News & Views | Published:

Genomic imprinting

Piece of cake

Knocking out a minor form of the Igf2 messenger RNA from the placenta in mice has surprisingly strong effects on nutrient transport to the fetus. This has implications for the theory of maternal–paternal genetic conflict.

About 2,500 years ago, Anaxagoras (500–428 bc) was the first, or so the record indicates, to perceive that embryos receive nutriment through the navel. (Some of his contemporaries had bizarre notions, including the idea that the embryo is nourished through its whole body, like a sponge.) Soon afterwards, Diogenes of Apollonia (460–400 bc) described some of the anatomical details of the placenta — although this Latin word, which literally means 'cake' and comes from the Greek 'plakous' (flat mass), had not been coined. In the second century ad, the Roman statesman Cato the Elder left a recipe for a popular pie called placenta ('pizza' evolved much later). It is unclear whether the simile went from bakers to anatomists or the other way around, but the fact is that the placenta started receiving serious attention after the beginning of modern embryology in the late nineteenth century. Its surprising secrets continue to be revealed today, as exemplified by a paper from Constância and co-workers1 on page 945 of this issue.

Although the details of placental structure (Fig. 1) differ between species, there is enough similarity to make data from mouse models relevant to humans2. In mice, the placental layer closest to the fetus is called the labyrinth — a network of mingled fetal and maternal blood vessels, separated by a layer of so-called labyrinthine trophoblast cells. This architecture leaves little doubt that the labyrinth is crucial to the exchange of nutrients, gases and waste between maternal and fetal blood. On the maternal side of the labyrinth is a poorly understood region called the junctional zone, composed of large spongiotrophoblast cells and another type of trophoblast cell, the glycogen cell. Next to the junctional zone are maternal decidual cells, which line the uterus. A discontinuous layer of trophoblast 'giant' cells contacts the decidual cells.

Figure 1: Cross-section of the mouse placenta.

Sp, spongiotrophoblast; Gly, glycogen cells; GC, trophoblast giant cells. Fetal blood enters via the large central blood vessel. In humans, the labyrinth is known as the fetal placenta and the junctional zone is called the basal plate. Drawing modified from ref. 16.

One determinant of the size of the placenta is the Igf2 gene, which encodes the insulin-like growth factor II (IGF-II) protein. Its significance for placental growth was recognized some time ago, from gene-knockout studies in mice: the lack of IGF-II resulted in small placentas3. Igf2 is expressed in the entire mouse placenta — in both spongiotrophoblast and labyrinth — and in the corresponding regions in humans. But in mice (not humans) one of the control regions of Igf2, dubbed the P0 promoter, controls labyrinth-specific expression to produce a messenger RNA copy with two small non-protein-coding regions ('mini-exons'), as well as the usual coding regions4. Previously5, Constância et al. described mutant mice that lacked 5,000 bases spanning the second mini-exon, eliminating the P0-derived mRNA; mutant fetuses showed retarded growth.

Constância et al.1 have now studied the same mutation in more detail. They find that, although the P0 version normally makes up only 10% or so of the total placental Igf2 mRNAs, its absence from mutants compromises the size of the entire placenta as well as the final size of the fetus. The total amount of Igf2 mRNA is somewhat reduced in the labyrinth but not in the spongiotrophoblast. And yet all placental regions — including, amazingly, the maternal component — are proportionately reduced in size. But there is more: the placental size at the end of gestation is the same as in mice that lack Igf2 entirely. This striking observation is difficult to interpret, unless one invokes the idea of differential translation of the various Igf2 mRNAs into protein6. This issue has yet to be addressed.

As with most genes, mammals have two copies (alleles) of Igf2, one from the mother and one from the father. But Igf2 belongs to the small club of 'imprinted' genes — those that are conditioned during egg or sperm development7 so that only one allele is expressed in offspring. Igf2 is imprinted in such a way that the paternal copy is expressed and the maternal allele is silenced. Given this fact, and evidence that parental-specific imprinting evolved in vertebrates along with the placenta8, the observations of Constância et al.1 may be hinting at something profound. But what?

One hypothesis that explains the co-evolution of imprinting and the placenta incorporates the idea that there is conflict between maternal and paternal alleles. This theory predicts that, in species in which the mother, but not the father, makes a major investment in offspring, and in which multiple males sire progeny by the same female, maternal alleles will have a selective advantage if they act to limit the resources allocated to any one conceptus. In contrast, paternal alleles will gain an advantage by commandeering as many maternal resources as possible9. In simple terms, then, paternally expressed imprinted genes, such as Igf2, should promote fetal growth, and maternally expressed genes should inhibit fetal or placental growth. Data from a growing number of imprinted genes10, including some that are unrelated to Igf2 signalling11, meet these predictions. But the theory has serious difficulties with falsifiability7, aggravated by confusing anthropomorphisms, unbecoming militaristic metaphors ("battle of the sexes" and the like), and the vexing fact that some imprinted genes do not control growth12.

