in expression with the Notch receptors in the granulosa cells. The transient expression of Rfng, the fact that Rfng null mutants have no phenotype, and the fact that Rfng/Lfng double null mutants have only defects associated with a lack of Lfng (Zhang et al., 2000), indicate that Lfng is likely to be the more important family member expressed in the adult mouse ovary.

Lunatic fringe-deficient ovaries have aberrant follicles

The role of Notch signaling in the ovary is unknown, and null mutants of most of the Notch receptors and Jagged ligands have embryonic or perinatal lethal phenotypes (Swiatek et al., 1994; Conlon et al., 1995; Hrabé de Angelis et al., 1997; Jiang et al., 1998; Hamada et al., 1999; Xue et al., 1999; McCright et al., 2001). Lfng is an important modifier of Notch signaling (del Barco Barrantes et al., 1999; Hicks et al., 2000; Moloney et al., 2000); in its absence, Notch signaling in the somites was impaired, as determined by the lack of Hes gene expression in Lfng–/– embryos (Evrard et al., 1997). Furthermore, Lfng null mutants have segmentation defects that are consistent with mutations in Dll1 and Notch1 (Zhang and Gridley, 1997; Evrard et al., 1997). Interestingly, of the two different Lfng–/– mutations, one results in complete embryonic lethality of Lfng–/– offspring (Zhang and Gridley, 1997), whereas the other has a 25% survival rate (Evrard et al., 1997); we studied the role of Lfng and the Notch signaling pathway in folliculogenesis using the latter mutant.

Initially studies were carried out to determine whether female Lfng-deficient mice would mate and produce litters. Lfng null and heterozygous mice were paired with Lfng+/– male mice. All heterozygous females (n=11) mated within 6 days, as determined by the presence of a copulatory plug. There were 26 litters from heterozygous pairings; by comparison, over the same time, null females demonstrated neither copulatory plugs nor pregnancies. These observations indicated the possibility of a fertility defect. However these mice have abnormalities of their axial skeletons, including fusions of the vertebrae and kyphosis, which may make lordosis and mating impossible (Evrard et al., 1997; Zhang and Gridley, 1997) (see Fig. S1 in the supplementary material).

In order to determine whether infertility was due to either defects in ovary development or folliculogenesis in the Lfng–/– mice, gross morphological and histological examination of ovaries from neonatal and adult Lfng–/– and Lfng+/– mice was done. At the gross morphological level, the ovaries and reproductive tracts of 4- and 7-week-old Lfng null mice were smaller than heterozygous littermates, but no abnormalities were noted. Protuberances on the surface of the ovaries indicated the presence of developing follicles (Fig. 2A,B). Lfng null mice have spinal defects that result in a shortened body axis, so smaller organs are not unexpected. The ovarian histomorphology of neonatal Lfng null mutants was disorganized compared with heterozygous littermates (Fig. 2, compare C with D), but primordial follicles were clearly present. Ovaries from sexually mature Lfng–/– mice had developing follicles of all sizes, but there were many abnormal follicles present. For example, there were polyovular follicles (Fig. 2F-H, red arrows in F). Furthermore, there were follicles that lacked a complete layer of theca developing next to the theca of other follicles or sharing theca, but not truly polyovular (Fig. 2E,F, black arrows in F). Many of these follicles appeared atretic. CL were noted, but, there were also many large lutealinized follicles with trapped oocytes (Fig. 2I,J).

Fig. 2.

