Immaculate misconception

In mammals, mother and father make an equal genetic, but an unequal 'epigenetic', contribution to offspring. Studies of humans and mice with no maternal epigenetic contribution reveal more about this asymmetry.

William Harvey, an anatomist and personal physician to two kings of England, was taking a gamble when he proposed in 1651 that 'Ex ovo omnia' — 'everything comes from an egg'. It wasn't until much later that the mammalian egg, or oocyte, was first detected and Harvey was proved right: a whole organism can develop from this remarkable cell. But what he did not know is that an input from sperm is also essential for mammalian oocytes to fulfil this potential. This is, of course, why sex is necessary and 'virgin' births are impossible in humans. The need for sperm lies in the fact that all genes are not equal. A fertilized egg contains two copies of every gene, one from sperm and one from oocyte. Usually, both copies are expressed in relevant cells in the embryo. But some genes are specifically labelled ('imprinted') so that only the maternal or paternal copy is active. So an embryo receives an equal genetic contribution from its parents, but the parental genomes are not functionally equivalent, explaining why both are needed. Indeed, embryos with two male or two female genomes cannot develop normally.

Three new papers1,2,3 provide another graphic demonstration of the importance of these unequal 'epigenetic' contributions from sperm and oocyte. On page 539 of this issue1, Judson and colleagues describe a unique case in which a mutation seems to prevent human oocytes from acquiring any maternal-specific imprints at all. The mutant eggs fail to develop even after fertilization. Meanwhile, Bourc'his et al.2 (writing late last year in Science) and Hata et al.3 (in a report in Development) have described a remarkably similar outcome in mice with a mutation in the Dnmt3L gene, which is required for imprinting in mouse oocytes.

Nearly 50 mammalian imprinted genes have been identified so far, and their significance for human development is clear from Judson et al.'s study1 of a woman in whom conceptuses (very early embryos) developing from normally fertilized eggs invariably died soon after they implanted into the wall of the uterus. Judson et al. analysed one such conceptus, and found that it had a normal genetic contribution from sperm and oocyte. The problem, it turns out, was with the epigenetic status of the oocyte: all the maternally inherited epigenetic marks (recognized as methyl groups on cytosine–guanine base pairs in particular DNA sequences associated with imprinted genes) were missing from the conceptus. Paternally inherited epigenetic marks were normal.

The physical characteristics of the conceptus were also revealing — they were similar to those that occasionally arise spontaneously after the embryonic loss of the maternal genome, leaving only the paternal genome. Why should a conceptus that lacks maternal imprints develop as if it lacks a maternal genome entirely? The reason is that, during the development of primordial germ cells, from which oocytes and sperm will form, all imprints are first erased (Fig. 1). New parental-specific imprints are introduced subsequently when sperm and oocytes begin to mature4, with oocytes receiving most known imprints5. So a maternal genome that lacks maternal imprints functions more like a paternal genome in the fertilized egg — although not entirely4,6. Although relatively sparse, the epigenetic marks in sperm are essential, as a conceptus with two maternal genomes also fails to develop.

Figure 1: Developing imprinting.

Imprinted genes are those that are 'marked' in a parental-specific way. a, As primordial germ cells enter the genital ridge of developing female and male embryos, all imprinting marks are erased10. b, Bottom, as female germ cells grow and develop into oocytes, the Dnmt3L gene, possibly with Dnmt3a and/or Dnmt3b, is involved in reintroducing maternal-specific imprints2,3. Top, these genes are also expressed during sperm development but how they function in paternal imprinting is unknown. c, After fertilization, a zygote is formed with both genetic and epigenetic (imprinting) contributions from the parental genomes. GV oocyte, germinal-vesicle oocyte.

How, then, are imprints laid down in developing oocytes? Here the studies by Bourc'his et al.2 and Hata et al.3 are important, as they show that the Dnmt3L gene is crucial to this process in mice. Bourc'his et al.2 engineered animals with a targeted mutation in both copies of Dnmt3L. The mutation had no immediate effect on development: the original, 'founder' males and females matured to adulthood, although the males did not produce sperm and were sterile.

The females lacking functional Dnmt3L protein did produce oocytes. But, after fertilization of these oocytes by normal sperm, conceptuses failed to develop after implantation, and showed several embryonic and placental abnormalities. These defects were clearly of maternal origin, as the embryos lacked all maternally inherited epigenetic modifications but, of course, not those from sperm2. Some of the abnormalities, for example in the placenta, were similar to those seen in conceptuses with paternal genomes only, or with genomes that lack all parental imprints4,6. Furthermore, because of the lack of maternal imprints, several imprinted genes were regulated abnormally, with some being repressed; this probably contributed to the observed abnormalities2.

