STAT
The seven human STAT proteins - STAT1, STAT2, STAT3, STAT4, STAT5A, STAT5B and STAT6 - sit latent in the cytoplasm until a receptor-associated JAK kinase phosphorylates a single C-terminal tyrosine. Two phosphorylated monomers then pair through reciprocal SH2-phosphotyrosine contacts and clamp around DNA at a gamma-interferon activation site (GAS): the short inverted repeat TTCNNNGAA. Spacing is almost the only sequence feature that distinguishes them, with STAT6 preferring a four-base spacer and the rest a three-base spacer, so a GAS match names the family and not the member. STATs convert essentially every cytokine, interferon and several growth factor signals into transcription, and germline or somatic STAT mutations cause immunodeficiency, allergy, growth failure and leukemia.
The factors
Seven genes: STAT1, STAT2, STAT3, STAT4, STAT5A, STAT5B and STAT6. All share one architecture - an N-terminal domain, a coiled-coil, a DNA-binding domain built on an immunoglobulin fold like those of NF-kappaB and p53 9630226, a linker, an SH2 domain, and a C-terminal transactivation domain carrying the conserved tyrosine. The protein is latent in the cytoplasm; a receptor-associated JAK phosphorylates that tyrosine, and two monomers pair by reciprocal contacts between the SH2 domain of one and the phosphotyrosine tail of the other, producing a C-shaped clamp that encircles the DNA 9630226,9671298. The phosphotyrosine site of the SH2 domain is structurally coupled to the DNA-binding domain, so dimerization and DNA binding are not independent 9630226. Which STAT is active is set by the cytokine, not by the DNA: STAT1 homodimers by interferon gamma, STAT1-STAT2 plus IRF9 (ISGF3) by type I interferons, STAT3 by the IL-6 family, STAT4 by IL-12, STAT5A and STAT5B by the IL-2 family, growth hormone, prolactin and erythropoietin, and STAT6 by IL-4 and IL-13 28323260. STAT5A and STAT5B have no measured difference in DNA-binding specificity 11053426.
The motif
The GAS element is TTCNNNGAA: two TTC/GAA half-sites facing each other across a three-base spacer, often written TT(N)5AA 7708771. Spacer width is the only substantial selectivity available. STAT6 prefers four spacer bases, TTCNNNNGAA, but still binds three-base sites reasonably well; STAT1 and STAT5 prefer three, and STAT5A tolerates four weakly; none of them binds a lone half-site 11053426. In reporter assays, a five-base TT-to-AA spacing is generally permissive, four favors STAT3-containing complexes and six favors IL-4-activated complexes 7708771. Within the three-base class, STAT1, STAT3, STAT4 and STAT5 are not separable by sequence, and the consensus that the models here agree on, TTCCTGGAA, is compatible with all of them. Two further arrangements matter in practice. First, tetramers: contacts between N-terminal domains let two dimers bind tandem, individually weak GAS sites cooperatively 8896455, so genuine STAT5 targets are often clusters of poor matches rather than one good one, and scanning for a single strong site will miss them. Second, ISGF3: when STAT1 and STAT2 act with IRF9 they read an ISRE rather than a GAS, with the STAT pair on the 5' GGAA and IRF9 on the second GAAA repeat 29317535. That is why several STAT-labelled models here look like ISREs - HOCOMOCO STAT1.H14CORE.1.P.B (TCAGTTTCRTTTTCC), HOCOMOCO STAT2.H14CORE.0.P.B (GGAAANHGAAACTGA) and JASPAR STAT1::STAT2.
Where it acts
STAT occupancy is signal-dependent and transient: sites become occupied within minutes of cytokine exposure and are vacated as the STAT is dephosphorylated and exported 28323260. Targets are both cytokine-inducible promoters and, more often, enhancers, where STATs cooperate with the lineage-defining factors of the responding cell. The practical consequence is that a GAS match in an unstimulated cell type usually corresponds to nothing: the motif is present in the genome whether or not the pathway is on. Type I interferon signaling shifts the element as well as the factor, from GAS to ISRE, so the same STAT1 protein leaves different footprints depending on which interferon acted.