Histological and morphological examination of Lfng–/– ovaries. (A) Ovaries from 4-week-old Lfng+/– and Lfng–/– mice; null ovaries are smaller and developing follicles are obvious in both. (B) The reproductive tracts of 7-week-old Lfng+/– and Lfng–/– mice; all structures were present and no gross abnormalities were noted. (C,D) Histological sections of neonatal ovaries stained with Hematoxylin and Eosin. Primordial follicles are evident; however, the Lfng null ovary (C) is smaller than the Lfng+/– ovary (D) and the morphology is not as well organized. (E-J) Histological sections of ovaries from 42-day-old Lfng–/– mice. Many abnormal follicles were noted. Some large follicles had smaller follicles within their boundaries (E, black arrow). There were follicles that shared theca or had incomplete theca (F, black arrowheads). There were polyovular follicles containing two or three oocytes (F, red arrows; G,H). There were also many large lutealinized follicles with trapped oocytes (I,J). Scale bars: 100 μm.

As there were many oocytes trapped in lutealinized follicles in the Lfng-deficient mice, and we never detected copulatory plugs, it was possible that these mice did not ovulate. Alternatively, if they did ovulate, the oocytes might not be competent for fertilization and subsequent development because folliculogenesis was aberrant. In order to examine these questions, 42-day-old Lfng null and control littermates were given PMSG and hCG to induce ovulation, and OCC were harvested from the oviduct. The OCC were incubated with capacitated heterozygous sperm. Only fertilized eggs, as determined by the presence of two pronuclei, were kept, and embryogenesis was scored daily.

Lfng null mice released approximately the same number of eggs with cumulus cells as controls in response to exogenous hormones, and cumulus expansion was evident (Table 1, Fig. 3C). After fertilization, 48.8% of wild-type eggs became two-cell embryos, but only 9.7% of null eggs did. Furthermore, whereas 41.9% of heterozygous embryos continued to develop to the four- to eight-cell stage, and 31.4% became blastocysts, only 2% of the null embryos became four- to eight-cell embryos, and none developed into blastocysts (Table 1). The Lfng null mutants can respond to exogenous hormone, but they had a very low fertilization rate and there may be a block in early development. A lack of fertilization and development suggests a defect in folliculogenesis.

View this table:
Table 1.

Embryonic development of Lfng-/- eggs

Fig. 3.

Lfng–/– females ovulate in response to exogenous hormones, but oocytes are not at meiotic metaphase II. (A) After hormone administration, OCC were collected from the oviduct. Oocytes were stained with anti-α-tubulin-FITC and Hoechst 33258, and visualized using confocal microscopy at 800 × final magnification. Genotypes are as indicated. In Lfng+/+ and Lfng+/– oocytes, note the presence of the polar bodies (white arrowhead) and chromosomes aligned on a metaphase spindle. Many Lfng–/– oocytes contained one or more bodies with scattered chromatin (white arrowheads). Some Lfng–/– oocytes were at anaphase/telophase I; note the lack of a polar body. (B) Representative kinase assays analyzed on 15% SDS-PAGE and visualized using a phosphorimager. Lanes are as marked. MBP, myelin basic protein. (C) Lfng+/– and Lfng–/– mice were induced to ovulate and OCC were collected from the oviducts. OCC demonstrated normal cumulus expansion in Lfng null mutant females, under light microscopy at 70 × magnification. (D) Graphs show relative kinase activity of MPF and CSF, the kinase assays were quantified and expressed as fold activity over control (n=9/genotype) ±s.d.

Meiotic maturation is compromised in Lfng female null mice

Oocyte maturation occurs in antral follicles in response to the surge of LH from the pituitary. The first step of maturation is GVB, after which the first meiotic spindle forms. The spindle is accentric at the oocyte cortex, resulting in conserved, limited cytokinesis with the outside spindle pole protruding through the oocyte surface into the perivitelline space. At the end of the first meiotic division, one set of homologous chromosomes remains within the oocyte and the other set is abstricted in the first polar body. After formation of the second meiotic spindle, meiotic maturation is completed (Robker and Richards, 1998; Erickson and Shimasaki, 2000; Matzuk et al., 2002). Eggs released into the oviduct should be at metaphase II (MII), so we determined which stage of meiosis Lfng null eggs had reached by ovulation.

