Download Natural selection and the function of genome imprinting:

Perspective
Origins of imprinting
Natural selection and
the function of genome
imprinting:
beyond the silenced minority
Most hypotheses of the evolutionary origin of genome imprinting assume that the biochemical character on
which natural selection has operated is the expression of the allele from only one parent at an affected locus.
We propose an alternative – that natural selection has operated on differences in the chromatin structure of
maternal and paternal chromosomes to facilitate pairing during meiosis and to maintain the distinction between
homologues during DNA repair and recombination in both meiotic and mitotic cells. Maintenance of differences
in chromatin structure in somatic cells can sometimes result in the transcription of only one allele at a locus.
This pattern of transcription might be selected, in some instances, for reasons that are unrelated to the original
establishment of the imprint. Differences in the chromatin structure of homologous chromosomes might
facilitate pairing and recombination during meiosis, but some such differences could also result in non-random
segregation of chromosomes, leading to parental-origin-dependent transmission ratio distortion. This
hypothesis unites two broad classes of parental origin effects under a single selective force and identifies a
single substrate through which Mendel’s first and second laws might be violated.
enome imprinting has come to be defined as the transcription of only one allele at a locus, dependent on
the parental origin of the allele1. The term is often
restricted further to describe a process that occurs only
in mammals1. Although the term was used first to
describe meiotic and mitotic chromosome segregation in
insects2, neither the casual reader of recent imprinting
literature, nor most investigators in the field, would take
issue with the notion that the true gist of the imprinting
process is parental-origin-dependent transcription in
mammals.
This consensus of opinion is surprising given the range
of biological phenomena and the phylogenetically diverse
collection of organisms in which epigenetic parental origin
effects occur (Table 1). Curiously, the exclusion of these
‘other’ parental origin effects (i.e. those that are nonmammalian or do not affect gene expression) from the
definition of imprinting has occurred in the absence of any
relevant data. It is as though one of two opposing armies,
realizing their superiority of number, decided to declare
victory and withdraw without ever engaging in battle.
Scientific decisions are not often made in this way, and it
is instructive to examine the circumstances surrounding
the consensus on the present use of the term.
G
0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(00)02134-X
The voice of gene silencing
The confluence of three factors probably led to the definition of imprinting as parental-origin-dependent transcription in mammals:
• the developmental failure of mouse embryos that are
gynogenetic (containing only maternal germline-derived
chromosomes) or androgenetic (containing only paternal
germline-derived chromosomes)3;
• the parental conflict hypothesis4;
• the demonstration of differences in DNA methylation
between maternal and paternal alleles at imprinted loci
(reviewed in Ref. 1).
This combination of a marked effect of the parental origin
of the genome on phenotype, a powerful argument for
how natural selection might exploit differences between
maternal and paternal genomes to result in parentalorigin-dependent gene expression, and the existence of a
strong candidate for the molecular identification mark by
which the transcriptional machinery might distinguish the
parental origin of an allele, drove most investigators
straight into the waiting arms of allele-specific DNA
methylation and RT–PCR. In retrospect, the field moved
so quickly in this direction that there was little serious
consideration of alternatives to the view that the primary
TIG December 2000, volume 16, No. 12
Fernando PardoManuel de Villena*
fernando@
unix.temple.edu
Elena de la CasaEsperón*
elena@
unix.temple.edu
Carmen Sapienza*†
sapienza@
unix.temple.edu
* Fels Institute for
Cancer Research and
Molecular Biology,
Temple University School
of Medicine, 3307 North
Broad Street,
Philadelphia, PA 19140,
USA.
†
Department of
Pathology and
Laboratory Medicine,
Temple University School
of Medicine,
Philadelphia, PA 19140,
USA.
