Re: AMBER: Simulating proteins with calcium ion

From: Qing Zhang <>
Date: Fri, 8 Dec 2006 16:31:02 -0800 (PST)

Thanks for the replies of Fenghui and Tom. I did further analysis including energy decompositions, and its points to the van der Waals parameters for Ca2+ in parm99.dat. I am giving a more detailed description of the system/problem and my analysis. The description will automatically answer some questions by Fenghui and Tom, and I will explicitly anwer the rest.

The system:
A protein-protein complex with a structural Ca2+. The Ca2+ ion binds to 6 oxygens from 5 residues (2 GLN, 2 ASP, and 1 GLY). The 6 oxygens form a binding pocket like a half-sphere, and Ca2+ is located nearly at the sphere center. The distances between the oxygens and Ca2+ range from 2.2 to 2.5 Angstrom.

The problem:
Energy minimizations of the complex crystal structure (using AMBER 9) cause Ca2+ to escape the half-sphere by about 1.9 Angstrom to the solvent but still binds to a few oxygens (Tom, there is no move/penetration through other atoms). I used 400 SD followed by 600 CG with igb=5 and a 3 kcal/mol (or 10) restraint on the heavy atoms of the proteins and Ca2+. Crystal waters are built into the system but there is no explicit sovlent. A sample input file is below.
Minimization with Cartesian restraints
    imin=1, maxcyc=1000, ntmin=1, ncyc=400,
    ntb=0, cut=16.0,
    ntr=1, restraint_wt=3.0, restraintmask=':1-297 & !.H=',

The analysis:
The first thing came to my mind is the radius of Ca2+ in parm99.dat. As the 6 oxygens (type O2 and O) have radii of 1.6612 (parm99.dat) and the distances between Ca2+ and the oxygens range from 2.2 to 2.5, the large raius for Ca2+ in parm99.dat (1.7131) will cause a large van der Waals penalty and make Ca2+ to be pushed out of the binding pocket. So I did energy composition analysis on Ca2+ (idecomp=2 and the restraint is removed to make idecomp work). The initial structure with only 1 SD gives a vdw of
nal |vdw |eel |pol |sas
        290 0.000 49.029 -353.216 32.657 0.000

Minimization with 400 SD (no CG as Tom suggested it might cause large jumps) leads to the same dislocation of Ca2+ and gives a vdw of 6 and eel of -350:
        resid |internal |vdw |eel |pol |sas
        290 0.000 5.994 -350.108 19.389 0.000

It indicates that Ca2+ is pushed out of the binding pocket during minimization to reduce van der Waals penalty.

In order to further prove it, I reduced the radius of Ca2+ to about 1.3. This is based on a post by Kenley Barrett on AMBER achieve. In this post, the vdw parameters for divalent ions computed based on the Aqvist paper (JPC 1990,k 94: 8021) are listed:

I took the vdw parameters of Ca2+ (1.3263, 0.4497) from the post, replaced those (for C0) in parm99.dat, and re-generated the topology file. The energy decompostion on the initial structure with 1 SD show a vdw of only 6 (reduced from 49):
        resid |internal |vdw |eel |pol |sas
        290 0.000 6.452 -353.216 32.657 0.000

Then I minimized the system for 400 SD (same minimization condition as the run without radius modification). The Ca2+ ion stays at its crystal location! The movement is only 0.13 Angstrom, and the energy decomposition shows a vdw of 12 and eel of -396:
        resid |internal |vdw |eel |pol |sas
        290 0.000 11.906 -395.934 61.374 0.000

>From these observations, the instability of Ca2+ during the minimizations is clearly due to the vdw parameters for Ca2+ in parm99.dat.

I am not familiar with divalent vdw parameterization. If someone has more insights on this, please feel free to raise them.


Qing Zhang, Ph.D.
Research Associate
Department of Molecular Biology, MB-5
The Scripps Research Institute
10550 North Torrey Pines Road
La Jolla, CA 92037-1000
Tel: (858) 784-2333
Fax: (858) 784-2860

Need a quick answer? Get one in minutes from people who know.
Ask your question on
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Dec 10 2006 - 06:07:07 PST
Custom Search