GLI
GLI factors are five-fingered C2H2 zinc-finger proteins that read the nine-base site GACCACCCA and form the transcriptional output of the Hedgehog pathway. Humans have three canonical members: GLI1, an activator that is itself a Hedgehog target gene, and GLI2 and GLI3, which are made as full-length activators but are partially degraded into N-terminal repressors when Hedgehog signaling is off. The related GLIS1, GLIS2 and GLIS3 proteins read a close variant, GACCCCCCAC. Binding is monomeric and the informative bases are concentrated at the 3' end of the site, which fingers four and five contact directly. GLI sites control limb and neural-tube patterning, and pathway mutations cause polydactyly syndromes, basal cell carcinoma and SHH-subgroup medulloblastoma.
The factors
Three canonical human GLI proteins, GLI1, GLI2 and GLI3, are the vertebrate counterparts of Drosophila Cubitus interruptus. Each carries five tandem C2H2 zinc fingers and binds DNA as a monomer. GLI2 and GLI3 are bifunctional: an N-terminal repressor domain and a C-terminal activation domain flank the finger array, and limited proteolysis removes the activation domain to leave a repressor 10693759,39576500. GLI1 lacks the N-terminal repressor region and is a dedicated activator; because GLI1 transcription is itself induced by Hedgehog signaling, GLI1 mRNA is widely used as a pathway readout. GLI2 and GLI3 are broadly expressed during development, whereas GLI1 accumulates only where the pathway is active.
The GLIS subfamily, GLIS1, GLIS2 and GLIS3, is the nearest relative: Kruppel-like proteins with five fingers whose DNA-binding domains are close enough to GLI that their motifs cluster together 14500813, but whose expression is largely non-overlapping (GLIS3 in thyroid, pancreatic islets and kidney, GLIS2 in kidney, GLIS1 in a restricted set of tissues and in oocytes). A fourth entry, GLI4 (ZNF767P), is annotated as a family member but the matrices assigned to it in these databases do not resemble a GLI site; treat it separately.
The motif
The consensus GACCACCCA was defined by selecting genomic fragments bound by recombinant GLI: the protein protected 23-24 bases at each site, and all recovered sites contained the same nine-base core 2105456. The crystal structure of the five GLI fingers on a high-affinity site explains the shape of the information content. Finger one makes no DNA contact; fingers two through five wrap into the major groove without the strictly periodic geometry of Zif268; and fingers four and five make extensive base contacts across a conserved nine-base-pair stretch whose conformation is intermediate between B- and A-form DNA 8378770. The consequence is that the 3' half of the site (reverse-strand TGGGTGGTC) carries most of the specificity while the 5' flank is tolerant.
Systematic affinity measurements of every single substitution show a wide dynamic range 16413481, and reporter assays with natural promoters show that many low-affinity variants still drive strong activation at moderate GLI concentrations 20070907. So "non-consensus" GLI sites are functional, and a strict consensus search under-counts real elements. The GLIS paralogues prefer GACCCCCCAC, with the central ACC replaced by CCC 23332764; the databases keep GLI and GLIS as separate matrices, but their cross-similarity is high enough that only strong matches can be assigned to one subfamily. The data consensus here, GACCACCCACgdcG, is the same core with a weakly informative 3' extension.
Where it acts
GLI binding is mostly at distal enhancers rather than core promoters, and GLI sites are not a CpG-island feature. ChIP of GLI activators in Hedgehog-responsive mouse neural tissue recovers on the order of a thousand high-confidence regions, enriched at known pathway components (Ptch1, Gli1, Hhip) and at patterning genes (Foxa2, Nkx2-2), with the motif recovered de novo 17442700. Because the informative core is nine bases with degenerate flanks, exact matches occur tens of thousands of times in the human genome and the great majority are never occupied. Occupancy tracks pathway activity and chromatin: the same sequence is silent in tissue where GLI2 and GLI3 are not stabilized. Frequent neighbors in Hedgehog-responsive enhancers include SOX2, FOXA2 and NKX factors, and ZIC proteins, which bind an overlapping sequence and can physically associate with GLI zinc fingers.