573
Perspective
Origins of imprinting
TABLE 1. Phylogenetic distribution of parental origin effects
Time (Myr)
1000
750
500
250
Organism
Parental origin effect
Refs
Mouse
Androgenote and gynogenote death
X-inactivation
Gene expression
DDK syndrome
Level of aneuploidy in sperm (Rob translocations)
Transgene methylation
3
41
1,6,7
40
29
27
Human
X-inactivation
Gene expression
Transmission ratio distortion
DNA methylation
CML translocations
Recombination
42
1
12,13,37
28
38
31
Sheep
Hindquarters development (Callipyge)
39
Kangaroo
X-inactivation
43
0
Eutherians
Marsupials
Opossum
Gene expression
44
Birds
Chicken
(Monoallelic expression of IGF2 )
44,45
Fish
Zebrafish
Transgene methylation
46
Drosophila
Gene expression
Position effect variegation
Chromosome breakage
11
10
47
Arthropods
Fungi
Scale insects Chromosome elimination
48
Sciara
Chromosome elimination
2,48
Yeast
Mating type
49
Maize
Aleurone pigmentation
Demethylation and expression of zein genes
50
50
Arabidopsis
Genome activation during seed development
Maternal control of embryogenesis by MEDEA
50
50
Marsilea
Non-random distribution of chromatids
51
Flowering plants
Ferns
The list of parental origin effects is not intended to be complete but to illustrate the diversity of organisms in which they are observed. Examples fulfil the criteria that the effect is
epigenetic, parent-of-origin dependent, linked to the nuclear genome and does not include classical maternal or paternal effects. Divergence times (in millions of years before present; Myr)
are indicated on the horizontal axis. Arrows, times at which natural selection as a result of ‘parental conflict’ might have arisen (excluding postnatal parental conflict). The large arrowhead
indicates the time at which the selective force discussed in the text may have arisen. The Igf2 locus exhibits imprinted expression in mammals but is ‘monoallelic’ in chickens45.
purpose of establishing epigenetic differences between
maternal and paternal genomes is the transcriptional control of gene expression. The early identification of several
genes that were transcribed from only one parental
allele5–7 provided experimental support for this view but it
is important not to underestimate the role of the parental
conflict hypothesis in its general acceptance.
Limitations of the parental conflict hypothesis in
explaining parental origin effects
The parental conflict hypothesis4 states that natural selection acted upon epigenetic differences between maternal
and paternal genomes such that the expression of at least
some genes involved in embryonic growth became
restricted to only the maternal or only the paternal allele.
The most relevant requirements of the hypothesis, with
respect to control of embryonic growth, are an unequal
investment in parental care and multiple paternity4. If
postnatal parental care is excluded, these requirements are
valid in only some phylogenetic groups, including mammals. The hypothesis makes the important predictions
that imprinted genes that enhance growth will be
expressed from the paternal allele, and that imprinted
genes that suppress growth will be expressed from the
maternal allele, as is observed at the mouse Igf2 and Igf2r
574
TIG December 2000, volume 16, No. 12
loci, respectively5,6 (Ref. 8 provides a detailed discussion
of the parental conflict hypothesis and parental-origindependent gene expression). The process appears to guard
against the perceived threat of parthenogenesis9, providing
additional support for the idea that there is something
peculiarly ‘mammalian’ about imprinting.
This interplay between embryological and biochemical
data, on the one hand, and evolutionary theory, on the
other, made it easy for many investigators (including us)
to discount or ignore an important but unstated requirement of the parental conflict hypothesis: generation of the
substrate upon which natural selection operates (i.e. epigenetic differences between maternal and paternal genomes)
must be, like genetic differences that arise between
genomes by mutation, a basic property of the system. In
other words, natural selection does not create epigenetic
differences between maternal and paternal genomes any
more than natural selection creates mutations. However,
once epigenetic differences have occurred, natural selection can act on those differences and also on any genetic
factors that enhance or suppress those differences, just as
natural selection can act on mutations and also on genetic
factors that increase or decrease the rate of mutation.
It is at the level of selection of the genetic factors that
modify the establishment of epigenetic differences between
Perspective
Origins of imprinting
maternal and paternal genomes, either in cis or in trans,
that the ‘function’ of imprinting is determined. This statement stems from a general tenet of evolutionary biology –
but one that is important, in this context, to state explicitly – that is, to define the function of a structure or
process, one must identify the selective force that has
directed the formation of that structure or process.
By this logic, if some aspect of ‘parental conflict’ constitutes a selective force that is unique to mammals and
this force favours the silencing of one parental allele at
loci involved in embryonic growth, then there could be
features of transcriptional silencing that are peculiar to
mammals. However, it is important to stress that this
phenomenon is predicted to be limited to an unknown,
but probably small, fraction of loci in those phylogenetic
groups in which this selective force has operated.
Imprinting and ‘other’ parental origin effects: do
they have anything in common?
The observation that parental origin effects occur in a
wide variety of organisms (Table 1) cannot be used as an
argument that parental origin effects, per se, have been
selected to serve some purpose because the generation of
epigenetic differences between maternal and paternal
genomes might be a consequence of sexual reproduction.
