|
 The
phenomenon of genomic imprinting might never have been recognized but
for its effects on gene expression. As was discussed in the last column,
certain chromosomal subregions undergo imprinting. Genes in those regions
are expressed in a manner that depends on the parent of origin for the
chromosome. Some genes are thus expressed only if they lie on one chromosome,
even though a second normal copy of the gene is present on the chromosome
derived from the other parent. At some loci, it is the maternal copy of
the gene that is expressed, while at other loci it is the paternal copy.
This is in contrast to what happens for the vast majority of genes in
which both the maternal and paternal copies of the gene are expressed.
The
first indication that imprinting was occurring in mammals arose from nuclear
transplantation experiments carried out in the early 1980s. The nuclei
from fertilized mouse oocytes were removed and replaced with either a
pair of sperm-derived or a pair of oocyte-derived haploid nuclei to reconstitute
a diploid chromosome number. Despite the normal chromosome number, the
androgenetic (sperm-derived) and gynogenetic (oocyte-derived) embryos
did not survive embryogenesis. The embryos survive only when half the
chromosomes derived from the mother and half from the father.
A
further notable finding in these experiments was the distinctive growth
patterns of the embryos. Those derived from androgenic cells formed extraembryonic
tissue well but embryonic tissue poorly, whereas those of gynogenic origin
formed embryonic tissue well and extraembryonic tissue poorly. These experiments
demonstrated that simply having the correct number of chromosomes was
not sufficient for normal mammalian development. Contrary to nearly 100
years of dogma, the results showed that maternally and paternally derived
chromosomes were not functionally equivalent.
In
the wake of these surprising results, efforts focused on identifying the
specific genomic regions responsible for these phenotypic effects. Mice
were bred to generate offspring with 2 copies of a specific chromosomal
region derived from a single parent. The effects on embryonic viability
or gross morphology could be studied in this manner. A number of genomic
regions were thus identified to be undergoing imprinting. One copy or
the other was silenced and did not express proteins.
In
subsequent experiments, many of these regions have been found to contain
one or more genes that are indeed imprinted. At present, more than 30
imprinted genes have been identified in mammals. Importantly, there is
no common functional characteristic to these genes, although many are
required for normal development.
Genomic
imprinting is not merely a gene expression phenomenon, however. When an
imprinted locus on the paternal and maternal chromosome is compared, differences
other than gene expression are also found. Chemical alterations may occur
on the maternal copy of a gene and not the paternal gene. For example,
methylation of cytosines can vary between the 2 genes. Nuclease accessibility
to DNA can also be different, indicating differences in the way the DNA
is packaged as chromatin. The timing of DNA replication at these loci
has been found to differ on the paternal and maternal chromosomes. These
processes are interesting, as they reflect the epigenetic organization
of the DNA. Epigenetic (from epi- above, upon, higher than) processes
are those that determine how the DNA is regulated to produce useful patterns
of gene expression.
We
have already discussed in these columns how one form of gene regulation
occurs through the promoter regions. Certain nucleotide sequences are
present that either enhance, or repress, transcription of the gene. Regulation
of gene expression is also mediated by epigenetic processes such as chromatin
organization and cytosine methylation. It is not surprising, then, that
differences in these regulatory elements are found at imprinted loci.
The contribution of certain epigenetic regulators to imprinting is discussed
below. Taking into account that gene expression is regulated by its epigenetic
organization, we are able to define genomic imprinting as follows: If
the epigenetic organization of a locus is dependent on its gamete of origin,
then that locus is subject to genomic imprinting.
Mutations
of imprinted genes cause disease in unusual ways. A mutation of an imprinted
gene may have no obvious effect. This will happen if it is inherited on
the usually silenced chromosome. However, if the mutation is inherited
on the normally active chromosome, then the individual will be unable
to produce functional protein. The mutated gene is unable to produce functional
protein, while the other nonmutated but imprinted or silent gene is unable
to compensate. As the parental origin of the mutation determines whether
there is a phenotypic effect, the resulting family histories can be distinctive.
