(Rev 12/2/99, Beta Version) Introduction to Molecular Modeling Lab X - Modeling Proteins Table of Contents Introduction ......................................................................... 2 Objectives .................................................................... 3 Outline ....................................................................... 3 Preparation of the pdb File .............................................................. 4 Using Leap to build the Biotin/Steptavidin Complex ........................................... 5 xLeap ....................................................................... 5 Building the biotin residue from pdb in Leap ..................................... 5 Carnal ............................................................................ 13 Sander ............................................................................ 13 summary .................................................................... 15 Carnal-II .......................................................................... 17 Appendices ........................................................................ 21 Appendix 1 .................................................................. 21 Appendix 2 .................................................................. 22 Appendix 3 .................................................................. 22 Appendix 4 .................................................................. 22 Appendix 5 .................................................................. 23 Appendix 6 .................................................................. 24 Appendix xx ................................................................. 24 Appendix xx ................................................................. 25
25
Embed
Introduction to Molecular Modeling Lab X - Modeling Proteins · Introduction to Molecular Modeling Lab X - Modeling Proteins Table of Contents Introduction ... “Proteins-Structure
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
(Rev 12/2/99, Beta Version)
Introduction to Molecular ModelingLab X - Modeling Proteins
The streptavidin/biotin system is of special interest because it has one of the largest free energies of associationas of yet observed for noncovalent binding of a protein and small ligand in aqueous solution (Kass = 1014). Thecomplexes are also extremely stable over a wide range of temperature and pH.
The streptavidin protomer is organized as an 8-stranded beta-barrel. Pairs of the barrels bind together to formsymmetric dimers, pairs of which in turn interdigitate with their dyad axes coincident to form thenaturally-occurring tetramer.
If you are interested in the Biotin/streptavidin system and would like to know more about the system in generalthen consult the two references given below. However, it is not necessary to do so to understand thislaboratory exercise.
Miyamoto S, Kollman PA. “What determines the strength of noncovalent association of ligands toproteins in aqueous solution.” PNAS, (1993), 90:8402-8406.
Miyamoto S, Kollman PA. “Absolute and relative binding free energy calculations of the interaction ofbiotin and its analogs with streptavidin using molecular dynamics free energy perturbation approaches. “Proteins-Structure Function and Genetics”, (1993) 16:226-245.
3
Objectives
In this tutorial, we massage a PDB file of the tetramer so that it can be understood by Amber, solvate theregion of interest (i.e. around one of the four complexed biotins), and run some dynamics keeping the restfrozen. The equilibration takes 2.3 hours per picosecond on moderately fast CPU (Convex 3820). Allowing thewhole system to move takes 4 hours/psec (the whole system would have to be solvated, thus still longer time,for this to be useful). The frozen part provides a more realistic electrostatic environment for the part that moves. Thus the main objectives are
1) Modifying Protein PDB files to suit specific needs.2) Partially hydrating proteins to decrease computational time yet obtain meaningful results.3) Run molecular dynamics on a protein4) Interpret the results of the molecular dynamics run.
The initial biotin/streptavidin tetramer was prepared by Richard Dixon from the momomer. The monitor wasobtained from the Brookhaven database, structure 1stp by P.C. Weber, D.H. Ohlendorf, J.J. Mendolowski,and F.R. Salemme (1992).
Outline
1) Prepare the required pdb file
2) xLeapa) Build biotin residue template. b) Load 'frcmod' file (extra force field params) for biotin.c) Load the streptavidin-biotin pdb. d) Add 'cap' of water. e) Save top and crd files for dynamics.
3) Carnal Figure out residues in biotin 'cap' region.
4) Sandera) Minimize and run dynamics to equilibrate.b) Check equilibration.c) Run more dynamics.
4
5) Carnal Analyze the trajectory.
