Downloaded from genesdev.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press Differences in DNA methylation durin.g oogenesis and spermatogenesis and their persistence during early embryogenesis in the mouse J.P. Sanford, 1,3 H.J. Clark, 2 V.M. Chapman, 1 and J. Rossant a ~Department of Molecular Biology, Roswell Park Memorial Institute, Buffalo, New York 14263 USA; 2Division of Molecular and Developmental Biology, Mount Sinai Hospital Research Institute and the Department of Medical Genetics, University of Toronto, Toronto, Ontario M5G 1X5, Canada We have examined the relative methylation levels of several dispersed repeated and low-copy-number gene sequences during gametogenesis and early embryogenesis. Southern blot analyses revealed that L1, intercisternal A particle (IAP), and major urinary protein (MUP) sequences were undermethylated extensively at Mspl sites in DNA from diplotene oocytes. In contrast, the same sequences were highly methylated in DNA from pachytene spermatocytes, round spermatids, and epididymal sperm. These results indicate that there are genome-wide DNA methylation differences between oogenesis and spermatogenesis. Repeated sequences in DNA from cleavage-stage embryos and inner cell masses (ICM) were methylated at intermediate levels, consistent with transient maintenance of gametic methylation levels during early embryogenesis. Gametic differences in DNA methylation observed here indicate that methylation could provide a mechanism for imprinting maternal and paternal genomes resulting in differential regulation of parental genomes during early development. [Key Words: Mouse; DNA methylation; gametogenesis; embryogenesis; genomic imprinting] Received August 28, 1987; revised version accepted October 6, 1987. Several lines of evidence suggest that maternal and pa- female genomes can be distinguished later in develop- ternal genomes do not play identical roles in mamma- ment. lian embryonic development. First, analysis of partheno- Genomic imprinting must involve epigenetic, heri- genetic development (Graham 1974)and nuclear trans- table differences between paternal and maternal pronu- plantation experiments has shown that embryos clef. Differential methylation of matemal and patemal containing two maternal or two paternal genomes die DNA could provide a mechanism for imprinting. DNA before or shortly after implantation (McGrath and Solter methylation at cytosine residues can be stably preserved 1984a; Surani et al. 1984), reflecting a requirement for through rounds of replication by maintenance methy- the presence of both maternal and paternal genomes lases {Razin and Riggs 1980)and has been postulated to during embryogenesis. Second, during normal develop- play a role in gene regulation {Cedar 1984}. Three recent ment, a paternal X chromosome is preferentially inacti- studies have provided strong supporting evidence that vated in extraembryonic tissues (Takagi and Sasaki DNA methylation could play a role in genomic im- 1975; West et al. 1977). Third, the hairpin tail mutation printing. These studies showed that the methylation (Thp) is lethal only when maternally inherited (McGrath pattems of exogenous DNA sequences in transgenic and Solter 1984b). Fourth, mice that receive both por- mouse lines sometimes varied according to whether the tions of certain chromosomes from one parent develop gene was inherited from the male or female parent {Reik abnormally (Cattanach and Kirk 1985; Searle and Bee- et al. 1987; Sapienza et al. 1987; Swain et al. 19871. In they 1985). None of these observations can be explained one study, expression of the transgene was also affected by a generalized delay in the onset of paternal gene ex- by the gamete of origin (Swain et al. 1987). Differences pression because gene activation from the zygotic in DNA methylation according to gamete of origin were genome occurs by the two-cell stage (Schultz 1986). detected in both mid-gestation (Reik et al. 1987)and Rather, these studies indicate that imprinting of the adult tissues (Reik et al. 1987; Sapienza et al. 1987; genome occurs during gametogenesis such that male and Swain et al. 1987) but were not directly compared be- tween oocytes and sperm. If DNA methylation were to act as the imprinting 3Present address: Department of Biological Sciences, State University of mechanism in mammalian development, one might ex- New York at Buffalo, Buffalo, New York 14260 USA. pect to find different patterns of DNA methylation in GENES & DEVELOPMENT 1:1039-1046 © 1987 by Cold Spring Harbor Laboratory ISSN 0890-9369/87 $1.00 1039 Downloaded from genesdev.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press Sanfotd et al. eggs and sperm and to observe perpetuation of these dif- ferences at least through early development. In fact, oocyte sperm liver little is known about the relative methylation levels of H M H M H M X kb endogenous gene sequences during oogenesis and sper- .,,, Q matogenesis. In a previous study, we noted that dis- persed repetitive L 1 sequences were undermethylated in DNA from fetal ovaries when compared with DNA from m - 23.1 mature sperm (Sanford et al. 1984). This comparison of somatic cell-contaminated tetraploid fetal oocytes and ~. -9.4 haploid sperm could not demonstrate conclusively, however, that extensive DNA methylation differences .