However, we can infer that the opportunity for selection
to act on these differences has existed for much of evolutionary history (Table 1). If a particular type of epigenetic
difference appears consistently, it suggests that the difference has persisted as a result of natural selection. The pertinent questions are what selective forces might act on
differences between maternal and paternal genomes and
whether any of these forces could result in many or all of
the parental origin effects observed.
Epigenetic parental origin effects fall into two general
classes: first, those that differentially affect the expression
of a particular gene or phenotype (violations of Mendel’s
first law; see glossary in Box 1), and second, those that
differentially affect the transmission of particular alleles
(violations of Mendel’s second law). The first class of
effects is mediated through transcriptional silencing or
activation of only one allele, and the second class might be
mediated through postmeiotic selection on the expression
or segregation of alleles, or through meiotic selection that
results in non-random segregation of chromosomes.
In considering selective pressures that might be common to all phylogenetic groups and all epigenetic parental
origin effects, it is reasonable to ask first whether these
two seemingly disparate types of effects, one a property of
somatic cells and the other a property of the germline,
need to be related in any way. If the two classes of effects
cannot be related in any mechanistic way, then they might
be subject to different selective forces, which might operate at completely different levels. However, if the two
classes reflect alternative effects of a common structure or
pathway, then it is possible that the same selective force
might act at both levels, through a common substrate.
In insects, both classes of effect are clearly related to
the establishment of epigenetic differences in chromatin
structure. There are many examples, but the most historically relevant concerns the original use of the term
‘imprinting’ in genetics: Crouse2 demonstrated that preferential segregation in Sciara of the maternal X chromosome to a functional meiotic product, rather than a
nonfunctional meiotic product, is associated with a
heterochromatic domain on the maternal X chromosome.
Epigenetic differences in chromatin structure also influence gene expression in the somatic cells of insects. All of
the variegating position effects that are sensitive to
parental origin in Drosophila10 are associated with
translocations that result in the juxtaposition of heterochromatic regions with euchromatic regions. Parentalorigin-dependent variability in phenotype is observed as
variation in mosaic expression of the affected allele and
many modifiers of position effect variegation encode proteins that influence chromatin structure. In fact, Lloyd et
al.11 have demonstrated a parental-origin-dependent gene
expression phenomenon in Drosophila that is indistinguishable from imprinting in mammals except for the lack
of detectable DNA methylation differences. These authors
propose that imprinting is an ancient and conserved mechanism of gene silencing based on the establishment of differences in chromatin structure.
Data implicating epigenetic differences in chromatin
structure in both types of parental origin effects in mammals are highly suggestive. Transmission ratio distortion
of maternal alleles at X chromosome loci in the human is
sensitive to the parental origin of the X chromosome in the
mother12, as well as the previous inactivation status of the
X chromosome (i.e. a heterochromatic versus a euchromatic state)13. And, of course, maternal and paternal alleles at loci that exhibit transcriptional imprinting are
known to exhibit parental-origin-dependent differences in
chromatin structure (reviewed in Ref. 14), as assayed
through sensitivity to DNase I or trichostatin A, replication timing and allele-specific binding of the chromatin
‘insulator’ protein CTCF (Refs 15, 16). Additional data
indicating that differences in chromatin structure, per se,
might directly affect meiotic chromosome segregation
come from the study of maternal meiotic drive of a variant
mouse chromosome 1 containing a large, homogeneously
staining region17 and ‘knob’ heterochromatin in maize18;
these differences in chromatin structure are not dependent
on parental origin, unlike the case in Sciara.
A common selective force in meiotic and mitotic
cells
We assume that parental origin effects that result in both
differential expression (a somatic cell phenomenon) and
differential transmission (a germline phenomenon) of
genes are mediated by differences in the chromatin structure of homologous chromosomes. If these differences in
chromatin structure are present in both somatic and
germline cells, then any biochemical process that occurs in
both the soma and the germline, and for which a distinction between maternal and paternal genomes is important,
could exert a selective force.
