CSL / RBPJ
CSL is the DNA-binding transcription factor at the end of the Notch pathway, named for its three orthologues: CBF1/RBPJ in mammals, Suppressor of Hairless in Drosophila and LAG-1 in C. elegans. Humans have one broadly expressed gene, RBPJ, plus the pancreas-restricted paralogue RBPJL. RBPJ binds the seven-base element GTGGGAA, often written on the other strand as TTCCCAC, and does so whether or not Notch is active: without a Notch intracellular domain it recruits corepressors, and with one it assembles an activating ternary complex with Mastermind-like coactivators. Some Notch targets carry two sites in a head-to-head pair with fixed spacing, an SPS, which allows two Notch complexes to dimerize. Notch signaling through RBPJ controls lateral inhibition, T-cell specification and vascular development, and is constitutively activated in T-cell acute lymphoblastic leukemia.
The factors
There is essentially one CSL protein per genome. In humans the gene is RBPJ (older names RBP-J, RBPSUH, CBF1); the paralogue RBPJL (RBPSUHL) is restricted to pancreatic acinar cells, where it replaces RBPJ in the PTF1 complex. RBPJ is expressed in nearly every tissue. It is not a Notch-specific protein in the sense of being present only when Notch is: RBPJ is constitutive, and Notch activation changes what RBPJ does rather than whether it exists. The DNA-binding module has three parts: an N-terminal domain with a Rel-homology-like fold, a beta-trefoil domain (BTD) and a C-terminal domain 15297877. The crystal structure of CSL on DNA showed that the Rel-like domain and the BTD grip the site from opposite faces, with the BTD doing double duty as the docking site for the RAM peptide of the Notch intracellular domain (NICD) 15297877,16530045. On Notch activation, NICD binds RBPJ and recruits a Mastermind-like coactivator (MAML1-3) to form the ternary complex 16530045; in the absence of NICD, RBPJ binds corepressors such as SPEN/SHARP and associated histone deacetylases and demethylases. So the same protein on the same sequence can repress or activate 15975935.
The motif
The literature consensus is GTGGGAA, usually written with a 5' pyrimidine as YGTGGGAA, and its reverse-strand form is TTCCCAC 8152928. Only about six or seven positions carry information, and the models in these databases are short accordingly: the representative motif here is TTCCCA, which is simply the reverse complement of TGGGAA with the 5' GT trimmed off, and JASPAR's RBPJ matrix (TGGGAA) is the same element on the forward strand. The GGG at the center is the least tolerant part of the site. Only one symbol, RBPJ, appears across all four databases, so there is no paralogue problem inside the family, and the four models agree closely (cohesion 0.93). The difficulty is external. TGGGAA and its complement TTCCCA sit inside the purine-rich sequence space used by ETS cores (GGAA), NFAT (GGAAA) and the half-sites of NF-kB and STAT models, and the label-agreement statistics for all four RBPJ motifs are low (0.3) precisely because their nearest neighbours across the four databases frequently belong to those families. A separate feature is site architecture rather than sequence: some Notch-responsive promoters carry two CSL sites in head-to-head orientation with a spacer of roughly 15-17 bp, the Su(H)-paired site or SPS, first described in the Drosophila Enhancer of split complex 7590239. The spacing and orientation matter because they allow the ankyrin repeat domains of two Notch molecules to contact each other across the DNA, which sharpens the response 17284587,16365048. No PWM captures an SPS; you have to look for two matches at the right distance and orientation.
Where it acts
RBPJ sites are found in both promoters and enhancers, and the best-studied targets are the HES and HEY families (HES1, HES5, HEY1, HEY2), DTX1, NRARP, and, in lymphoid cells, PTCRA, IL2RA and MYC. Genome-wide, RBPJ and Notch1 occupy thousands of sites, and occupancy is dynamic: in Drosophila, Su(H) binding at target enhancers is transient and only partly overlaps its Notch-independent binding 17545467, and in mammalian cells RBPJ occupancy at many sites increases on Notch activation rather than being fixed 23651858. Conversely, a subset of sites is occupied by CSL without Notch and functions in default repression, keeping targets off in cells that have not received a signal 15975935,23651858. RBPJ co-occurs with lineage factors that determine which of its sites are used: TCF/LEF and GATA3 in T cells, ETS and RUNX factors at lymphoid enhancers 21737748, and PTF1A through RBPJL in pancreas.
