Genomic Imprinting

Genomic imprinting is an epigenetic phenomenon in which a subset of autosomal genes is expressed monoallelically from one parental chromosome—either from the maternally inherited allele or from the paternally inherited allele—while the opposite allele is transcriptionally silenced. In classical Mendelian genetics, autosomal alleles function equivalently regardless of whether they are transmitted through the sperm or the oocyte (Mendel's law of equivalence of reciprocal crosses). Genomic imprinting represents a systematic departure from this rule: although a diploid zygote inherits two sequence-identical copies of an imprinted gene, sex-specific epigenetic marks—principally DNA methylation and chromatin remodeling—established during gametogenesis instruct the somatic transcriptional machinery which parental copy to transcribe and which to silence.

Restrictively distributed across the tree of life, canonical genomic imprinting has evolved independently in therian mammals (marsupials and eutherian placental mammals, where ~150 to 200 imprinted genes—representing ~1% of protein-coding loci—have been verified in mice and humans) and in flowering plants (angiosperms) (specifically within the triploid nutritive endosperm of developing seeds), as well as in certain scale insects. Because imprinted loci are functionally hemizygous, a single point mutation, deletion, or epigenetic defect on the active parental allele cannot be compensated by the silent wild-type allele on the homologous chromosome, rendering imprinted clusters hotspots for congenital human growth, metabolic, and neurodevelopmental syndromes.

Pronuclear Transplantation and the Barrier to Parthenogenesis

The non-equivalence of maternal and paternal mammalian genomes was demonstrated in 1984 through independent nuclear-transfer experiments conducted by Davor Solter (with James McGrath) and M. Azim Surani (with Sheila Barton). Using micromanipulation of fertilized 1-cell mouse zygotes before pronuclear fusion, the investigators constructed diploid mouse embryos with two identical chromosome complements (2*n* = 40) derived from a single parental sex:

• Gynogenetic (or parthenogenetic) embryos (2m:0p): Containing two maternal pronuclei, these embryos developed recognizable embryonic tissues up to the early somite stage (embryonic day E10.5) but exhibited stunted extraembryonic membranes, deficient trophoblast/placental development, and universal mid-gestation lethality.
• Androgenetic embryos (0m:2p): Containing two paternal pronuclei, these embryos exhibited markedly retarded embryonic proper structures alongside hyperplastic overgrowth of the extraembryonic trophoblast and yolk sac. In humans, an androgenetic conception (typically arising when an enucleated oocyte is fertilized by a haploid sperm that duplicates its genome, or by two sperm) produces a complete hydatidiform mole—a placental trophoblastic tumor devoid of fetal tissue.

These reciprocal phenotypes proved that viable mammalian development requires both a maternal and a paternal genome: paternal-specific gene expression promotes extraembryonic placental resource extraction, whereas maternal-specific gene expression prioritizes embryogenesis and restrains placental invasion.

Erasure, Establishment, and Somatic Maintenance

Unlike permanent DNA sequence mutations, genomic imprints must switch polarity whenever an allele passes through the germline of the opposite sex (for example, a paternal allele inherited by a daughter must be converted into a maternal imprint within her oocytes). This dynamic imprint cycle unfolds in three developmental phases:

1. Erasure in Primordial Germ Cells (PGCs): As embryonic PGCs migrate into the developing gonadal ridge (between embryonic days E10.5 and E13.5 in mice; weeks 5–9 in human embryos), genome-wide epigenetic reprogramming strips away all parental 5-methylcytosine (5mC) marks—including those at Imprinting Control Regions (ICRs)—via passive replication-coupled dilution and active oxidation to 5-hydroxymethylcytosine (5hmC) catalyzed by TET1 and TET2 dioxygenases.
2. Sex-Specific Establishment during Gametogenesis: De novo DNA methylation is re-established asynchronously by the methyltransferase DNMT3A complexed with its regulatory cofactor DNMT3L (and in rodents, DNMT3C). In the male germline, paternal imprints (only 3–4 known ICRs, such as H19/IGF2 and DLK1/DIO3, predominantly situated in intergenic regions) are methylated prenatally in prospermatogonia prior to meiosis. In the female germline, maternal imprints (>20 known ICRs, such as SNRPN, IGF2R, and KCNQ1OT1, typically overlapping intragenic CpG island promoters) are methylated postnatally in growing oocytes arrested in prophase I, guided by read-through transcription and H3K36me3 histone marks deposited by SETD2.
3. Post-Fertilization Resistance and Somatic Maintenance: Shortly after fertilization, the pre-implantation blastocyst undergoes a global wave of demethylation that erases >80% of gametic methylation across non-imprinted loci. Differentially methylated ICRs are selectively shielded from this erasure by the Kruppel-associated box (KRAB) zinc-finger protein ZFP57 (and ZNF445 in humans), which binds the methylated hexanucleotide motif TGCC^mGC and recruits the KAP1 (TRIM28) scaffold, heterochromatin protein 1 (HP1), histone methyltransferase SETDB1 (depositing H3K9me3), and maintenance DNA methyltransferase DNMT1, preserving monoallelic expression across all subsequent somatic cell divisions.

