PAX3 (splash white)

Crossing two of them

Deafness

A homozygote at this locus (no N copy) is deaf, for the same reason as at MITF: the pigment cells that never reached the coat never reached the inner ear either. The description on this page used to say deafness was “described, not modelled”. It is modelled now — as a named, INFORMATIONAL condition that costs the horse nothing, because the mod has no hearing to take away.

It is the same Condition object MITF declares, so a horse homozygous at both loci is told once. See the horse’s body.

Its gene carrot

Gene key

horsegenetics.pax3

Priority

79

Alleles

SW2 SW4 N

Combinations

6, of which 5 a horse can carry, landing on 3 outcomes

Outcomes

wild, splash, splash-bold

Default allele

N

Wild frequency

Pax3Gene.WILD_SW2_PERCENT carry one SW2, WILD_SW4_PERCENT one SW4 — heterozygotes only. The breed tables carry most of it

Founder draws

1 × nextFloat()

Deterministic

no

Epi draws

nextLong() waterline seed, nextFloat() waterline height, then the shared face-marking draw (1 long + 8 floats)

Painter

WhitePattern.splash — the same shape MITF uses

Two loci, one pattern

SW1 sits on MITF and SW2 on PAX3. They are different genes on different chromosomes, they produce the same-looking pattern — which is exactly why they were long taken for one gene — and a horse carrying one copy of each has considerably more white than a horse carrying either alone.

In this model that falls out for free. Both genes paint, one after the other, and white finds white: each painter raises its own strength by how much of the horse is already de-pigmented. There is no interaction rule anywhere, and the measured result is 38% white for SW1/N + SW2/N against 12% and 14% for the two on their own — more than either, and more than the two simply added, because each raises the other’s strength.

The alleles

The combination table

SW4 is an Appaloosa allele: it was described in one Appaloosa family and nowhere else, and one copy can be a splash or little more than a broad blaze. Its homozygote has never been detected, so the model does not let one occur — the same precaution MITF applies to SW3/SW3, and for the same reason. “Never seen” is a weaker claim than “lethal”, and it is a breeding hazard either way.

SW2 is the rare one — and it used not to be

This locus used to carry the mod’s near-ubiquitous minimal splash allele: 90% of founders were SW2/N, on the reasoning that a mild splash allele is what gives an ordinary horse its socks and its blaze, so most horses should have one.

The reasoning was right and the allele was wrong. The widespread, several-hundred-year-old, minimally-expressed splash allele is SW1, on MITF. SW2 is a coding change — PAX3 p.Cys70Tyr — with a much narrower distribution: Quarter Horses and American Paints above all, with reports in Lipizzaners and Norikers. So the common-allele job moved next door, and this locus is now what its own evidence says it is.

Why the table is still written out

Both splash loci list their heterozygotes explicitly rather than deriving a table from allele frequencies. Two reasons, and either one on its own rules out FounderTable.hardyWeinberg:

• The doubled combinations are the reward for breeding. SW2/SW2 is the bold outcome and it is deaf; it must not turn up in a wild-caught horse — the same rule the health loci and magic body size follow.
• Hardy-Weinberg cannot state a carriage rate. Its heterozygote share is 2pq, which peaks at 50% at p = q = 0.5, so a table that says “this many founders carry one copy” has to say it directly rather than derive it. That mattered most when this locus held the common allele; MITF holds it now, and uses the same shape of table for the same reason.

What the split still buys is what it was built for. Because both splash loci read the coat they are handed (white finds white), a horse carrying SW1 and SW2 is whiter than either alone with no interaction rule anywhere — and in the Quarter Horse and the American Paint, which carry both, that horse is common enough to meet.

The face

The same shared face-marking vocabulary that MITF uses, at the same jag of 0.11 and with the same face boost (SPLASH_FACE_BOOST = 0.34) — so a single-copy PAX3 splash is a blaze, not a star, and splash-bold is a bald face. That the vocabulary is shared is the point: a horse carrying one copy at each splash locus draws two bold markings from one vocabulary, compounding the way the body white does, instead of two hand-written blazes landing on top of each other and reading as one.

