Supporting Information

Files in this Data Supplement:

SI Methods

QM/MM free-energy (potential of mean force) simulations were applied for the study of the mono- and dimethylation processes catalyzed by SET7/9 and its mutants (Y305F and Y335F). The CHARMM (version c32b2) program (1) was used for the simulations. The self-consistent charge density functional tight binding (SCC-DFTB) (2) method was used for the QM description in the present study. The results of the SCC-DFTB and B3LYP/6-31G** methods for the description of the methyl transfer along the reaction coordinate [R = r(Sd...C_M) - r(Nz...C_M)] in a small-model system containing trimethylsulfonium and methylamine were first compared by using the energy minimization-based approach. It was found that, although the SCC-DFTB optimized geometries along the reaction pathway are rather close to those from B3LYP/6-31G**, there are some systematic deviations in using the SCC-DFTB method to describe the energetics of the methyl transfer compared to the B3LYP calculations. These differences can be corrected by applying a simple linear function DE_corr = k_c´ R + E_c for the region where the deviations between the SCC-DFTB and B3LYP/6-31G** energies were observed. Here, k_c = 18.84 kcal·mol^-1·Å^-1, E_c = -3.77 kcal/mol, and R is the reaction coordinate for the methyl transfer (see below). This correction function was found to be similar for the both mono- and dimethylations and was applied to the free-energy profiles for the methyl transfers in SET7/9 and the mutants. In addition, the correction based on MP2/6-31G** calculations was also derived, and free-energy curves based on the MP2 correction were obtained. Because the same conclusions were obtained based on the B3LYP or MP2 corrections, the results based on the MP2 correction were not given here. AdoMet, AdoHcy, Tyr-335 (for the deprotonation), and lysine/methylated lysine side chains of the substrates and products were treated by QM and the rest of the system by MM. The all-hydrogen potential function (PARAM22) (3) was used for the MM atoms. The link-atom approach (4) available in the CHARMM program was used to separate the QM and MM boundaries. A modified TIP3P water model (5, 6) was used for the solvent. The stochastic boundary molecular dynamics method (7) was used for the QM/MM MD and free-energy simulations. The reference center for partitioning the system was chosen to be the Cb atom of the target lysine residue or the methylated Lys-4. The resulting systems contain 5,500≈5,700 atoms, including ≈800-900 water molecules (200 crystal water molecules). The initial structures for the entire stochastic boundary system were optimized by adopted basis Newton-Rhaphson method. The systems were gradually heated from 50.0 to 310.15 K in 50 ps and equilibrated at 310.15 K for 500 ps. A 1-fs time step was used for integration of the equation of motion, and the coordinates were saved every 50 fs for analyses.

The initial coordinates were based on the crystallographic ternary complex (PDB ID code 1O9S) containing SET7/9, AdoHcy, and a histone H3 MeLys-4 peptide (8). After inspection of the x-ray structure, it was noticed that the OT oxygen atom of the terminal carboxylate of AdoHcy (AdoMet) was missing from the structure. This atom was manually built. The free-energy simulations were performed starting from this modified product structure to generate the reactant complex containing SET7/9, AdoMet, and the H3-K4 peptide (i.e., by moving the methyl group along the reaction coordinate in the free-energy simulations). For the study of the second methyl transfer from AdoMet to the monomethylated Lys-4 (H3-K4me), a methyl group was manually added to AdoHcy and changed it into AdoMet. This leads to the reactant complex of dimethylation. The Y305F or Y335F mutant was generated simply by changing Tyr-305 or Tyr-335 in wild type to Phe. In addition, AdoHcy in the Y305F product complex from monomethylation and AdoMet in the wild-type reactant complex for monomethylation (with a proton added to H3-K4) were removed to produce the models without the cofactor bound and to generate the states before deprotonation of H3-K4me and H3-K4, respectively.