We propose that natural selection caused the establishment and maintenance of epigenetic differences between
the maternal and paternal genomes of sexually reproducing organisms for the purpose of homologous pairing of
chromosomes at meiosis and the associated processes of
DNA repair and recombination during both meiosis and
mitosis. In meiotic cells, homologous pairing and repair of
DNA double-strand breaks has been selected to take place
between maternal and paternal chromosomes (meiotic
recombination), rather than sister chromatids19. Unless
these processes occur between homologues, segregation
of homologous chromosomes is adversely affected20. In
TIG December 2000, volume 16, No. 12
575
Perspective
Origins of imprinting
BOX 1. Glossary
Knob heterochromatin: Knobs are cytological features of maize chromosomes that can influence chromosome segregation. They are heterochromatic and consist of thousands to millions of 180- and 350-basepair repeats.
Microsatellite instability: A phenomenon in which errors made during the
replication of mono- and di-nucleotide repeats are not corrected. The process
has been associated with defective DNA mismatch repair. Microsatellite
instability is a quantitative character detected as increases or decreases in
the length of alleles at some portion of tested loci.
Mendel’s 1st law: Although not formally defined by Mendel, the laws of
segregation and independent assortment require that the genetic material is
stable between generations and functionally equivalent when inherited
either maternally or paternally. Historically, new mutations and rare meiotic
recombination events within genes have violated this law, and imprinting
provides a third exception.
Mendel’s 2nd law: Each pair of homologous chromosomes segregate at
meiosis in each generation, ensuring maintenance of proper chromosome
number in sexually reproducing organisms and resulting in equal probability
of transmitting either allele at any locus.
Replication asynchrony: Differential timing of the replication of the two
alleles at a locus during S phase of mitosis. The phenomenon is commonly
observed for alleles at loci on the active versus inactive X chromosome in
females but has also been reported for alleles at imprinted loci.
Gene conversion: The phenomenon was first described in fungi as a
recombination process in which the transfer of information between homologues
is non-reciprocal. The process results in tetrads in which the segregation of
alleles at a locus is not 1:1.
Position effect variegation: Mosaic expression of a phenotype caused by a
chromosomal rearrangement in which the new euchromatic–heterochromatic
chromosomal boundary influences the expression of adjacent genes.
Transmission ratio distortion: A significant departure from the Mendelian
inheritance ratio expected, regardless of the origin of the distortion.
Uniparental disomy: The abnormal inheritance of both members of a homologous chromosome pair from only one parent in a chromosomally balanced
individual. Both uniparental isodisomy (two copies of the same chromosome)
and uniparental heterodisomy (both copies of a homologous pair from the
same parent) have been observed.
mitotic cells, repair of DNA double-strand breaks and
other types of DNA damage, and any associated mitotic
recombination, has been selected to take place between
sister chromatids rather than between maternal and paternal homologous chromosomes19. Potentially large-scale
functional hemizygosity as a result of gene conversion or
mitotic recombination is thus avoided.
The selective pressure to distinguish maternal and paternal homologous chromosomes from each other and from
sister chromatids during DNA repair and recombination is
universal and occurs in both the soma and the germline of
all organisms that reproduce sexually. The selective pressure to maintain this distinction will be least pronounced in
organisms such as yeasts that have only a transient diploid
stage whose purpose is the creation of the ‘prevalent’, haploid, form. However, the selective pressure to maintain epigenetic differences between maternal and paternal genomes
will be strong in organisms in which there are a large number of mitotic divisions between the time of fertilization and
576
TIG December 2000, volume 16, No. 12
the establishment of the germline. In such cases, the identity
of maternal and paternal homologous chromosomes must
be maintained from the time of fertilization until meiotic
recombination takes place. This would explain a number
of unexpected observations surrounding the analysis of
parental origin effects (discussed below).
Observations on imprinting in mammals that are
not predicted by natural selection at the level of
gene silencing
Most studies of imprinted genes in mammals have focused
on the description of patterns of gene expression and
the identification of factors responsible for parentalorigin-dependent transcription of only one allele1,4–7.
Interestingly, when such studies extended beyond the
analysis of transcription, several observations were made
that are not predicted by models in which the role of
imprinting is to control gene expression. These observations provide support for our hypothesis that imprinting
was established as a general mechanism for distinguishing
homologous chromosomes from each other, from nonhomologous chromosomes and from sister chromatids.
The following examples illustrate this point.
Somatic cells
The loss of epigenetic differences between alleles in
somatic cells has been found to affect relationships
between alleles and between homolgous chromosomes
that were, themselves, unexpected. For example, pairing
of homologous chromosomes is not thought to be a common occurrence in the somatic cells of mammals.