Biology and disease
Human genetics has produced an unusually clear catalogue. Gain-of-function STAT1 mutations impair IL-17 immunity and cause chronic mucocutaneous candidiasis 21727188. Dominant-negative mutations in the STAT3 DNA-binding domain cause hyper-IgE (Job) syndrome - a disease of DNA binding specifically, since the mutant protein dimerizes but cannot read its site 17676033. Somatic STAT3 mutations, mostly in the SH2 domain, occur in large granular lymphocytic leukemia 22591296, and activating STAT5B mutations recur in T-cell leukemias. Loss-of-function STAT5B mutations cause growth hormone insensitivity with severe short stature and immune dysfunction 13679528. Germline gain-of-function STAT6 mutations cause severe early-onset allergic disease 36758835. Together these establish that STAT dosage and DNA-binding competence, not just pathway activation, are rate-limiting in vivo.
Reading a hit
Be conservative. TTCNNNGAA specifies only six bases, so it is expected on the order of several hundred thousand times per strand in a 3.1-Gb genome; STAT matches are among the least specific in this atlas, and this family is the least internally cohesive of the ETS/IRF/STAT group. The evidence behind the models is also weaker than elsewhere: every HOCOMOCO model assigned here is ChIP-derived (evidence letter P, grade B) with no in vitro HT-SELEX or methyl-HT-SELEX model, all nine JASPAR models are ChIP-seq, six of them from mouse rather than human experiments (Stat2, Stat4, Stat5a, Stat5b, Stat6, Stat5a::Stat5b), and two CIS-BP entries are transferred from mouse proteins on DNA-binding-domain similarity - one of which is the motif this atlas selects as the family representative. "This protein binds this sequence" therefore rests largely on in vivo co-occupancy plus older in vitro studies. Expect confusion with IRF: the ISRE-shaped STAT models cross-match IRF motifs, and in this atlas most nearest neighbours of STAT1.H14CORE.1.P.B, STAT2.H14CORE.0.P.B and JASPAR STAT1::STAT2 are labelled IRF, while the JASPAR mouse Stat2 model's best match anywhere lies outside the STAT family. The discriminator is orientation: a GAS is an inverted repeat (TTC facing GAA), an ISRE is a direct repeat of GAAA. Finally, no STAT paralogue can be named from sequence - STAT5A versus STAT5B is impossible - so use the cytokine, the time point and the cell type.
How this was worked out
Natural STAT elements vary but share the form TT(N)5AA; synthetic sites show that the 5-bp TT-to-AA spacing binds STAT complexes generally, 4 bp selects complexes containing Stat3 and 6 bp selects the IL-4-activated complex, with matching effects on transcription.
Still open then: How the dimer read the half-sites, and whether spacing was the only source of selectivity, was unknown.
- 1996mediumIn vitro DNA binding with purified phosphoprotein, mass spectrometry, domain truncation8896455
Purified phosphorylated Stat1 binds strong and weak sites with similar equilibrium affinity but leaves weak sites far faster; two dimers bind tandem sites (not head-to-head sites) more stably than one site, and this cooperativity needs the N-terminal domain.
Still open then: How often natural targets depend on tandem weak sites was not measured.
Crystal structures of a phosphorylated STAT1 dimer and a STAT3beta dimer on DNA show an immunoglobulin-fold DNA-binding domain and a C-shaped clamp around the DNA held by reciprocal SH2-phosphotyrosine contacts structurally coupled to the DNA-binding domain.
- 2001mediumIn vitro selection from random oligonucleotides, quantitative affinity measurements, comparison with natural sites11053426
Side-by-side selection and affinity measurements for STAT1, STAT5A, STAT5B and STAT6 confirm near-identical specificity: STAT6 prefers four-base spacers but binds three-base sites fairly well, STAT1 and STAT5 prefer three, STAT5A weakly tolerates four, none binds a lone half-site, and STAT5A and STAT5B do not differ.
Still open then: STAT3 and STAT4 were not part of the comparison.