Exogenous hormones were used to induce ovulation in Lfng null and control littermates, and OCC were harvested from the oviduct, as described above. Eggs were fixed and stained with anti-α-tubulin antibody conjugated to FITC; the chromatin was stained with Hoechst 33258. We found that 77.8% and 88.2% of heterozygous and wild-type eggs, respectively, were in MII, as determined by the presence of a barrel-shaped meiotic spindle with chromosomes on the metaphase plate and a polar body, but only 5% of Lfng–/– eggs were in MII (P<0.0001, null compared with controls). Most of the oocytes from Lfng–/– mice were at metaphase I (MI), anaphase/telophase I, or had multiple bodies with chromatin fragments throughout (Fig. 3A). These eggs were not characteristically parthenogenic, there were no obvious nuclei, nor polar bodies, none were two cells (Hirao and Eppig, 1997), and the diffuse chromatin was indicative of apoptosis. These data indicated that in Lfng-deficient follicles, oocytes were not completing meiotic maturation prior to induced ovulation. Interestingly, these oocytes have a normally expanded cumulus (Fig. 3C): the mean diameter of the OCC from Lfng heterozygous and null mice was not different when compared (mean diameter heterozygous: 3.8±0.4 mm; null: 3.4±0.4 mm; P=0.1) (see Fig. S2 in the supplementary material).

To further examine oocyte maturation in the Lfng null ovary, the level of maturation promoting factor (MPF) and cytostatic factor (CSF) kinase activities was determined. MPF is necessary for GVB, and its activation precedes CSF activation. Mos activates MAP kinase, which is a component of CSF and is necessary for the MII block. MPF and CSF activity both peak in MII eggs (Zhao et al., 1991; Verlhac et al., 1993; Gebauer and Richter, 1997; Sagata, 1997). Eggs were collected from oviducts post-hormone administration, and kinase activity determined in null and control eggs (n=9, representative data in Fig. 3B). Both MPF and CSF kinase activity was evident in null oocytes, but it was less than in controls (Fig. 3D). Consistently, 75% of Lfng+/– (n=364) and 74.9% of Lfng–/– (n=355) (P=1) oocytes, underwent GVB in vitro within 2 hours, thus null oocytes resume meiosis normally.

It is possible that null oocytes have a defect that blocks progression through meiosis, so GV stage oocytes were collected from large follicles post-PMSG administration and allowed to mature in vitro, as described by LeMaire-Adkins et al. (LeMaire-Adkins et al., 1997). Only oocytes that underwent GVB within 2 hours were followed, in order to have a synchronous cohort. After 10, 12 and 16 hours in culture, oocytes were fixed and stained and the stage of meiosis was determined. Oocytes from heterozygous ovaries progressed through meiosis comparably to previously published data (LeMaire-Adkins et al., 1997), with 40.4% at MII at 10 hours, 57% at 12 hours and 93.6% at 16 hours (Table 2). Oocytes from null ovaries reached MI quickly (75% at 10 hours), but few progressed to MII. At early timepoints, there were oocytes progressing through telophase/anaphase I, but by 16 hours in culture this stage was not detected; possibly these cells die quickly if progression to MII does not occur. We did not detect an increase in parthenogenically activated oocytes (4.3% of control and 6.5% of null, P=0.13).

View this table:
Table 2.

In vitro maturation of oocytes

Notch family gene expression in Lunatic fringe-deficient follicles

In the somites of Lfng–/– embryos, the expression of Notch1 and Dll1 were not altered; however, the Notch downstream target gene Hes5 was not expressed (Evrard et al., 1997). Thus, a lack of Lfng blocked Notch activation in the presence of ligand in the somites. Also, disruption of the somites and somite-derived tissues in Lfng-deficient mice was consistent with the phenotypes of Notch1–/– and Dll1–/– mutants (Evrard et al., 1997; Zhang and Gridley, 1997). As Lfng is the predominant family member expressed in granulosa cells, it was logical to ask whether Notch signaling pathway genes were altered in expression and whether Notch signaling was inhibited in the Lfng-deficient granulosa cells.

