SALL
SALL1 to SALL4 are the human Spalt-like proteins: large C2H2 zinc-finger factors built from several clusters of paired fingers, with a short N-terminal motif that recruits the NuRD corepressor complex. Their DNA binding is unusual in being defined by base composition rather than by a sequence. A single C-terminal pair of fingers in SALL4 reads AT-rich DNA, and a crystal structure shows why it is promiscuous: the contacts use small side chains in the major groove, so many different AT-rich combinations are accepted. There is consequently no reliable consensus string for this family. SALL1 mutations cause Townes-Brocks syndrome and SALL4 mutations Okihiro syndrome; SALL4 is also the cereblon substrate whose degradation explains thalidomide limb teratogenicity.
The factors
The four human SALL genes, SALL1 to SALL4, are homologues of Drosophila spalt. Each protein is large and built from several clusters of C2H2 zinc fingers, usually a set of paired fingers (three or four double-finger clusters) plus one unpaired finger, separated by long low-complexity linkers. A short motif at the extreme N terminus recruits the nucleosome remodeling and deacetylase (NuRD) complex, and NuRD recruitment accounts for much of the repressive activity attributed to these proteins; the same interaction is required for SALL4's contribution to reprogramming 37208322. SALL4 is highly expressed in embryonic stem cells and in germ-cell and hepatocellular tumors but silenced in most adult tissues, which is why it is used as an immunohistochemical marker. SALL1 is prominent in kidney and limb, SALL3 in the nervous system. Mouse work shows Sall1, Sall2 and Sall4 are each required for neural tube closure 18818376.
The motif
Only one of the SALL clusters binds DNA in SALL4: a pair of fingers near the C terminus (ZFC4). Inactivating that one cluster drastically reduces SALL4's genome occupancy and reproduces the phenotype of a complete Sall4 knockout, including precocious differentiation of embryonic stem cells and embryonic lethality, whereas deleting two of the other clusters is phenotypically neutral 33406384. What that cluster reads is AT-richness rather than a defined word: SALL4 was recovered in a screen for proteins that bind AT-rich motifs, and genes are de-repressed on SALL4 loss in proportion to their AT content 33406384. Protein-binding-microarray and CUT&RUN experiments define an AT-rich consensus 33406418, and the crystal structure of the C-terminal fingers on AT-rich DNA explains the flexibility: recognition uses small hydrophobic and polar side chains in the major groove, so many different combinations of A and T are accepted 36635047. We therefore give no consensus string for this family; any single IUPAC word overstates what the protein specifies.
The database situation reflects this awkwardly. Of the three matrices assigned here, only the SALL3 model (consensus TTATGAATTATGATAT, HOCOMOCO grade A with HT-SELEX, GHT-SELEX and ChIP-seq support) is the AT-rich motif the literature describes, and it has no close neighbor among the other family members. The SALL1 and SALL4 models are ChIP-only (grades C and B) and are GC-rich, with the consensus GGGTGGGG, which is also the family's data consensus and its medoid. GGGTGGGG is a GC-box- and KLF/SP-like sequence, and the most economical explanation is that these ChIP-derived models capture a co-occupancy or promoter signal rather than SALL binding preference. Cohesion in this family is 0.51, essentially because the literature motif and the database medoid disagree.
Where it acts
SALL4 occupancy is broad and follows base composition: it accumulates over long AT-rich domains and at AT-rich genes, a distribution unlike the punctate enhancer occupancy of most transcription factors, and it acts largely to hold such genes down 33406384. That makes the relevant genomic feature a region rather than a site. Reported partners include NuRD and, in pluripotent cells, the core pluripotency factors 37208322. SALL1 in kidney and limb, and SALL4 in limb and heart development, act at developmental enhancers, but no well-validated composite element carrying a named SALL site is established in the way that exists for, say, GATA or ETS composites.
