K50 homeodomain (OTX/PITX/CRX-type, TAATCC)
A small group of homeodomains carries lysine instead of glutamine at position 50 of the domain, and that single substitution changes the two base pairs read 3' of the invariant TAAT core from TA to CC. The result is TAATCC, the bicoid-type site. In human the true K50 set is nine genes: OTX1, OTX2, CRX and DMBX1, PITX1, PITX2 and PITX3, and GSC and GSC2. They are tightly consistent - 34 motifs, all converging on TAATCC, with cooperative dimers on TAATCCGATTA - and unusually well supported in vitro. Paired-like factors that share their gene families but carry glutamine at position 50, such as RAX, VSX2, PHOX2B and NOBOX, read TAATTA and are covered by HD_Q50 instead.
The factors
The homeodomain is a 60-residue three-helix module whose recognition helix sits in the major groove while a flexible N-terminal arm reads the adjacent minor groove. Bicoid, the Drosophila anterior morphogen, is the founding member of the K50 class, defined by lysine at position 50 - the last residue of the recognition helix - and unique among homeodomains in also carrying arginine at position 54, which it needs for RNA binding 16406070.
In human the class is small and well defined: OTX1, OTX2, CRX and DMBX1 (the OTX group), PITX1, PITX2 and PITX3, and GSC and GSC2. These nine genes contribute 34 motifs across JASPAR, HOCOMOCO, CIS-BP and the footprint archetypes. The only other member is ANHX, a divergent homeobox gene that has alanine rather than lysine at position 50 and whose two reported motifs (TTGACAAG) are not K50 sites at all.
Membership here follows measured specificity, not gene family. Many proteins classified as paired-like alongside OTX, PITX and GSC - VSX1/2, RAX/RAX2, PHOX2A/2B, NOBOX, MIXL1, VENTX, PROP1, DRGX, BSX, OTP and SEBOX - carry glutamine at position 50 and their motifs are TAATTA-type; they are grouped with the Q50 homeodomains in HD_Q50. Expression of the K50 proteins is narrow and lineage-defining: CRX and OTX2 in photoreceptors and retina, OTX1/OTX2 in forebrain and eye, PITX1-3 in pituitary, anterior eye segment and hindlimb, GSC and GSC2 in craniofacial mesenchyme and mesendoderm, DMBX1 in midbrain.
The motif
The consensus is TAATCC (equivalently GGATTA on the opposite strand). Every homeodomain recognizes the 5'-TAAT-3' core through a hydrogen bond from the invariant Asn51 to its second adenine plus minor-groove contacts from N-terminal arm arginines; residue 50 discriminates the two base pairs immediately 3' of it 8692913. Site selection with six different residues at position 50 gave six distinct specificities differing at exactly those positions, through water-mediated hydrogen bonds and van der Waals contacts 8692913; lysine contacts the C:G pairs that glutamine cannot 15895993. Swapping residue 50 in Bicoid converts it into an Antennapedia-class TAATTA binder 2500253; the reverse swap in Paired confers Bicoid specificity 2572327. Solution structures of the native Bicoid homeodomain on TAATCC and of PITX2 on DNA show the K50, N51 and R54 side chains remaining mobile within the interface 16406070,15895993, and the PITX2 structure rationalizes the homeodomain substitutions that cause Axenfeld-Rieger syndrome 15895993.
Only about six base pairs carry information, so monomeric models are short: all nine genes give 5-7 bp TAATCC/GGATTA matrices, and the family medoid is the footprint archetype AC0061 (PITX/OTX) at exactly TAATCC. This is a tight family: cohesion 0.90, data consensus exactly TAATCC, no member flagged as resembling another family.
The longer models are dimeric. Paired-like homeodomains bind cooperatively on palindromes whose half-sites are three base pairs apart (the P3 site, TAATNNNATTA), and residue 50 sets the central bases 7901121, so the K50 version is TAATCGGATTA or TAATCCGATTA. That is exactly what the OTX1, OTX2 and OTX archetype models show; the PITX1 and PITX3 models instead carry two GGATTA half-sites. Residue-level rules for cooperativity across the 36 human paired-like factors have recently been derived 42675056, and blindness-causing CRX homeodomain mutations have been shown to alter cooperative dimerization rather than monomeric affinity 39715683.
Within the family, individual paralogues are not separable: no matrix distinguishes OTX2 from CRX, or PITX1 from PITX2 from PITX3. K50 versus Q50 is the only specificity distinction that sequence supports.