Biology and disease
Hedgehog ligand binding to PTCH1 relieves inhibition of SMO, which blocks the phosphorylation of GLI2 and GLI3 by PKA, GSK3 and CK1 and thereby blocks their processing 39576500. In the unstimulated state GLI3 is converted into a truncated repressor, GLI3R; in the developing limb this produces a graded anterior-to-posterior distribution of repressor that patterns the digits 10693759. SUFU sequesters GLI proteins and opposes the pathway.
Gli2-null mice lack a floor plate and show reduced Shh responses 9636069; Gli1-null mice are viable and only reveal defects in combination with a Gli2 mutation 10725236, consistent with GLI2 being the principal activator and GLI1 an amplifier. Truncating GLI3 mutations cause Greig cephalopolysyndactyly syndrome, while frameshifts in the middle third of the gene, which leave a repressor-like product, cause Pallister-Hall syndrome 20301619,15739154. GLI1 was originally cloned as an amplified, highly expressed gene in a human glioma 3563490, and Hedgehog-driven tumors (basal cell carcinoma, SHH-subgroup medulloblastoma) signal through GLI even when the initiating lesion is in PTCH1 or SMO 39576500. On the GLIS side, biallelic GLIS3 loss causes neonatal diabetes with congenital hypothyroidism 16715098, GLIS2 (NPHP7) loss causes nephronophthisis 17618285, and GLIS1 improves the efficiency of somatic-cell reprogramming 21654807.
Reading a hit
A GACCACCCA match identifies the GLI/GLIS class and nothing finer. GLI1, GLI2 and GLI3 have effectively the same matrix, so no sequence argument distinguishes them; assignment to a member comes from expression or ChIP data. More importantly, a hit does not tell you the sign of the output, because GLI3R binds the same site as GLI2 full-length activator. With nine informative bases the site is common enough to occur by chance in any few-kilobase window, so an isolated match is weak evidence; conservation, open chromatin in a Hedgehog-responsive tissue, and proximity to other pathway targets all help. Conversely, the functional importance of low-affinity variants means absence of a consensus match does not exclude GLI regulation. In HOCOMOCO, the GLI2 and GLI3 models carry in vitro support (HT-SELEX and methyl-SELEX, grade A), whereas the GLI1 model is ChIP-only at grade C; the GLI4 entries are ChIP-only and do not look like a GLI site at all.
How this was worked out
GLI is identified as a gene amplified more than 50-fold and highly expressed in a malignant glioma; its sequence predicts a zinc-finger protein.
Still open then: Whether the protein bound DNA, and what sequence, was unknown.
Binding-site selection from total human DNA with a recombinant GLI protein recovers three sites, each footprinted over 23-24 bases and each containing GACCACCCA.
Still open then: The consensus rested on three natural sites; which positions were essential was not tested.
Crystal structure of the five GLI fingers on a high-affinity site: finger 1 makes no DNA contact and fingers 4 and 5 make most base contacts across a nine-base-pair stretch, explaining why the 3' half of the site carries most of the specificity.
Still open then: A single site was solved; how variant sites are tolerated was not addressed.
- 2000mediumBiochemical analysis of GLI3 processing in chick and mouse limb; human syndrome and chick talpid2 mutants10693759
Hedgehog-regulated, PKA-dependent processing of GLI3 produces a repressor distributed as an anterior-posterior gradient in the limb, so the same site can receive activator or repressor input.
Still open then: How an enhancer distinguishes activator from repressor bound to the same sequence.
- 2006mediumHigh-throughput in vitro binding-affinity assay, computational enhancer prediction, transgenic mouse reporters16413481
High-throughput affinity measurements for GLI1-3 across single-substitution variants of the site give a quantitative specificity model, used to predict Hedgehog-responsive enhancers genome-wide.
Still open then: Many natural GLI target sites scored as low affinity, raising the question of whether they were functional.
ChIP of an epitope-tagged GLI activator in differentiating mouse neural progenitors identifies direct targets, including enhancers of Nkx2-2 and Nkx2-1, confirmed in transgenic embryos.
Still open then: Coverage was limited to candidate loci rather than the whole genome.
- 2010mediumLuciferase reporters in the PTCH1 promoter context, natural non-consensus promoters20070907
Reporters carrying every single substitution of the site show that many low-affinity variants activate strongly at moderate GLI concentration, revising the view that only consensus GACCACCCA sites are functional.