Biology and disease
Heterozygous truncating SALL1 mutations cause Townes-Brocks syndrome (imperforate anus, thumb malformations, dysplastic ears and hearing loss, with variable renal and cardiac involvement) 9425907,20301618. Heterozygous SALL4 mutations cause Okihiro (Duane-radial ray) syndrome 12393809, which overlaps the radial-ray phenotype of Holt-Oram syndrome and includes Duane anomaly. The SALL4 phenotype is mechanistically informative in two directions. First, the Okihiro missense mutations that fall inside the C-terminal fingers reduce overall DNA binding without abolishing the AT preference and shift the protein's association with AT-rich genomic sites, which supports their pathogenicity 36635047. Second, SALL4 is a thalidomide-dependent cereblon neosubstrate: thalidomide and its analogs recruit SALL4 to cereblon for degradation, and this provides a molecular explanation for thalidomide limb teratogenicity, since chemical SALL4 loss phenocopies genetic SALL4 loss 30190590,30067223. A crystal structure of the SALL4-pomalidomide-cereblon-DDB1 complex shows how the degron is presented 32251415.
Reading a hit
There is little to read. This family specifies base composition rather than a word, so a PWM hit is close to meaningless on its own: an AT-rich match is expected anywhere in AT-rich DNA, and the same matrices score homeodomain, forkhead and poly-A sequence. The four paralogues are not separable, and in these databases only SALL1, SALL3 and SALL4 have models at all. The specific trap is the opposite of the usual one: if your hit came from the GC-rich GGGTGGGG matrices labelled SALL1 or SALL4, you have almost certainly matched a GC-box-like sequence, and a KLF, SP or other GC-box factor is the more likely direct binder; those two models are ChIP-only and disagree with every in vitro measurement of SALL4 specificity. A credible SALL4 claim rests on AT-rich domain-level enrichment, on the effect of degrading or mutating the ZFC4 cluster, or on the clinical genetics, not on a motif match.
How this was worked out
Heterozygous SALL1 mutations are shown to cause Townes-Brocks syndrome, the first human phenotype for a vertebrate spalt homologue.
SALL4 mutations are shown to cause Okihiro (Duane-radial ray) syndrome, whose radial-ray limb phenotype overlaps that of thalidomide embryopathy.
Still open then: The resemblance to thalidomide embryopathy had no mechanistic explanation.
Mouse mutants show Sall1, Sall2 and Sall4 are each required for neural tube closure, establishing partially overlapping developmental requirements across the family.
- 2018highQuantitative proteomics of drug-treated cells, CRBN binding and ubiquitination assays, zebrafish and chicken limb models30190590 30067223
Two groups identify SALL4 as a thalidomide-dependent cereblon neosubstrate, so the drug degrades the protein whose mutation causes the overlapping limb syndrome; chemical loss phenocopies genetic loss.
A crystal structure of the SALL4-pomalidomide-cereblon-DDB1 complex shows how a SALL4 zinc finger is presented as a drug-induced degron, explaining the species and analog specificity of the effect.
- 2021highProteomic screen with AT-rich baits, mutant mouse and ESC genetics, genome-wide occupancy; protein-binding microarray and CUT&RUN33406384 33406418
SALL4 is recovered in a screen for proteins that read AT-rich motifs: a single C-terminal pair of zinc fingers accounts for genome occupancy, inactivating it reproduces the null phenotype, and genes are de-repressed in proportion to their AT content, so the protein reads base composition rather than a word; a parallel study defines an AT-rich consensus by protein-binding microarray and CUT&RUN.
Still open then: Whether the AT-rich consensus derived in vitro adds anything beyond composition was not resolved.
The crystal structure of the SALL4 C-terminal zinc fingers on AT-rich DNA shows recognition through small hydrophobic and polar side chains in the major groove, which explains why many A/T combinations are accepted, and Okihiro missense mutations in these fingers reduce binding without changing the AT preference.
The N-terminal NuRD-interacting motif is shown to be required for SALL4's contribution to reprogramming, tying the family's repressive output to a recruited complex rather than to its own binding site.
Still unresolved
- Whether SALL4 recognizes anything more specific than base composition is unsettled: protein-binding microarray and CUT&RUN data are reported as an AT-rich consensus, while the structural and genomic work describes flexible recognition of many A/T combinations over long domains, and the two readings differ on whether a position weight matrix is meaningful here at all.