A
particularly interesting and unusual mechanism of disease-causing mutation
involving imprinted genes is that of uniparental disomy (UPD). UPD refers
to the inheritance of both copies of a chromosome (or chromosomal region)
from a single parent. The mechanism by which this occurs is presumed to
involve an initial trisomy that would be lethal if transmitted. The loss
of one of the trisomic chromosomes at random in totipotent cells may occur
early in embryogenesis and allows for the further development of the embryo.
However, if 2 paternal chromosomes and 1 maternal chromosome comprised
the initial trisomy and the maternal chromosome were to be lost, the resulting
fetus would have inherited 2 normal-appearing chromosomes, both of which
would be from the father. A similar mechanism would give rise to a maternal
UPD.
In
the situation of paternal UPD, difficulties would arise for those genes
that are normally expressed only if they lie on the maternal chromosome.
Although 2 normal copies of the gene are present, both are on paternally
derived chromosomes and in this scenario these genes are silenced. No
functional protein can be synthesized.
Maternal
UPD would conversely give rise to silencing of paternally expressed imprinted
genes on that chromosome. The translocation-bearing mice referred to earlier
were used to generate offspring with UPD of either parental origin for
chromosomal subregions, indicating that certain defined regions contain
imprinted genes. Cases of UPD in humans have been serendipitously identified,
allowing subregions of the human genome to be likewise defined as likely
or unlikely to contain imprinted genes.
Mutations
of single imprinted genes, as well as deletions or UPD for whole or parts
of chromosomes containing imprinted genes, have been found to give rise
to human diseases. Just as there is no common family of genes that are
imprinted, neither is there a typical category of human disease caused
by mutations involving imprinted genes. The broad categories of human
diseases involving imprinted genes include neoplasia, neurodevelopmental
disorders, metabolic disorders, dysmorphic conditions, and possibly psychiatric
disorders (for instance, various authors have proposed that Tourette¹s
disorder and autism may involve imprinted genes).
These
disorders are due to inherited mutations involving imprinted loci. A separate
category of disease involves the disruption of the gametic imprint in
somatic cells. In other words, the imprint is established normally in
gametes but is disrupted later in somatic cells, so that genes that were
expressed only from one chromosome either fail to be expressed from either
chromosome or are expressed from both chromosomes. The phenotypic consequence
of such an event is typically found to be neoplasia.
There
are several current models for the mechanism of genomic imprinting. The
first hypotheses arose from the observed differences of methylation between
imprinted loci on homologous chromosomes. This chemical modification of
the DNA (the addition of a methyl group to cytosine nucleotides) is reversible
and stable and was therefore an attractive candidate for being what was
³imprinted² in genomic imprinting. The disruption of imprinting in mice
that lacked the enzyme necessary to add methyl groups to DNA added support
to this idea.
However,
a number of more recent observations suggest that methylation, while necessary,
is not sufficient to govern imprinting in mammals. X inactivation, a process
that has many similarities to imprinting, has been studied extensively
in somatic cells during the period when the inactivation occurs. It has
been found that methylation was established only subsequent to inactivation,
and therefore it appeared to be involved in consolidating some other primary
process rather than being the primary process itself. Second, the discovery
that imprinting could occur in domains of hundreds of kilobases changed
the focus from processes such as methylation of promoter regions of individual
genes to much larger domains of DNA that affect multiple genes. Finally,
the methylation patterns seen at imprinted genes are rarely found to be
stably maintained throughout development, thus casting doubt on the notion
that methylation is the primary determinant of the epigenetic regulation
of an imprinted region.
A
second proposed mechanism for imprinting involves competition for regulators
of gene expression. This phenomenon has been found recently to occur for
2 imprinted regions, the H19/Igf2 region of mouse chromosome 7
and the Igf2r region of mouse chromosome 17. In each case, there
are 2 genes close to each other on the same chromosome. One of the genes
is expressed on one chromosome and is imprinted on the other. The second
gene is expressed only when it, and not the first, can use local enhancers.