At a glance

What PAX3 does

PAX3 encodes a transcription factor — a protein that binds DNA and regulates other genes during embryonic development. Its relevant roles are in neural-crest development; the migration and survival of melanoblasts, the embryonic precursors of melanocytes; some nervous-system and muscle lineages; and the activation of pigment pathways including regulation connected to MITF.

Melanocytes migrate from the neural crest to skin, hair follicles, eyes and inner ear. Where fewer precursors arrive, or fewer survive, that tissue has no pigment: white hair and pink skin in the coat, blue or partly blue eyes in the iris, and a predisposition to congenital deafness in the inner ear. That is one mechanism producing three apparently unrelated symptoms — and it is why the mod’s eye-colour hooks and its deafness condition both hang off the white-pattern genes rather than off anything of their own. Hauswirth et al., PMC3325211

SW2 does not switch PAX3 off. One copy creates a partial, dosage-sensitive developmental effect, and the visible result is a patchy absence of pigment cells rather than a whole-body one.

The mutation

The substitution sits in the paired domain, one of PAX3’s key DNA-binding domains. The cysteine at position 70 is highly conserved across PAX-family proteins and across animals, and structural comparison with human PAX6 suggests it normally helps contact the DNA backbone — so replacing it with tyrosine should alter DNA binding and downstream regulation.

The case for causality is strong, and much of the mechanism is inferred rather than measured in equine embryonic tissue: linkage in a Quarter Horse family pointed to ECA6, where PAX3 sits; the allele was found in splash-white horses and not in the solid-coloured controls examined; PAX3’s biological role fits abnormal pigment-cell development directly; comparable variants in humans and mice cause pigmentary and hearing phenotypes; and the affected residue is conserved and functionally crucial.

Coat and eye

SW2 is a white spotting pattern, not a dilution. The base colour still exists in the pigmented areas; SW2 removes pigment where melanocytes are missing. Typical features: a large broad blaze; a bald face, often over or past the eyes and onto the cheeks; white lower legs, sometimes carried high; one or two blue eyes, or partial blue and sectoral heterochromia; small belly spots or more extensive body white; and the impression of a horse dipped in white from below.

Expression is highly variable. One N/SW2 horse has a bald face and four high whites; another has ordinary socks and a modest blaze. Visual identification is unreliable. Horses carrying SW2 and SW1 averaged more facial depigmentation than horses carrying either alone, and chestnut horses in that study showed more extensive white face than bays of the same splash genotype — so base colour and other modifiers matter.

visible white = SW2 effect + other white-pattern alleles + base colour and modifiers + developmental variation

So markings cannot tell you whether a horse has SW2, or whether it has one copy or two.

Deafness

The well-supported health association is congenital sensorineural deafness. Some splash-white horses are deaf; many are not, and the research and laboratory guidance do not support treating it as inevitable.

The mechanism is developmental. Melanocytes matter to normal inner-ear physiology: in the cochlea, melanocyte-like cells in the stria vascularis help maintain the ionic environment hearing depends on. If the pigment pathway leaves the inner ear without enough functional melanocytes, the hearing loss is permanent.

• It is usually present from birth, not progressive.
• A blue eye is a clue to altered pigment-cell distribution and is not a hearing test.
• Extensive facial white, particularly around the ears, raises suspicion; appearance cannot diagnose it.
• BAER testing is the objective method where hearing matters for safety, training, breeding or sale disclosure.
• A deaf horse can live and work successfully with predictable routines, visual cues and safety-aware handling.

What SW2 is not known to cause

The caveat that matters: deafness changes management, not inherent athletic ability. A deaf horse may need training adapted to vibration, body language, touch, visual signals and an awareness of startling risk. No study establishes an SW2-specific reduction in athletic capacity. People infer physical traits from the striking look of a high-white horse; there is no basis for it. SW2 affects embryonic patterning of melanocytes and is not a quantitative trait locus for conformation or sport.