A STAT5B mutation is found in a patient with growth hormone insensitivity, linking STAT5B to growth hormone signaling in humans.
De novo heterozygous mutations in the STAT3 DNA-binding domain cause hyper-IgE syndrome; the mutants are non-functional alone and dominant-negative with wild-type STAT3, and patient cells show greatly reduced STAT3 DNA binding.
Heterozygous STAT1 mutations in the coiled-coil domain cause chronic mucocutaneous candidiasis by gain of function: they impair nuclear dephosphorylation and impair IL-17 immunity. Previously described disease-causing STAT1 alleles were all loss-of-function.
Structures of the IRF9 association domain and its complex with the STAT2 coiled-coil domain explain the selective STAT2-IRF9 pairing in ISGF3; a model of ISGF3 on an ISRE puts the STAT pair on the 5' GGAA repeat and IRF9 on the second GAAA repeat.
Still open then: The full ISGF3 complex on DNA has not been solved.
Still unresolved
- No high-throughput in vitro model exists for any STAT in these databases: every HOCOMOCO model is ChIP-only and every JASPAR model ChIP-seq, so whether the spacer-base preferences in the TTCCNGGAA consensus are intrinsic, or reflect partners and chromatin, has not been checked against the older selection and affinity data on a common footing.
- How ISGF3 contacts the ISRE is inferred from a docked model rather than a structure, so whether the ISRE-shaped STAT1 and STAT2 ChIP motifs record STAT base contacts or only IRF9 binding within the complex is unresolved.
- Whether STAT3 has a spacer preference of its own is unsettled: reporter and binding work found that a 4-bp TT-to-AA spacing selects STAT3-containing complexes, while the systematic comparison that found the STATs near-identical did not include STAT3, and the ChIP-derived STAT3 models here read the ordinary three-base-spacer TTCNNNGAA.
A GAS match names the family, never the member: STAT1, STAT3, STAT4 and STAT5 read the same three-spacer inverted repeat, STAT5A and STAT5B have no measured specificity difference, and only STAT6's preference for a four-base spacer gives partial leverage. The element is short enough to occur hundreds of thousands of times by chance, and occupancy is signal-dependent, so a hit in an unstimulated cell type usually means nothing. Evidence is weaker here than for most families: all HOCOMOCO models assigned to STAT are ChIP-only, all JASPAR models are ChIP-seq and two-thirds of them are mouse, and two CIS-BP entries - including the family representative - are inferred from mouse proteins. Expect overlap with IRF, since ISGF3 (STAT1-STAT2-IRF9) reads the ISRE, not a GAS.
References
- Chen X, Vinkemeier U, Zhao Y, et al. Crystal structure of a tyrosine phosphorylated STAT-1 dimer bound to DNA. Cell 1998;93:827-39. PMID 9630226
- Becker S, Groner B, Müller CW. Three-dimensional structure of the Stat3beta homodimer bound to DNA. Nature 1998;394:145-51. PMID 9671298
- Ehret GB, Reichenbach P, Schindler U, et al. DNA binding specificity of different STAT proteins. Comparison of in vitro specificity with natural target sites. J Biol Chem 2001;276:6675-88. PMID 11053426