Setup
1) Create a subdirectory in your home directory call biotin
mkdir ~/biotin
2) Copy all the files from /usr/people/amberlab11 into the ~/biotin directory
cp /usr/people/amberlab4/* ~/biotin/The files that should be copied are:
Hydrogen naming conventions in the given.pdb file are ‘wrong’ - a not uncommon experience. Oneoption is to try to correct all the hydrogen atoms (about 4000). However, it is much quicker to simply delete allthe hydrogren atoms and then let xleap add them back when you load up the pdb file. This can be achieved byusing 'egrep -v' to exclude lines matching a pattern with H in either of the 1st or 2nd columns of the atomname.
where the '̂ starts the pattern at the beginning of the line and the .'s are wild-card single characters Note thatthere are 13 ‘.’ in the first line and 12 ‘.’ in the second.
5
Using Leap to build the Biotin/Steptavidin Complex
xLeap
Building the biotin residue from pdb in Leap The amber force field lacks parameters for biotin. Therefore to getAmber to understand biotin will require you to load a coordinate file for biotin, then make bonds between theconnected atoms, and then input parameters for each atom of biotin. While this is a bit tedious, we will nothave to actually determine what the parameters are, which is really the hard part.
Start xleap and load the residue:
> BTN = loadpdbbioti n.pdb> edit BTN
When you look at the biotinmolecule in the Unit editoryou will see each of theatoms of biotin appear asdiamonds. They will becolored according toatom type but the structure willlack any bonds.
6
To add the atoms, select all of the atoms by snapping a box that encloses all of the atoms. Then, in the Worldeditor, type the following command:
> bondbydistance BTN
If you now view the biotin molecule you will see bonds have been added.
7
Now, select all of the atoms by dragging a boxabout the structure with the left mouse button, theywill turn purple.
Select all atoms by dragging a box with left mouse button and pull down 'Editselected atoms':
Note: Editing the table that appears can be frustrating. The program driving it is notvery stable and tends to hang. When it hangs, you lose all of your work. If thiscrashes more than once on you, let me know and I will provide you with what youneed to by-pass this step. Also, if it does crash, try to note exactly what you weredoing when it crashed. I am trying to collect this information to help theprogrammers discover the problem and then, hopefully, fix it.
8
ThistablelistsalltheatomspresentinBiotin. Youm
ust input an atom type, charge, Pert.name and Pert.type for each atom as shown on the next table.
9
Note that for most of the atoms the Name/PERT.name and the Type/PERT.type entries are identical. Theexceptions are H71, H81, H91 and H92.
When you have finished entering the data, pull down 'Operations / Check table' to make sure all is ok, and'Save and quit' the table.
10
In the Unit Editor, deselect the atoms by holding down the Shift key and clicking in the background (i.e. not onany atom). This is mainly cosmetic, so that the saved state will not have selection on.
Finally, save the residue in xleap format at the command line. The .lib file contains a description of eachatom - atom name, type, charge, coordinates, and connectivity information. However, it does not contain forcefield information.
> saveoff BTN btn.lib
The residue can be loaded in later sessions by
> loadoff btn.lib
Load the premade 'frcmod' file for biotin. This file is shown in Appendix 1. It contains the force fieldparameters which were missing from the Amber force field. When this file is loaded, the parameters in this fileare merged with the Amber force field. If there is duplication, the .frcmod file parameters take precedence.
> fmod = loadamberparams btn.frcmod
Load prepared pdb of streptavidin/biotin complex:
> stbt = loadpdb start.pdb> edit stbt
11
To manipulate the structure, hold down the two right buttons and push forward/back to zoom in/out. Themiddle button alone rotates, the right button translates, and the space bar recenters the molecule.
Add a 'cap' of waters around the site of the 1st biotin. This is done by estimating a median x, y, z coordinate byeyeballing the coordinates of the BTN at the beginning of start.pdb, which can be viewed by jotstart.pdb:
> solvatecap stbt WATBOX216 { 35 12 -6 } 20
12
The number of waters in the cap varies slightly, depending on machine; about 265 is to be expected on SGI.
