Brownian motions of charged and
uncharged probes of different radii (0.1-1.4 A) were simulated in the presence
of rigid protein.
Starting point for simulations
was a 3A vicinity of the geometric center of the following atoms of HisH
active site residues:
SG_CYS_B_84, NE2_HIS_B_178, OE1_GLU_B_180
and OE2_GLU_B_180 (or chains D or F for other copies of HisHF).
End (target) point of simulations
was variable (defined by reaction distance) vicinity of the geometric center
of the following atoms of HisF active site residues:
ND2_ASN/ASP_A_11, OD1_ASP_A_130 and OD2_ASP_A_130
(or chans C or E).
To simulate only the passage
of a substrate through the channel (exclude trajectories reaching the end
point through the solvent), the trajectories were stopped when the substrate
moves further than 20 A from the center of HisHF . This ensures that
no trajectory approaches the end point without passing the channel.
Additional simulations were done with
trajectory stopping distance of 400 A. The rates from these simulations
were 3-4 times larger than the rates when trajectories truncated at 20
A, i.e. there is ca 1 channeling trajectory per 3-4 successfull trajectories.
A few simulations were done with the
different choice of the starting point (NE2 of bound Gln, CD of bound Glu,
active site center defined by PASS) - no significant changes in channeling
rates.
Ammonia has a dipole moment 1.47 D
(comparable to the dipole moment of water molecule,1.855 D). Modeling
it as dipole (as having 2 or 4 charges) did not result in large differences
from neutral probe.
Channeling rate
- approaching the target point to within 14 A (K99 is passed) and 5 A (S101
is passed)
Structure
chains
Comments
largest channeling probe (A)
channeling rate for a neutral probe
(%)
channeling rate for a charged probe
(%)
aicar
AB
1.1 / 0.8
11.3 / 2.7
54.4 / 1.2
CD
ligand removed
0.9 / 0.8
11.3 / 2.1
16.9 / 0.3
EF
1.2 / 0.8
13.0 / 2.9
5.8 / 0.2
gln
AB
ligand removed
0.8 / 0.7
11.8 / 2.3
10.7 / 1.1
CD
0.7 / 0.7
11.1 / 2.3
6.0 / 0.3
EF
ligand removed
0.7 / 0.7
10.4 / 2.2
7.8 / 1.7
glut
AB
ligand removed
1.1 / 0.7
10.9 / 2.5
42.7 / 1.6
CD
ligand removed
1.2 / 0.8
10.8 / 2.2
43.4 / 1.2
EF
1.1 / 1.1
11.8 / 3.9
43.4 / 7.3
imgp-gln
AB
ligand removed
0.8 / 0.8
11.8 / 2.5
10.3 / 2.3
CD
0.7 / 0.7
10.3 / 2.1
17.1 / 0.2
EF
ligands removed
0.7 / 0.7
11.2 / 2.6
12.5 / 7.5
1jvn
AA
acivicin left
1.0 / 0.8
29.0 / 10.6
-
BB
acivicin left
0.8 / 0.7
27.2 / 10.9
-
aicar
AB
T78L
1.1 / 0.8
10.9 / 2.2
-
AB
T78L+S101I
1.1 / 0.5
10.6 / 1.6
-
AB
R5A
1.0 / 0.8
11.0 / 1.9
-
The difference between the structures
seen from simulations :
channel in the aicar and glut structures
is more open than in gln and imgp+gln
aicar and glut structures have open
salt link tetrad for ammonium ion, which passes salt link 4 times more
often than neutral ammonia does. Aicar EF structure is different
from AB and CD (has different residues 23-24);
in the gln structure salt link tetrad
seems to be closed for ammonium. This correlates with the observation of
hydrogen bondng between OH_Tyr_B138 and NZ_Lys_A99 in the gln structure..
Yeast structure 1jvn has larger channeling
rate, because the starting point is more buried (but still accessible for
the probes ~ 0.5 A) .
Mutation influence is different for
different probes. Maximal influence of mutations are the rate
decrease by a factor of 1.4 forT78L (probe radius 0.4 A), 4.6 for T78L+S101I
(probe radius 0.5 A), 1.7 for R5A (probe radius 0.7 A)
Electrostatic
potentials at GLN and GLU binding sites
Below: electrostatic potential computed
for gln-AB structure and mapped onto the surface. GLN substrate (right)
is not used in calculations, but put back for visualisation purposes.
GLU substrate (left) is taken from glu-AB structure (superposed to gln-AB).
Below: electrostatic potential computed
for glu-AB structure and mapped onto the surface. GLU substrate is
not used in calculations, but put back for visualisation purposes.
GLN substrate is taken from gln-AB structure (superposed to glu-AB):
From these images it is not clear why GLU
is moved to its pocket, which is even more negative than GLN's pocket.
From hydrogen donor/acceptor pairs analysis
one can see that there is 1 hydrogen bond donor for GLU substrate in GLU
pocket: NE2_Gln_A123 (at 3.4 A from OXT of GLU) and one potential donor
ND1_His_B178 (at 3.43 A from OE2 of GLU, it is the only reason for blue
spot in the GLU binding pocket).
In the GLN pocket there are more donors
for GLU : N_Gly_B52 (for OE2), NE2_GlnB88 (for OXT), N_ThrB142 (for O),
N_Tyr_B143 (for OXT). I.e. from this viewpoint, GLN pocket is more
favourable for glutamate than GLU pocket.
I have done electrostatic binding free
energy calculations for GLU bound to HisHF. GLU bound to GLN pocket of
gln structure was modeled simply replacing NE2 to OE2 .
Structure
Chain
Pocket
Energy (van der Waals dielectric surface)
(kcal/mole)
Energy (connoly dielectric surface) (kcal/mole)
gln
AB
GLN
-3.6
+1.5
EF
GLN
-1.9
+7.3
glut
AB
GLU
+2.8
+22.0
CD
GLU
+3.7
+27.1
Again, with any dielectric surface treatment,
for glutamate, GLU pocket is less favourable than GLN pocket.
Brownian dynamics
simulation trajectories
There are 2 trajectories shown as yellow
and red taken from ca 1-2 ns long simulations. Trajectories start at the
lower part
