MARF1 Regulates Essential Oogenic Processes in Mice

See allHide authors and affiliations

Science  23 Mar 2012:
Vol. 335, Issue 6075, pp. 1496-1499
DOI: 10.1126/science.1214680


Development of fertilization-competent oocytes depends on integrated processes controlling meiosis, cytoplasmic development, and maintenance of genomic integrity. We show that meiosis arrest female 1 (MARF1) is required for these processes in mammalian oocytes. Mutations of Marf1 cause female infertility characterized by up-regulation of a cohort of transcripts, increased retrotransposon expression, defective cytoplasmic maturation, and meiotic arrest. Up-regulation of protein phosphatase 2 catalytic subunit (PPP2CB) is key to the meiotic arrest phenotype. Moreover, Iap and Line1 retrotransposon messenger RNAs are also up-regulated, and, concomitantly, DNA double-strand breaks are elevated in mutant oocytes. Therefore MARF1, by suppressing levels of specific transcripts, is an essential regulator of important oogenic processes leading to female fertility and the development of healthy offspring.

Oogenic processes essential for producing a “good” egg competent to support production of healthy offspring include accurate completion of meiosis (1), cytoplasmic maturational events that provide competence for fertilization and embryogenesis (1, 2), and maintenance of genomic integrity by protection against disruptive factors such as retrotransposon activation (3). Precise control of these processes is critical for successful reproduction. Abnormalities in any of these events can lead to infertility, miscarriage, and/or birth defects and endanger future generations. We report here the identification of a regulator in mouse oocytes that is required for all of these oogenic processes and acts by suppressing levels of specific transcripts.

By using N-ethyl-N-nitrosourea (ENU) mutagenesis, we generated a line of mutant mice with a female-only autosomal recessive infertility phenotype (fig. S1) characterized initially by oocyte meiotic arrest; the mutation was designated ENU375-18. Ovaries of the mutant mice appeared normal (fig. S2, A to C), but oocytes did not resume meiosis even after a superovulatory regimen of gonadotropins and were ovulated at the immature germinal vesicle (GV) stage (Fig. 1A and fig. S2D). In contrast, both wild-type (WT) and heterozygous (HET) oocytes were ovulated at mature metaphase II (MII) stage (Fig. 1A and fig. S2D). Therefore, the hallmark of infertility in mutant females is oocyte meiotic arrest at the GV stage.

Fig. 1

Meiotic arrest and loss of Marf1 expression in Marf1 mutant oocytes. (A) Number of oocytes (graph) and histology of cumulus-oocyte complexes (photographs) ovulated in oviducts by WT, HET, and mutant Marf1ENU females 14 hours after hCG injection. Pb1 indicates first polar body; scale bars, 50 μm; arrowheads, MII spindle and Pb1; Arrow, GV. (B) qRT-PCR of Marf1 exon 16 (left) and 5′- (right) mRNA expression in WT, HET, and mutant Marf1ENU FGOs. In this and subsequent figures, graphs show mean ± SEM. *P < 0.05 compared with WTs. Bars indicated with different letters (a, b, c) are significantly different, P < 0.05. (C) Western blot of MARF1 and beta actin (ACTB) protein in WT, HET, and mutant Marf1ENU FGOs (left); WT, HET (GT/+), and mutant Marf1GT (GT/GT) (middle); and Marf1GT/+ (GT/+), Marf1ENU/+ (ENU/+), and Marf1GT/ENU (GT/ENU) (right).

Positional cloning of the mutated gene revealed a G→T transition at the 5′-splice donor site of the 16th intron of the 4921513D23Rik gene (fig. S3, A to C). This mutation causes the skipping of exon 16 during pre-mRNA splicing (fig. S3, B and D), resulting in expression of negligible levels of exon 16 mRNA in mutant fully grown oocytes (FGOs) (Fig. 1 B) and a frameshift in the coding sequence with creation of a premature stop codon (fig. S3, B and E). The 4921513D23RIK protein was not detected in mutant FGOs by an antibody targeting its C terminus (Fig. 1C). Only a trace amount of 5′-4921513D23Rik mRNA was detected in mutant FGOs by quantitative reverse transcription polymerase chain reaction (qRT-PCR) (Fig. 1B). This may be caused by nonsense-mediated decay of the truncated 4921513D23Rik mRNA, suggesting that 4921513D23RikENU375-18 is unlikely to be a neomorph but rather an extreme hypomorph.

