NFI (CTF/NF-1)
Nuclear factor I proteins bind a 15-bp palindrome, TTGGC followed by five variable bases and GCCAA. Four human genes, NFIA, NFIB, NFIC and NFIX, encode DNA-binding domains that are about 85% identical, so they bind as homodimers or heterodimers in any combination and cannot be told apart from sequence. The domain adopts a SMAD MH1-like fold with a bound zinc ion, which is why the family is classified with SMAD rather than with any CCAAT-box factor. The full palindrome is highly informative, but NFI also occupies single TTGGC half-sites. NFI factors keep stem cell populations in glia, retina, skin and muscle on track, and are mutated in human overgrowth and brain malformation syndromes.
The factors
Four human genes, NFIA, NFIB, NFIC and NFIX, each produce several splice isoforms 10831836. The N-terminal region carries both the DNA-binding and the dimerization functions; the C-terminal regions differ and carry the transactivation or repression activity 10831836. Because the DNA-binding domains are roughly 85% identical 41261119, all six pairwise heterodimers form in addition to the four homodimers, and the effective factor in a cell is usually a mixture. A crystal and solution study of NFI-X bound to its consensus palindrome showed a MH1-like fold with a structural Zn2+ site, an obligate dimeric assembly, and the base contacts that set specificity; the authors present it as prototypic for the whole family given the DBD identity 41261119. This places NFI in the same structural class as the SMADs, which is why TFClass and the archetype labels used here call the group "SMAD/NF-1". Expression is broad but graded: NFIA and NFIB dominate in developing CNS and glia, NFIX in neural stem cells, muscle and hematopoiesis, NFIC in tooth and kidney. NFI was originally purified as a host factor required for adenovirus type 2 DNA replication, and independently as "CTF" (CCAAT box transcription factor) and as the chicken "TGGCA protein"; all three are the same family 4040852,3398920.
The motif
The consensus is the 15-bp palindrome TTGGC(N5)GCCAA, equivalently written TGGC(N5)GCCA 4040852,3039460. Each monomer reads one TTGGC half-site, so the information sits almost entirely on the two half-sites; the five central bases are a spacer and contribute little. Spacing matters more than spacer sequence, though it is not rigid: early binding studies accepted 6- or 7-bp gaps between the motifs TGG and GCCAA, 6 bp being the canonical spacing in that notation 3039460, and a later oligonucleotide study described a flexible binding mode that lets site length vary 10736221. The NFI-X structure explains the strong preference for TTGGC(N5)GCCAA: the five-base spacer places both DNA-binding domains on the same side of the DNA, and with a spacer one base shorter only a single monomer bound, at one half-site 41261119. All the JASPAR and HOCOMOCO full-site models fix the five-base spacer and run 14-17 columns with the spacer degenerate. Full-site models reach 15-26 bits, among the most informative in these collections. Half-site models also exist and are legitimate: NFI binds single TTGGC/TGCCAA sites with lower affinity, and several database entries for all four paralogues are 5-7 bp half-sites carrying only 8-12 bits. No paralogue is distinguishable from any other by sequence. The name "CCAAT box transcription factor" is a historical trap: the NFI half-site GCCAA contains the letters of a CCAAT box read on one strand, and NFI was long listed among CCAAT-binding proteins 3398920, but NFI and NF-Y are unrelated proteins with different folds and different optimal sites. If a scan returns both, they are separate hypotheses, not confirmations of each other.
Where it acts
NFI sites are found at promoters and at enhancers, and NFI proteins are enriched at distal regulatory elements in stem and progenitor populations. NFI binding is closely tied to chromatin accessibility: in skin stem cells NFI factors occupy super-enhancers and provide access for lineage factors while suppressing alternative fates 32393888, and in retinal progenitors NFI factors control mitotic exit and late-born cell specification 31128945. NFI also binds nucleosomal DNA more readily than many factors, consistent with a role early in opening elements. Recurring partners include FOXA1 in prostate, the glucocorticoid and progesterone receptors at the MMTV promoter (where NFI and the receptor compete for overlapping DNA) 2162764, SOX factors in glia, and MYOD in muscle. Because the full palindrome is long, genome-wide scans recover a manageable number of strong matches, but half-site matches are effectively everywhere.
Biology and disease
Mouse knockouts give distinct phenotypes despite the shared motif: Nfia loss causes corpus callosum agenesis and hydrocephalus, Nfib loss impairs lung maturation and forebrain development, Nfic loss blocks tooth root formation 12529411, and Nfix loss affects brain, muscle and hematopoiesis. The human syndromes mirror them. NFIA haploinsufficiency causes a CNS malformation syndrome with corpus callosum hypoplasia, macrocephaly and urinary tract defects 17530927. NFIX variants cause two different disorders depending on whether the transcript escapes nonsense-mediated decay: missense or deletion alleles affecting the DNA-binding/dimerization domain cause Malan (Sotos-like) overgrowth syndrome, while alleles producing stable truncated protein cause Marshall-Smith syndrome 20673863. NFIB is amplified and drives progression and metastasis in small cell lung cancer 27373156, and NFIB works with CARM1 in that setting; NFIB fusions also occur in adenoid cystic carcinoma via MYB-NFIB translocation.
