To setup our QM/MM simulations of AChE it is necessary, as explained in Chapter 2, to choose an existing X-ray structure. The chosen structure is the PDB structure 3DL7, obtained
and described by Carletti et al.[1] This structure has the advantage of being phosphorylated by
aged tabun instead of being inhibitor free. It also has a good resolution of 2.5 Å. Finally, since this was structure used in the doctoral work of Ophélie Kwasnieski, it was initially thought that the new results could more easily be compared to those obtained in her PhD project if we kept
using the same structure.[2]
The chosen PDB structure contains an AChE dimer with some ligands used to crystallize AChE. Those ligands are hexaethylene glycol, tetraethylene glycol, N-acetyl-D-glucosamine, and a chloride ion (see Figure I-1A). The ligands as well as one of the enzyme dimer were deleted from the structure to obtain the structure presented in Figure I-1B. This structure contains the enzyme chain A, the chloride ion and all 256 crystalline water molecules present in the X-ray data (Figure I-2).
Figure I-1. A) PDB structure 3DL7 with both chains of the AChE dimer in ribbon representation and secondary
structure colouring and the ligands in van der Waals radius representation. B) Chain A of 3DL7 with the chloride ion.
[1] E. Carletti, J.-P. Colletier, F. Dupeux, M. Trovaslet, P. Masson, F. Nachon, J. Med. Chem. 2010, 53, 4002– 4008.
Figure I-2. Chain A of AChE structure 3DL7 with the crystal water molecules represented as red spheres.
In 3DL7 the X-ray data is inconclusive on the structure of Arg522, His447. The B chain of Arg522 was chosen over the A chain because the positive charge it carries appears to be better compensated by its environment (Figure I-3). Both A and B chains are exposed to the solvent but the chain B is located near a glutamine residue that create the potential for an hydrogen bond which is absent from chain A. For His447, the A chain was chosen because it is in a reactive position as opposed to the B chain where the hydrogen bond between His447 and Ser203 is broken, disrupting the triad (Figure I-4).
Figure I-3. A and B chains of Arg522 with some nearby residues highlighted in Ball&Stick representation and
Figure I-4. Chains A and B of His447 in the catalytic triad.
The final heavy atom modifications were to replace the aged tabun adduct by VX. To do so, the N-dimethyl substituent of tabun is replaced by a methyl functional group.
After the heavy atoms modifications of the structure, the next step is the addition of hydrogens. To do so, the protonation state of various protonatable residues of the enzyme
must be discussed before moving forward with the setup. The standard pKa of protonatable
residues does not account for the environment of the side chain of a specific residue which
might influence its pKa. To assess this local pKa of a specific side chain, the software PROpKa
can be used.[3] In the case presented here, PROpK
a was used on the structure with all heavy
atom modifications described earlier. All the pKas calculated using PROpKa are not presented
here because for the sake of brevity as there is over a hundred protonatable residues in AChE.
The most likely protonation state based on the calculated pKa was used for most protonatable
residues. The PROpKa calculated value of a few important residues can be found in Table I-1.
Table I-1. pKa calculated using PROpKa for a selection of important residues.
Residue Glu202 Glu334 His447 Glu450 Glu452 PROpKa
calculated pKa 10.10 6.42 7.12 6.87 7.27
An interesting feature of these residues whose protonation states will be discussed along the entire manuscript is their ability, except for Glu202, to exchange protons at
[3] a) M. H. M. Olsson, C. R. Søndergaard, M. Rostkowski, J. H. Jensen, J. Chem. Theory Comput. 2011, 7, 525–537. b) C. R. Søndergaard, M. H. M. Olsson, M. Rostkowski, J. H. Jensen, J. Chem. Theory Comput.
physiological pH of 7.2,[4] based on the calculated pK
a values. For the initial setup, Glu334 was
kept unprotonated in line with the extensive research of its role in the catalytic triad.[5] Glu450
and Glu452 were kept unprotonated as it is more likely at pH 7.2. Finally, in this chapter the protonation state of His447 and Glu202 will be made to vary to study the role of proton exchange between these residues in the reactivation process. For histidines there is another level of complexity in that there two possible protonation states. Thus a histidine with a protonate imidazole ring can be protonated at site ε, or at site δ (see Scheme I-1). A histidine can also be protonated at both ε and δ sites and have an imidazolium side chain. The chosen protonation site for every histidine of AChE is indicated in Table I-2. The choice was based on a careful analysis of the environment of these histidines and especially the possibilities for hydrogen bonds with both sites.
Scheme I-1. Possible protonation states for the side chain of histidine residues
Table I-2. Chosen protonation state for histidine residues. Protonation states are designated as per the names
defined in Scheme I-1.
Residue His212 His223 His284 His287 His381 His387 His393 His405 His432 His447 Protonation
state HSE HSE HSD HSE HSD HSE HSE HSD HSE HSP
PROpKa
calculated pKa
2.60 5.77 6.48 5.97 3.50 6.76 6.39 2.14 6.08 7.12
Once these questions had been resolved, hydrogens were added to the rest of the
enzyme structure using the tools built in the software CHARMM.[6] The chosen force field for
this step as well as the molecular dynamics and the QM/MM simulations described in this
chapter is CHARMM22.[7] A 24 Ångström thick shell of water molecules was added around the
[4] R. Vroman, L. J. Klaassen, M. H. C. Howlett, V. Cenedese, J. Klooster, T. Sjoerdsma, M. Kamermans, PLoS
Biol. 2014, 12, e1001864.
