Test the performance of new CHARMM torsion potential for GBSW implicit-solvent simulations
New CMAP (par_mod24_2_6.prm): recently optimized by the Feig group at MSU to correct for the helical bias vs extended regions
par31_2_6: same CMAP as above but with the new force field for alkanes from the MacKerell group at UMBT
GBSW CMAP (par_all22_prot_gbsw.inp): an analytic function was added to the old CMAP to correct for the helical bias for GBSW implicit-solvent simulations. GBSW CMAP was able to fold both helical and beta-sheet proteins (Chen et al, JACS 2006)
Old CMAP (par_all22_prot_cmap.inp): Phi, Psi grid map to correction for overstabilization of pi helix
(Feig et al, JPCB 2003, MacKerell et al, JCC 2004, MacKerell et al, JACS 2004)
Ala dipeptide
20 ns REX-GBSW simulations, data collected in the last 5 ns at 298 K
|
par22_2_6 (Phi, Psi) PMF
|
par31_2_6
|
|
Comparison to explicit-solvent results: Diff(AlphaR, PPII) is too small by 1 kcal/mol. PPII is not enough deep or AlpaR is not enough high.
GBSW CMAP
PPII region is just as stable as the alpha region. However, the stabilization is less than sufficient as compared to the explicit-solvent results using the new CMAP.
|
Comparison to explicit-solvent results: Diff(AlphaR, PPII) too small by 1 kcal/mol.
Diff(AlphaL,PPII): too small by 0.5 kcal/mol
par31_2_6 with GBMV
Very similar to the GBSW map!!
|
|
Ensemble averaged J-coupling data
|
Ensemble averaged J-coupling data
|
Ala3 (free N-ter and protonated C-ter)
20 ns REX-GBSW simulations, data collected in the last 5 ns at 298 K.
|
par22_2_6 residue #2
|
par31_2_6 residue #2
|
|
Same as Ala dipeptide, the PPII region is more stable by 0.5 kcal/mol.
|
Similar to Ala dipep map. Alpha-L is too high as compared to explicit-solvent data (consistent with the ala-dipep map).
|
|
Ensemble averaged NMR coupling
simulation expt
3JHNHa 5.074219 5.68
3JHNCp 1.376013 1.13
3JHACp 1.728351 1.84
3JCpCp 0.503688 0.25
3JHNCb 2.141917 2.39
1JNCa 11.256388 11.34
2JNCa 8.264948 8.45
Much better agreement with experiment!
|
Ensemble averaged NMR coupling
simulation
|
Ala5
20 ns REX-GBSW simulations, data collected in the last 5 ns at 295 K
|
par22_2_6 res 2
|
par31_2_6 res 2
|
|
Same problem as in Ala dipep.
|
.Alpha-L region is a little improved.
|
|
par22_2_6 res 3
|
par31_2_6 res 3
|
|
AlphaR region is slightly more stabilized.
|
AlphaR region is improved
|
|
par22_2_6 res 4
|
par31_2_6 res 4
|
|
AlphaR region is as stable as PPII region. Helicity is building up as the peptide lengths increases. This has to do with overstabilization of hydrogen bonding in GB.
|
Almost no change as compared to par22.
|
|
J-coup res 2 res 3 res 4
3JHNHa 5.014 5.128 4.966
3JHNCp 1.319 1.360 1.356
3JHACp 1.586 1.758 1.612
3JCpCp 0.488 0.507 0.487
3JHNCb 2.190 2.130 2.190
1JNCa 11.136 11.041 10.928
2JNCa 8.118 7.995 7.824
|
J-coup res 2 res 3 res 4
|
(AAQAA)3
30 ns REX-GBSW simulations, data collected in the last 5 ns at 275 (red) and 300 K (green). Convergence verified by evaluating 30-40 ns data and compare it with 25-30 ns.
|
New CMAP (helicity based on hydrogen bonding)
|
Old CMAP (helicity based on hydrogen bonding)
|
|
Residue-averaged helix content is comparable to experiment although helicity for residues in the middle is significantly higher. We will investigate the backbone hydrogen bonding.
|
GBSW CMAP
Good agreement because the backbone atomic radii were optimized to match the experimental data.
|
HP36 fragments (test influence from side chains):
helicity assigned by DSSP
HP13 (N-terminal 13 residues)
With the new CMAP, overall helicity is significantly reduced. NMR J coupling data suggests sparsely populated helical states. We will compare with J coupling data.
C-terminal 14 residues
New CMAP (red) gave much lower helicity in agreement with experiment. We need to compare with NMR data which suggests low helicity.
Middle 10 residues
This is an exception: new CMAP (red) gives slightly higher helicity for residues in the middle of the sequence. NMR data suggests no helicity.