However, analysis of the position of alleles within interphase nuclei by fluorescence in situ hybridization indicates
that homologous chromosomes do associate in the vicinity
of the imprinted loci for Prader–Willi and Angelman syndromes on human chromosome 15 and the H19 region of
human chromosome 11 (Ref. 21). This association
appears to be mediated by differential epigenetic marking
of these regions (rather than sequence homology) because
the association does not occur in Prader–Willi patients in
which both chromosomes 15 are maternally derived. The
proximity of maternal and paternal alleles is restricted to
late in S phase, when only one allele of the asynchronously
replicating pair22 is likely to be undergoing DNA synthesis. The presumed concomitant ‘loss’ of imprinting (or
failure to inherit a paternal chromosome, in this case),
allelic pairing and replication asynchrony suggests that the
maintenance of epigenetic differences is required for this
previously unrecognized association between homologues
at the time of DNA replication. This is consistent with a
role for imprinting in distinguishing homologues during
post-replication repair. Most importantly, no hypothesis
in which imprinting evolved to control gene expression
predicts a physical association between alleles at
imprinted loci. However, if imprinting evolved as a mechanism of chromosome recognition and distinction, then
such interactions are not unexpected.
A more direct indication that there is a relationship
between imprinting and DNA repair comes from the association between ‘microsatellite instability’ (see Ref. 23 for
review) and ‘loss of imprinting’ in colon tumours24. If loss
of imprinting occurs in the precursor cell of a colon
tumour (as appears to be often the case24), homologous
chromosomes could become indistinguishable from each
other and/or from sister chromatids. If the cell then
Perspective
Origins of imprinting
(c)
(b)
(a)
Mat
Mat
Pat
Female germline
Recombination
—
Pat
+
Male germline
Recombination
+
—
Centromeres
Telomeres
trends in Genetics
attempts to correct a substantial replication error (such as
the expansion or contraction of a microsatellite array) and
chooses the homologous chromosome as a template,
rather than the sister chromatid, the DNA repair machinery would probably signal an apoptotic response. This
would be caused by an apparently high level of DNA
replication errors25 perceived by the DNA repair machinery by comparison of homologous chromosomes rather
than sister chromatids. (The estimated DNA sequence
diversity (expected heterozygosity across all sites) in the
human population is 0.2% (Ref. 26). This difference
between homologues is orders-of-magnitude greater than
the normal error rate of DNA synthesis.) Thus there will
be selection for cells that have inactivated the relevant
repair/apoptotic signalling pathways and these cells might
also give rise to tumours.
These unexpected observations are most simply interpreted if the establishment and maintenance of epigenetic
differences between homologous chromosomes in somatic
cells reflects a selective pressure that is uncoupled from
transcriptional control. Many early observations surrounding parental-origin-dependent methylation of transgene loci in the mouse could also be similarly interpreted:
the pattern of differences in methylation between mater-
nally and paternally derived transgene arrays is similar in
almost all cell types and tissues examined, regardless of
whether the transgene was expressed in a specific tissue,
in multiple tissues or not at all (reviewed in Ref. 27).
Some epigenetic differences between alleles at endogenous
loci could also be interpreted this way; the mannose6-phosphate/insulin like growth factor 2 receptor (IGF2R),
which is expressed from only the maternal allele in mice, is
expressed biallelically in most humans. However, differential methylation of maternal and paternal alleles is maintained in both human and mouse28.
Meiotic cells
The second type of unexpected observation surrounding
imprinting indicates a requirement for the maintenance
(rather than the erasure) of differential marking of chromosomes in the germline well beyond the point at which
one might expect such differences to have been erased if
imprinting functions only to silence genes. In some cases,
differences between maternal and paternal chromosomes
appear to be maintained until the onset of the first meiotic division. For example, there are significant differences in the frequency of aneuploid sperm found among
carrier males when the same robertsonian translocation
TIG December 2000, volume 16, No. 12
577
Perspective
Origins of imprinting
chromosome is inherited maternally as opposed to paternally29.
Perhaps the most unexpected meiotic effect reported30 is
that methylation differences between maternal and paternal
H19 alleles are distinguishable far into meiotic prophase
during spermatogenesis. If epigenetic differences between
maternal and paternal homologues are required to facilitate
homologous pairing (Fig. 1 and discussion below) and
DNA repair associated with meiotic recombination, then
the epigenetic differences between homologues must survive
until at least the zygotene stage of meiosis (when pairing of
homologous chromosomes is observed). Consistent with
this expectation, Davis et al.30 observed differences in DNA
methylation between maternal and paternal H19 alleles at
the preleptotene stage (when chromosomes begin to condense) and found that such differences begin to disappear at
the pachytene stage (after pairing has occurred and when
crossing over begins).