Inheritance

Dominant genetically; variable in expression. A horse can inherit SW2 and show only modest markings. So SW2 is dominant with variable expressivity, not “one allele equals a fixed paint pattern”. Other loci add or alter white: SW1 and the other MITF splash alleles, tobiano, frame overo, sabino 1 and the other KIT variants, W20 and the dominant whites, and the base colour itself.

The homozygote, and a hypothesis that did not survive

Early work found no SW2/SW2 horses. Because PAX3 matters in neural development, the 2012 discovery paper reasonably raised the possibility that homozygosity might be embryonic or fetal lethal, partly on mouse biology. That was a hypothesis.

Large-scale genotyping settled it. A UC Davis dataset of 11,281 horses found nine SW2/SW2 animals — three Paints and six Quarter Horses. So the genotype is not universally lethal. The six homozygous Quarter Horses with photographs were all described as all-white, and five also carried one or more additional white-patterning alleles. PMC9498372

Careful conclusion: homozygotes can be born and survive; they tend toward far more extensive depigmentation than heterozygotes; the data are too limited to predict a uniform SW2/SW2 pattern; many known examples have confounding stacked white alleles; and the frequency and severity of deafness among them is not well quantified. UC Davis describes homozygosity as likely non-lethal while noting such horses may be deaf and warrant veterinary evaluation.

Breeds and frequency

In the 11,281-horse, 28-breed study, SW2 was found in Quarter Horses and Paint Horses only.

That is an allele frequency near 0.8%, not a carrier percentage. For a rare dominant allele, heterozygotes are about 2p(1−p), so 0.008 corresponds to roughly 1.5–1.7% heterozygotes under random mating, before selection and relatedness. Quarter Horse and Paint registries have historical gene flow and dual registration, so the two figures are not independent breed estimates, and the original research treated them as one connected population.

The mod carries WILD_SW2_PERCENT = 3.0 and WILD_SW4_PERCENT = 0.5 — SW2 roughly twice the real stock-horse heterozygote rate, and SW4 rarer still, which is right: SW4 is one Appaloosa family and nothing else.

SW2 against the rest of the splash family

“Splash white” is a phenotype category, not a gene. At least two genes produce splash-like patterning:

So a horse can look splashy and test N/N for SW2 because it has SW1, a different splash allele, a KIT white allele, frame overo, an untested variant, or a combination — and an SW2-positive horse can look only subtly marked. The mod carries SW2 and SW4 on this locus and SW1 on its MITF twin, which is the correct split: they are different chromosomes’ worth of gene, so a horse can be splash twice over, and one carrying both is whiter than either alone with no special case anywhere.

What is settled, and what is not

Well supported. SW2 is the PAX3 p.Cys70Tyr variant on ECA6, inherited as an autosomal dominant white-pattern allele; it affects melanocyte development and produces variable splash-white spotting with broad facial white, leg white, blue or partly blue eyes and sometimes extensive body white; some splash-white horses have congenital deafness; SW2 interacts with SW1 to increase average white; homozygotes exist, so the genotype is not universally lethal; and the documented distribution is Quarter Horse and Paint.

Still uncertain. The exact molecular change in DNA binding and downstream expression caused by p.C70Y in living equine pigment cells; the exact penetrance of deafness for heterozygotes and homozygotes; whether SW2 alone accounts for the all-white homozygote phenotypes, given that most known homozygotes carry additional white alleles; the precise influence of base colour, MITF and KIT alleles and unknown modifiers; and the real frequency in individual sublines, regions and programmes.

Not supported. SW2 as a speed, racing, jumping, endurance or conformation gene; that it improves or impairs athletic potential; that a blue eye proves deafness; that an SW2-positive horse is automatically deaf; that a splashy horse is automatically SW2-positive; or that homozygous SW2 is categorically lethal — that concern was superseded by living homozygotes.