DNA-binding sequence specificity of DUX4

BACKGROUND: Misexpression of the double homeodomain transcription factor DUX4 results in facioscapulohumeral muscular dystrophy (FSHD). A DNA-binding consensus with two tandem TAAT motifs based on chromatin IP peaks has been discovered; however, the consensus has multiple variations (flavors) of unknown relative activity. In addition, not all peaks have this consensus, and the Pitx1 promoter, the first DUX4 target sequence mooted, has a different TAAT-rich sequence. Furthermore, it is not known whether and to what extent deviations from the consensus affect DNA-binding affinity and transcriptional activation potential. RESULTS: Here, we take both unbiased and consensus sequence-driven approaches to determine the DNA-binding specificity of DUX4 and its tolerance to mismatches at each site within its consensus sequence. We discover that the best binding and the greatest transcriptional activation are observed when the two TAAT motifs are separated by a C residue. The second TAAT motif in the consensus sequence is actually (T/C)AAT. We find that a T is preferred here. DUX4 has no transcriptional activity on “half-sites”, i.e., those bearing only a single TAAT motif. We further find that DUX4 does not bind to the TAATTA motif in the Pitx1 promoter, that Pitx1 sequences have no competitive band shift activity, and that the Pitx1 sequence is transcriptionally inactive, calling into question PITX1 as a DUX4 target gene. Finally, by multimerizing binding sites, we find that DUX4 transcriptional activation demonstrates tremendous synergy and that at low DNA concentrations, at least two motifs are necessary to detect a transcriptional response. CONCLUSIONS: These studies illuminate the DNA-binding sequence preferences of DUX4. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (doi:10.1186/s13395-016-0080-z) contains supplementary material, which is available to authorized users.