To examine the effect of a lack of Lfng on the Notch pathway in the ovary, the expression of Notch receptor and ligand genes was determined by sqRT-PCR, using whole ovary RNA, and compared with the results obtained from ISH. Representative sqRT-PCR from three replicates is presented in Fig. 4. Notch2 and Jagged2 demonstrated no change in expression level (Fig. 4), and no change in their cell or follicle stage-specific pattern of expression when examined by ISH (data not shown). Notch3 had a slightly reduced level of expression by sqRT-PCR, but no change in expression was detected by ISH (Figs 4, 5). Jagged1 was expressed at greatly reduced levels in Lfng-deficient ovaries. This gene was still oocyte restricted, but was expressed only in very small follicles (Figs 4, 5). No change in the expression level of Notch2 and Jagged2, and a relatively small reduction in Notch3, is consistent with observations reported in the somites of Lfng–/– embryos (Evrard et al., 1997).

Fig. 4.

sqRT-PCR examination of gene expression in Lfng–/– ovaries. sqRT-PCR with gene-specific primers was performed to detect transcripts in total ovary RNA from 42-day-old mice of all three genotypes (Lfng+/+, Lfng+/– and Lfng–/–). Three samples per genotype were normalized against the ribosomal gene L7 and representative data are presented. N2, Notch2; N3, Notch3; J1, Jagged1; J2, Jagged2; EP2, prostaglandin type 2 receptor (Ptger2 – Mouse Genome Informatics); Fshr, follicle stimulating hormone receptor.

Fig. 5.

Expression of Notch family genes in the Lfng–/– ovary by ISH. Gene-specific digoxigenin labeled probes were used, as indicated; representative data from a minimum of three replicates of Lfng+/– and Lfng–/– are presented. Notch3 demonstrated no change in follicle stage or cell-type-specific expression in null ovaries; black arrows indicate granulosa cell staining and the red arrowhead indicates expression in an early CL. Jagged1 was limited to oocytes of small growing follicles in the Lfng–/– ovary (arrowhead). Expression of Notch downstream target genes Hes5, Hes1, Hesr1 and Hesr2 was not detected in the Lfng–/– follicles (only representative family members Hes5 and Hesr2 are presented). Note positive follicles indicated by arrows in +/– and compare with similarly sized follicles indicated by arrowheads in– /–.

Lfng has been demonstrated to potentiate interactions between Notch2 and Jagged1 and Dll1 (Hicks et al., 2000). As stated previously, in the somites, where Lfng is the only fringe gene expressed, the downstream target gene Hes5 was not expressed. No data regarding Hes1 or the Hesr family of Notch target genes is available. As the ovary primarily expresses Lfng (Fig. 1), the expression of the downstream target genes was assessed as a measure of Notch signaling activity. The downstream target genes Hes1, Hes5, Hesr1 and Hesr2 were greatly reduced by sqRT-PCR and were undetectable by ISH (Figs 4, 5). Interestingly, Hesr3 was unchanged in its expression level (Fig. 4), and there was no change in its expression pattern (data not shown).

Expression of other follicle genes in the Lfng–/– ovary

Reciprocal signaling between the oocyte and granulosa cells is crucial for folliculogenesis and oocyte maturation (Gosden et al., 1997; Erickson and Shimasaki, 2000; Su et al., 2002; Su et al., 2003). Lfng-deficient oocytes do not mature normally, so possibly other important signaling pathways may be altered in this mutant. Kit ligand (KL) is expressed in granulosa cells and the Kit receptor (previously known as c-Kit) is expressed in oocytes, in follicles from primordial to preovulatory in size. In antral follicles, KL is only expressed in the mural population. Signaling through this pathway has been implicated in differentiation of the theca, proliferation of granulosa cells and meiotic maturation (for a review, see Rawls et al., 2001). KL did not change in expression, and there was no change detected in Kit receptor expression in the Lfng null ovary when compared with controls (Fig. 6).