Where it acts
These sites concentrate in tissue-specific distal enhancers and in promoters of terminal differentiation genes. The best-mapped case is photoreceptors, where CRX occupies thousands of sites at rod and cone gene promoters and enhancers 20693478, typically with NRL and NR2E3 motifs nearby. TAATCC is common in absolute terms - roughly two million expected copies in 3.1 Gb at human base composition - so location and cell type do the discriminating, not the 6-mer. The dimeric TAATCCGATTA is about a thousand-fold rarer, with a few thousand expected copies, and is correspondingly more informative.
Biology and disease
This small family carries a heavy disease load, mostly dominant. CRX mutations cause cone-rod dystrophy 9390563 and Leber congenital amaurosis; heterozygous OTX2 mutations cause microphthalmia and anophthalmia 15846561; PITX2 mutations cause Axenfeld-Rieger syndrome 8944018; PITX1 mutation causes asymmetric lower-limb malformation 18950742; and homozygous predicted-null GSC mutations or deletion cause the autosomal-recessive SAMS syndrome (short stature, auditory canal atresia, mandibular hypoplasia and skeletal abnormalities) 24290375. Many of these substitutions fall inside the homeodomain and alter DNA binding or dimerization directly 42675056,39715683, which is why the in vitro motifs and cooperativity rules have clinical relevance.
Reading a hit
A TAATCC match is one of the better-behaved homeodomain calls, but it is still a class-level call: it says a K50 homeodomain could bind, not whether OTX2, CRX, PITX2 or GSC bound it, and in retina at least two of those are usually present. Evidence quality here is uniformly good - all 12 HOCOMOCO models have HT-SELEX or methyl-HT-SELEX support, none is ChIP-only, none of the 34 motifs is a cross-species transfer, and no member is flagged as belonging to a different family. Two things to watch. The ANHX entries read TTGACAAG and should not be scanned as K50 sites. And one CIS-BP CRX model carries only 2.3 bits of information, with consensus TAAT, so it will match almost any AT-rich sequence; prefer the JASPAR or HOCOMOCO CRX models. Finally, if your hit is TAATTA rather than TAATCC, you are looking at the Q50 family - which includes paired-like proteins whose names (RAX, VSX2, PHOX2B, NOBOX, PROP1) might otherwise lead you to expect a K50 site.
How this was worked out
- 1989highYeast transcription-activation assay; in vitro DNA binding of mutant homeodomains2500253 2572327
Residue 50 (residue 9 of the recognition helix) is shown to set Bicoid's specificity. One substitution makes Bicoid recognize Ultrabithorax and Fushi tarazu sites, and lysine at residue 50 of Paired confers Bicoid specificity. Lysine 50 becomes the defining feature of the class.
Still open then: How lysine 50 contacts the bases 3' of TAAT was unknown.
Paired-class homeodomains bind cooperatively as dimers to palindromic sites, with up to 300-fold enhancement. Residue 50 sets the half-site spacing and the identity of the central base pairs, so the preferred dimer site changes with residue 50.
In vitro selection with six different residues at position 50 gives six distinct specificities differing at the two bases 3' of TAAT, attributed to water-mediated hydrogen bonds and van der Waals contacts.
The first human K50 disease genes: RIEG (PITX2) mutations cause Rieger syndrome, and in 1997 CRX mutations, including E80A at the first residue of the recognition helix, are found to cause dominant cone-rod dystrophy. The CRX alleles were attributed to haploinsufficiency or a dominant-negative effect.
Still open then: How the homeodomain substitutions change DNA binding was not known.
The first structure of a native K50 homeodomain on DNA: in PITX2 on TAATCC, lysine 50 contacts the two guanines on the opposite strand next to TAAT, as in the crystal structure of the engineered Engrailed Q50K mutant, but makes fluctuating rather than fixed contacts. The native Bicoid structure (2006) shows K50, N51 and R54 all flexible in the interface.
Still open then: How far native K50 proteins differ in affinity and specificity from the Q50K mutant used as the earlier model.
CRX ChIP-seq in mouse retina maps thousands of bound regions around photoreceptor genes; bound regions carry several CRX sites whose spacing and orientation tune output, and comparison with Nrl-/- retina separates rod, cone and shared elements.