Additional support for the hypothesis that parental-origin-specific chromatin structures facilitate meiotic pairing
of homologous chromosomes comes from the observation
that imprinted regions of human chromosomes display
sex-specific recombination frequencies31. Moreover, these
sex-specific differences in recombination depend on the
parental identity of the imprint in subregions of the
imprinted domains. Regions containing genes expressed
from the paternal chromosome recombine at higher frequencies through males, whereas regions containing genes
expressed from the maternal chromosome show higher
recombination rates through females (Fig. 1). It seems
improbable that expression of imprinted genes is both
parental-origin-specific and sex-specific during the first
meiotic prophase. Therefore it is unreasonable to relate
these differences in recombination to transcriptional
states. Such differences might be expected, however, if
these regions are intimately involved in meiotic pairing.
Indirect support for the maintenance of epigenetic differences during meiosis is also provided by maternal transmission ratio distortion at loci that are either known to be
transcriptionally imprinted32 or show very strong parental
origin effects on phenotype33. The asymmetry of female
meiosis provides a unique opportunity for the direct selection of epigenetic differences in chromosome structure as a
result of preferential segregation of one homologue to the
polar body and the other to the ovum (female meiotic
drive). Transmission ratio distortion in this instance in the
mouse is a consequence of preferential segregation of chromosomes at meiosis33, but this is not proven in humans.
Parental origin effects: a unifying view
Given the antiquity of epigenetic differences between
maternal and paternal genomes (Table 1), it is worth
recalling that the raison d’être of the sexual mode of
reproduction is the meiotic pairing and recombination of
homologous chromosomes. Although meiotic pairing
occurs between homologous chromosomes, initiation of
pairing is thought not to be mediated directly by homologous DNA sequence but by epigenetic factors34. These
factors have not been identified, but a complementary
References
1 Tilghman, S.M. (1999) The sins of the fathers and mothers:
genomic imprinting in mammalian development. Cell 96,
185–193
2 Crouse, H.V. (1960) The controlling element in sex chromosome
578
code of maternal and paternal imprints has the potential
to fulfil this role. A conceptually simple system is possible,
in which oppositely imprinted domains are templates by
which homologous chromosomes might recognize each
other and, equally importantly, discriminate against pairing of non-homologous chromosomes (Fig. 1). This mechanism explains why imprinted loci are non-randomly distributed on chromosomes and why oppositely imprinted
genes are found in clusters35; multiple imprinted domains
will be required to provide specificity of pairing and pairing will be initiated in specific regions of chromosomes.
Once pairing and recombination are complete, the
maternal or paternal ‘pairing code’ (imprint) can then be
established on both homologues and transmitted to the next
generation. The maintenance or additional modification of
these codes in somatic cells can lead to variable amounts of
allelic silencing36, as well as distinguish homologues from
sister chromatids during DNA repair. However, the maintenance of these differences could also engender a tendency
for transient pairing of homologues during mitosis, occasionally resulting in mitotic recombination.
There are a number of potentially interesting implications of this hypothesis. For example, the observation
of parental-origin-dependent transmission ratio distortion
during the mapping of complex genetic diseases has generally been interpreted to imply that a transcriptionally
imprinted gene is involved in the etiology of the disease37.
However, it is also possible that these parental origin
effects reflect biases in meiotic chromosome segregation,
rather than transcription of an allele. In the same vein, if
some non-homologous chromosomes carry complementary codes of sufficient similarity, specific translocations
could exhibit a parental origin bias, even though the genes
involved in the translocation might not be transcriptionally
imprinted38. Another interesting possibility with respect to
human disease etiology, is that reactivation of DNA repair
pathways in tumours that show microsatellite instability
and loss of imprinting24, mighty result in an apoptotic
response. The hypothesis presented here also predicts that
uniparental disomy (for any chromosome) will lead to
alterations in meiotic recombination as well as increased
aneuploidy of the uniparentally disomic chromosome.
Although we agree that ‘parental conflict’ might have
been an important selective force in the evolution of
transcriptional control, we believe that the selective force
that has maintained most epigenetic differences between
maternal and paternal genomes, throughout evolution,
has not operated at the level of gene silencing in somatic
cells. Consideration of ‘imprinting effects’ from this viewpoint could provide new insights into several phenomena,
including mechanisms of chromosome pairing and segregation34, sex determination2 and the unusual genetics of
some parental origin effects39,40.