To verify that mutation of 4921513D23Rik is responsible for the ENU-induced phenotype, we produced mice carrying a gene-trapped allele of 4921513D23RikGt(AS0671)Wtsi (fig. S4, A to B) and mice with heteroallelic combination of 4921513D23RikGt(AS0671)Wtsi and 4921513D23RikENU375-18 (4921513D23RikGt(AS0671)Wtsi/ENU375-18). No 4921513D23RIK protein was detected in FGOs of 4921513D23RikGt(AS0671)Wtsi/Gt(AS0671)Wtsi or 4921513D23RikGt(AS0671)Wtsi/ENU375-18 mice (Fig. 1C), and both types of mice phenocopied 4921513D23Rik ENU375-18 /ENU375-18 (fig. S4, C to F). This noncomplementation of the two alleles indicates that the 4921513D23Rik ENU375-18 allele contains the causative mutation underlying the ENU-induced infertile phenotype. Hereafter, the 4921513D23Rik gene will be referred to as meiosis arrest female 1 (Marf1), the Marf1ENU375-18 allele as Marf1ENU, and the gene-trapped allele as Marf1GT.

Marf1 contains 27 exons that encode a 7765–base pair (bp) mRNA and 1736–amino acid (aa) peptide (fig. S5A). There is 86% amino acid identity between mouse and human MARF1. cDNA cloning and sequencing revealed that mouse oocytes express a unique variant of Marf1 different from the isoform expressed in cumulus cells. This oocyte-specific variant lacks 537-bp nucleotides at the 3′ end of exon 3 (fig. S5B). By sequence analysis, MARF1 has three major domains: an N-terminal LK-Nuc domain belonging to the 5′→3′ nuclease domain superfamily of proteins having ribonuclease (RNase) activity, two RRM domains, and a C-terminal tandem repeat of LOTUS or OST-HTH novel domains also present in Drosophila Oskar and mammalian tudor domain–containing proteins (TDRD) 5 and 7 (4, 5) (fig. S5C).

Marf1 mRNA is highly expressed in oocytes relative to other cell types (fig. S6A). There is ~60-fold higher expression of Marf1 mRNA detected in oocytes than in granulosa cells (Fig. 2, A and B). MARF1 protein is also expressed predominantly by oocytes and barely detectable in granulosa cells (Fig. 2B). This was further confirmed by a β-galactosidase (GAL) reporter assay in Marf1GT/GT ovaries, with positive β-GAL staining in oocytes of follicles at all developmental stages but not in other ovarian cell types (Fig. 2C and fig. S6B). Moreover, the mutant meiotic arrest defect is oocyte-autonomous. Marf1ENU/ENU oocytes, when developed in vivo within reaggregated chimeric follicles composed of Marf1ENU/+ somatic cells and Marf1ENU/ENU oocytes, displayed the meiotic arrest phenotype, whereas Marf1ENU/+ oocytes acquired meiotic resumption competency when grown in reaggregated ovarian follicles composed of Marf1ENU/ENU somatic cells and Marf1ENU/+ oocytes (table S1).

Fig. 2

Expression of Marf1 by mouse oocytes. (A) In situ hybridization of Marf1 mRNA in ovaries of 22-day-old B6SJLF1 mice 46 hours after equine CG (eCG) injection. Bright and dark field images are at the top and bottom, respectively. Arrowheads, oocytes; arrow, cumulus; and asterisks, mural granulosa cells (MGC). (B) qRT-PCR (left) and Western blot (right) analyses of Marf1 mRNA and protein levels in oocytes, cumulus cells (CC), and MGC of eCG-primed F1 mice. (C) X-gal staining of Marf1GT/GT ovaries from eCG (46 hours)-primed 22-day-old mice. Scale bars, 50 μm.

High levels of cyclic adenosine monophosphate (cAMP), produced in normal FGOs, maintain meiotic arrest (see fig. S7 for diagram illustrating meiotic control in FGOs). However, alleviation of cAMP inhibition, through pharmacological (fig. S8A) or genetic approaches (fig. S8, B and C), did not reverse meiotic arrest in Marf1ENU/ENU oocytes. Therefore, loss of MARF1 function affects processes downstream of relief from cAMP inhibition. No activation of oocyte maturation-promoting factor (MPF) was detected in Marf1ENU/ENU oocytes after administration of human chorionic gonadotropin (hCG) (fig. S9A), nor was there reduction of expression of key cell cycle regulators that activate MPF—Cdk1; Ccnb1; and Cdc25a, b, and c—in Marf1ENU/ENU FGOs (fig. S9B). However, microinjection of mRNA encoding an active form of CDC25B into Marf1ENU/ENU FGOs reversed the meiotic arrest, with ~56% of injected oocytes resuming meiosis and activated MPF (Fig. 3C), of which ~42% progressed to MII with normal appearing spindles (Fig. 3, A and B, and fig. S9C). Therefore, defects in oocytes caused by Marf1 mutation lie upstream of MPF activation.

