Evolving Catalytic Properties of the MLL Family SET Domain

Abstract

Methylation of histone H3 lysine-4 is a hallmark of chromatin associated with active gene expression. The activity of H3K4-specific modification enzymes, in higher eukaryotes the MLL (or KMT2) family, is tightly regulated. The MLL family has six members, each with a specialized function. All contain a catalytic SET domain that associates with a core multiprotein complex for activation. These SET domains segregate into three classes that correlate with the arrangement of targeting domains that populate the rest of the protein. Here we show that, unlike MLL1, the MLL4 SET domain retains significant activity without the core complex. We also present the crystal structure of an inactive MLL4-tagged SET domain construct and describe conformational changes that account for MLL4 intrinsic activity. Finally, our structure explains how the MLL SET domains are able to add multiple methyl groups to the target lysine, despite having the sequence characteristics of a classical monomethylase.

Graphical abstract

Figure 1

The KMT2 Family of Histone H3K4 Methyltransferases (A) The domain architecture of the KMT2 proteins found in yeast, Drosophila, and humans. (B) Phylogenetic analysis based on the sequence of only the SET domains of KMT2 proteins. The proteins segregate into the same groups in both sets of proteins. See also the extended sequence alignment in Figure S1.

Figure 2

Activity and WRAD Binding of the MLL1 and MLL4 SET Domains Details of the selection of assay conditions are provided in Figure S2. (A) Methyltransferase activity of MLL1 and MLL4 SET domain constructs with saturating SAM and increasing H3 peptide substrate. MLL4 shows much higher methyltransferase activity than MLL1 with respect to peptide substrate. (B) Methyltransferase activity of MLL1 and MLL4 SET domain constructs plus WRAD complex with saturating SAM and increasing H3 peptide substrate. Both MLL1 and MLL4 show significantly higher activity in the presence of WRAD complex. Quantification of methyl transfer was calculated from liquid scintillation measurement of recovered peptide, assuming 1 dpm is equivalent to 5.68⁻⁹ nM CH₃. Error bars represent the standard deviation of triplicate measurements. (C and D) Binding kinetics of immobilized MLL1 SET domain (C) and MLL4 domain (D) to WRAD complex. Both MLL1 and MLL4 SET domain constructs bind tightly to the WRAD complex.

Figure 3

Structure of the MLL4 SET Domain (A) Schematic representation of MLL4 constructs used in this study, indicating the subdomains of the MLL4 SET domain and providing a key for colors used in (C) and (D). (B) Methyltransferase activity of SET domain constructs used in this study. Activities are presented relative to the MLL1 (WIN) construct. Further details of the constructs used and their relative activity are presented in Figure S3. Error bars represent the standard deviation of triplicate measurements. (C) Cartoon representation of MLL4 in two orthogonal orientations. SET domain regions are colored as indicated in the key. The protein is shown in cartoon representation with the C-terminal 6xHis tag shown in yellow. The cofactor product, S-adenosyl homocysteine, is shown in stick representation, and the single coordinated Zn²⁺ ion as a sphere. The substrate binding channel, inferred from the structure of MLL1, is indicated in gray. (D) Orthogonal view of the MLL4 SET domain indicating the relative positions of the SET-I region and postSET loop. The box indicates the position of the postSET loop.

Figure 4

Comparison of the MLL1 and MLL4 SET Domain Structures (A) Superposition of the MLL1 (darker colors) and MLL4 SET domains in carton representation. Superposition performed in Coot on the cofactor molecule. (B) Detail of the position of the channel tetrapeptide segments of MLL1 and MLL4, showing how the MLL4 construct better constrains the target lysine substrate due to a shift of approximately 3.3 Å toward the Tyr5510 residue (MLL4 numbering). (C) Schematic representation showing the relationship of the channel tetrapeptide with the cofactor and SET-C strand. The Ala residue on the SET-C strand forms a closer interaction with the tetrapeptide Phe in MLL4. (D) Mutation of the MLL4 Ala5484 residue to Ser reduces the intrinsic methyltransferase activity of the construct relative to the wild-type. Further structural details are presented in Figure S4. Error bars represent the standard deviation of triplicate measurements.

Figure 5

Role of the PostSET Loop in Enzyme Activation (A) Sequence of the postSET regions of the MLL methyltransferases. Position 5522 in MLL4, indicated by an asterisk, is Lys in some database entries. The boxed sequence indicates the motif swapped between MLL1 and MLL4. (B) The effect of swapping the postSET loop regions of MLL1 and MLL4 on methyltransferase activity, relative to respective wild-type (WT). Error bars represent the standard deviation of triplicate measurements. (C) The MLL4 postSET loop (orange) makes potential interactions with the SET-I region (cyan). The C-terminal tag is indicated in yellow. (D) Superposition of the MLL1 and MLL4 postSET regions, showing that the MLL1 postSET loop (gray) is located further from the SET-I region than that of MLL4 (orange). (E) Site-specific mutations to the MLL4 postSET loop have a detrimental effect on their activity relative to the WT construct. However, activity is recovered in the presence of WRAD complex. (F) The effect of reverse mutations in MLL1 on activity.

Figure 6

Product Specificity of the MLL4 SET Domain (A) Superposition of the active sites of MLL1 (blue), MLL4 (cyan), and PR-Set7 (orange) indicating the positions of the channel tetrapeptide residues and Phe/Tyr switch residues. Conserved hydrogen bonds are indicated by gray dots, while the tetrapeptide main chain to active-site tyrosine hydrogen bond is shown by orange dots. This interaction is affected by the position of the channel tetrapeptide in MLL1 and MLL4. Here the main-chain carbonyl of the next tetrapeptide residue is within hydrogen bonding distance of the active-site tyrosine hydroxyl (3.5 Å in both structures), probably contributing to a displacement of this residue compared with the position observed in PR-Set7. (B) MLL4 shows intrinsic monomethylase specificity. Error bars represent the standard deviation of triplicate measurements. (C) Analysis of the reaction products at specified time points following incubation of MLL4 with unmodified peptide substrate. Different methyl species are indicated by the shading in the adjacent key. Representative MALDI-TOF spectra are shown to the right, and the full range of spectra is presented in Figure S5. (D) Analysis of the generation of H3K4 methyl species using MLL4/WRAD complex, shaded as in (C). (E) Gel-based methyltransferase assay with MLL4/WRAD reconstituted complex using peptide and nucleosome substrate. Reactions were stopped at the indicated times by boiling in sample buffer. The resulting blot was probed with antibodies specific to mono-, di-, or trimethylated H3K4 (Abcam). (F and G) Addition of WRAD to the MLL4 or MLL1 SET domain increases the overall methyltransferase activity, but not the pattern of substrate preference. Note that these assays were performed at an enzyme concentration of 25 μM.