Genomic Imprinting

Why genomic imprinting matters

• Mendelian genetics is not the whole story. Classical recessive inheritance assumes the two alleles at a locus are functionally interchangeable. Imprinting refutes this for ~150 mammalian loci where the parental origin of an allele dictates whether it speaks at all. A paternal-deletion at 15q11-q13 produces Prader-Willi while a maternal-deletion at the same DNA stretch produces Angelman — same DNA, opposite phenotypes.
• Mammalian uniparental development fails. The 1984 Surani/McGrath/Solter experiments showed gynogenetic mouse embryos (two maternal pronuclei) develop fetuses with poor placentas, while androgenotes (two paternal pronuclei) develop hyperdeveloped placentas with stunted fetuses. Mammals categorically cannot reproduce parthenogenetically, in contrast to many fish, reptiles, and invertebrates.
• Imprinting-disorder syndromes affect ~1 in 5,000 births collectively. Prader-Willi, Angelman, Beckwith-Wiedemann (~1 in 10,000), Silver-Russell (~1 in 50,000), and transient neonatal diabetes mellitus (~1 in 400,000) all arise from imprinting errors. Diagnostic methylation-specific PCR or methylation arrays distinguish deletion from uniparental disomy from imprinting defect, with implications for recurrence risk.
• Cancer co-opts imprinting. Loss of imprinting at IGF2 (LOI — both alleles expressed) is detected in ~10–40% of Wilms' tumor, colorectal, breast, ovarian, and esophageal cancers. Andy Feinberg's group reported in 2003 that LOI of IGF2 in normal colon mucosa is associated with ~5x elevated risk of personal and familial colorectal cancer, suggesting imprinting status as a cancer-risk biomarker.
• Assisted reproduction modestly elevates risk. Multiple cohort studies report ~3–6x elevated relative risk of Beckwith-Wiedemann in IVF/ICSI conceptions, attributed to embryo-culture or ICSI disturbance of the imprint-establishment window. Absolute risks remain small (~1 in 4,000 vs background ~1 in 14,000) but the effect is reproducible across registries.
• Cloning efficiency is bottlenecked by imprinting. ~50–80% of somatic-cell nuclear transfer mammals show large-offspring syndrome or placental abnormalities traceable to dysregulated H19/IGF2 and other imprinted loci. This is why Dolly-the-sheep-style cloning has remained at single-digit success rates three decades after the proof of concept.
• Window into mammalian evolution. Imprinting is largely confined to therian mammals (placentals + marsupials) and absent from monotremes, birds, reptiles, and invertebrates. The phylogenetic distribution coincides with placental nourishment of offspring, supporting Haig's kinship theory that placentation creates the parent-of-origin conflict that selects for imprinting.

Common misconceptions

• Imprinting silences the gene completely. Many imprinted genes show partial monoallelic expression — 80/20 rather than 100/0 — and the bias can be tissue-specific. UBE3A is biallelic in most tissues but maternal-only in mature neurons, which is why Angelman is primarily a neurological disorder.
• The mark is on histones, not DNA. The primary inherited mark at almost all imprinting control regions is DNA methylation at CpG dinucleotides, deposited by DNMT3A/DNMT3L in the germline. Histone modifications and noncoding RNAs (like KCNQ1OT1, AIRN) are downstream effectors, not the primary inherited tag.
• Parent-of-origin and X-inactivation are the same kind of silencing. X-inactivation chooses one X at random per cell, producing mosaics; imprinting consistently silences the same parental copy in every cell. The mechanisms differ — XIST coats one chromosome in cis; imprinting uses locus-specific methylation at ICRs.
• You inherit your imprints from your parents and pass them unchanged. Imprints are erased in primordial germ cells and re-established according to the sex of the new germline. A man whose mother carried a methylated H19 ICR will, in his sperm, deposit unmethylated H19 ICRs (the paternal pattern). Imprints are reset every generation.
• Only 150 genes are affected, so it can't matter much. Many imprinted genes are master regulators of growth, metabolism, and behavior — IGF2, IGF2R, GNAS, MEST, PEG3, NESP55. Their dysregulation maps to obesity, diabetes, autism, and cancer pathways, so the per-gene impact is disproportionately large.
• Imprinted means inherited as a Mendelian trait. Imprinting disorders show non-Mendelian inheritance precisely because phenotype depends on which parent transmitted the allele. A paternal SNRPN mutation produces Prader-Willi only when transmitted by the father; the same mutation in the mother is silent because the maternal allele was already off.

How imprints are written, read, and erased

The imprint is a DNA methylation mark on cytosines within CpG-rich imprinting control regions (ICRs) of typically 1–2 kb. ICRs are categorized as either maternally methylated (about 21 known in mouse) or paternally methylated (about 4 known) — the asymmetry mirrors the fact that mammalian sperm undergo prenatal de novo methylation while oocytes accumulate methylation during postnatal growth. DNMT3A together with the catalytically inactive cofactor DNMT3L lays down the methylation in the germline, with sex-specific transcription factors guiding it to the right ICRs.

Once fertilization occurs, the embryo's preimplantation genome undergoes a wave of global demethylation, but ICRs are protected — by ZFP57 in conjunction with TRIM28/KAP1 and the maintenance methylase DNMT1 — preserving the parental asymmetry through somatic divisions. The reading machinery then differs by locus: at the H19/IGF2 cluster, the maternal-only unmethylated ICR1 binds the insulator protein CTCF, blocking enhancer access to IGF2 and routing the enhancers to H19 instead; on the paternal copy, methylation prevents CTCF binding so enhancers reach IGF2 and silence H19. At the KCNQ1OT1 cluster on chromosome 11p15, the methylated maternal ICR2 silences the long noncoding RNA KCNQ1OT1, allowing nearby genes (CDKN1C, KCNQ1, SLC22A18) to be maternally expressed; on the paternal copy, KCNQ1OT1 is transcribed and silences those neighbors in cis.

Erasure happens in primordial germ cells migrating to the gonads (mouse E8.5–E13.5). DNA methylation across the genome drops to ~5%, including ICRs. Re-establishment then follows the new germline's sex: in the male, prenatal DNMT3A/DNMT3L deposit paternal-pattern methylation at paternal ICRs; in the female, postnatal oocyte growth deposits maternal-pattern methylation at maternal ICRs. The cycle ensures that a man passes paternal imprints regardless of his own mother's marks, and vice versa for women. Failures of this cycle produce imprinting-defect cases of Prader-Willi, Angelman, BWS, and SRS.

Imprinting vs X-inactivation

Property	Genomic imprinting	X-inactivation
Allele choice	Always the same parental copy	Random in each cell at early embryogenesis
Cellular pattern	Uniform across body	Mosaic — calico-cat patches
Affected loci	~150 loci genome-wide	~all genes on inactive X
Primary mark	DNA methylation at ICRs	XIST RNA coating + histone marks
Inheritance window	Erased + reset in germline each gen	Reset at fertilization (paternal X reactivated in ICM)
Sex specificity	Affects both sexes equally	Confined to XX females
Disease examples	Prader-Willi, Angelman, BWS, SRS	X-linked disorder severity in females (e.g. Rett)
Theoretical driver	Haig's parent-of-origin conflict	Dosage compensation between XX and XY

Famous case studies

• Prader-Willi vs Angelman (15q11-q13). Same ~5 Mb chromosomal region, opposite parental losses. Paternal deletion (or maternal UPD15) of 15q11-q13 silences SNRPN, MAGEL2, NDN and produces Prader-Willi (~1 in 15,000–25,000): neonatal hypotonia, hyperphagia from age ~2, obesity, intellectual disability. Maternal deletion (or paternal UPD15, or UBE3A mutation) silences neuronal UBE3A and produces Angelman (~1 in 12,000–20,000): severe intellectual disability, ataxia, frequent laughter, seizures. Same DNA, opposite parental origin, completely different syndromes — the textbook proof of imprinting.
• IGF2/H19 cluster on 11p15.5. Insulin-like growth factor 2 is paternally expressed; the long noncoding RNA H19 immediately downstream is maternally expressed. They share ICR1, whose methylation status governs CTCF binding. Hypomethylation of paternal ICR1 (loss of paternal IGF2 expression) causes Silver-Russell syndrome — severe intrauterine growth restriction, ~1 in 30,000–100,000. Hypermethylation or paternal UPD11p15 (overexpression of paternal IGF2 plus loss of CDKN1C) causes Beckwith-Wiedemann syndrome — overgrowth, macroglossia, abdominal-wall defects, ~5–10% childhood tumor risk including Wilms' tumor and hepatoblastoma, ~1 in 10,000.
• Surani/McGrath/Solter 1984 nuclear transfer. Two independent labs constructed mouse embryos by pronuclear transfer to produce gynogenotes (two maternal pronuclei) and androgenotes (two paternal). Both classes failed to develop normally despite carrying complete diploid genomes. Gynogenotes built reasonable embryos but defective placentas; androgenotes had hyperdeveloped placentas (the embryological analog of human hydatidiform moles) with stunted fetuses. The asymmetry showed that maternal and paternal genomes carry distinct, non-redundant information — the empirical foundation of the imprinting field.
• Callipyge sheep. A 1983 mutation in the DLK1-DIO3 imprinted cluster on sheep chromosome 18 produces a "polar overdominance" phenotype: only sheep that inherit the mutation paternally and a wild-type allele maternally show muscular hindquarter hypertrophy. Homozygotes don't show the phenotype because the mutation only acts in cis with the paternal expression program. The first mammalian "polar overdominance" example, traced to imprinting in 1996 and explained in molecular detail by 2002.
• Cancer LOI of IGF2. Loss of imprinting — biallelic expression — of IGF2 occurs in ~10–40% of Wilms' tumor, colorectal, breast, ovarian, esophageal cancers. Andy Feinberg's group reported in Science (2003) that ~30% of normal colon mucosa from individuals with personal or familial colorectal cancer carries IGF2 LOI, versus ~10% of controls — a roughly 3–5x elevated risk and a candidate biomarker for prevention strategies.