Fig. 6.

Expression of other follicle-specific genes. Connexin43 (Cx43) was detected by IHC using an alkaline phosphatase-conjugated secondary antibody and DAB substrate, as indicated by the red stain in granulosa cells and oocytes. There was no alteration in Cx43 expression. Kit ligand is expressed normally in Lfng–/– granulosa cells. Kit ligand transcripts were detected by ISH, note the purple stain in granulosa cells, dark field microscopy. Kit receptor (c-Kit) was also unaffected in Lfng null follicles. Kit receptor was detected by IHC, arrowhead in +/– indicates a group of primary follicles with positive oocytes. Controls included no primary antibody, note the lack of red stain.

Signaling through gap junctions is crucial for folliculogenesis (Anderson and Albertini, 1976; Gilula et al., 1978; Simon et al., 1997; Juneja et al., 1999; Ackert et al., 2001). Connexin43 (Cx43) is the major connexin in gap junctions between granulosa cells and between granulosa cells and oocytes, and folliculogenesis is arrested very early in Cx43 null mutant mice (Juneja et al., 1999; Ackert et al., 2001). An anti-Cx43 antibody was used for IHC and Cx43 was detected in the granulosa cells and oocytes of follicles of both Lfng+/– and Lfng–/– ovaries (Fig. 6), indicating that gap junctions are present. The presence of antral follicles in the mutant ovaries and the observation of ovulation in response to exogenous hormones would further indicate that the gap junctions are functional.

Signaling through the FSH receptor (Fshr), which is expressed by every granulosa cell, is necessary for normal development of the ovary and for normal sexual maturation. Mice with null mutations in the Fshr gene (FORKO) demonstrate hypergonadotropic hypogonadism (Danilovich et al., 2000). When expression of Fshr was examined by sqRT-PCR, no change was detected (Fig. 4). Because the FORKO mice demonstrate only primary and small preantral follicles and no CL (Danilovich et al., 2000), these data are consistent. As there was no change in the expression of many genes that are important for folliculogenesis, we conclude that the defects are due to a lack of Lfng and decreased Notch signaling in the follicles of these mutants.


We determined that Lfng had overlapping expression with Notch2 and Notch3 in granulosa cells, but not in oocytes, of stages 3-8 follicles (Fig. 1). There is no information regarding the role of Lfng in interactions between any Notch receptor and Jagged2, or between Notch3 and any ligand. But, with the exception of Hesr3, all of the downstream target genes of Notch were downregulated in the absence of Lfng (Figs 4, 5). Hesr3 expression may be induced by the transient expression of Rfng in follicles or through non-Lfng-dependent Notch signaling. The early downregulation of Jagged1 expression in oocytes may indicate that a lack of Notch signaling either activates an inhibitory signal that turns off Jagged1 or that Notch activity is necessary to maintain expression of Jagged1. It has been demonstrated that signaling between Notch and its ligands positively regulates the expression of both genes, i.e. an increase in Notch and the ligand in their respective cells (Piddini and Vincent, 2003).

We have demonstrated that Lfng–/– female mice are infertile. Furthermore, there is an interesting disconnection between cumulus expansion and GVB and completion of meiotic maturation. Observations of null oocytes indicated a lack of progression from MI to MII (Table 2; Fig. 3). These data indicate an important role for Lfng, and thereby for Notch signaling, in folliculogenesis.