- Seidel HM, Milocco LH, Lamb P, et al. Spacing of palindromic half sites as a determinant of selective STAT (signal transducers and activators of transcription) DNA binding and transcriptional activity. Proc Natl Acad Sci U S A 1995;92:3041-5. PMID 7708771
- Vinkemeier U, Cohen SL, Moarefi I, et al. DNA binding of in vitro activated Stat1 alpha, Stat1 beta and truncated Stat1: interaction between NH2-terminal domains stabilizes binding of two dimers to tandem DNA sites. EMBO J 1996;15:5616-26. PMID 8896455
- Rengachari S, Groiss S, Devos JM, et al. Structural basis of STAT2 recognition by IRF9 reveals molecular insights into ISGF3 function. Proc Natl Acad Sci U S A 2018;115:E601-E609. PMID 29317535
- Villarino AV, Kanno Y, O'Shea JJ. Mechanisms and consequences of Jak-STAT signaling in the immune system. Nat Immunol 2017;18:374-384. PMID 28323260
- Minegishi Y, Saito M, Tsuchiya S, et al. Dominant-negative mutations in the DNA-binding domain of STAT3 cause hyper-IgE syndrome. Nature 2007;448:1058-62. PMID 17676033
- Liu L, Okada S, Kong XF, et al. Gain-of-function human STAT1 mutations impair IL-17 immunity and underlie chronic mucocutaneous candidiasis. J Exp Med 2011;208:1635-48. PMID 21727188
- Kofoed EM, Hwa V, Little B, et al. Growth hormone insensitivity associated with a STAT5b mutation. N Engl J Med 2003;349:1139-47. PMID 13679528
- Koskela HL, Eldfors S, Ellonen P, et al. Somatic STAT3 mutations in large granular lymphocytic leukemia. N Engl J Med 2012;366:1905-13. PMID 22591296
- Baris S, Benamar M, Chen Q, et al. Severe allergic dysregulation due to a gain of function mutation in the transcription factor STAT6. J Allergy Clin Immunol 2023;152:182-194.e7. PMID 36758835
26 motifs in this family — JASPAR 9 · HOCOMOCO 8 · CIS-BP 5 · Vierstra 4
| motif | database | consensus | bits |
|---|---|---|---|
| STAT1.H14CORE.1.P.BPB | HOCOMOCO | TcAGTTTCrtTTTCC | 17.1 |
| M08019_2.00 | CIS-BP | ThAGTTTCanTTTC | 16.6 |
| STAT1::STAT2 (MA0517.2)dimer | JASPAR | AGTTTCanTTTCc | 15.8 |
| STAT2.H14CORE.0.P.BPB | HOCOMOCO | gGAAAnhGAAACTgA | 15.6 |
| M08015_2.00 | CIS-BP | TTTCCcGGAAatCA | 15.5 |
| AC0669:STAT5A:STAT | Vierstra | TTCccaGAAttnnnnTTCCnvG | 14.9 |
| M08013_2.00 | CIS-BP | catTTCChGGAA | 14.7 |
| STAT1 (MA0137.4) | JASPAR | TTCcaGGAA | 14.1 |
| M08173_2.00Mus musculus 0.96 | CIS-BP | TTCCtgAGAA | 14.1 |
| Stat6 (MA0520.2) | JASPAR | TTCCtgAGAA | 13.6 |
| STA5B.H14CORE.0.P.BPB | HOCOMOCO | TTCttaGAA | 13.5 |
| M08174_2.00Mus musculus 0.94 | CIS-BP | tTTCCtgGAA | 13.5 |
| STA5A.H14CORE.0.P.BPB | HOCOMOCO | TTCtaaGAA | 13.2 |
| Stat5a::Stat5b (MA0519.2)dimer | JASPAR | TTCCaAGAA | 13.1 |
| STAT6.H14CORE.0.P.BPB | HOCOMOCO | TTCtcaGGAA | 13.1 |
| Stat4 (MA0518.2) | JASPAR | TTCcaGGAAa | 13 |
| Stat5b (MA1625.2) | JASPAR | TTCCCAGAA | 13 |
| Stat2 (MA1623.2) | JASPAR | gAAACAGAAA | 12.9 |
| STAT3 (MA0144.3) | JASPAR | TTCTgGGAA | 12.6 |
| Stat5a (MA1624.2) | JASPAR | TTCCAAGAA | 12.5 |
| STAT1.H14CORE.0.P.BPB | HOCOMOCO | ccaTTTCCtggAA | 12.2 |
| STAT3.H14CORE.0.P.BPB | HOCOMOCO | TTCCcGGAA | 11.4 |
| AC0668:STAT/ETV:STAT | Vierstra | TTCCnGGAA | 11.3 |
| STAT4.H14CORE.0.P.BPB | HOCOMOCO | TTCCcgGAA | 10.5 |
| AC0684:STAT:STAT | Vierstra | TTCCtgaGAA | 10.4 |
| AC0667:STAT/STAT5A:STAT | Vierstra | TTCc | 5.3 |
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.