X-Inactivation

What X-inactivation is

Mammalian sex is set by chromosomes: females are XX, males are XY. That creates a math problem. The X chromosome carries roughly 800–900 protein-coding genes, and almost none of them have anything to do with sex — they run metabolism, neurons, blood clotting, vision. If both X chromosomes in a female were left fully active, she would make twice the dose of every one of those gene products compared with a male, who has a single X. Across hundreds of dosage-sensitive genes, a flat 2-fold imbalance is not survivable.

X-inactivation is the fix. Early in development, every female cell transcriptionally silences one of its two X chromosomes, condensing it into a compact, gene-dead lump of heterochromatin. From then on that cell — and every cell it divides into — runs on a single active X, exactly like a male cell. The chromosome that gets silenced is chosen randomly in each cell, and the choice is permanent and heritable: once a cell has shut off, say, the paternal X, all of its millions of descendants do too. So a female body is not uniform. It is a mosaic — a patchwork of two clonal populations, one expressing the maternal X and one the paternal X. When the two X chromosomes carry different alleles for a visible trait, you can literally see the mosaic. That is what a calico cat's coat is: a map of which X won the coin-flip in each early skin cell.

The British geneticist Mary Lyon proposed this whole picture in a one-page 1961 Nature letter, reasoning from mottled coat patterns in mice. It is still sometimes called lyonization in her honor.

How it works, step by step

The decision-making region is a stretch of the X called the X-inactivation center (Xic), and the master switch inside it is a gene that does not code for a protein at all. It codes for a long non-coding RNA called Xist (X-inactive specific transcript), about 17 kb in humans and 17–18 kb in mouse. The process runs in distinct phases:

1. Counting and choosing. The cell first establishes that it has two X chromosomes (one is the trigger; cells with a single X never inactivate it). Then it picks one X to keep active and one to silence. The choice involves an antisense lncRNA, Tsix, transcribed across the same locus in the opposite direction: high Tsix keeps Xist off and protects an X from inactivation, so the future active X is the one where Tsix wins. Trans-acting factors and pairing of the two Xic regions help make the choice mutually exclusive — exactly one X is selected.
2. Xist coats the chosen X in cis. Xist is up-regulated only from the X destined for silencing and, crucially, it never leaves that chromosome. It spreads along the chromosome it was made from — coating in cis — over minutes to hours, first reaching gene-rich regions that are physically close in 3D, then filling in the rest. Under microscopy this looks like an "Xist cloud" enveloping one X.
3. Recruiting the silencing machinery. Xist is modular. Its A-repeat binds the protein SPEN (SHARP), which recruits the NuRD complex and histone deacetylases to strip activating acetyl marks. Other regions recruit the Polycomb complexes PRC1 and PRC2, which deposit the repressive histone marks H2AK119ub and H3K27me3, plus hnRNPK and the matrix protein SAF-A / hnRNPU that tether the RNA to the chromosome scaffold.
4. Heterochromatin and the Barr body. Activating marks disappear, repressive marks accumulate, the histone variant macroH2A is incorporated, and the whole chromosome compacts and moves to the nuclear periphery. It becomes the Barr body, a dense ~1 µm blob of facultative heterochromatin visible in interphase nuclei.
5. Locking it in with DNA methylation. Finally, CpG islands of the promoters on the inactive X are methylated by DNA methyltransferases, making the silent state extraordinarily stable. At this point the chromosome no longer needs Xist to stay off — silencing has become Xist-independent and self-maintaining through every subsequent cell division. The inactive X also replicates late in S phase, a hallmark that distinguishes it from its active partner.

The whole cascade is a beautiful example of an RNA-driven epigenetic switch: a non-coding RNA reads out a chromosomal decision and converts it into a permanent change in chromatin state, without altering the DNA sequence itself.

The players and conditions

• Xist (~17 kb lncRNA). The initiator. Expressed only from the future inactive X; coats it in cis; recruits silencers via its repeat domains. Knock out Xist on one X and that X cannot be inactivated.
• Tsix (antisense lncRNA). The antagonist. Overlaps Xist in the opposite orientation and represses it; the X with high Tsix becomes the active X.
• SPEN/SHARP + NuRD/HDACs. The first wave of silencing — removes histone acetylation, the earliest and one of the most essential steps.
• PRC1 and PRC2 (Polycomb). Write the repressive H2AK119ub and H3K27me3 marks that blanket the inactive X.
• DNA methyltransferases (DNMT1/3). Methylate promoter CpG islands to lock the silent state for the long term.
• macroH2A, SAF-A/hnRNPU, hnRNPK. Structural and scaffolding factors that build and stabilize the Barr-body heterochromatin.
• The window of opportunity. Xist can establish full silencing only during a narrow developmental window (around implantation in mouse). After differentiation, the same Xist over-expression silences far fewer genes — the chromosome has become resistant.

