Serines 4.53 and 4.57 of the Cannabinoid CB2 Receptor Form Intrahelix Hydrogen Bonds that Affect the Conformation of TMH4
Dow P. Hurst, Diane L. Lynch, Kurt Buehner and Patricia H. Reggio
| Acknowledgements | References |
Introduction:
The cannabinoid
CB1 and CB2 receptors are members of the Rhodopsin (Rho) family of G protein
coupled receptors (GPCRs). Gouldson and co-workers reported that the CB2
affinity of the CB2 inverse agonist/antagonist SR144528 Ki
= 0.67 ± 0.09 nM dropped to Ki > 1,000 nM upon mutation of
either of two serine residues (S4.53A or S4.57A) in TMH4 of CB21
(See Fig. 1). These investigators proposed
that each of these serines may be direct interaction sites for SR144528
at CB2 and attributed the loss of affinity in each mutant to loss of a
hydrogen bond.
Hypothesis:
In the work presented here, we test the alternate hypothesis that the
severe loss in SR144528 affinity at CB2 S4.53A or S4.57A mutants is due
to steric clashes at the SR144528 binding site produced by a change in
the conformation of TMH4 upon the loss of a serine residue.
In their
study of crystal structures of transmembrane proteins in the the PDB data
base, Ballesteros et. al.,2 documented the effects of Serine
residues found in a g- or +60o 
1.
For serines in the common g+ 1
value, the backbone phi (
)
and psi (
) dihedrals
have values of -62.5 ± 4.9o and -43.9 ± 9.9o. However,
Ballesteros found that with a g- 1
conformation, the and
values were -64.5
± 9.3o and -33.3 ± 8.4o. This change induces a local
bend of 4.3o which can result in a 3.3 angstrom displacement
of the helix end (see Figs. 2 and 3).
Methodology:
In order to test our hypothesis, we applied the Monte Carlo/simulated
annealing Conformational Memories 3 (CM) method as implemented
in Macromodel to TMH4 in WT CB2 (L4.37 to C4.66) and to the CB2 S4.53A
and S4.57A TMH4 mutants. CM has been shown to converge in a very practical
number of steps and to be capable of overcoming energy barriers efficiently.3
The calculation is performed in two phases, the exploratory phase and
the biased annealing phase.
Exploratory Phase. In the Exploratory phase, a random walk
is used to identify the region of conformational space that is populated
for each torsion angle studied. Starting at a temperature of 3000K, 12,000
steps were applied to the rotateable bonds with cooling in 18 steps to
a final temperature of 310K. All calculations were performed using a distance
dependent dielectric. Initial structures were not minimized prior to CM
calculations. 84 torsion angles of the WT TMH4 and 83 torsion angles of
the mutant TMHs 4 were allowed to vary during the CM runs. For L4.37 to
L4.62, these included the entire helix backbone f’s and j’s (allowed to
vary ± 10o), as well as the amino acid side chain torsion angles
(allowed to vary ± 180o). The backbone f’s and j’s for V4.56
through P4.60 (i.e., the turn before P4.60) were allowed to vary ± 50o.
For WT TMH4, Ser f’s and j’s were set to the values in the range reported
by Ballesteros 2 (f = -73.8o j = -24.9o).
For the S4.53A and S4.57A mutants, the Ala f’s and j’s were set to values
in the range reported by Ballesteros for Ala residues (f = -54.4o;
j = -53.0o). The c1 and c2 dihedrals on beta branched residues
( V, I or T) were excluded from all the runs, except the c2 dihedrals
for isoleucines.
Trial conformations were generated at each temperature by randomly picking three torsion angles from the set of 83 or 84 and changing each angle by a random value within the range set in the calculation. A modification of the original CM method was used to restrain the helix backbone. Instead of applying harmonic constraints, a set of biased memories were created for the first phase. The dihedrals which required restriction were given memories with populations distributed evenly over the allowed region of conformational space. For non-biased dihedrals, the populations were normalized across all population wells ensuring equal sampling of all phase space. During the exploration of these “biased” memories, a new memory file is written.
Biased Annealing Phase. In the second phase of the CM calculation, only torsion angle moves that would keep the angle in the “populated conformational space” mapped in the Exploratory phase are attempted. The Biased Annealing phase began at a temperature of 722 K with cooling to 310 K in 9 steps.
Analysis. The resulting 100 structures from CM were analyzed using the program Xcluster as implemented in the molecular modelling program Macromodel. Xcluster clusters the CM output into conformational familes.
Results:
The CM representative structure for WT TMH4 as identified by X-cluster
was found to be accommodated well within the CB2 bundle of our existing
CB2 model. This representative structure was used as a template to which
the TMH4 output from the mutant runs was compared. All 100 output structures
of the S4.53A and S4.57A runs were superimposed onto WT TMH 4 at its intracellular
end from residue L4.41 to V4.51 (See Fig. 4).
The CB receptors share with Rho a structural motif at the extracellular end of TMH4 stabilized by an aromatic stacking interaction between W175 and F203 in Rho (W4.64 and Y5.39 in CB1; see Fig. 5). TMH4’s output by CM were first screened for the ability to maintain this interaction. Remaining helices were then screened for their abilities to avoid steric clashes with TMHs 3 and 5. Seventy-seven S4.57A helices and forty-three S4.53A helices were found to meet both criteria. Fewer TMH4 S4.53A helices were acceptable due to the larger loss of induced bending by the removal of the g- Ser influence near the center of TMH4.
A representative conformer from each set of S4.57A (77 conformers) and S4.53A ( 43 conformers) was incorporated into a CB2 model in preparation for ligand docking.
SR144528 Binding Pocket at CB2. In WT CB2, the TMH 3-4-5-6 region represents the binding pocket for SR144528 and aromatic stacking interactions appear to be the primary interaction for this ligand at CB2. At the binding site, Ring A (see drawing below right) interacts with F5.46 and W5.43; the pyrazole ring interacts with W5.43; Ring B interacts with Y5.39; and the fenchyl ring of SR fits into a pocket near TMH6. Figure 6-left illustrates that at WT CB2, the Ring B methyl group of SR is near P4.60. Figures 6-middle and right show that this methyl group overlaps P4.60 in both the S4.57A and S4.53A mutants. This overlap should preclude SR binding in the TMH3-4-5-6 binding pocket in either mutant.
Figure 7 illustrates that the steric clash seen in Figs 6B and C can be overcome by a des-methyl analog of SR (see drawing below right) in which the Ring B methyl has been removed.
Conclusions :
CM results reported here suggest that the greater than 1000-fold reduction
in SR144528 affinity upon mutation of either S4.57 or S4.53 to alanine
1 is likely due to steric clashes between P4.60 with the Ring
B methyl group of SR144528. The des-methyl analog of SR144528 illustrated
in Figure 8 and the drawing at right has
been synthesized and is currently under evaluation in WT CB2 and S4.57A
and S4.53A mutants.
Acknowledgements:
This work was supported by NIDA RO1 Grant DA03934 and KO2 Grant DA000489.
References:
1. Gouldson et. al., Eur. J. Pharmacol. 401,17-25, 2000.
2. Ballesteros et.al., Biophys. J. 79, 2754-2760, 2000.
3. Guarnieri and Weinstein, J. Am. Chem. Soc. 118, 5580-5589, 1996.
4. Javitch et.al., Biochemistry 39, 12190-12199, 2000.