[5] a) H. Tsukada, D. M. Blow, J. Mol. Biol. 1985, 184, 703–711. b) M. A. Massiah, C. Viragh, P. M. Reddy, I. M. Kovach, J. Johnson, T. L. Rosenberry, A. S. Mildvan, Biochemistry 2001, 40, 5682–5690.
[6] B. R. Brooks, C. L. Brooks, A. D. Mackerell, L. Nilsson, R. J. Petrella, B. Roux, Y. Won, G. Archontis, C. Bartels, S. Boresch, et al., J. Comput. Chem. 2009, 30, 1545–1614.
[7] a) A. D. Mackerell, M. Feig, C. L. Brooks, J. Comput. Chem. 2004, 25, 1400–1415. b) MacKerell A. D., D. Bashford, M. Bellott, Dunbrack R. L., J. D. Evanseck, M. J. Field, S. Fischer, J. Gao, H. Guo, S. Ha, et al., J.
system and in the channel of AChE using Chimera 1.10.1[8] to model a solvated enzyme. The
overall charge of the enzyme being -8, the system, with the crystallized chloride ion, has an
overall charge of -9. Nine water molecules were substituted with sodium ions using VMD 1.9[9]
to have a system with a total neutral charge.
The next step is the minimization, heating, and equilibration of the system using NAMD
2.9.[10] First the full system, acetylcholinesterase, water molecules and ions is minimized with
a frozen backbone. The system is then gradually heated to 310 K and finally a 5ns equilibration molecular dynamic is performed. During both the heating and the equilibration, only the inner 8 Å thick shell is mobile, the rest has been kept frozen to prevent the system from bursting. From this dynamic, a snapshot is extracted. The QM/MM simulations will be performed on this snapshot. The outer 16 Å of water molecules used to keep the molecular dynamics in a bubble are removed.
Figure I-5. AChE with QM region in red and in licorice representation, the optimized MM region in orange and
ball and stick representation and the static MM region in wire representation and in grey.
[8] E. F. Pettersen, T. D. Goddard, C. C. Huang, G. S. Couch, D. M. Greenblatt, E. C. Meng, T. E. Ferrin J.
Comput. Chem. 2004, 25, 1605–1612.
[9] W. Humphrey, A. Dalke, K. Schulten, J. Mol. Graphics 1996, 14, 33–38.
[10] J. C. Phillips, R. Braun, W. Wang, J. Gumbart, E. Tajkhorshid, E. Villa, C. Chipot, R. D. Skeel, L. Kalé, K. Schulten, J. Comput. Chem. 2005, 26, 1781–1802.
The QM/MM scheme used is additive QM/MM with three regions (see Figure I-5). The QM region, is composed of the catalytic triad, the reactivator ad the oxyanionic hole. The precise composition of this region has been changed at several points during the simulations described in this chapter. Thus, the precise composition is specified at the beginning of the relevant subsections. The MM region includes all other residues but a distinction is made between a fixed MM region and an optimized MM region.
The optimized MM region includes the QM region and most residues surrounding the QM region (Figure I-6). More specifically, residues of the enzymatic channel, catalytic anionic site, acyl pocket are included as well as all the nearby water molecules. A chain of residues and water molecules forming hydrogen bonds with Glu334, involving Glu202, Glu450, Glu452, Ser229, and Try428 are also included in the optimized MM region.
Figure I-6. Residues included in the optimized MM region. The residues of the standard QM region are coloured
by chemical element. In red are the residues of the channel and acyl pocket. The water molecules as well as the tyrosine that structures the hydrogen bond network in AChE’s active site are in blue. Finally, the residues in yellow are those that form a network of hydrogen bonds with Glu334.
The influence of the MM region on the QM region is taken into account with electrostatic embedding. The link atom with the charge shift scheme is used to handle the boundary between the QM and the MM regions. The geometries are optimized using the
HDLCOpt module included in Chemshell 3.1b1.[11] The QM energetic data is provided by
[11] P. Sherwood, A. H. de Vries, M. F. Guest, G. Schreckenbach, C. R. A. Catlow, S. A. French, A. A. Sokol, S. T. Bromley, W. Thiel, A. J. Turner, et al., J. Mol. Struct.: THEOCHEM 2003, 632, 1–28.
Turbomole V6.5[12] and calculated using B3LYP[1] with the D3 dispersion correction[13] and the
def2-SV(P) basis set[14]. The MM data is provided by the DL_POLY_3 interface of Chemshell
based on CHARMM22 parameters with modifications for the covalent adduct VX and 2-PAM (Appendix 2). A script was written to generate the list of atom index for the QM and MM region necessary to set up the Chemshell input. It can be found in (Appendix 6, Section I). In some important cases, single points were performed using Turbomole at the B3LYP-D3/def2-
TZVP level and using ORCA[15] at the DLPNO-CCSD(T)/def2-TZVPP level.[16] In both cases single
charges corresponding to the MM region are added to have consistent single points with electrostatic embedding.
The DLPNO-CCSD(T) method, for domain based local pair natural orbital coupled cluster method, is an approximation of the highly accurate CCSD(T) method. It was shown to
be consistently within 1 kcal.mol-1 of standard CCSD(T) with the NormalPNO setting which is
the one used in this work.[17] It is a particularly fast method compared to standard CCSD(T)
with a very limited accuracy loss.[18]