Active X vs inactive X

Property	Active X (Xa)	Inactive X (Xi)
Transcription	Full expression of ~800–900 genes	Silenced (except ~15–25% escapees)
Xist coating	None (Tsix-high)	Coated by Xist RNA cloud
Chromatin state	Open euchromatin	Compact facultative heterochromatin
Key histone marks	H3K4me3, H3/H4 acetylation	H3K27me3, H2AK119ub, hypoacetylation
DNA methylation	CpG islands unmethylated	Promoter CpG islands methylated
Histone variant	Canonical H2A	macroH2A enriched
Replication timing	Early S phase	Late S phase
Nuclear position	Interior, intermingled	Periphery / perinucleolar (Barr body)
Microscopy	Indistinct	Dense ~1 µm Barr body

The numbers

• Genes affected. The human X has about 800–900 protein-coding genes; silencing one X removes roughly one full copy's worth of expression for the ~75–85% that are subject to inactivation.
• Xist length. ~17 kb in humans, transcribed but never translated; it works as RNA.
• Timing. Random inactivation is established around the 50–200 cell blastocyst stage — roughly embryonic day 5.5 in mouse, the first week or so in humans. Imprinted (paternal) inactivation in mouse extra-embryonic tissue begins even earlier, at the 4–8 cell stage.
• Mosaic ratio. On average about 50:50 maternal-X-active vs paternal-X-active cells, but the founding population is small (tens of cells), so by chance the ratio in any individual can drift; "skewed" inactivation (e.g. 80:20 or more extreme) occurs in a sizable minority of women.
• Barr body size. Roughly 1 µm across — a dense heterochromatic mass at the nuclear edge.
• Barr body count rule. Number of Barr bodies = (number of X chromosomes) − 1. 46,XX → 1; 46,XY → 0; 47,XXY → 1; 47,XXX → 2; 49,XXXXX → 4.
• Escape. About 15–25% of human X genes escape silencing, including all genes in the pseudoautosomal regions PAR1 and PAR2; the figure is much lower in mouse (~3–7%).
• Stability. Once locked by DNA methylation, the inactive state is maintained through an effectively unlimited number of cell divisions across the entire lifespan — decades in humans.

Where it shows up — organisms, disease, examples

• Calico and tortoiseshell cats. The orange/black coat-color gene is X-linked. A heterozygous female becomes a coat-color mosaic; an XY male, with one X, is a single color. Cloning a calico (the cat "CC", 2001) produced a kitten with a different coat pattern from the donor — because mosaicism is set anew during each animal's development and is not in the genome.
• X-linked disease carriers. Women heterozygous for X-linked recessive disorders are mosaics of normal and mutant cells. In Duchenne muscular dystrophy, hemophilia A/B, and red-green color blindness, carriers are usually mildly or unaffected — but skewed inactivation can leave a carrier with mostly mutant-X-active cells and overt symptoms (a "manifesting carrier").
• X-linked dominant lethals. Disorders like incontinentia pigmenti and Rett syndrome (MECP2) appear almost exclusively in females: the mutation would kill a hemizygous male embryo, but in a female, X-inactivation creates a survivable mosaic in which roughly half the cells silence the mutant allele. The swirled skin lesions of incontinentia pigmenti trace the underlying clonal mosaic — visible "lines of Blaschko."
• Aneuploidy phenotypes. Because all but one X is silenced, people with Klinefelter (47,XXY) or Triple X (47,XXX) have relatively mild phenotypes. Turner syndrome (45,X) is more severe largely because there is no second X to supply the dose of the ~15–25% of genes that normally escape inactivation and need two copies.
• Marsupials and the placenta. In kangaroos and opossums, the paternal X is always the one silenced (imprinted X-inactivation) — and the same is true in the placenta of mice. This is an evolutionarily older form of the same mechanism.
• Stem-cell biology. The inactive X is reactivated when cells are reprogrammed to pluripotency (and in the germ line and early epiblast). Faithful X-reactivation is a quality benchmark for induced pluripotent stem cells.

Dosage compensation across animals

System	Strategy	Mechanism
Mammals (XX/XY)	Silence one X in females	Xist lncRNA + Polycomb + DNA methylation → Barr body
Fruit fly Drosophila (XX/XY)	Double the single X in males	MSL complex + roX RNAs hyperacetylate (H4K16ac) the male X
C. elegans (XX/XO)	Halve both X chromosomes in hermaphrodites	Dosage compensation complex (condensin-like) downregulates X 2-fold
Birds (ZZ/ZW)	Incomplete / gene-by-gene	No chromosome-wide system; local, partial compensation of the Z

Common misconceptions and pitfalls

• "The whole X chromosome is switched off." No — about 15–25% of human X genes escape and stay biallelic, including the pseudoautosomal regions shared with the Y. That residual two-dose expression is exactly why Turner syndrome (45,X) has a phenotype.
• "It's always the father's X that's silenced." Only in marsupials and in the mouse placenta. In the body proper of humans and mice, the choice is random and roughly 50:50 between maternal and paternal X.
• "The silenced X is deleted or lost." It is fully present, replicated every cell cycle (late in S phase), and faithfully transmitted — it is merely transcriptionally silenced and compacted. It can even be reactivated during reprogramming and in the germ line.
• "Xist is a protein." Xist is a long non-coding RNA. It is transcribed, spliced and polyadenylated like an mRNA but is never translated — the RNA molecule itself is the functional product.
• "A male calico cat is impossible." Rare but real — almost always an XXY (Klinefelter) cat whose two X chromosomes permit the same mosaicism; such males are typically sterile.
• "Identical twins / clones will have the same X-inactivation pattern." They will not. The random choice is made independently in each twin and each clone during development, so a cloned calico has a brand-new coat pattern and even monozygotic twin girls can differ in X-linked carrier phenotypes.
• "X-inactivation is the only dosage compensation needed." Mammals also up-regulate the single active X about 2-fold in both sexes (Ohno's hypothesis) so the X balances against the diploid autosomes. Silencing the second X and boosting the active one are two distinct layers.