All MD simulations were performed using the GROMACS 3.2.1 simulation suite of pro-
grams (www.gromacs.org)235and the GROMOS96 43a2 united atom force field (ffG43a2).
Simulations were performed either on University of Warwick Centre for Scientific Comput-
ing Argus task farm or Cluster of Workstations (COW) between April 2005 and December
2006. The source code for GROMACS 3.2.1 was compiled on each machine by the author
of this thesis. Each simulation was run on a single node of one of the above machines. The
approximate run time for a 1 ns simulation was 180 hours or 7.5 days. Docking simulations
were carried out using the AutoDock 3.0.5 program 220with the Lamarckian genetic algo-
rithm (LGA). Version 8.1 of Modeller was used to model the missing active site loop (A3)
Molecular dynamics simulations of the
phenylalanine activating adenylation
3.1
Introduction
In this chapter the results of a molecular dynamics (MD) simulation study of the L-Phenylalanine
activating gramicidin S synthetase (GrsA) A domain (PheA) fromBacillus brevis62are pre-
sented. Although the A domains have been studied extensively and various sequence sub-
strate specificity prediction models developed, understanding of the mechanism of substrate
selectivity and the dynamics of the protein is still relatively rudimentary.
The NRPS A domain specifically selects and activates the amino/hydroxyl acid substrate
through a two step reaction. In the first half reaction, a highly reactive aminoacyl adenylate
is formed by reaction with Mg-adenosine triphosphate (ATP) resulting in the release of
pyrophosphate. In the second half reaction the A domain binds the phosphopantetheinyl
(PPant) arm of the downstream domain, the Peptidyl Carrier Protein (PCP) domain.
The PheA protein chain is folded into two distinct domains, a large Acore domain (1–412 pdb:17–428) and a smaller Asub domain (413–514 pdb:429–530), which can be further divided into three and two sub-domains respectively, as outlined in chapter 1 section 1.4.2.
The active site is located at the junction of the two structural domains. The L-Phe substrate
is bound in a pocket accessible from the concave surface of theAcoredomain near where the threeAcorel sub-domains intersect62. The ten residues that line the L-Phe substrate binding pocket (pdb numbering in brackets) are: Asp 219 (235), Ala 220 (236), Trp 223 (239), Thr
262 (278), Ile 283 (299), Ala 285 (301), Ala 306 (322), Ile 314 (330) and Cys 315 (331)
contributed by the Acoresub-domain, and Lys 501 (517), by the Asubdomain62. The Mg2+ ion was positioned in the structure by the authors. As PheA was the first “in module” A
domain structure to be determined it has been used as a model for all subsequent A domain
structural studies.
As outlined in Chapter 1 in section , “domain alternation” has been proposed as a strat-
egy exploited by members of the adenylate-forming superfamily to reconfigure the single
active site of the enzyme to perform the two half reactions. Structures of members of the
adenylate-forming superfamily have been determined in the presence of the first and second
site where the reactions take place, however the conformation of the structures differed with
respect to the orientation of the Acore domain relative to the Asub domain52.
While A domains have only been determined in the adenylate-forming conformation, sim-
ilarities between members of the adenylate-forming superfamily suggest NRPS A domains
may exploit a similar strategy of domain alternation to reconfigure the enzyme’s single ac-
tive site. Members of the Adenylate forming superfamily contain highly conserved motifs
and adopt a conserved fold. Residues from the core 5 motif (A8 motif) are are highly con-
served in the A domains. These residues are critical for binding of the pantetheine portion
of Coenzyme A (CoA) in the second half-reaction structures of members of the Adenylate
forming superfamily. Limited proteolysis studies of the tyrocidine synthetase 1 A domain
(TycA)75,113have indicated intrinsic flexibility of the protein in the region linking the Acore
and Asub domain, the first Arginine residue of the A8 motif, which was reduced in the pres- ence of the first half-reaction ligands.
One way to probe conformation and examine the interaction between proteins and ligands
is by using computer simulation, especially Molecular Dynamics (MD), which can provide
information at the molecular level that is complementary to experiment and can, therefore,
further the understanding of a system. To date, no molecular simulation study of the A
domains has been reported in the literature. The MD simulations reported in this chapter
were designed to explore the dynamics of the PheA A domain. Of particular interest was
to probe the dynamical behaviour of PheA in the presence and absence of the hydrolysed
products of the first half reaction.
These simulations of PheA reveal motion of the Asub domain relative to the Acore domain. The principal modes of motion have been determined for PheA in each simulation. In each
apo simulation the principal motion, described by eigenvector 1, describes the Asub domain twisting clockwise, and tilting to the right, towards the Acore domain, and away from the A3 motif loop. In each holo simulation the principal motion shows the tilting and rotation
of the Asub domain (PheA2-holo), or part of the Asub domain (PheA1-holo) towards the A3 motif loop. This loop is thought to play a key role in stabilising the phosphate atoms of
the enzyme active site following the adenylation reaction. This domain motion is more pro-
nounced in the PheA2-holo simulation where the A3 motif loop exhibits less flexibility and
residues Thr 190 from the A3 motif loop form strong interactions with the highly conserved
key L-Phe binding pocket residues Asp 235.
The rotation of the Asub domain in the PheA1 and PheA2-holo simulations results in increased exposure between the domains on the right side of the protein. The extreme
conformations of this motion were overlayed with a representative structure of the second
half-reaction conformation (acetyl-CoA synthetase (bAcS) from Salmonella entericapdb
1PG4) to identify the PheA phosphopantetheinyl binding site, and with the modular NRPS
structure from the SrfAC synthetase to indicate the positioning of the PCP domain. This
overlay indicates the principal motion of the Asub domain in each holo simulation widens an opening between the domains on the right side of PheA which the flexible PCP domain
and phosphopanteinyl arm could utilise to access the enzymes active site. The interaction
between Thr 190 from the A3 motif loop and Asp 235 may be required to maintain the
opening between the Acore and Asub domain through which the PPant arm may access the PheA active site, or this interaction may be an intermediate stabilising interaction required
to facilitate further rotation of the Asub domain.