Reading a hit
A full TTGGC(N5)GCCAA match is good evidence that some NFI dimer can bind, and it is specific enough that chance matches are uncommon. It is no evidence at all about which paralogue, and because heterodimers are the rule, "which paralogue" may not even be a well-posed question in a cell expressing three of them; assign from expression data. Half-site hits are a different proposition: TTGGC or TGCCAA alone carries under 12 bits and will occur by chance many times in any window you look at, so half-site models should be treated as weak. Two confusions are worth flagging. First, the GCCAA half-site overlaps the letters of a CCAAT box, and NFI's old name CTF invites the false inference that an NFI hit is an NF-Y hit; the true NF-Y site is CCAAT read in a different context and the proteins are unrelated. Second, NFI's MH1-like fold means archetype and class labels place it with SMAD, which can look like a mislabeling but is structurally correct 41261119. HOCOMOCO NFI models graded A generally rest on both ChIP-seq and in vitro selection (evidence codes PS or PSM), so the full-site matrices here are well supported.
How this was worked out
Nuclear factor I, a HeLa protein that stimulates adenovirus DNA replication, protects nucleotides 19-42 of the adenovirus 2 terminal repeat, and deletion mapping places its minimal recognition site at 15-16 bp.
Still open then: Which bases inside the site were read, and whether the site was symmetrical.
Binding to synthetic and selected sites defines the NFI site as two motifs, TGG and GCCAA, separated by a 6- or 7-bp spacer, with spacer-position biases that differ between the two spacings.
Still open then: Whether the 7-bp spacing is a genuine second site class used in cells.
Cloning shows that the transcription factor CTF and the replication factor NF-I are one family of differentially spliced proteins that bind GCCAAT-containing sites and act both as promoter-selective activators and as replication initiation factors.
Still open then: The CCAAT-box label grouped NFI with unrelated CCAAT-binding proteins such as NF-Y.
At the MMTV promoter NFI is needed for efficient hormone-induced transcription, yet purified NFI and the glucocorticoid or progesterone receptor compete for overlapping DNA rather than binding cooperatively.
Haploinsufficiency for NFIA in five individuals causes corpus callosum hypoplasia or agenesis, hydrocephalus or ventriculomegaly and urinary tract defects, matching the Nfia knockout mouse and giving the first human NFI syndrome.
Ablating NFIB and NFIX in hair follicle stem cells decommissions the super-enhancers of stem cell identity genes and de-represses other lineages, casting NFI as a chromatin-access factor.
Structures of the NFI-X DNA-binding domain, alone and on its palindromic site, reveal an MH1-like fold with a structural Zn2+ site and a dimeric assembly, placing NFI structurally with the SMADs; the authors present it as prototypic for all four paralogues.
Still open then: How heterodimers and alternative spacings are accommodated was not addressed.
Still unresolved
- Whether NFI has a second spacing class is unresolved: binding studies in 1987 recovered TGG...GCCAA sites with either 6-bp or 7-bp spacers and a 2000 study described a binding mode flexible enough to accommodate variation in site length, whereas the NFI-X structure explains a strong preference for TTGGC(N5)GCCAA (the 6-bp form in the 1987 notation), the only spacing that the full-site models in the current collections use.
- Why NFIA, NFIB, NFIC and NFIX knockouts differ so sharply despite one shared motif is not settled: the NFI-X structure is presented as prototypic for all four DNA-binding domains, which points to expression, heterodimer composition and C-terminal domains rather than DNA specificity, but paralogue-specific binding has not been excluded directly.
- How much genomic NFI binding uses single half-sites rather than full palindromes is not established; half-site models (TGCCAA, TTGGCA) exist for all four paralogues, including in vitro-supported ones, alongside full-site models above 15 bits.
The four paralogues share one matrix and freely heterodimerize, so a hit identifies NFI, never NFIA versus NFIB versus NFIC versus NFIX. Full-palindrome matches are trustworthy; half-site models (TTGGC or GCCAA alone, under 12 bits) are not, and several database entries for this family are half-sites. Spacer length matters more than spacer sequence: a gap one base shorter than TTGGC(N5)GCCAA supports only half-site binding, and how well other gaps are tolerated is less clear. Do not equate the GCCAA half-site with the CCAAT box of NF-Y despite NFI's historical name CTF; the two proteins are unrelated. The SMAD/NF-1 class label reflects the shared MH1-like fold and is not an error.