Two dominant CRX retinopathy alleles act through cooperativity. E80A keeps binding at monomeric sites but loses cooperative binding at dimeric sites; K88N becomes highly cooperative and moves to ectopic sites rich in dimeric motifs. This gives a specific mechanism for an allele described in 1997 only as haploinsufficient or dominant-negative.
- 2026mediumStructural, biochemical and bioinformatic analysis; quantitative DNA-binding assays42675056
Analysis of the 36 human paired-like homeodomains derives 11 rules for cooperative binding to the P3 site, and 11 of 12 tested disease-associated variants change cooperativity but not DNA-binding affinity.
Still unresolved
- How much K50 function depends on cooperative P3 dimer sites rather than monomeric TAATCC sites is unsettled. The dimeric site is about a thousand-fold rarer than the monomer, yet CRX alleles that change only cooperativity (E80A loses it, K88N gains it) cause dominant retinopathy.
- How OTX2 and CRX, which read the same TAATCC site and overlap in the developing retina, divide targets is unresolved: shared sites predict redundancy, and no sequence feature in these models separates the two.
A TAATCC match identifies a K50-type homeodomain site, not a protein: OTX1/2, CRX, DMBX1, PITX1-3 and GSC/GSC2 share one motif. Expect roughly two million matches genome-wide, so condition on expression and accessibility; the dimeric TAATCCGATTA is about a thousand-fold rarer and far more informative. Evidence is unusually clean for a homeodomain family - all 12 HOCOMOCO models have in vitro SELEX support, none is ChIP-only, nothing is a cross-species transfer and nothing is flagged as belonging elsewhere. Two exceptions: the ANHX motifs (TTGACAAG) are not K50 sites, and one CIS-BP CRX model carries only 2.3 bits. A TAATTA hit belongs to HD_Q50, which holds the paired-like factors that carry Q50.
References
- Hanes SD, Brent R. DNA specificity of the bicoid activator protein is determined by homeodomain recognition helix residue 9. Cell 1989;57:1275-83. PMID 2500253
- Treisman J, Gönczy P, Vashishtha M, et al. A single amino acid can determine the DNA binding specificity of homeodomain proteins. Cell 1989;59:553-62. PMID 2572327
- Wilson DS, Sheng G, Jun S, et al. Conservation and diversification in homeodomain-DNA interactions: a comparative genetic analysis. Proc Natl Acad Sci U S A 1996;93:6886-91. PMID 8692913
- Baird-Titus JM, Clark-Baldwin K, Dave V, et al. The solution structure of the native K50 Bicoid homeodomain bound to the consensus TAATCC DNA-binding site. J Mol Biol 2006;356:1137-51. PMID 16406070
- Chaney BA, Clark-Baldwin K, Dave V, et al. Solution structure of the K50 class homeodomain PITX2 bound to DNA and implications for mutations that cause Rieger syndrome. Biochemistry 2005;44:7497-511. PMID 15895993
- Wilson D, Sheng G, Lecuit T, et al. Cooperative dimerization of paired class homeo domains on DNA. Genes Dev 1993;7:2120-34. PMID 7901121
- Cain B, Wasmund C, Rowan FC, et al. Decoding the mechanisms of cooperative DNA binding by the Paired-like homeodomain family. Nat Commun 2026;17. PMID 42675056
- Zheng Y, Stormo GD, Chen S. Aberrant homeodomain-DNA cooperative dimerization underlies distinct developmental defects in two dominant CRX retinopathy models. Genome Res 2025;35:242-256. PMID 39715683
- Freund CL, Gregory-Evans CY, Furukawa T, et al. Cone-rod dystrophy due to mutations in a novel photoreceptor-specific homeobox gene (CRX) essential for maintenance of the photoreceptor. Cell 1997;91:543-53. PMID 9390563
