Other / unclassified
This is not a motif family. It is the residual category for motifs whose factor has no assigned structural family in this ontology, and in the current data it holds three matrices for three proteins. Two are the paralogous corepressors LCOR and LCORL, which carry a psq-type helix-turn-helix rather than a homeodomain and whose models are 4-5 bit A-rich strings that carry almost no information. The third is LRRFIP1, cloned as GC-binding factor 2, whose only model is a GC-rich matrix derived from ChIP-seq alone and resembles neither of the others. Internal cohesion is 0.60 and no member has a same-family neighbour that it matches. Read a hit here as unassigned rather than as shared biology, and re-report it to whichever named family the sequence actually matches.
The factors
Three motifs, three proteins. The definition is broader than the contents: it also covers chromatin regulators without a classical DNA-binding domain and unlabelled archetypes, but both of those have entries of their own, so what lands here is proteins believed to bind DNA whose domain has no family of its own.
Two of the three are paralogues. LCOR, ligand-dependent corepressor, is recruited to agonist-bound nuclear receptors through a single LXXLL motif and represses both by binding histone deacetylases directly and by recruiting CtBP corepressors through two consensus motifs; it attenuates receptor signalling rather than activating a programme of its own 12535528. LCORL is its larger relative. Both carry a psq-type helix-turn-helix - the pipsqueak fold, a member of the homeodomain-like superfamily - which is a real DNA-binding domain 29425488, but this atlas has no pipsqueak family and the fold is not a homeodomain, so neither protein has anywhere else to go.
The third, LRRFIP1, entered the literature as GCF2, found by cross-hybridisation with the cDNA of the GC-rich repressor GCF and shown to bind and repress epidermal growth factor receptor promoter fragments through a major site at nucleotides -249 to -233 2556218,9705290. It has no recognised DNA-binding domain 29425488; the part of it that has been solved is a dimeric coiled coil, and its best-documented activity is cytosolic, as a nucleic-acid-binding protein that drives interferon-beta production through beta-catenin 20453844.
The motif
There is no consensus for this bucket and none should be quoted. The CnAAa string the pipeline reports is only the medoid of three unrelated matrices, and the medoid is the weakest of the three.
The three entries: CIS-BP M02433, labelled LCOR, reads CnAAa and carries 4.1 bits; its provenance record marks it as taken from a Meleagris gallopavo - wild turkey - protein whose DNA-binding domain is 100 percent identical to the human one 25215497. CIS-BP M01304, labelled LCORL, reads cchAAa and carries 4.9 bits. The two match each other at 0.97, but each also matches a motif in another family at 0.83-0.84: at five bits an A-rich matrix is not distinguishable from anything else A-rich. LCORL turns up again in the forkhead entry as part of the Vierstra archetype AC0014:FOXR/LCORL, a 3.4-bit AAa matrix pooled with genuine forkhead models 32728250. The third entry, HOCOMOCO LRRF1.H14CORE.0.P.B for LRRFIP1, reads aaGCTCGC with 9.6 bits. Its evidence code is P alone - ChIP-seq, with no in vitro assay - and grade B means two experiments of that one type 37971293. It has no same-family neighbour anywhere in its top thirty; its best match outside the bucket is only 0.66. Family cohesion is 0.60.
The pipeline treats residual buckets as places where membership says nothing about what a matrix should look like, so it suppresses the suspect flag and reports a resemblance instead; the reported zero suspect motifs is a property of that rule, not a sign that these three belong together.
Where it acts
Not established as a cis-element for any of the three. LCOR and LCORL act as corepressors recruited by nuclear receptors, so their functional footprint on the genome is at oestrogen, glucocorticoid and related receptor sites 12535528, and finding CNAAA under a peak tells you nothing about which receptor brought the corepressor there. The one mapped element for LRRFIP1 is the GC-rich region of the EGFR promoter around -240 9705290; its ChIP-derived matrix will pick out promoter-proximal GC-rich sequence, as most ChIP-only GC-rich models do whatever protein they are labelled with. A five- or six-base A-rich matrix of the kind LCOR and LCORL contribute matches several million times in 3.1 Gb on either strand, so a scan against them returns noise at any threshold that returns anything.