Still open then: Reporter assays in cultured cells do not show how affinity is used to read the Hedgehog gradient in tissue.
A large HT-SELEX survey of human transcription factors gives the GLIS paralogues a GACCCCCCAC site, which differs from the GLI1-3 site GACCACCCA at the central ACC.
Still unresolved
- Why many functional Hedgehog-responsive enhancers use low-affinity GLI sites is unresolved: one view is that weak sites restrict the response to cells with high activator levels and so sharpen the readout of the Hedgehog gradient, the other that weak sites are simply sufficient at moderate GLI concentration and are not selected against.
- Whether GLI activators and the GLI3 repressor occupy the same genomic sites in vivo, or whether some sites are read preferentially by one form, is not settled; the three proteins share one matrix in vitro, so any difference would have to come from chromatin, partners or site arrangement.
- Whether GLI4 binds a GLI-type site at all is unknown: the only matrices assigned to it are ChIP-derived and share a TGGCATTCAAGGCC consensus unrelated to GACCACCCA, and a Vierstra archetype labeled GLI carries the same sequence.
A hit tells you "GLI/GLIS", never which paralogue and never whether the factor present is an activator or the GLI3 repressor. The nine-base core occurs frequently by chance, so isolated matches carry little weight without conservation or Hedgehog-responsive chromatin. GLI and GLIS matrices differ at only one or two central positions, so weak matches cannot be assigned to a subfamily. Low-affinity, non-consensus sites are demonstrably functional, so consensus searching both over-calls inert sites and misses real ones. The GLI4/ZNF767P matrices in these databases are ChIP-derived and unlike the GLI site; the Vierstra archetype carrying that motif is named GLI but groups poorly with the rest of the family.
References
- Kinzler KW, Vogelstein B. The GLI gene encodes a nuclear protein which binds specific sequences in the human genome. Mol Cell Biol 1990;10:634-42. PMID 2105456
- Pavletich NP, Pabo CO. Crystal structure of a five-finger GLI-DNA complex: new perspectives on zinc fingers. Science 1993;261:1701-7. PMID 8378770
- Hallikas O, Palin K, Sinjushina N, et al. Genome-wide prediction of mammalian enhancers based on analysis of transcription-factor binding affinity. Cell 2006;124:47-59. PMID 16413481
- Winklmayr M, Schmid C, Laner-Plamberger S, et al. Non-consensus GLI binding sites in Hedgehog target gene regulation. BMC Mol Biol 2010;11:2. PMID 20070907
- Wang B, Fallon JF, Beachy PA. Hedgehog-regulated processing of Gli3 produces an anterior/posterior repressor gradient in the developing vertebrate limb. Cell 2000;100:423-34. PMID 10693759
- Vokes SA, Ji H, McCuine S, et al. Genomic characterization of Gli-activator targets in sonic hedgehog-mediated neural patterning. Development 2007;134:1977-89. PMID 17442700
- Ding Q, Motoyama J, Gasca S, et al. Diminished Sonic hedgehog signaling and lack of floor plate differentiation in Gli2 mutant mice. Development 1998;125:2533-43. PMID 9636069
- Park HL, Bai C, Platt KA, et al. Mouse Gli1 mutants are viable but have defects in SHH signaling in combination with a Gli2 mutation. Development 2000;127:1593-605. PMID 10725236