~ -6.6 existed during comparable stages of male and female ga- metogenesis nor that such differences could persist into -4.4 embryogenesis. In the present study, we have compared om O I the methylation status of purified fetal oocytes with that of tetraploid spermatocytes and haploid spermatids. We have shown that several repeated and low copy se- - 2.3 quences that are heavily methylated at all stages of sper- - 2.0 matogenesis are undermethylated in the oocyte genome. Furthermore, we have shown that DNA methylation patterns in preimplantation embryos and inner cell masses (ICMs) are consistent with perpetuation of dif- ferences in gametic DNA methylation into early devel- opment. Results 1 2 3 4 5 6 Comparison of methylation levels of repeated and low copy sequences m fetal oocytes and mature sperm Figure 1. Methylation of L1 sequences in oocyte and sperm DNA. In each lane, 2 ~tg of genomic DNA, digested with HpaII (H, odd-numbered lanes} or MspI (M, even-numbered lanes), Tetraploid dictyate oocytes arrested in prophase I of was loaded. [Lanes 1,2)Oocyte; {lanes 3,4)sperm; {lanes 5,6) meiosis were dissected from 16- to 17-day female fe- liver. tuses. At this stage the number of oocytes recovered from the ovary is maximal because oocytes become This result indicates that these MspI sites are heavily closely associated with contaminating somatic follicle methylated in sperm. cells at later stages (Whittingham and Wood 1983). Oo- To determine whether hypomethylation of oocyte cyte samples used in this study were collected using a DNA was limited to highly repeated sequences, the procedure that gives a high yield of oocytes with min- HpaII digestion patterns of other DNA sequences were imum somatic cell contamination (DeFelici and examined. Intracistemal A particle {lAP)sequences rep- McLaren 1982). One hundred and fifty fetuses yielded 30 resent a retroviral-like class of sequence repeated ap- ~g of oocyte DNA. Cytological analysis of oocyte prepa- proximately 1000 times per haploid mouse genome rations indicated that only 1-5% of the cells were so- (Lueders and Kuff 1980). IAP transcripts are made in the matic (not shown). Using this pure population of oo- oocyte and early embryo but not in later development cytes, we reexamined the methylation at MspI restric- (Piko et al. 1984). We observed significant HpaII diges- tion sites in L1 sequences in oocyte DNA. The tion of MspI sites in IAP sequences in oocyte DNA and abundance of L1 fragments generated by HpaII digestion determined by densitometry that 75-100% of the major was comparable to that in the MspI digest (Fig. 1, lanes 1 MspI IAP sites were cleaved by HpalI (Fig. 2, lanes 1 and and 2). Densitometric analysis indicated that more than 2). In contrast, IAP MspI sites were not cleaved by HpaII 95% of the L1 MspI sites were digested by HpaII in oo- in sperm and liver DNAs. We also examined methyl- cyte DNA. These results indicate that the MspI sites ation of MspI sites in a tandemly repeated low-copy gene that generate the 3.5- and 5.0-kb fragments are unmeth- family of 35 genes encoding mouse urinary proteins ylated in nearly all of the estimated 30,000 copies of L1 (MUPs), which, unlike the lAP sequences, are not MspI sites present in the oocyte genome. Our previous known to be transcribed during oogenesis (Shaw et al. observation of a sizable proportion of methylated L1 se- 1983). Fragments recognized by the MUP cDNA clone quences in fetal oocytes (Sanford et al. 1984) is most represent both coding and intragenic noncoding regions likely due to extensive somatic cell contamination of in the MUP locus (Clark et al. 1984). We found no signif- the oocytes following the trypsin dissociation procedure icant differences between HpaII and MspI digestion pat- used previously. In contrast to the results obtained using terns for MUP sequences in oocyte DNA (Fig. 3, lanes 1 oocytes, L1 sequences were virtually undigested by and 2), and 90-100% of the MspI sites were recognized HpaII in DNA from sperm and somatic tissue (Fig. 1). by HpaII according to densitometric analysis. These re- 1040 GENES & DEVELOPMENT Downloaded from genesdev.cshlp.org on September 23, 2021 - Published by Cold Spring Harbor Laboratory Press DNA methylation in gametogenesis sults suggested that nearly all MUP MspI sites examined oocyte sperm liver were unmethylated in oocyte DNA. In contrast, HpaII H M H M H M digestion of sperm DNA revealed negligible digestion of MUP sequences. Methylation status of repeated and low-copy genes at - 23.1 different stages of spermatogenesis The results described in the previous section demon- -9.4 strated that oocyte DNA was undermethylated consis- tently when compared with DNA from mature sperm.