- The GC-rich GGGTGGGG models labeled SALL1 and SALL4 in HOCOMOCO contradict every in vitro measurement of SALL4 specificity; whether they capture an unrecognized GC-box-like binding mode or simply promoter co-occupancy in ChIP has not been tested directly.
- Whether SALL1, SALL2 and SALL3 share SALL4's AT preference is unknown: only SALL3 has an in vitro-supported model in these databases, and it is not close to any other family member's motif.
- Whether degradation of SALL4 alone accounts for thalidomide limb teratogenicity remains open, since other cereblon neosubstrates have been proposed to contribute and the drug degrades many zinc-finger proteins.
SALL proteins read base composition, not a word: a C-terminal finger pair in SALL4 binds AT-rich DNA promiscuously, and the crystal structure shows the contacts accept many A/T combinations. We give no consensus for that reason, and an AT-rich hit is expected throughout AT-rich DNA and overlaps homeodomain, forkhead and poly-A signals. The paralogues are indistinguishable, and SALL2 has no model here. Worse, the family's medoid and data consensus come from ChIP-only SALL1 and SALL4 matrices that are GC-rich (GGGTGGGG, a GC-box/KLF-like sequence) and contradict every in vitro measurement; cohesion is 0.51. Matches to those two models are most likely GC-box factors, not SALL.
References
- Kohlhase J, Wischermann A, Reichenbach H, et al. Mutations in the SALL1 putative transcription factor gene cause Townes-Brocks syndrome. Nat Genet 1998;18:81-3. PMID 9425907
- Kohlhase J, Heinrich M, Schubert L, et al. Okihiro syndrome is caused by SALL4 mutations. Hum Mol Genet 2002;11:2979-87. PMID 12393809
- Adam MP, Bick S, Mirzaa GM, et al. SALL1-Related Townes-Brocks Syndrome. 1993. PMID 20301618
- Pantier R, Chhatbar K, Quante T, et al. SALL4 controls cell fate in response to DNA base composition. Mol Cell 2021;81:845-858.e8. PMID 33406384
- Kong NR, Bassal MA, Tan HK, et al. Zinc Finger Protein SALL4 Functions through an AT-Rich Motif to Regulate Gene Expression. Cell Rep 2021;34:108574. PMID 33406418
- Watson JA, Pantier R, Jayachandran U, et al. Structure of SALL4 zinc finger domain reveals link between AT-rich DNA binding and Okihiro syndrome. Life Sci Alliance 2023;6. PMID 36635047
- Böhm J, Buck A, Borozdin W, et al. Sall1, sall2, and sall4 are required for neural tube closure in mice. Am J Pathol 2008;173:1455-63. PMID 18818376
- Wang B, Li C, Ming J, et al. The NuRD complex cooperates with SALL4 to orchestrate reprogramming. Nat Commun 2023;14:2846. PMID 37208322
- Matyskiela ME, Couto S, Zheng X, et al. SALL4 mediates teratogenicity as a thalidomide-dependent cereblon substrate. Nat Chem Biol 2018;14:981-987. PMID 30190590
- Donovan KA, An J, Nowak RP, et al. Thalidomide promotes degradation of SALL4, a transcription factor implicated in Duane Radial Ray syndrome. Elife 2018;7. PMID 30067223
- Matyskiela ME, Clayton T, Zheng X, et al. Crystal structure of the SALL4-pomalidomide-cereblon-DDB1 complex. Nat Struct Mol Biol 2020;27:319-322. PMID 32251415
3 motifs in this family — HOCOMOCO 3
| motif | database | consensus | bits |
|---|---|---|---|
| SALL3.H14CORE.0.PSG.APSGA | HOCOMOCO | gTTATgAATtATgAtat | 18.5 |
| SALL1.H14CORE.0.P.CPC | HOCOMOCO | rGGGtGGGG | 10.3 |
| SALL4.H14CORE.0.P.BPB | HOCOMOCO | GGGTGgGG | 10.1 |
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.