The Insulator Paradigm: IGF2 and H19 at Chromosome 11p15.5

Most imprinted genes reside in 100-kb to 3-Mb chromosomal clusters controlled in cis by a single Imprinting Control Region (ICR), primarily operating through either an insulator model or a long non-coding RNA (lncRNA) silencing model (such as the paternal antisense lncRNA Airn silencing maternal Igf2r, or KCNQ1OT1 silencing CDKN1C).

The archetype of the chromatin insulator mechanism is the IGF2–H19 domain on human chromosome 11p15.5. IGF2 encodes insulin-like growth factor 2 (a fetal mitogen expressed from the paternal chromosome), whereas H19 encodes a non-coding RNA situated ~90 kb downstream (expressed from the maternal chromosome). Both genes compete for shared downstream mesodermal and endodermal enhancers:

• On the Maternal Chromosome: The intergenic ICR (~2–4 kb upstream of H19) is unmethylated. The zinc-finger architectural protein CTCF binds to tandem target sites within the unmethylated ICR and interacts with cohesin to form a chromatin loop that acts as a topological insulator, blocking the downstream enhancers from contacting the IGF2 promoter and instead directing enhancer activity to transcribe maternal H19.
• On the Paternal Chromosome: The ICR is hypermethylated during spermatogenesis. CpG methylation sterically prevents CTCF binding and spreads across the H19 promoter, silencing paternal H19 and allowing the downstream enhancers to loop past the inactive ICR to drive robust transcription of paternal IGF2.

Disruptions of the 11p15.5 imprinted cluster cause two reciprocal congenital growth disorders: Beckwith–Wiedemann syndrome (BWS) (characterized by macrosomia, macroglossia, abdominal wall defects, and elevated risk of Wilms embryonal kidney tumor, resulting from biallelic IGF2 expression or paternal uniparental disomy) and Silver–Russell syndrome (SRS) (characterized by marked intrauterine and postnatal growth retardation, resulting from hypomethylation of the paternal H19/IGF2 ICR and loss of IGF2 expression).

Reciprocal Neurodevelopmental Syndromes at Chromosome 15q11–q13

The clinical consequences of parent-of-origin imprinting are vividly illustrated on human chromosome 15q11–q13, which houses a 2-Mb imprinted cluster containing paternally expressed genes (MKRN3, MAGEL2, NDN, SNRPN, and the SNORD116 snoRNA cluster) alongside the maternally expressed gene UBE3A (which encodes the E6-AP E3 ubiquitin-protein ligase and is imprinted specifically within cerebral cortex and cerebellar Purkinje neurons, where the paternal UBE3A allele is silenced in cis by a giant antisense transcript initiated from the unmethylated paternal SNRPN promoter).

Although an interstitial deletion of 4–6 Mb at 15q11–q13 involves the same chromosomal interval, the resulting clinical syndrome depends on the parental origin of the deleted chromosome:

• Prader–Willi Syndrome (PWS): Caused by loss of the paternally expressed 15q11–q13 genes (via ~70% paternal 15q11–q13 deletion, ~25–28% maternal uniparental disomy [UPD], or 1–3% imprinting center defect). Infants present with neonatal hypotonia and poor suckling, followed in early childhood by insatiable hyperphagia, morbid obesity, hypogonadism, and mild-to-moderate cognitive impairment.
• Angelman Syndrome (AS): Caused by loss of the maternally expressed UBE3A gene in neurons (via ~70% maternal 15q11–q13 deletion, ~10–15% loss-of-function point mutations in maternal UBE3A, ~3–5% paternal UPD, or ~3% imprinting center defect). Patients exhibit intellectual disability, absence of speech, ataxic gait with tremulous limb movements, microcephaly, seizures, and a characteristic behavioral phenotype of frequent paroxysmal laughter and smiling.