- Adam MP, Bick S, Mirzaa GM, et al. GLI3-Related Greig Cephalopolysyndactyly Syndrome. 1993. PMID 20301619
- Maekawa M, Yamaguchi K, Nakamura T, et al. Direct reprogramming of somatic cells is promoted by maternal transcription factor Glis1. Nature 2011;474:225-9. PMID 21654807
- Kim YS, Nakanishi G, Lewandoski M, et al. GLIS3, a novel member of the GLIS subfamily of Krüppel-like zinc finger proteins with repressor and activation functions. Nucleic Acids Res 2003;31:5513-25. PMID 14500813
- Yoshida S, Yoshida K. Regulatory mechanisms governing GLI proteins in hedgehog signaling. Anat Sci Int 2025;100:143-154. PMID 39576500
- Kinzler KW, Bigner SH, Bigner DD, et al. Identification of an amplified, highly expressed gene in a human glioma. Science 1987;236:70-3. PMID 3563490
- Jolma A, Yan J, Whitington T, et al. DNA-binding specificities of human transcription factors. Cell 2013;152:327-39. PMID 23332764
- Johnston JJ, Olivos-Glander I, Killoran C, et al. Molecular and clinical analyses of Greig cephalopolysyndactyly and Pallister-Hall syndromes: robust phenotype prediction from the type and position of GLI3 mutations. Am J Hum Genet 2005;76:609-22. PMID 15739154
- Senée V, Chelala C, Duchatelet S, et al. Mutations in GLIS3 are responsible for a rare syndrome with neonatal diabetes mellitus and congenital hypothyroidism. Nat Genet 2006;38:682-7. PMID 16715098
- Attanasio M, Uhlenhaut NH, Sousa VH, et al. Loss of GLIS2 causes nephronophthisis in humans and mice by increased apoptosis and fibrosis. Nat Genet 2007;39:1018-24. PMID 17618285
27 motifs in this family — JASPAR 6 · HOCOMOCO 11 · CIS-BP 7 · Vierstra 3
| motif | database | consensus | bits |
|---|---|---|---|
| GLIS1 (MA0735.2) | JASPAR | aGACCCCCCACGAaG | 25.2 |
| M08394_2.00 | CIS-BP | TgggCnTGGCATTCAAGGCCc | 24.6 |
| AC0433:GLI:C2H2_ZF | Vierstra | ggCnTGGCATTCAAGGCC | 22.9 |
| GLI3 (MA1491.3) | JASPAR | GACCACCCACGtcG | 22.3 |
| GLIS3 (MA0737.1) | JASPAR | GACCCCCCACGAAG | 21 |
| M02880_2.00 | CIS-BP | CTTCGTGGGGGGTC | 21 |
| GLI3.H14CORE.0.PS.APSA | HOCOMOCO | CgwCGTGGGTGGTC | 20.6 |
| GLI2.H14CORE.0.PSM.APSMA | HOCOMOCO | CdaCGTGGGTGGTC | 19.7 |
| GLIS1.H14CORE.2.SM.BSMB | HOCOMOCO | CttcGTGGGGGGTa | 19.5 |
| M04402_2.00 | CIS-BP | GACCACCCACGwcG | 18.4 |
| AC0527:GLIS:C2H2_ZF | Vierstra | CttCGTGGGGGGT | 17.1 |
| GLIS2 (MA0736.1) | JASPAR | gACCCCCCGCGaaG | 16.9 |
| M04416_2.00 | CIS-BP | GACCACCCACgdcG | 16.6 |
| M02643_2.00 | CIS-BP | GACCACCCAcG | 16.1 |
| GLIS2.H14CORE.0.SM.BSMB | HOCOMOCO | CttCGtGGGGGGTc | 16 |
| M04449_2.00 | CIS-BP | ACCCCCCgCGahG | 15.6 |
| AC0526:GLI:C2H2_ZF | Vierstra | GACCACCCAC | 14.8 |
| GLI1.H14CORE.0.P.CPC | HOCOMOCO | cTGGGTGGTCt | 14.7 |
| GLI3.H14CORE.1.M.CMC | HOCOMOCO | GTGGTCt | 13.7 |
| Gli2 (MA0734.4) | JASPAR | GACCACCCA | 13.1 |
| Gli1 (MA1990.2) | JASPAR | aGACCACCCA | 12.4 |
| GLIS3.H14CORE.0.P.CPC | HOCOMOCO | CCTGGGgGG | 12.1 |
| GLIS3.H14CORE.1.S.CSC | HOCOMOCO | GtGGGGGGT | 12.1 |
| GLIS1.H14CORE.0.P.BPB | HOCOMOCO | CtCtgcgGGaGGc | 11.9 |
| M08345_2.00 | CIS-BP | CnCtcCgGGgGGc | 11.6 |
| GLI4.H14CORE.0.P.CPC | HOCOMOCO | TgnbchTggCaTTCaaGGCC | 11.3 |
| GLIS1.H14CORE.1.P.BPB | HOCOMOCO | cCCTGGGgGG | 11.1 |
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.