Save system in xleap and pdb formats for future reference:
Save top and prm files for dynamics & perturbation:
> saveamberparm stbt stbtcap.top stbtcap.crd
13
Carnal
For dynamics, we only want the region of interest to move - the rest is there to provide a more lifelikeenvironment.
Use a carnal feature to figure out the residues around the biotin molecule that the water cap is on:
carnal < carnal_grp.in > carnal_grp.out
Appendix 2 contains a listing of the carnal_grp.in. If you take a look at the input file you will see thatfirst the stbtcap.top and stbtcap.crd files are read. Next the file says to find all of the residues thatare within 15 A of group 479. Group 479 is the biotin residue. The output then contains the list of groupsfound with these constraints and it would be a good idea to examine this file.
GROUP format is described in Appendix B of the Amber manual.
Sander
Introduction
When the biotin/stretavidin complex was built, the hydrogens were deleted and then xleap added them back. They may have not all been added back in a way that was sterically favorable. In addition, a water cap wassubsequently added. This is a pre-equilibrated volume of water and unfavorable steric interactions between thewaters and the biotin/streptavidin complex may exist. Therefore, we first need to do a quick molecularmechanics minimization to remove any particularly bad steric interactions.
The file called min.in is the control file for the run and it is listed in the Appendix (3). The key parametersin this file are
14
IMIN=1, IMIN determines whether the run is a molecular mechanics (=1) or molecular dynamics run(=0).MAXYCY=100, we will let the minimizer run for 100 cycles.NTPR=20SCEE=1.2. This is the 1,4-VDW scaling term. When using the all_nuc94.dat data base it should beset to this value.NSNB=999999NTMIN=2
Look at the output from this run (located in min.out). Note that the first part of this file lists what all theinput parameters were, files used, date, etc., which can be very helpful once you start to accumulate a lot ofminimization runs. After that, for each cycle, is listed various values such as total energy and each of itscomponents. If a run blows up on you, this can help you diagnose where the problem may lie. Finally, there issome information about the length of the run.
Molecular Dynamics
Next we are going to gradually warm, using 'belly' option to restrict motion to region of interest as determinedby carnal, above. (The group defined by carnal is modified in md0.in to include all the waters.)
This equilibration immediately warms to 300K; this may be ok in this case since a major part of the system isfrozen - if this were not the case, too-rapid warming could disrupt the structure. Similarly, if this was a constantpressure simulation in a periodic box of solvent, too-rapid warming could lead to excessive pressurefluctuations.
Although no such critical problems can happen here, we still need to see if the system has truly equilibrated. Todetermine this we need to take a look at the variantion in temperature energy and its components. This
15
information is numerically present in the md0.out file and the relevant numbers can be sorted into files forplotting with a perl script called process_md_perl and then the plots can be viewed with xmgr.
process_md_perl < md0.out
This script will create a bunch of files called summary_xxx where xxx is temperature (TEMP), totalenergy (ETOT), kinetic energy (KETOT), etc
To plot these files use xmgr.
Equilibration: Temperature
The temperature seems fairly equilibrated, but this does not necessarily mean the system is fully equilibrated,especially because the temperature coupling we use in AMBER forces the temperature to the desired value.(Note: since this setup uses an effectively infinite cutoff, once it is equilibrated the temperature scaling could beturned off; see NTT in the Sander manual.)
Equilibration: Energy
Looking at the total energy of the system we see a definite trend to lower energy that presumably has notcompleted. The total energy of the system can be subdivided into its main components, kinetic and potentialenergies and are shown below.
16
Equilibration: Kinetic Energy Component
Equilibration: Potential Energy Component
Thus after the initial warming period, kinetic energy is relatively constant (it scales with temperature so is notreally worth checking, in fact), but the system is gradually relaxing to a lower-energy conformation. Furtherequilibration is called for. Note: in a periodic system, we would also want to check pressure fluctuations anddensity.
We restart from the saved coordinates of the previous run, using the same conditions but setting theappropriate variables to use the saved final velocities (IREST=1, NTX=5) and letting it run twice as long(NTSLIM=2000) (see Appendix 5 for a listing of the control file md1.in):
Looking at the potential energy for the combined runs leads to the plots shown below. These were generatedare for the first run. You should generate these plots following the procedure given above using the perl scriptand xmgr.