Because silencing one functional copy of an autosomal gene sacrifices the primary evolutionary advantage of diploidy—the masking of deleterious recessive mutations—evolutionary biologists long sought to explain why genomic imprinting evolved specifically in viviparous therian mammals and endosperm-bearing angiosperms, while remaining absent in oviparous (egg-laying) vertebrates such as birds, reptiles, amphibians, and monotremes.

Formulated by David Haig and colleagues in 1989–1991, the kinship theory of genomic imprinting (widely known as the parental conflict hypothesis) extends W. D. Hamilton's theory of inclusive fitness and kin selection to intragenomic asymmetry. Whenever females mate with more than one male across their reproductive lifetime (polyandry) and nourish offspring post-zygotically through an invasive placenta (or seed endosperm), a relatedness asymmetry arises between the maternal and paternal alleles inside a fetus:

• A maternally inherited allele in a fetus has a relatedness coefficient of r = 0.5 to every past, current, and future offspring of that mother. Natural selection on maternal alleles therefore favors conserving maternal metabolic resources to maximize total lifetime maternal reproductive output, promoting growth-suppressing genes (such as maternal IGF2R—a decoy receptor that binds and targets IGF2 to lysosomes for degradation—and maternal CDKN1C, a cyclin-dependent kinase inhibitor).
• A paternally inherited allele in that same fetus has r = 0.5 to full siblings, but r = 0 to maternal half-siblings sired by other males. Natural selection on paternal alleles therefore favors maximizing placental resource extraction from the mother for the current fetus even at the expense of future maternal litters, driving paternal-specific expression of fetal and placental growth enhancers (such as paternal IGF2, PEG1/MEST, and PEG3).

This evolutionary tug-of-war predicts that loss of paternally expressed genes causes fetal growth restriction (as seen in Silver–Russell syndrome and Igf2 knockout mice, which are born at 60% of normal birth weight), whereas loss of maternally expressed genes causes fetal overgrowth (as seen in Beckwith–Wiedemann syndrome and Igf2r knockouts).

What is uniparental disomy (UPD) and how does it cause imprinting diseases without any DNA sequence mutation?

Uniparental disomy (UPD) occurs when an individual inherits both copies of a specific chromosome pair from one parent and zero copies from the other parent—most often via trisomy rescue (loss of one chromosome from a trisomic zygote) or monosomy duplication. Even though the child possesses a normal 46-chromosome karyotype and zero DNA sequence mutations, inheriting two maternal copies of chromosome 15 (mUPD15) means neither copy expresses the paternally active SNRPN/SNORD116 cluster, resulting in Prader–Willi syndrome.

Why do egg-laying vertebrates such as birds and monotreme mammals lack genomic imprinting?

In oviparous animals—including birds, reptiles, and egg-laying monotreme mammals (the platypus and echidna)—the mother provisions the oocyte with a fixed quantity of yolk before fertilization occurs. Because the paternal genome enters at fertilization after resource allocation is already sealed inside a shelled egg, paternal alleles in the embryo cannot manipulate maternal nutrient transfer, eliminating the selective pressure described by Haig's parental conflict hypothesis.

How does genomic imprinting differ from X-inactivation?

In random X-inactivation, an entire sex chromosome (~850 genes) is silenced in XX somatic cells with a 50:50 probability of choosing the maternal or paternal X, creating a somatic mosaic of maternal-active and paternal-active clones. In genomic imprinting, silencing targets localized autosomal gene clusters (1–15 genes per cluster) in both males and females, and the choice is strictly deterministic by parent of origin in every cell (e.g., IGF2 is expressed only from the paternal allele in 100% of cells).

1. McGrath, J., & Solter, D. (1984). "Completion of mouse embryogenesis requires both the maternal and paternal genomes." Cell, 37(1), 179–183. https://doi.org/10.1016/0092-8674(84)90313-190313-1)
2. Surani, M. A. H., Barton, S. C., & Norris, M. L. (1984). "Development of reconstituted mouse eggs suggests imprinting of the genome during gametogenesis." Nature, 308(5959), 548–550. https://doi.org/10.1038/308548a0
3. Bartolomei, M. S., & Ferguson-Smith, A. C. (2011). "Mammalian genomic imprinting." Cold Spring Harbor Perspectives in Biology, 3(7), a002592. https://doi.org/10.1101/cshperspect.a002592
4. Moore, T., & Haig, D. (1991). "Genomic imprinting in mammalian development: A parental tug-of-war." Trends in Genetics, 7(2), 45–49. https://doi.org/10.1016/0168-9525(91)90230-N90230-N)