Acknowledgements
We thank M. Bartolomei, M. Hansen and K. Latham for
their comments on an earlier version of this manuscript.
We apologize to the authors of many relevant papers that
could not be cited because of space limitations.
behavior in Sciara. Genetics 45, 1429–1443
3 McGrath, J. and Solter, D. (1984) Completion of mouse
embryogenesis requires both the maternal and paternal
genomes. Cell 37, 179–183
4 Moore, T. and Haig, D. (1991) Genomic imprinting in
TIG December 2000, volume 16, No. 12
mammalian development: a parental tug-of-war. Trends
Genet. 7, 45–49
5 DeChiara, T.M. et al. (1991) Parental imprinting of the mouse
insulin-like growth factor II gene. Cell 64, 849–859
6 Barlow, D.P. et al. (1991) The mouse insulin-like growth factor
Perspective
Origins of imprinting
7
8
9
10
11
12
13
14
15
16
17
18
19
20
type-2 receptor is imprinted and closely linked to the Tme
locus. Nature 349, 84–87
Bartolomei, M.S. et al. (1991) Parental imprinting of the mouse
H19 gene. Nature 351, 153–155
Hurst, L.D. and McVean, G.T. (2000) Do we understand the
evolution of genome imprinting? Curr. Opin. Genet. Dev. 8,
701–708
Varmuza, S. and Mann, M. (1994) Genomic imprinting –
defusing the ovarian time bomb. Trends Genet. 10, 118–123
Spofford, J.B. (1961) Parental control of position-effect
variegation. II. Effect of sex of parent contributing white
mottled rearrangement in Drosophila melanogaster. Genetics
46, 1151–1167
Lloyd, V.K. et al. (1999) Genomic imprinting and position-effect
variegation in Drosophila melanogaster. Genetics 151,
1503–1516
Naumova, A.K. et al. (1998) Parental origin-dependent, male
offspring-specific transmission-ratio distortion at loci on the
human X chromosome. Am. J. Hum. Genet. 62, 1493–1499
Naumova, A.K. et al. (1995) Transmission-ratio distortion of X
chromosomes among male offspring of females with skewed
X-inactivation. Dev. Genet. 17, 198–205
Wolffe A.P. (1995) Epigenetic inheritance: the chromatin
connection. In Genomic imprinting. Causes and consequences.
(Ohlsson, R. et al., eds.), pp. 49–70, Cambridge University
Press
Bell, A.C. and Felsenfeld, G. (2000) Methylation of a CTCFdependent boundary controls imprinted expression of the Igf2
gene. Nature 405, 482–485
Hark, A.T. et al. (2000) CTCF mediates methylation-sensitive
enhancer-blocking activity at the H19/Igf2 locus. Nature 405,
486–489
Agulnik, S.I. et al. (1990) Meiotic drive in female mice
heterozygous for the HSR inserts on chromosome 1. Genet. Res.
55, 97–100
Peacock, W.J. et al. (1981) Highly repeated DNA sequence
limited to knob heterochromatin in maize. Proc. Natl. Acad. Sci.
U. S. A. 78, 4490–4494
Haber J.E. (2000) Recombination: a frank view of exchanges
and vice versa. Curr. Opin. Cell Biol. 12, 286–292
Koehler, K.E. and Hassold, T.J. (1998) Human aneuploidy:
lessons from achiasmate segregation in Drosophila
melanogaster. Ann. Hum. Genet. 62, 467–479
21 LaSalle, J.M. and Lalande, M. (1996) Homologous association
of oppositely imprinted chromosomal domains. Science 272,
725–728
22 Kitsberg, D. et al. (1993) Allele-specific replication timing of
imprinted gene regions. Nature 364, 459–463
23 Lengauer, C. et al. (1998) Genetic instabilities in human
cancers. Nature 396, 643–649
24 Cui, H. et al. (1998) Loss of imprinting in normal tissue of
colorectal cancer patients with microsatellite instability. Nat.