- Semina EV, Reiter R, Leysens NJ, et al. Cloning and characterization of a novel bicoid-related homeobox transcription factor gene, RIEG, involved in Rieger syndrome. Nat Genet 1996;14:392-9. PMID 8944018
- Ragge NK, Brown AG, Poloschek CM, et al. Heterozygous mutations of OTX2 cause severe ocular malformations. Am J Hum Genet 2005;76:1008-22. PMID 15846561
- Corbo JC, Lawrence KA, Karlstetter M, et al. CRX ChIP-seq reveals the cis-regulatory architecture of mouse photoreceptors. Genome Res 2010;20:1512-25. PMID 20693478
- Gurnett CA, Alaee F, Kruse LM, et al. Asymmetric lower-limb malformations in individuals with homeobox PITX1 gene mutation. Am J Hum Genet 2008;83:616-22. PMID 18950742
- Parry DA, Logan CV, Stegmann AP, et al. SAMS, a syndrome of short stature, auditory-canal atresia, mandibular hypoplasia, and skeletal abnormalities is a unique neurocristopathy caused by mutations in Goosecoid. Am J Hum Genet 2013;93:1135-42. PMID 24290375
34 motifs in this family — JASPAR 9 · HOCOMOCO 12 · CIS-BP 10 · Vierstra 3
| motif | database | consensus | bits |
|---|---|---|---|
| M03197_2.00 | CIS-BP | TAATCGGATTA | 16.5 |
| AC0060:OTX:Homeodomain | Vierstra | TAATCGGATTA | 16.1 |
| PITX1.H14CORE.1.S.BSB | HOCOMOCO | GGATTAAngGatTa | 15.9 |
| OTX1.H14CORE.1.S.BSB | HOCOMOCO | TAAtCGGATTA | 15.8 |
| PITX3.H14CORE.1.S.BSB | HOCOMOCO | attangGGATTA | 14.2 |
| PITX2 (MA1547.2) | JASPAR | tTAATCC | 12.1 |
| Dmbx1 (MA0883.2) | JASPAR | cGGATTA | 11.9 |
| OTX1.H14CORE.0.SM.BSMB | HOCOMOCO | GGATTA | 11.3 |
| PITX1.H14CORE.0.PSM.APSMA | HOCOMOCO | GGATTA | 11 |
| PITX1 (MA0682.3) | JASPAR | TAATCC | 10.9 |
| PITX3.H14CORE.0.SM.BSMB | HOCOMOCO | GGATTA | 10.9 |
| M04997_2.00 | CIS-BP | TAATCC | 10.9 |
| Crx (MA0467.3) | JASPAR | GGATTA | 10.8 |
| OTX2.H14CORE.0.PSM.APSMA | HOCOMOCO | gGGATTA | 10.8 |
| CRX.H14CORE.0.PSM.APSMA | HOCOMOCO | GGATTA | 10.7 |
| M05022_2.00 | CIS-BP | TAATCC | 10.4 |
| OTX2 (MA0712.3) | JASPAR | GGATTA | 10.3 |
| M00310_2.00 | CIS-BP | tTAATCC | 10.3 |
| PITX2.H14CORE.0.SM.BSMB | HOCOMOCO | GGATTA | 10 |
| M05134_2.00 | CIS-BP | TAATCC | 9.5 |
| AC0061:PITX/OTX:Homeodomain | Vierstra | TAATCC | 9.4 |
| GSC.H14CORE.0.SM.BSMB | HOCOMOCO | GATTA | 9.3 |
| M02096_2.00 | CIS-BP | TTGACAaG | 9.3 |
| M03082_2.00 | CIS-BP | TAATCC | 9.3 |
| AC0398:ANHX:Homeodomain | Vierstra | TTGACAaG | 9.3 |
| GSC (MA0648.2) | JASPAR | TAATCC | 9.1 |
| OTX1 (MA0711.2) | JASPAR | TAATCC | 9.1 |
| PITX3 (MA0714.2) | JASPAR | TAATCC | 8.6 |
| M05383_2.00 | CIS-BP | TAATCC | 8.2 |
| GSC2.H14CORE.0.SM.BSMB | HOCOMOCO | GATTA | 8 |
| M03080_2.00 | CIS-BP | TAATCC | 7.9 |
| DMBX1.H14CORE.0.SM.BSMB | HOCOMOCO | GATTA | 7.6 |
| GSC2 (MA0891.2) | JASPAR | TAATCC | 7.4 |
| M00270_2.00 | CIS-BP | taAt | 2.3 |
HOCOMOCO tags read as assay + quality grade: P ChIP-seq, S HT-SELEX, M methyl-HT-SELEX, G GHT-SELEX, I SMiLE-seq, B PBM; grade A means two or more assay types agreed. A P-only matrix is a record of where the protein was crosslinked, which need not be a sequence it binds itself.
The factors, elsewhere
These entries are written from the primary literature and every citation is checked against PubMed, but this one can still be out of date, too confident, or simply wrong — especially where a motif’s published site and its database matrices disagree. Corrections and additions are welcome, and they are the fastest way to make this page better for everyone.