References
- Tiberi M, Lapi M, Gourlay LJ, et al. Structural basis of Nuclear Factor 1-X DNA recognition provides prototypic insight into the NFI family. Nat Commun 2025;16:10170. PMID 41261119
- Leegwater PA, van Driel W, van der Vliet PC. Recognition site of nuclear factor I, a sequence-specific DNA-binding protein from HeLa cells that stimulates adenovirus DNA replication. EMBO J 1985;4:1515-21. PMID 4040852
- Gronostajski RM. Site-specific DNA binding of nuclear factor I: effect of the spacer region. Nucleic Acids Res 1987;15:5545-59. PMID 3039460
- Santoro C, Mermod N, Andrews PC, et al. A family of human CCAAT-box-binding proteins active in transcription and DNA replication: cloning and expression of multiple cDNAs. Nature 1988;334:218-24. PMID 3398920
- Gronostajski RM. Roles of the NFI/CTF gene family in transcription and development. Gene 2000;249:31-45. PMID 10831836
- Roulet E, Bucher P, Schneider R, et al. Experimental analysis and computer prediction of CTF/NFI transcription factor DNA binding sites. J Mol Biol 2000;297:833-48. PMID 10736221
- Brüggemeier U, Rogge L, Winnacker EL, et al. Nuclear factor I acts as a transcription factor on the MMTV promoter but competes with steroid hormone receptors for DNA binding. EMBO J 1990;9:2233-9. PMID 2162764
- Adam RC, Yang H, Ge Y, et al. NFI transcription factors provide chromatin access to maintain stem cell identity while preventing unintended lineage fate choices. Nat Cell Biol 2020;22:640-650. PMID 32393888
- Clark BS, Stein-O'Brien GL, Shiau F, et al. Single-Cell RNA-Seq Analysis of Retinal Development Identifies NFI Factors as Regulating Mitotic Exit and Late-Born Cell Specification. Neuron 2019;102:1111-1126.e5. PMID 31128945
- Lu W, Quintero-Rivera F, Fan Y, et al. NFIA haploinsufficiency is associated with a CNS malformation syndrome and urinary tract defects. PLoS Genet 2007;3:e80. PMID 17530927
- Malan V, Rajan D, Thomas S, et al. Distinct effects of allelic NFIX mutations on nonsense-mediated mRNA decay engender either a Sotos-like or a Marshall-Smith syndrome. Am J Hum Genet 2010;87:189-98. PMID 20673863
- Steele-Perkins G, Butz KG, Lyons GE, et al. Essential role for NFI-C/CTF transcription-replication factor in tooth root development. Mol Cell Biol 2003;23:1075-84. PMID 12529411
- Semenova EA, Kwon MC, Monkhorst K, et al. Transcription Factor NFIB Is a Driver of Small Cell Lung Cancer Progression in Mice and Marks Metastatic Disease in Patients. Cell Rep 2016;16:631-43. PMID 27373156
21 motifs in this family — JASPAR 7 · HOCOMOCO 8 · CIS-BP 4 · Vierstra 2
| motif | database | consensus | bits |
|---|---|---|---|
| NFIA.H14CORE.1.PS.APSA | HOCOMOCO | TTGGCACGgTGCCAA | 26.4 |
| NFIB.H14CORE.0.PSM.APSMA | HOCOMOCO | TTGGCAhcgTGCCAA | 20.6 |
| NFIC::TLX1 (MA0119.1)dimer | JASPAR | TGGCAccatGCCAA | 19.7 |
| M03480_2.00 | CIS-BP | tTGGCAnnnTGCCAA | 19.5 |
| NFIB (MA1643.2) | JASPAR | CCTGGCAtnGTGCCAAG | 18.8 |
| NFIC (MA1527.2) | JASPAR | TTGGCdnngTGCCAa | 18.2 |
| NFIC.H14CORE.0.PSM.APSMA | HOCOMOCO | TTGGCahcdtGCCAA | 17.7 |
| NFIX.H14CORE.0.SM.BSMB | HOCOMOCO | TTGGCanCgtGCCAA | 16.7 |
| NFIX (MA1528.2) | JASPAR | tTGGCanngTGCCA | 15.6 |
| M03475_2.00 | CIS-BP | tTGGCnnnnTGCCAA | 15.3 |
| M05747_2.00 | CIS-BP | tTGGCnnnnTGCCA | 14.4 |
| NFIA (MA0670.2) | JASPAR | TGCCAA | 12 |
| AC0480:NFIX/NFIC:SMAD | Vierstra | TGGCannnTGCCA | 11.3 |
| NFIX.H14CORE.1.S.BSB | HOCOMOCO | TGCCAA | 10.7 |
| M08164_2.00 | CIS-BP | CTTGGCA | 10.6 |
| NFIC (MA0161.3) | JASPAR | CTTGGCA | 10.5 |
| AC0479:NFIA/NFIC:SMAD | Vierstra | TGCCAA | 10.2 |
| NFIA.H14CORE.0.P.BPB | HOCOMOCO | tTGGCA | 9.6 |
| NFIB.H14CORE.1.PS.APSA | HOCOMOCO | TGCCAA | 9.4 |
| NFIX (MA0671.2) | JASPAR | GCCAA | 7.8 |
| NFIC.H14CORE.1.PSM.APSMA | HOCOMOCO | tGCCAa | 7.5 |
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.