Normally, what occurs is that on one of the chromosomes the first gene
is not imprinted and therefore is expressed and competes successfully
for use of the local enhancers. The second gene is therefore silenced.
On the other chromosome, however, the first gene is not expressed, allowing
the second gene to use the local enhancers and be expressed.
Deletion
of the primarily imprinted gene in each of the cases mentioned (either
by knockout or by deletion) leads to alteration in expression of the secondarily
imprinted gene, confirming this interrelationship. While this explains
the interrelationship of imprinting of these specific genes, it neither
explains how the first gene is imprinted to establish the process nor
explains how the many other nearby imprinted genes are regulated.
A
third model of imprinting is based on the imprint switch mutations observed
in Prader-Willi and Angelman syndrome patients. A male should derive half
of his chromosomes from his mother. When he passes them on to the next
generation, these same chromosomes must now know that they derive from
the paternal side. Normally, a male will ³switch² his maternally inherited
chromosomes to a paternal imprint in his gametes, with the converse for
females (Fig. 1).
Imprint
switch mutations have been found that interfere with this process and
prevent the establishment of the correct imprint in the gamete. Very small
deletions can lead to these imprint-switching problems, suggesting that
there are discrete gamete-responsive segments of DNA in imprinted regions
that interact with the nuclear environment to initiate the imprinting
process. While the mechanism by which such elements might work is not
yet known, the effect of this ³imprinting center² is to regulate chromatin
structure and methylation, resulting in imprinting of the adjacent hundreds
of kilobases of DNA. Such an idea represents a hierarchical model in which
the activity of an imprinting center dictates a number of biochemical
and structural changes, including methylation, resulting in the silencing
of genes in the immediate region.
It
is not known why imprinting occurs in mammals. Several explanations have
been proposed for its biological necessity. For instance, it has been
suggested that imprinting allows the mother to tolerate the fetal growth
without a catastrophic expense of energy. This theory is based on the
observation that genes promoting embryonic growth are frequently paternally
expressed, while genes that appear to be required to moderate fetal growth
are maternally expressed. As the first imprinted genes to be identified
fit this pattern, this theory carried a great deal of weight, but subsequently
a large number of imprinted genes were identified with less immediately
obvious links to embryonic growth. Moreover, an assumption underlying
this model is that genomic imprinting is purely a mammalian phenomenon,
as egg-laying species would not be subject to the same energy expenditure.
However, there is accumulating evidence that insects such as Drosophila,
Sciara, and Planococcus have an imprinting process with
similar effects on gene expression, but, being insects, they lay eggs
and do not carry fetuses.
The
simplest conclusion regarding the biological importance of genomic imprinting
is based on some of the earliest experimental observations: the male and
female genomes are rendered complementary to each other, and certain genes
are reduced in dosage by up to 50%. These observations alone suggest that
significant reasons for imprinting to occur in mammals include the prevention
of asexual reproduction and the repression of expression of certain dosage-sensitive
genes. The failure of asexual reproduction is clearly illustrated by the
experiments described earlier involving androgenetic and gynogenetic mouse
embryos, which are derived from one parent¹s gametes only. In addition,
the notion that imprinting is maintained in order to regulate levels of
gene expression is supported by the example of certain imprinted genes
that cause overgrowth and malformations if overexpressed.
It
is clear that imprinting plays an essential role in gene expression in
very diverse species. The discovery of imprinting helped explain patterns
of transmission of human diseases that had confounded generations of geneticists.
It was one of a handful of advances over the past 2 decades that challenged
long-held notions about how genetic traits are passed from one generation
to the next. The further clarification of the underlying mechanisms of
imprinting is likely to have an equally dramatic effect on our understanding
of how genes and large regions of the genome are regulated and how disruptions
in this process contribute to human disease.
top of page
|