Lfng deficiency results in morphological defects of follicles

The defects noted in developing follicles are unusual, and indicate that there is a problem with the definition of borders between follicles. In mice, early germ cells are found in cysts with interconnecting ring canals. At embryonic day (E) 13.5 to E19.5, germ cells are found in cysts of eight cells or more, by postnatal day 3, almost all germ cells are single oocytes in primordial follicles (Pepling and Spradling, 1998; Pepling and Spradling, 2001). In the process of becoming single oocytes, the cysts break down through apoptosis of single germ cells. The current model posits that this results in cysts of three, and each of these germ cells will become a primordial follicle. The somatic cells of the ovary are important for ring canal breakdown and formation of primordial follicles (Pepling and Spradling, 2001). In Lfng-deficient ovaries there are polyovular follicles, this may indicate a role for Notch signaling in the organization of primordial follicles. Furthermore, the ring of Lfng transcripts that we observed around small growing follicles (Fig. 1) may indicate a process of early boundary formation that is not occurring correctly in mutant follicles. Similarly, during oogenesis in Drosophila, the polar cells are required to separate germ cell cysts and enclose them appropriately in somatic cells. Fringe-deficient mutants had compound follicles containing more than one egg. They also had a disorganized polar epithelium; instead of a single layer there were multiple layers of cells (Grammont and Irvine, 2001). Thus, these data point to evolutionary conservation of function for this pathway in the development of follicles.

Signaling through the Notch pathway is necessary for completion of meiotic maturation

It is possible that the trapped oocytes in lutealinized follicles reflect a pituitary defect. However, in the hypogonadal (hpg) mouse, FSH and LH are not secreted by the pituitary, and there is a lack of postnatal ovary and follicle development (Halpin et al., 1986a; Halpin et al., 1986b). Grafting neural tissue or providing exogenous hormones to hpg female mice completely reverses this phenotype; folliculogenesis is initiated and results in MII eggs that are competent for fertilization and embryonic development (Charlton, 1987; Hashizume et al., 1995). The hpg phenotype is significantly different than the Lfng null females in several ways. First, Lfng null mice demonstrated postnatal ovary development and had developing, albeit abnormal, follicles (Fig. 2). Second, providing exogenous hormone to Lfng–/– females induced ovulation, cumulus expansion and GVB, but did not induce complete meiotic maturation, and these eggs had a low fertilization rate (Tables 1, 2; Fig. 3). These observations indicate that Lfng–/– mice have a defect at the level of the follicle.

Reciprocal signaling between the oocyte and the somatic granulosa cells is necessary for meiotic maturation of the oocyte and differentiation of the cumulus cells. Work by several groups has demonstrated that these events are regulated by the interplay of several different pathways. For example, recombinant Gdf9 can induce the expression of has2 and cox2, and cumulus expansion, but not GVB (Elvin et al., 1999; Su et al., 2002; Su et al., 2003). Furthermore, gonadotropin induced activation of ERK1/2 in granulosa cells is necessary for cumulus expansion and meiotic maturation, and yet both of these events also require a signal from the oocyte (Su et al., 2002; Su et al., 2003). This suggests that there is some signal that is switched on or off in oocytes during the final stages of folliculogenesis that must activate a signal(s) from the cumulus cells to the oocyte that is necessary for GVB and meiotic maturation (Su et al., 2003). It is possible that activation of Notch, in the granulosa cells, by Jagged1 in the oocyte, regulates some of these signals. This idea is further supported by the observation that oocytes from preantral follicles cultured with cumulus cells can activate ERK1/2, but that cumulus expansion does not occur, whereas fully grown oocytes can activate ERK1/2 and induce expression of cox2 and cumulus expansion when cultured with cumulus cells (Vanderhyden et al., 1990; Joyce et al., 2001). Normally, expression of Jagged1 in the oocyte is downregulated as follicles develop an antrum (Johnson et al., 2001). This is the point at which the oocyte is acquiring the competence to resume meiosis and cumulus cells would begin differentiating. But, in oocytes from Lfng null mutants, the expression of Jagged1 is restricted to very small growing follicles, indicating a change in the reciprocal signaling between the germ and somatic cells (Fig. 5).