Biology and disease
None of the three has a Mendelian disease attributed to it, but the biology is real. LCOR primes interferon responses in mammary epithelium; miR-199a represses it to protect normal and cancerous mammary stem cells from interferon-induced differentiation and senescence, and that axis is active in poorly differentiated ER-negative breast tumours and associated with poor outcome 28530657. LCORL sits in the NCAPG-LCORL interval, one of the most reproducible loci for body size and stature in human and livestock association studies; knockout mice are born small, stay lean, eat less, have reduced circulating IGF-1 and show improved glucose tolerance 39467326. LRRFIP1 represses the EGFR and RhoA promoters as GCF2 9705290,30709060, regulates Wnt/beta-catenin signalling through a specific spliced isoform 28322931, and acts in innate immunity as a cytosolic sensor of double-stranded RNA and DNA 20453844.
Reading a hit
Do not report a match here as a match to a family motif. Report the matrix instead, then ask what it is. If the hit is M02433 or M01304, the honest reading is that a four- to five-bit A-rich matrix matched, which is no evidence, and the LCOR matrix is in any case a cross-species transfer from a bird. If it is LRRF1.H14CORE.0.P.B, read it as a GC-rich model built from ChIP-seq alone and compare the sequence against the GC-box families - KLF/SP, MAZ, EGR, NRF1 - before crediting LRRFIP1 with the site. More generally, a residual category exists so that well-supported families are not diluted by unresolved single cases, and a hit in it should be read as unassigned. For LCOR and LCORL the informative question is not what they bind but which nuclear receptor recruited them; for LRRFIP1 it is which GC-box family the sequence belongs to, and whether the protein was in the nucleus at all.
How this was worked out
- 1998mediumcDNA cloning, promoter-fragment binding assays, reporter cotransfection with site mutation9705290
GCF2, the protein now called LRRFIP1, is cloned by differential hybridization with the cDNA of the GC-binding repressor GCF; it binds epidermal growth factor receptor promoter fragments at a major site between -249 and -233 and represses the promoter only when that site is intact.
Still open then: Whether the protein occupies that site, or any GC-rich site, in cells was not tested.
- 2003mediumProtein interaction assays in vitro and in cells, reporter assays, HDAC inhibitor treatment12535528
LCoR is defined as a ligand-dependent corepressor: it is recruited to agonist-bound nuclear receptors through a single LXXLL motif and represses by binding HDACs directly and by recruiting CtBP corepressors, so its genomic placement is set by the receptor.
Still open then: Whether LCoR makes any sequence-specific DNA contact of its own, rather than riding on the receptor, was not addressed.
- 2010mediumKnockdown and overexpression in macrophages, viral and bacterial infection, protein interaction assays20453844
LRRFIP1 is recast as a cytosolic nucleic-acid sensor that binds exogenous double-stranded RNA and DNA and drives interferon-beta production through beta-catenin and IRF3, moving its best-supported activity out of the nucleus.
Still open then: How this activity relates to the GC-rich promoter binding reported for GCF2 was not addressed.
The CIS-BP survey measures DNA preferences for more than 1,000 transcription factors from 131 eukaryotes and finds that closely related DNA-binding domains almost always share them. That rule is what lets CIS-BP file a matrix measured on a wild turkey protein, with a DNA-binding domain 100% identical to the human one, under human LCOR.
The human transcription factor census keeps LCOR and LCORL as likely sequence-specific factors with a pipsqueak DNA-binding domain and high-throughput in vitro motifs, LCOR's taken from an ortholog at 100% identity. It judges LRRFIP1 unlikely to be a sequence-specific factor and records nothing that would indicate DNA binding.
Still open then: The census note on LRRFIP1 does not engage with the 1998 report of EGFR-promoter binding by GCF2, so the disagreement was recorded rather than resolved.
The reference footprint atlas clusters matrices from several collections into archetypes named after their members. LCORL's low-information A-rich matrix is pooled with forkhead models in AC0014:FOXR/LCORL.
Still open then: An archetype name records which matrices clustered, not which protein binds.
HOCOMOCO is rebuilt from motif discovery in more than 14,000 ChIP-seq peak sets plus HT-SELEX data. In the current release LRRFIP1's only matrix is a GC-rich ChIP-seq-only model, LRRF1.H14CORE.0.P.B (grade B), even though the census had found no evidence that the protein binds DNA.
