FIGURE 1.

CREBBP sequence mutations in relapsed ALL. a, Most variants are missense mutations in CREBBP domains involved in histone acetylation or transcription factor recruitment, or result in protein truncation. NRID, nuclear-receptor-interaction domain; TAZ1/2, transcriptional-adaptor zinc-finger 1/2; KIX, KID-binding domain; Bromo, bromodomain; HAT, histone acetyltransferase domain; ZZ, zinc-binding domain near the dystrophin WW domain; NCBD, nuclear-receptor coactivator-binding domain. b, The locations of CREBBP HAT mutations are shown using the crystal structure of the EP300 HAT domain complexed with its bisubstrate inhibitor, Lys-CoA (blue)²². CREBBP R1446 (equivalent to EP300 R1410) contacts phosphates of the CoA moiety of the inhibitor (salt bridges are shown as dashed lines), and the R1446H mutation is predicted to disrupt substrate binding. Q1500P (EP300 Q1464) is predicted to disrupt the alpha 4 helix, which stabilizes the substrate binding loop L1 between the beta 5 strand and the alpha 4 helix. C1408H (EP300 C1372) is predicted to disrupt the hydrophobic core that involves both the N and C termini of the HAT domain. R1563 cannot be shown as this residue lies in a proteolytically sensitive autoacetylation loop that was deleted in order to generate the crystal structure²². c, Duplication of the R1446H mutation at relapse (R). This mutation was heterozygous at diagnosis (D) and absent in the matched normal sample (N). There is copy neutral loss of heterozygosity of 16p at relapse but not at diagnosis. d, CREBBP mutations are present in subclones at diagnosis, and emerge in the predominant clone at relapse. The S1761* mutation is heterozygous in the relapse sample, absent in the matched normal sample, and appears as a minor peak in the diagnosis sample. Presence of this mutation in a subpopulation of cells at diagnosis was confirmed by PCR, cloning and bidirectional sequencing of multiple colonies of the diagnosis sample (data not shown).