Figure 1.
EZH2 adopts the canonical fold of SET domain methyltransferases.
(A) Linear domain architecture of EZH2 showing the crystallized construct. Residue numbers according to GenBank isoform C (Uniprot isoform 1). (B) The catalytic SET domain (yellow) is folded as previously described for other histone methyltransferases such as EHMT1/GLP and MLL, but the post-SET domain is largely unresolved and its first five residues (blue) are oriented away from its expected position. The unique CXC domain adopts a novel conformation including two clusters of three Zn ions (light blue spheres). (C) A mesh representation of the EZH2 structure in the same orientation. The cofactor is expected to bind at the junction of the SET, post-SET and I-SET (cyan) domains. (D) Residues forming the substrate lysine-binding channel in EHMT1/GLP (beige – PDB code 2RFI) are structurally conserved in EZH2 (color coding as in A-C).
Figure 2.
The substrate binding site of EZH2 is occluded.
The substrate binding groove is too wide in MLL (right) and too narrow in EZH2 (left), compared with the catalytically competent state observed in the EHMT1/GLP ternary complex (center). Color-coding as in Figure 1.
Figure 3.
The cofactor binding site of EZH2 is incomplete.
(A) Superimposition of the EZH2 structure (colored mesh; post-SET shown as blue ribbon) with a ternary complex of EHMT1/GLP (white ribbon) shows that the cofactor binding site is only partially formed in EZH2, due to an atypical orientation of the post-SET domain. (B) The cofactor site of EZH2 is occupied by the CXC domain of a second molecule within the crystal lattice. (C) Mapping of the location of lysine-mediated cross-links detected in the purified PRC2 complex [53]. Cross-links between Lys735 and Lys569 as well as Lys713 indicate that the post-SET domain of EZH2 (yellow) can project towards the CXC domain in solution, consistent with the conformation seen in our structure.
Figure 4.
Structural basis for altered activity of mutations recurrent in lymphomas.
Hydrogen bonding between Tyr 641 and the substrate lysine’s ε-nitrogen, and steric envelope of the tyrosine hydroxy group impose rotational constraints that penalize proper alignment with the cofactor’s scissile bond, required for displacement of a third methyl group. A677 stabilizes the conformation of Y641 hydrogen-bonded to the substrate lysine. The cofactor and substrate lysine are from a superimposed ternary structure of EHMT1/GLP (2RFI).
Figure 5.
Catalytic activity and substrate/cofactor binding of EZH2 (520-746) and the trimeric (EZH2-EED-SUZ12) complex.
(A) The full-length trimeric complex (●) was active, and the crystallized EZH2 construct (○) was not. Activity assay conditions were optimized for the full length EZH2 in complex with EED and SUZ12 as a control. Kinetic analysis shows that the trimeric complex binds SAM (B) and a histone peptide (C) (Km SAM: 900 ± 100 nM; Km peptide: 205 ± 25 nM; kcat: 24 ± 2 h-1). Apparent kinetic parameters are the average of three measurements ± standard deviation. ITC shows that the crystallized construct binds neither SAM (D) nor the peptide substrate (E).