Biology and disease
Rbpj-null mouse embryos die around embryonic day 10 with defects in somitogenesis and vasculature, phenocopying loss of Notch1 more closely than any single ligand mutant. Through Notch, RBPJ drives lateral inhibition in neurogenesis, T-cell versus B-cell specification, arterial-venous specification and intestinal secretory-versus-absorptive choice 27507209. Activating NOTCH1 mutations that stabilize NICD are the commonest genetic lesion in T-cell acute lymphoblastic leukemia, and the resulting Notch1/RBPJ complexes bind a largely conserved set of enhancers in human and mouse leukemia cells 21737748. Notch loss of function causes disease as well: NOTCH3 in CADASIL, JAG1 and NOTCH2 in Alagille syndrome, and inactivating NOTCH1 mutations in squamous carcinomas, where the pathway is tumor-suppressive. RBPJ itself is mutated in Adams-Oliver syndrome.
Reading a hit
The important caution is interpretive, not statistical: a match to the RBPJ motif, and even measured RBPJ occupancy, does not mean Notch signaling is active at that site. RBPJ binds constitutively and represses in the unsignaled state, so the sequence marks a potential Notch response element, not a signaled one. To claim active signaling you need evidence for NICD or MAML at the site, or a change in output on gamma-secretase inhibition. The site is also short: six or seven informative bases occur often by chance, and the purine-rich core is easily confused with ETS, NFAT or NF-kB half-sites, which is what the low label-agreement scores in these data reflect. Two matches in head-to-head orientation about 15-17 bp apart are much stronger evidence than one 17284587.
How this was worked out
RBP-Jkappa, first isolated as a protein binding the immunoglobulin Jkappa recombination signal, turns out not to read that signal: mutational gel-shift analysis and selection from random oligonucleotides define its core as CGTGGGAA, matching a site in the Drosophila Enhancer of split m8 promoter. This overturned the original recombination-signal assignment.
Still open then: What the protein did at its sites, and its link to Notch, were not addressed.
Su(H) directly activates Enhancer of split complex genes in response to Notch through multiple binding sites in their proximal upstream regions, making the site a Notch response element.
Crystal structure of CSL on DNA: Rel-like N- and C-terminal domains with an inserted beta-trefoil domain; the N-terminal domain and the beta-trefoil domain contact the DNA, and a hydrophobic pocket on the beta-trefoil domain is proposed as the Notch docking site.
Still open then: How Notch binding converts a DNA-bound repressor into an activator.
Su(H) represses Notch targets by default, through the same binding sites it uses for activation, in cells that receive no Notch signal; Hairless, Groucho and CtBP confer the repressive activity.
Structure of the CSL-NotchIC-Mastermind ternary complex on DNA: the RAM domain binds the beta-trefoil domain, the C-terminal domain unexpectedly organizes the interface with the ankyrin repeats and Mastermind, and CSL changes conformation on complex formation.
Head-to-head paired CSL sites in HES/E(spl)-type promoters require cooperative dimerization of two Notch transcription complexes to activate transcription.
Still open then: How many Notch targets genome-wide depend on paired rather than single sites.
Su(H) occupancy at E(spl) target genes rises transiently after Notch activation, contradicting the prevailing model of CSL statically bound to its enhancers.
The first RBPJ ChIP-seq in mammalian cells shows binding that rises with Notch activation at essentially all NICD-co-occupied sites, plus a separate class of sites bound statically regardless of Notch status. This revised the static-occupancy model in mammals as well.
Still unresolved
- How much of RBPJ's genomic occupancy is Notch-independent is unsettled: the default-repression model has RBPJ sitting on its sites and exchanging corepressors for NICD, whereas ChIP in Drosophila and in mouse myogenic cells finds binding at Notch-responsive sites rising on signaling, with static binding confined to a separate set of sites.