The meiosis defect and trapped oocytes reflect alterations in the granulosa cells caused by a lack of Notch signalling, which leads to temporal changes in, or a lack of, some signal(s) from these cells. As Notch signaling can regulate lineage decisions, it is possible that the defects observed are due to alterations in mural and cumulus cell populations in Lfng-deficient follicles. In the absence of normal Notch signalling, the granulosa cells could become predominantly cumulus at the expense of mural cells, as expansion occurs in response to hormone. Alternatively, it could be a temporal change in cell differentiation: a lack of Notch signaling may allow for differentiation of the cumulus lineage in small growing follicles instead of early antral follicles. This could alter the timing of a crucial signal, and the oocyte may be unable to appropriately detect or respond to it. This would be consistent with the noted loss of Jagged1 expression in small growing instead of early antral follicles that we observed (Figs 4, 5). This is also consistent with data from other systems, changes in lineage choice mediated by Lfng alteration of Notch signaling has been demonstrated in lymphoid progenitors. Transgenic expression of Lfng inhibited Notch1 activation and caused the cells to choose the B lineage at the expense of the T lineage (Koch et al., 2001). Future studies will be necessary to determine whether there are changes in the mural and cumulus populations.

We determined that oocytes from Lfng-deficient ovaries resume meiosis in response to exogenous hormones, but do not complete it. MPF and CSF kinase activity were decreased with respect to control oocytes. However, there is enough MPF activity for meiosis to resume (Fig. 3, Table 2). There are no data that demonstrate that a reduction of CSF activity will block progression of meiosis. On the contrary, mos null oocytes do not arrest at MII, they continue to divide and activate parthenogenically. Mos is a kinase that is necessary for MAPK and CSF activity in mouse oocytes (O'Keefe et al., 1989; Zhao et al., 1991;Verlhac et al., 1993; Verlhac et al., 1996; Colledge et al., 1994; Hashimoto et al., 1994). Other germ-cell-specific genes also have roles in progression through meiosis. For example, cks2, a Cdk1 binding protein, is necessary for entry to anaphase I, and cks2–/– oocytes do not progress past MI (Spruck et al., 2003). There are many genes necessary for progression of meiosis that participate in synapsis and recombination (Hunt and Hassold, 2002). Importantly, in all cases these are oocyte-specific genes, Lfng is not expressed in the oocyte (Fig. 1) it is expressed in granulosa cells, as are Notch2 and Notch3, its targets. Our data indicate a fundamental change in the signaling between somatic granulosa cells and oocytes in Lfng-deficient follicles. Disregulation of meiosis through a somatic-cell-specific signal is a novel observation. These data reinforce the importance of reciprocal signaling between the granulosa cells and the oocyte during folliculogenesis. Altered Notch signaling results in the loss of normal meiotic maturation, but not cumulus expansion and GVB. Collectively, these data demonstrate that Lfng and the Notch signaling pathway play important roles in mammalian folliculogenesis.


We thank Drs Barbara Vanderhyden, Patricia Hunt, Robert McGaughey and Alan Rawls for helpful discussions and advice, Drs Randy Johnson and Shirley Tilghman for the Lfng mice, Carolyn Christel for help with statistical analysis and Dr Barbara Vanderhyden for the c-KIT probe. Confocal microscopy was done in the W. M. Keck Bioimaging Laboratory at ASU. This work was supported in part by a grant from the NIH to J.W.-R. Authors have no competing financial interests. All animal work was done in accordance with AALAAC standards.


  • Supplementary material

    Supplementary material for this article is available at http://dev.biologists.org/cgi/content/full/132/4/817/DC1

  • * Present address: Vincent Center for Reproductive Biology, Department of Obstetrics and Gynecology, Massachusetts General Hospital, Harvard Medical School, Room 6607, Building 149, 149 13th Street, Charlestown, MA 02129, USA

    • Accepted December 1, 2004.


View Abstract