The work by Constância et al.1 not only addresses maternal–paternal conflict but also has some truly exciting implications for fetal physiology. One of the most novel aspects of the authors' work is that they took the bull by the horns and measured the passive permeability and active transport of nutrients across the mutant placentas. Using radiolabelled tracers, they showed a decrease in passive, but an increase in active, transport early in development, when normalized to placental weight. But this compensatory attempt by the stunted placentas to increase the provision of nutrients to the embryos failed. Net delivery of an amino-acid analogue, not adjusted for placental weight, was below normal.

These results imply that the paternally expressed Igf2 acts in the placenta to channel maternal resources to the fetus; when it is mutated, other mechanisms try, but fail, to compensate. So this gene seems to score a home run for the conflict hypothesis (at least for the 10% of Igf2 mRNA that is affected; how the other 90% fits into the picture is unclear). Still, given that Igf2 was one of the two genes that motivated the hypothesis in the first place9, there is more than a little circularity in this argument. Evidence from genetic manipulations of other imprinted genes will be grist for the mill.

Do these results relate to human health? A low birth weight correlates with health problems both soon after birth and, more perniciously, in later life13. The mechanism behind the late complications is still wanting, but Smith et al.14, as they report on page 916, have taken a step towards a simpler objective: to predict low birth weight by a practical screen. They measured levels of the pregnancy-associated plasma protein A (PAPP-A) in maternal blood in a large human population, early in pregnancy. Mothers with very low PAPP-A levels were more likely to have small babies. As PAPP-A is expressed in early trophoblast tissue, these data imply that the health and abundance of the early trophoblast correlates with birth weight. Might these data directly implicate the Igf2 pathway? That remains to be seen. PAPP-A cleaves IGF-II-binding proteins in an IGF-II-dependent manner, but the function of these proteins in vivo remains uncertain. In fact, single — and some double — knockouts of the genes encoding these proteins in mice had no startling effects on growth (ref. 3 and J. Pintar, personal communication).

Normally, mice consume the placenta after giving birth (yes, they have their cake and eat it too). Like deletion of Igf2, and consistent with conflict, knocking out another paternally expressed imprinted gene, Peg1, causes fetal and placental growth retardation15. But a finding apparently unrelated to conflict is that Peg1-lacking mothers also show a behavioural quirk — they no longer eat the placenta. Food for thought.


  1. 1

    Constância, M. et al. Nature 417, 945–948 (2002).

  2. 2

    Georgiades, P., Ferguson-Smith, A. C. & Burton, G. J. Placenta 23, 3–19 (2002).

  3. 3

    Efstratiadis, A. Int. J. Dev. Biol. 42, 955–976 (1998).

  4. 4

    Moore, T. et al. Proc. Natl Acad. Sci. USA 94, 12509–12514 (1997).

  5. 5

    Constância, M. et al. Nature Genet. 26, 203–206 (2000).

  6. 6

    Nielson, F. C. et al. J. Biol. Chem. 265, 13431–13434 (1990).

  7. 7

    Solter, D. Int. J. Dev. Biol. 42, 951–954 (1998).

  8. 8

    John, R. M. & Surani, M. A. Cell 101, 585–588 (2000).

  9. 9

    Haig, D. & Graham, C. Cell 64, 1045–1046 (1991).

  10. 10

  11. 11

    Frank, D. et al. Proc. Natl Acad. Sci. USA 99, 7490–7495 (2002).

  12. 12

    Hurst, L. D. & McVean, G. T. Curr. Opin. Genet. Dev. 8, 701–708 (1998).

  13. 13

    Gluckman, P. D. & Harding, J. E. Horm. Res. 48, 11–16 (1997).

  14. 14

    Smith, G. C. S. et al. Nature 417, 916 (2002).

  15. 15

    Lefebvre, L. et al. Nature Genet. 20, 163–169 (1998).

  16. 16

    Sapin, V. et al. Dev. Biol. 191, 29–39 (1997).

Download references

Author information

Correspondence to Benjamin Tycko or Argiris Efstratiadis.

Rights and permissions

Reprints and Permissions

About this article

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.