- What fraction of Notch targets depend on paired sites is unclear: cooperative dimerization on head-to-head SPS elements is required at HES-type promoters, yet most RBPJ sites are single, and the Notch paralogues differ in how well they use paired versus single sites.
- How RBPJ shares sites with ZNF143 is unresolved: in T-ALL cells many RBPJ sites lie inside ZNF143 motifs and the two proteins bind such DNA mutually exclusively in vitro, which suggests exchange in cells, but the exchange was inferred rather than observed.
One factor, RBPJ, so a hit does not have a paralogue problem inside the family; the risk is outside it. The informative core is only six or seven bases and overlaps ETS, NFAT and NF-kB purine-rich half-sites, which is why all four RBPJ models here score low label agreement against the other databases. The HOCOMOCO model (SUH.H14CORE.0.P.B) is ChIP-derived, so it reports where RBPJ was found, not what RBPJ binds in isolation. Most importantly, RBPJ occupies sites constitutively and represses without Notch, so neither a sequence match nor a ChIP peak licenses the conclusion that Notch signaling is active.
References
- Kovall RA, Hendrickson WA. Crystal structure of the nuclear effector of Notch signaling, CSL, bound to DNA. EMBO J 2004;23:3441-51. PMID 15297877
- Wilson JJ, Kovall RA. Crystal structure of the CSL-Notch-Mastermind ternary complex bound to DNA. Cell 2006;124:985-96. PMID 16530045
- Nam Y, Sliz P, Pear WS, et al. Cooperative assembly of higher-order Notch complexes functions as a switch to induce transcription. Proc Natl Acad Sci U S A 2007;104:2103-8. PMID 17284587
- Bailey AM, Posakony JW. Suppressor of hairless directly activates transcription of enhancer of split complex genes in response to Notch receptor activity. Genes Dev 1995;9:2609-22. PMID 7590239
- Castro B, Barolo S, Bailey AM, et al. Lateral inhibition in proneural clusters: cis-regulatory logic and default repression by Suppressor of Hairless. Development 2005;132:3333-44. PMID 15975935
- Ong CT, Cheng HT, Chang LW, et al. Target selectivity of vertebrate notch proteins. Collaboration between discrete domains and CSL-binding site architecture determines activation probability. J Biol Chem 2006;281:5106-19. PMID 16365048
- Krejcí A, Bray S. Notch activation stimulates transient and selective binding of Su(H)/CSL to target enhancers. Genes Dev 2007;21:1322-7. PMID 17545467
- Castel D, Mourikis P, Bartels SJ, et al. Dynamic binding of RBPJ is determined by Notch signaling status. Genes Dev 2013;27:1059-71. PMID 23651858
- Wang H, Zou J, Zhao B, et al. Genome-wide analysis reveals conserved and divergent features of Notch1/RBPJ binding in human and murine T-lymphoblastic leukemia cells. Proc Natl Acad Sci U S A 2011;108:14908-13. PMID 21737748
- Bray SJ. Notch signalling in context. Nat Rev Mol Cell Biol 2016;17:722-735. PMID 27507209
- Weirauch MT, Yang A, Albu M, et al. Determination and inference of eukaryotic transcription factor sequence specificity. Cell 2014;158:1431-1443. PMID 25215497
- Vierstra J, Lazar J, Sandstrom R, et al. Global reference mapping of human transcription factor footprints. Nature 2020;583:729-736. PMID 32728250
- Tun T, Hamaguchi Y, Matsunami N, et al. Recognition sequence of a highly conserved DNA binding protein RBP-J kappa. Nucleic Acids Res 1994;22:965-71. PMID 8152928
4 motifs in this family — JASPAR 1 · HOCOMOCO 1 · CIS-BP 1 · Vierstra 1
| motif | database | consensus | bits |
|---|---|---|---|
| SUH.H14CORE.0.P.BPB | HOCOMOCO | CgTGGGAA | 11.6 |
| M08105_2.00 | CIS-BP | TTCCCA | 11.2 |
| RBPJ (MA1116.2) | JASPAR | TGGGAA | 11 |
| AC0543:RBPJ:CSL | Vierstra | TTCCCA | 6.4 |
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.