Fig. 3

Meiotic reversal by CDC25B and alteration of transcriptomes in Marf1ENU/ENU oocytes. (A) Effect of microinjection of Cdc25b mRNA (Cdc25b) on meiotic progression (GVB and Pb1 emission) in Marf1ENU/ENUoocytes. (B) Marf1ENU/ENUoocytes matured after Cdc25b injection. (Left) Differential image contrast (DIC) image. Scale bar, 50 μm. (Right) Chromosomes (blue) and spindles (green) staining. Scale bar, 10 μm. (C) MPF and mitogen-activated protein kinase (MAPK) activity in Marf1ENU/ENUoocytes with or without Cdc25b mRNA microinjection (Cdc and Ctl group, respectively) or treated with or without 2.5 μM OA (Ctl and OA group, respectively). (D) Distribution of significantly changed transcripts [represented by Affymetrix (Affymetrix, Incorporated, Santa Clara, CA) probe sets] at various magnitudes of difference in expression levels between Marf1ENU/ENU (Mutant) and WT FGOs detected by microarray analysis. The number and the percentage of each group of transcripts in total changed transcripts are indicated above each bar. (E) Levels of hnRNA and mRNAs measured by qRT-PCR in WT and Marf1ENU/ENU growing oocytes.

To further determine the mechanisms of meiotic arrest in Marf1 mutant oocytes, we examined the Marf1ENU/ENU oocyte transcriptome by using microarray analysis. A cohort of transcripts was markedly elevated in Marf1ENU/ENU FGOs, with 377 transcripts expressed ≥4-fold higher than in WT oocytes and only 27 transcripts down-regulated to the same extent (Fig. 3D and tables S3 and S4). This increased transcript expression and/or stability is consistent with proposed RNase activity of MARF1. Posttranscriptional control is also indicated by a lack of difference in the levels of unprocessed heterogeneous nuclear RNA (hnRNA) in mutant oocytes, despite dramatic increases in mRNAs (Fig. 3E).

We did not detect elevated levels of Wee2 mRNA (fig. S9D); however, we found a profound, ~34-fold, up-regulation of Ppp2cb mRNA encoding the beta-isoform catalytic subunit of PPP2 [also known as protein phosphatase 2A (PP2A)] in Marf1ENU/ENU oocytes (Fig. 4A), and PPP2C protein was up-regulated ~50% in Marf1 mutant oocytes (fig. S9E). Brief treatment of Marf1 mutant oocytes with 2.5 μM okadaic acid (OA), an inhibitor of PPP2, was sufficient to reverse meiotic arrest in the majority (~70%) of Marf1 mutant oocytes (Fig. 4B) and induce MPF activation (Fig. 3C). Moreover, microinjection of Marf1ENU/ENU oocytes with Ppp2cb small interfering RNA (siRNA) knocked down Ppp2cb mRNA by ~77% (fig. S9F) and induced about 46% GV breakdown (GVB) (Fig. 4C). Microinjection with Ppp2cb morpholinos also induced GVB in ~43% of mutant oocytes (Fig. 4D). Therefore, up-regulation of Ppp2cb is a key to meiotic arrest in Marf1 mutant oocytes, although other factors are not excluded. Indeed, microarray analysis also revealed elevation of other transcripts in mutant oocytes that could be related to the meiotic arrest phenotype. For example, gene ontology (GO) analysis ( indicates that ADNP, ADORA2B, and IMPDH2 participate in cyclic nucleotide production; HORMAD1 and MLH3 are meiotic proteins involved in synaptonemal complex assembly and recombination; CCNE1, CDK9, CDK19, CDKN2B, CSPP1, HEXIM1, ID2, ID3, PSMG2, and RAD9 participate in regulation of cell cycle processes; and CDK9, OBFC2A, UBE2V2, and YY1 are involved in DNA repair. Moreover, among elevated transcripts there is a high representation of those encoding proteins that participate in regulation of cellular metabolism and gene expression (table S5). It is not known whether all of these transcripts are direct targets of regulation by MARF1 or whether some mRNA levels are abnormally high as a consequence of disruption of oogenic processes.

Fig. 4

Up-regulation of Ppp2cb, defective developmental competency, and dysregulation of retrotransposons and DNA DSBs in Marf1ENU/EMU oocytes. (A) Rpl19 and Ppp2cb mRNA expressed by WT and Marf1ENU/ENU FGOs. (B) Effect of OA on GVB in Marf1ENU/ENU FGOs. (C and D) Effect of microinjecting Marf1ENU/ENU FGOs with siRNAs (C) and morpholinos (MO) (D) on GVB. (E) Mature (induced by Cdc25b-injection) Marf1ENU/ENUoocytes 24 hours after in vitro fertilization. Scale bar, 50 μm. (F) Levels of retrotransposon mRNAs measured by qRT-PCR in WT and Marf1ENU/ENU FGOs. (G) Confocal microscopy of FGOs stained with anti-γH2AX (red) and 4′,6′-diamidino-2-phenylindole (DAPI, blue). Scale bars, 5 μm. Bar graph shows the number of oocytes with various numbers of foci in the nucleus.

Consistent with the up-regulation of a cohort of transcripts potentially affecting oocyte development, after in vitro insemination there was no cleavage or embryonic development of the Marf1 mutant oocytes induced to mature to metaphase II by microinjection with mRNA encoding an active form of CDC25B (Fig. 4E and fig. S10A). Therefore, defects in Marf1 mutant oocytes are not restricted to those that directly influence meiotic progression; they include those crucial for acquisition of competence to undergo fertilization and embryogenesis.

Steady-state levels of transcripts for intracisternal A particles (Iap) and long interspersed repetitive element (Line) 1 (but not other retrotransposons analyzed) were also significantly higher in Marf1ENU/ENU mutant oocytes than in WT (Fig. 4F). Retrotransposons exert deleterious effects on genomic integrity, in part because their dysregulated insertion into the genome produces DNA double-strand breaks (DSBs) (6). Excess DSBs can affect meiotic progression in oocytes (7), and up-regulation of retrotransposons is associated with increased nuclear DSBs and meiotic arrest in mouse spermatocytes (8, 9). Therefore, the effect of Marf1 mutations on nuclear DSBs was examined by using γ histone 2AX (also known as H2AFX) immunolabeling. The numbers of DSBs in both Marf1ENU/ENU and Marf1GT/GT mutant FGOs were significantly elevated compared with those of WT controls (Fig. 4G and fig. S10B). Increased DSBs may contribute to the meiotic arrest phenotype of Marf1 mutant oocytes, as does overexpression of PPP2CB. In somatic cells, DSBs trigger G2/M checkpoints that inhibit entry of cells into mitosis. This checkpoint activation requires PPP2 and results in inhibition of CDC25 and subsequently CDK1 (10). It is unknown whether a similar mechanism regulates meiotic progression in mammalian oocytes, because oocytes from mice carrying mutations that affect DSB repair usually die before follicular development (11). Although it is not clear whether the increase in nuclear DSBs directly affects meiosis or activates a checkpoint control mechanism, MARF1 is shown to be involved in establishing both retrotransposon mRNA levels and competence to resume meiosis in mammalian oocytes.

Together, these observations reveal a pivotal role for MARF1 in regulating oogenic processes essential for meiotic progression, genomic integrity, acquisition of developmental competence, and female fertility. The phenotype of meiotic arrest was a window through which other functions of this unique regulator were identified, and the mutant phenotypes are likely linked through up-regulation of mRNAs. Aberrant mRNA expression levels can be due to transcriptional or posttranscriptional control. Although MARF1 does not possess domains typical of transcription factors, it contains a predicted RNase domain. The fact that hnRNA primary transcript levels are not affected in Marf1 mutants, whereas their respective mRNAs are elevated provides support for misregulation of transcript processing and/or stability, suggesting involvement of MARF1 in RNA homeostasis. Indeed, posttranscriptional control of RNA levels is emerging as a regulatory mechanism of considerable importance in germ cell development (12), and its involvement in oocyte meiotic progression has been demonstrated by analyses of oocyte-specific conditional knockouts of Dicer1 (13, 14). Through its direct and indirect effects, MARF1 is pivotal in establishing the network of pathways essential for the development of a “good” egg.

Supporting Online Material

Materials and Methods

Figs. S1 to S10

Tables S1 to S5

References (1529)

References and Notes

  1. Acknowledgments: We thank M. Conti for providing the Cdc25b plasmid and Gpr3−/− mice and K. Wigglesworth and M. O’Brien for technical assistance. Supported by NIH grant HD42137 (Y.-Q.S., K.S., F.S., J.K.P., M.A.H., J.C.S., and J.J.E.) and Scientific Services by grant CA34196 from the National Cancer Institute. Microarray data are deposited in the Gene Expression Omnibus (, data set GSE 31985).
View Abstract

Stay Connected to Science

Navigate This Article