Med. 4, 1276–1280
25 Fishel, R. (1999) Signaling mismatch repair in cancer. Nat.
Med. 5, 1239–1241
26 Nickerson, D. A. et al. (1998) DNA sequence diversity in a 9.7kb region of the human lipase gene. Nat. Genet. 19, 233–240
27 Sapienza, C. et al. (1989) Epigenetic and genetic factors affect
transgene methylation imprinting. Development 107, 165–168
28 Riesewijk, A.M. et al. (1996) Maternal-specific methylation of
the human IGF2R gene is not accompanied by allele-specific
transcription. Genomics 31, 158–166
29 Baulch, J.E. et al. (1996) Evidence for a parent-of-origin effect
on sperm aneuploidy in mice carrying robertsonian
translocations as analysed by fluorescence in situ
hybridization. Mutat. Res. 372, 269–278
30 Davis, T.L. et al. (1999). Acquisition of the H19 methylation
imprint occurs differentially on the parental alleles during
spermatogenesis. Genomics 58, 18–28
31 Paldi, A. et al. (1995) Imprinted chromosomal regions of the
human genome display sex-specific meiotic recombination
frequencies. Current Biol. 5, 1030–1035
32 Bennett, S.T. et al. (1997) Insulin VNTR allele-specific effect in
type 1 diabetes depends on identity of untransmitted paternal
allele. The IMDIAB Group. Nat. Genet. 17, 350–352
33 Pardo-Manuel de Villena, F. et al. (2000) A genetic test to
determine the origin of maternal transmission ratio distortion.
Meiotic drive at the mouse Om locus. Genetics 154, 333–342
34 Sybenga, J. (1999) What makes chromosomes find each other
in meiosis? A review and an hypothesis. Chromosoma 108,
209–219
35 Beechey, C.V. (1999) Appendix: Imprinted genes and regions in
mouse and human. Results Probl. Cell Differ. 25, 303–323
36 Lee, M.P. et al. (1997) Human KVLQT1 gene shows tissue-
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
specific imprinting and encompasses Beckwith–Wiedemann
syndrome chromosomal rearrangements. Nat. Genet. 15,
181–185
Paterson, A.D. et al. (1999) The analysis of parental origin of
alleles may detect susceptibility loci for complex disorders.
Hum. Hered. 49, 197–204
Haas, O.A. et al. (1992) Parental origin of chromosomes
involved in the translocation t(9;22). Nature 359, 414–416
Cockett, N.E. et al. (1996) Polar overdominance at the ovine
Callipyge locus. Science 273, 236–238
Wakasugi, N. (1974) A genetically determined incompatibility
system between spermatozoa and eggs leading to embryonic
death in mice. J. Reprod. Fertil. 41, 85–96
Takagi, N. and Sasaki, M. (1975) Preferential inactivation of
the paternally derived X chromosome in the extraembryonic
membranes in the mouse. Nature 256, 640–642
Harrison, K.B. (1989) X-chromosome inactivation in the human
cytotrophoblast. Cytogenet. Cell Genet. 52, 37–42
Sharman, G.B. (1971) Late DNA replication in the paternally
derived X chromosome of female kangaroos. Nature 230,
231–232
O’Neill, M.J. et al. (2000) Allelic expression of IGF2 in
marsupials and birds. Dev. Genes Evol. 210, 18–20
Koski, L.B. et al. (2000) Monoallelic transcription of the
insulin-like growth factor-II gene (Igf2) in chick embryos. Mol.
Reprod. Dev. 56, 345–352
Martin, C.C. and McGowan, R. (1995) Parent-of-origin effects
on the methylation of a transgene in the zebrafish, Danio rerio.
Dev. Genet. 17, 233–239
Levitan, M. and Verdonck, M. (1986) 25 years of a unique
chromosome-breakage system. I. Principal features and
comparison to other systems. Mutat. Res. 161, 135–142
Herrick, G. and Seger, J. (1999) Imprinting and paternal genome
elimination in insects. Results Probl. Cell Differ. 25, 41–71
Nakayama, J. et al. (2000) A chromodomain protein, Swi6,
performs imprinting functions in fission yeast during mitosis
and meiosis. Cell 101, 307–311
Messing, J. and Grossniklauss, U. (1999) Genomic imprinting
in plants. Results Probl. Cell Differ. 25, 23–40
Tourte, Y. et al. (1980) Different behaviour of maternal and
paternal genomes during embryogenesis in the fern, Marsilea.
Eur. J. Cell Biol. 21, 28–36
Hanno Bolz is a clinical genetics counsellor and research assistant in the Faculty of Medicine, University of Hamburg, Germany.
TIG December 2000, volume 16, No. 12
579