Still open then: Whether aaGCTCGC is LRRFIP1's own preference or that of a co-bound GC-box factor has not been tested in vitro.
Still unresolved
- Whether LRRFIP1 binds DNA sequence-specifically at all is unresolved. The 1998 GCF2 study reported repression through a mapped site in the EGFR promoter; the human transcription factor census found nothing indicating DNA binding; and the only matrix for the protein is ChIP-derived, with no in vitro assay behind it.
- Whether the pipsqueak-type helix-turn-helix of LCOR and LCORL has a sequence preference worth modelling is not settled. The in vitro assays recorded for both yield only 4-5-bit A-rich matrices, while the biochemistry of LCOR describes recruitment by agonist-bound nuclear receptors rather than direct recognition of a DNA sequence.
Treat this as an administrative category, not a biological one. It holds three matrices for three proteins whose medoid similarity is 0.60, and the group has no shared consensus: two are 4-5 bit A-rich matrices for the corepressors LCOR and LCORL, one of them transferred from a wild turkey protein, and the third is a 9.6-bit GC-rich model for LRRFIP1 supported by ChIP-seq alone at grade B. No member has a same-family neighbour it actually matches. The reported zero suspect motifs reflects a pipeline rule that suppresses the flag inside residual buckets, not internal agreement. Do not build a consensus, a logo or a representative matrix for this group, and re-report any hit to the named family the sequence matches.
References
- Kageyama R, Pastan I. Molecular cloning and characterization of a human DNA binding factor that represses transcription. Cell 1989;59:815-25. PMID 2556218
- Reed AL, Yamazaki H, Kaufman JD, et al. Molecular cloning and characterization of a transcription regulator with homology to GC-binding factor. J Biol Chem 1998;273:21594-602. PMID 9705290
- Fernandes I, Bastien Y, Wai T, et al. Ligand-dependent nuclear receptor corepressor LCoR functions by histone deacetylase-dependent and -independent mechanisms. Mol Cell 2003;11:139-50. PMID 12535528
- Yang P, An H, Liu X, et al. The cytosolic nucleic acid sensor LRRFIP1 mediates the production of type I interferon via a beta-catenin-dependent pathway. Nat Immunol 2010;11:487-94. PMID 20453844
- Weirauch MT, Yang A, Albu M, et al. Determination and inference of eukaryotic transcription factor sequence specificity. Cell 2014;158:1431-1443. PMID 25215497
- Labbé P, Faure E, Lecointe S, et al. The alternatively spliced LRRFIP1 Isoform-1 is a key regulator of the Wnt/β-catenin transcription pathway. Biochim Biophys Acta Mol Cell Res 2017;1864:1142-1152. PMID 28322931
- Celià-Terrassa T, Liu DD, Choudhury A, et al. Normal and cancerous mammary stem cells evade interferon-induced constraint through the miR-199a-LCOR axis. Nat Cell Biol 2017;19:711-723. PMID 28530657
- Lambert SA, Jolma A, Campitelli LF, et al. The Human Transcription Factors. Cell 2018;172:650-665. PMID 29425488
- Takimoto M. Multidisciplinary Roles of LRRFIP1/GCF2 in Human Biological Systems and Diseases. Cells 2019;8. PMID 30709060
- Vierstra J, Lazar J, Sandstrom R, et al. Global reference mapping of human transcription factor footprints. Nature 2020;583:729-736. PMID 32728250
- Vorontsov IE, Eliseeva IA, Zinkevich A, et al. HOCOMOCO in 2024: a rebuild of the curated collection of binding models for human and mouse transcription factors. Nucleic Acids Res 2024;52:D154-D163. PMID 37971293
- Wyler SC, Gahlot S, Bideyan L, et al. LCoRL Regulates Growth and Metabolism. Endocrinology 2024;165. PMID 39467326
3 motifs in this family — HOCOMOCO 1 · CIS-BP 2
| motif | database | consensus | bits |
|---|---|---|---|
| LRRF1.H14CORE.0.P.BPB | HOCOMOCO | aaGCTCGC | 9.6 |
| M01304_2.00 | CIS-BP | cchAAa | 4.9 |
| M02433_2.00Meleagris gallopavo 1 | CIS-BP | CnAAa | 4.1 |
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.