QM/MM MD simulations (1 ns) were carried out for each of the reactant and product complexes of mono- and dimethylations after 500 ps of equilibration was performed. The distributions of r(C_M-Nz) and q [defined by the C_M-Sd bond (r₂) and the direction (r₁) of the electron lone pair; see Fig. 1B] were monitored during the MD simulations for the reactant complexes and used for the calculation of the free energies required to generate the active structures. The histogram method was used to calculate the probability density distributions of r(C_M-Nz) and q. For r(C_M-Nz), histograms with bin width of 0.1 Å were used, and the probability density in the ith histogram is as follows: r_i = N_i/N (N = 20,000, the total number of the configurations from the 1-ns MD simulations, and N_i is the occurrence number in the ith histogram). For a given q value, the direction of r₂ dwells in a circle with the r₁ direction fixed (see Fig. 1B). Thus, the histograms were enclosed by homocentric circuits with different radius. To calculate the probability density distribution of q, the histograms with width of 10° were used, and N_i was weighted by 1/A_i, where A_i is the area of the ith histogram of q. Thus, the probability density in the ith histogram of q is as follows: N_i/(N ´ A_i). The relative free energy of the ith histogram of r(C_M-Nz) and q were calculated through W_i = -k_BT ´ ln r_i (9), where k_B is the Boltzmann's constant, and T is the temperature (310.15 K or 37°C). The MD simulations were also performed for the wild-type complex without the cofactor bound (see article). To test the hypothesis whether the negatively charged Y335 could act as the general base to deprotonate the positively charged methyl lysine and lysine, Y335 was manually deprotonated in each of the models and MD simulations were performed. A similar approach has been used previously in the study of chorismate mutase (10).

The umbrella sampling method (11) implemented in the CHARMM program along with the weighted histogram analysis method (12) was applied to determine the change of the free energy (potential of mean force) as a function of the reaction coordinate for the methyl transfer from AdoMet to H3-K4 (H3-K4me) in each case. The reaction coordinate is defined as a linear combination of r(C_M-Nz) and r(C_M-Sd) [i.e., R = r(C_M-Sd) - r(C_M-Nz)]. Twenty-one windows [corresponding to a change of r(C_M-Nz) ≈ 0.1 Å from one window to the next] were used for each methylation process. Within each window, 100-ps productive runs were performed after the equilibration. The force constants of the harmonic biasing potentials used in the potential of mean force simulations were 100-500 kcal·mol^-1·Å^-2.

1. Brooks BR, Bruccoleri RE, Olafson BD, States DJ, Swaminathan S, Karplus M (1983) J Comput Chem 4:187-217.

2. Cui Q, Elstner M, Kaxiras E, Frauenheim T, Karplus M (2001) J Phys Chem B 105:569-585.

3. Mackerell AD, Jr, Bashford D, Bellott M, Dunbrack RL, Jr, Evanseck JD, Field MJ, Fischer S, Gao J, Guo H, Ha S, et al. (1998) J Phys Chem B 102:3586-3616.

4. Field MJ, Bash PA, Karplus M (1990) J Comput Chem 11:700-733.

5. Jorgensen WL, Chandrasckhar J, Madura JD, Impey RW, Klein ML (1983) J Chem Phys 79:926-935.

6. Neria E, Fisher S, Karplus M (1996) J Chem Phys 105:1902-1921.

7. Brooks III CL, Brunger A, Karplus M (1985) Biopolymers 24:843-865.

8. Xiao B, Jing C, Wilson JR, Walker PA, Vasisht N, Kelly G, Howell S, Taylor IA, Blackburn GM, Gamblin SJ (2003) Nature 421:652-656.

9. Brooks CL, III, Karplus M, Pettitt BM (1988) Proteins: A Theoretical Perspective of Dynamics, Structure, and Thermodynamics (Wiley, New York).

10. Ma JP, Zheng XF, Schnappauf G, Braus G, Karplus M, Lipscomb WN (1998) Proc Natl Acad Sci USA 95:14640-14645.

11. Torrie GM, Valleau JP (1974) Chem Phys Lett 28:578-581.

12. Kumar S, Bouzida D, Swendsen RH, Kollman PA, Rosenberg JM (1992) J Comput Chem 13:1011-1021.