Equilibration 2: Potential Energy Component
17
The potential energy clearly has still not fully equilibrated. At this point, it appears that the assumption that fastwarming was ok could have been wrong, and a gradual warming protocol or use of cartesian restraints in theinitial warmup may have been in order (this more conservative approach is recommended in general). On theother hand, 6-10 psec of equilibration is not an excessive requirement.
In effect, we have performed a mild form of simulated annealing on the initial structure, and in any case need toknow how far the solute has moved from the crystallographic starting position. For this analysis, we turn tocarnal.
Carnal-II
To get a rough measure of how far the structure has drifted from the crystal form during the equilibration, weperform a root-mean-square (RMS) comparison of the solute atoms that were allowed to move, using the firstset as a reference. (If this system were not held in place by the stationary atoms, we would want to get the bestfit of each coordinate set to the first using carnal's RMS FIT option.)
Three types of RMS are measured: the entire moving part of the system; the moving biotin to get a rough ideaof the changes that it experiences in its pocket; and the biotin using the FIT option to see how much of its variation is due to internal (vs. orientational) changes (See appendix 6 for a listing of carnal_rms.in).
carnal < carnal_rms.in > carnal_rms.out
As might be expected, the average RMS decreases in each successive case:0.9, 0.7 and 0.2 Angstromsrespectively. The time results:
Equilibration: RMS Deviation from Initial Structure
18
Equilibration: RMS Deviation of Backbone from Initial Structure
Although the equilibration is not quite complete based on the potentialenergy above, the overall RMS indicates that the structure is not driftingprogressively (as virtually guaranteed by the frozen part). The overall RMSof ca. 1 Angstrom is well within the norm for simulations; beyond 2Angstroms would be alarming. The backbone RMS is also reasonable:NMR-derived structures tend to differ from X-ray by about 0.8 Angstroms.(Different X-ray structures vary by about 0.4 Angstroms.)
Turning to the biotin:
Equilibration: RMS Deviation of Biotin fromInitial Structure
Equilibration: Internal RMS Deviation of Biotin
19
The first graph indicates a quasi-periodic motion relative to the initial structure that may be interesting to exploreonce the trajectory has equilibrated. The internal variations are faster and smaller, as one would expect.
Since the solute RMS is not drifting appreciably, a progressive change in water structure presumably is thecause of the progressive dropping of potential energy. This stands to reason, since we simply superimposed asphere cut from a periodic system of pure water onto the solute and trimmed away any waters that overlapped,then warmed the local system rapidly to 300K, with nothing to prevent the sphere of waters from expanding.Therefore one would expect the drop in potential energy to correlate with the waters settling in toward thesolute. Measuring the distance between the geometric centers of two groups of atoms, the biotin and the wateroxygens, illustrates:
carnal < carnal_dist.in > carnal_dist.out
resulting in md01.dist.
Treating the trajectory as if it were at equilibrium for didactic purposes, we run carnal again to analyse hydrogenbonding between the ligand and its receptor:
carnal < carnal_hbond.in > carnal_hbond.out
The summary data from the .out file shows that 6 hbonds persist throughout the trajectory ('grep 100carnal_hbond.out'). They are:
Thus all of the potential hbonding atoms of the biotin except for the sulfur are 100% engaged with the acceptor. As one might expect, these hbonds are not weak; e.g. the first in detail:
63 (SER 13 OG )_(SER 13 HG )..(BTN 479 O3 ) % occupied: 100.000000 distance avg 2.592 dev 0.113 max 2.971 min 2.410 N 150
20
angle(deg) avg 7.999 dev 4.300 max 21.185 min 0.890 N 150
21
Appendices
Appendix 1-btn.frcmod file
BTNMOD Modified paramaters for Biotin (RWD)MASSSD 32.06DH 1.008DC 12.01DN 14.01