So it seems that Dnmt3L is crucial in laying down imprints in the female germ line, at least in mice. But in the human conceptus studied by Judson et al.1, there was no identifiable mutation in the human DNMT3L gene (D. Bonthron, personal communication). Mutations in another gene must account for their observations.

Two possibilities are the DNMT3A and DNMT3B genes, which share similarities with DNMT3L. The encoded proteins are two DNA methyltransferase enzymes, which can initiate de novo DNA methylation2,7,8. Indeed, they might be required for DNMT3L to work, as DNMT3L does not encode the catalytic domain required for DNA methylation. Consistent with this idea, Hata et al.3 show that the mouse Dnmt3L protein can interact with these enzymes, and that the absence of Dnmt3a, like the lack of Dnmt3L, results in the loss of maternal imprints in mice. But the methyltransferases cannot themselves detect specific DNA sequences that require de novo methylation, so some other modification might be needed first. This might involve the methylation of histone proteins — part of the packaging that enables DNA to be squeezed into the nucleus. For example, an interaction between methylated histone H3 protein and another packaging protein, HP1β, can lead to de novo DNA methylation9.

Whatever the answers, the new papers1,2,3 stress the importance of the epigenetic asymmetry between parental genomes for normal mammalian development10 (Fig. 2). Even a slight deviation in the epigenetic marks on some chromosomal regions, such as on chromosomes 11 or 15, can result in devastating human diseases such as Beckwith–Wiedemann and Prader–Willi/Angelman syndromes11. Moreover, the cloning of mice and other mammals is possible because the donor nuclei, derived from adult tissues, possess the appropriate parental imprints10.

Figure 2: The importance of imprinting.

a, Epigenetic asymmetry between paternally and maternally derived genomes, as shown here, is essential for normal development. b, The absence of the Dnmt3L and probably Dnmt3a genes in developing mouse oocytes prevents maternal-specific imprinting2,3. After fertilization, the zygote contains a maternal genome with no imprints and a normal paternal genome; it cannot develop normally. Judson et al.1 have described a similar outcome in a human patient, although the genetic basis is unclear. c–f, These combinations of genomes in different epigenetic states are also incompatible with full development4.

Why imprinting exists at all remains enigmatic. There are signs that, during evolution, a war broke out between the sexes that left the maternal genome with most of the DNA-methylation marks associated with imprinted genes5. Even now we observe the preferential demethylation of the paternal genome after fertilization; the maternal genome remains largely unaffected12. This suggests that maternally inherited proteins in the oocyte are used to demethylate, and so regulate the function of, the paternal genome after fertilization. Nevertheless, even the relatively few epigenetic marks that have survived in the paternal genome are crucial for development and hence a significant barrier to virgin birth. So for the time being men are indispensable. But with the increasing pace of research, even this barrier might be breached in the future. I for one will be following the unfolding story with fascination — and apprehension.


  1. 1

    Judson, H., Hayward, B. E., Sheridan, E. & Bonthron, D. T. Nature 416, 539–542 (2002).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Bourc'his, D., Xu, G.-L, Lin, C.-S, Bollman, B. & Bestor, T. H. Science 294, 2536–2539 (2001).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Hata, H., Okano, M., Lei, H. & Li, E. Development (in the press).

  4. 4

    Obata, Y. et al. Development 125, 1553–1560 (1998).

    CAS  PubMed  Google Scholar 

  5. 5

    Reik, W. & Walter, J. Nature Genet. 27, 255–256 (2001).

    CAS  Article  Google Scholar 

  6. 6

    Kato, Y. et al. Development 126, 1823–1832 (1999).

    CAS  PubMed  Google Scholar 

  7. 7

    Aapola, U. et al. Genomics 65, 293–298 (2000).

    CAS  Article  Google Scholar 

  8. 8

    Okano, M., Bell, D. W., Haber, D. A. & Li, E. Cell 99, 247–257 (1999).

    CAS  Article  Google Scholar 

  9. 9

    Tamaru, H. & Selker, E. U. Nature 414, 277–283 (2001).

    ADS  CAS  Article  Google Scholar 

  10. 10

    Surani, M. A. Nature 414, 122–128 (2001).

    ADS  CAS  Article  Google Scholar 

  11. 11

    Nicholls, R. D. et al. J. Clin. Invest. 105, 423–418 (2000).

    Article  Google Scholar 

  12. 12

    Mayer, W. et al. Nature 403, 501–502 (2000).

    ADS  CAS  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to M. A. Surani.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Surani, M. Immaculate misconception. Nature 416, 491–493 (2002).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Sign up for the Nature Briefing newsletter for a daily update on COVID-19 science.
Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing