Biological Coherence · System 04 of 12

DNA
Methylation

A methyl group — three hydrogen atoms bound to a carbon — attached to cytosine in DNA. This single atomic modification silences genes for decades, can be inherited across generations, and is precisely reversible by dedicated enzyme families.

Robin Holliday
b. 1932 · Proposed DNA methylation as epigenetic memory · 1975
I. The Machine

A Mark That Remembers

In 1975, Robin Holliday (b. 1932) and John Pugh independently proposed that methylation of cytosine in DNA — adding a methyl group to the 5 position of the cytosine ring to produce 5-methylcytosine (5mC) — could serve as a heritable epigenetic memory. When DNA is replicated, the methyl groups on the parental strand are on one strand of the new hemimethylated double helix. DNMT1 — the maintenance methyltransferase — recognizes hemimethylated CpG dinucleotides and methylates the daughter strand, faithfully copying the parental pattern.

This mechanism allows a cell to remember its epigenetic state across divisions without any genetic information being passed. The methyl mark propagates the information that a gene should be silenced — and it does so through a chemical mechanism that operates independently of the DNA sequence, exploiting the symmetry of CpG dinucleotides on opposite DNA strands.

"5-methylcytosine is the fifth base of the mammalian genome. Its placement — at CpG dinucleotides, at specific promoters, across imprinted regions — is as precisely specified as the four bases of the genetic code." — Bird, 2002

The silencing mechanism: methylated CpGs recruit methyl-CpG binding domain (MBD) proteins, which in turn recruit histone deacetylases (HDACs) and other chromatin compaction machinery. Methylated promoters are wrapped in condensed, inaccessible chromatin. Transcription factors cannot bind. RNA polymerase cannot initiate. The gene is effectively switched off — for the lifetime of the cell and its progeny.

II. Writer / Reader / Eraser

The Three-Enzyme System

W
Writers (DNMT)
DNMT1/3A/3B

DNMT3A and DNMT3B: de novo methyltransferases — write new marks. DNMT1: maintenance methyltransferase — copies hemimethylated patterns after replication with ~97% fidelity. DNMT3L: regulatory cofactor enhancing DNMT3A activity at imprinted loci. Three enzymes, three distinct functions.

R
Readers (MBD)
MeCP2, MBD1-4

Methyl-CpG binding domain proteins recognize 5mC and recruit silencing complexes (NuRD, Sin3A). MeCP2 mutations cause Rett syndrome — X-linked neurodevelopmental disorder. Illustrates how tightly neural function depends on correct methyl mark reading.

E
Erasers (TET)
TET1/2/3

TET dioxygenases oxidize 5mC → 5hmC → 5fC → 5caC, enabling active demethylation. TET2 is one of the most commonly mutated genes in hematological malignancies — its loss of function causes clonal hematopoiesis. Active erasure is as essential as writing.

III. The Goldilocks Explorer

CpG Methylation Windows

DNA methylation must be precisely placed and maintained. Too much silences essential genes; too little fails to suppress transposable elements and oncogenes. Explore the narrow windows that define correct methylation function.

DNA Methylation System Explorer
Adjust CpG island methylation, gene body methylation, repeat element silencing, and TET activity to observe system integrity.
CpG Island Methylation at Promoters (%)5%
drag
5% CpG island methylation at active promoters: Correct. Active gene CpG islands are hypomethylated — protected from DNMT by CXXC domain proteins that read unmethylated CpG. This creates a default active state for CpG island genes, silenced only by targeted DNMT3A/3B recruitment.
Gene Body Methylation (%)65%
drag
65% gene body methylation: Biological range. Paradoxically, actively transcribed genes have highly methylated bodies (exons and introns). Gene body methylation suppresses spurious intragenic transcription initiation and is correlated with, not causative of, gene expression level.
Repeat Element Silencing (%)85%
drag
85% repeat silencing: Correct. Transposable elements (LINEs, SINEs, ERVs) — ~45% of the human genome — silenced by DNA methylation + H3K9me3. If transposons mobilize, they insert into new genomic locations causing insertional mutagenesis.
TET Oxidation Activity (%)20%
drag
20% TET activity: Appropriate for somatic cell. Targeted active demethylation at regulatory regions (enhancers, CTCF sites). Not global — high TET activity confined to specific contexts (zygote, primordial germ cells, neural differentiation).
Methylation
System Score
95%
Methylome fidelity
DNA methylation correctly placed. Active genes accessible, silent genes and repeats suppressed. Imprinting maintained. Genomic stability preserved.
IV. The Inference

Memory Without Neurons

DNA methylation is epigenetic memory at the chemical level. A methyl group placed on a cytosine in a germ cell can determine which genes are expressed in every cell of the organism derived from it — and in some cases, in the next generation's cells as well. This transgenerational inheritance of methylation patterns (demonstrated in plants, invertebrates, and some mammalian studies) means that environmental information can be encoded in DNA chemistry and transmitted to offspring — without altering the DNA sequence.

The system requires three enzyme families with different specificities, a structural feature of DNA (CpG symmetry) that enables inheritance, and reader proteins that translate the chemical mark into chromatin architecture. The mark, the inheritance mechanism, and the reader-effector system must all be present simultaneously: a methyl mark without reader proteins has no function; reader proteins without methyl marks have nothing to read; the maintenance mechanism without the original writer cannot establish new marks. Three independent molecular systems forming one coherent epigenetic memory system.

Primary Source
Holliday, R. & Pugh, J.E. (1975). "DNA modification mechanisms and gene activity during development." Science 187(4173):226–232.
The founding paper proposing DNA methylation as heritable epigenetic memory. Holliday correctly predicted the hemimethylation maintenance mechanism half a decade before DNMT1 was characterized — one of the most prescient predictions in molecular biology.
Read at Science (DOI) ↗