Structural and functional characterization of a cell cycle associated HDAC1/2 complex reveals the structural basis for complex assembly and nucleosome targeting

Abstract

Recent proteomic studies have identified a novel histone deacetylase complex that is upregulated during mitosis and is associated with cyclin A. This complex is conserved from nematodes to man and contains histone deacetylases 1 and 2, the MIDEAS corepressor protein and a protein called DNTTIP1 whose function was hitherto poorly understood. Here, we report the structures of two domains from DNTTIP1. The amino-terminal region forms a tight dimerization domain with a novel structural fold that interacts with and mediates assembly of the HDAC1:MIDEAS complex. The carboxy-terminal domain of DNTTIP1 has a structure related to the SKI/SNO/DAC domain, despite lacking obvious sequence homology. We show that this domain in DNTTIP1 mediates interaction with both DNA and nucleosomes. Thus, DNTTIP1 acts as a dimeric chromatin binding module in the HDAC1:MIDEAS corepressor complex.

Figure 1.

Interaction of DNTTIP1 with histone deacetylase complexes. (A) Domain structures of MIDEAS, TRERF1 and ZNF541; three co-repressor proteins associated with DNTTIP1. (B) Clustal alignment of ELM2 (magenta and cyan) and SANT (green) domains of TRERF1, MIDEAS and ZNF541 compared with the ELM2-SANT domain of MTA1. The blue rectangles are the α-helices in MTA1 and the orange rectangles are the predicted α-helices in MIDEAS. (C) Small scale co-transfection of HEK293F cells and purification of various FLAG tagged constructs of MIDEAS (marked with an asterisk) with full-length DNTTIP1 and full-length HDAC1. FLAG-tagged DNTTIP1 and MTA1 (also marked with an asterisk) with untagged HDAC1 and TEV cleaved FLAG-tagged DNTTIP1 are shown as controls.

Figure 2.

The crystal structure of the dimerization domain of DNTTIP1. (A) Domain structure of DNTTIP1. The protease digestion sites are shown as blue triangles. (B) Cartoon representation of the dimer of DNTTIP1 (62–130) in two different orientations 90° apart. The proteins are colored from the N- to C- terminus from blue to red. (C) The extensive dimeric interface between the two monomers with one molecule shown as a surface representation bound to a cartoon and sticks representation of the other molecule. (D) An alternative orientation of the dimerization domain. (E and F) Water molecules (shown in cyan) stabilizing the two kinks in the long helix 1. The hydrogen bonds from the peptide backbone of the α-helix are shown in green. The water molecules are shown in cyan.

Figure 3.

The NMR structure of the DNA binding domain of DNTTIP1. (A) Two views of DNTTIP1 (197–316) oriented at 90° to each other of a superposition of the Cα atoms for the 20 lowest energy structures of the 65 that converged are shown colored N to C terminus from blue to red. (B) Structural alignment of the primary sequence of the SKI/SNO/DAC domains of human DNTTIP1, SKI and DACH1. Strands are shown as pink arrows and helices as blue rectangles. (C) Superposition of DNTTIP1 (197–316) (red) with the SKI/SNO/DAC domain of the Ski oncogene (cyan). The elements of DNTTIP1 (H-1), and of SKI (S1), that do not superimpose are slightly transparent. (D) Two views oriented at 180° to each other of the electrostatic potential of the surface of DNTTIP1 (197–316) calculated in APBS (45) (blue is positive and red is negative).

Figure 4.

Function of the domains of DNTTIP1. (A) EMSA assay of DNTTIP1 (197–316) binding to a mixture of double stranded oligonucleotides. The DNA concentration and the ratio of DNTTIP1 to DNA are indicated. (B) Melting curves for DNTTIP1 (197–316), a mixture of double stranded oligonucleotides together with observed and calculated melting curves of DNTTIP1 bound to DNA. The circular dichroism was monitored at 222 nm. (C) SDS-PAGE stained with Coomassie of the binding of nucleosomes to GST-tagged DNTTIP1, GST-tagged PPARg LBD and just the GST-tag. (D) Schematic representation of the binding of DNTTIP1 (green) to the HDAC1 (orange) /MIDEAS (blue) complex. (E) Fractions from a Superdex-200 column of a complex of HDAC1, the ELM2-SANT domain of MIDEAS (717–887) and the N-terminus of DNTTIP1 containing the dimerization domain (1–150). (F) HDAC assay of the purified complex showing activation with inositol 1,4,5,6-tetraphosphate.

Figure 5.

Chromatin binding of the complex with full-length HDAC1, full-length DNTTIP1 and MIDEAS (650–887). Electrophoretic Mobility Shift Assays of complex at the indicated concentrations binding to: (A) 0.38 μM 147 base pairs double stranded DNA; (B) 0.62 μM 167 base pairs of 601 sequence reconstituted with histone octamer (MN – mononucleosome) using pUC DNA as competitor (PN – polynucleosome)); (C) 0.18 μM 501 base pairs (167 × 3) of 601 sequence reconstituted with histone octamer (TN – tri nucleosome) using 147 bp DNA as competitor; (D) 0.4 A260/ml of long chromatin extracted from HEK293F cells and purified using a 5–50% sucrose gradient. DNA markers (1 kb Hyperladder Bioline) are marked with M.