Virtual Scanning Mutagenesis of Myosin for
Spectroscopic Studies of Muscle
Introduction
In this project, computational molecular dynamics calculations were used to begin predicting the most and least appropriate sites on a myosin light chain to spin-label for electron paramagnetic resonance (EPR) experiments. Recent advances in molecular-modeling technology have provided new opportunities for enhancing biochemistry research. Myosin movement in relationship to muscle contraction has been studied using EPR, but some obstacles remain. EPR provides a direct observation of the movement of the spin label but only gives indirect information on protein movement. The angle and magnitude of myosin movement must be calculated from the relative location of the spin label. Since only one appropriate residue location for spin-label attachment has been found thus far (residue 108), the EPR work on myosin is limited to the use of one spin label (Thomas, 1998). This means that researchers have data in only one plane of rotation for the myosin.
Muscle contraction is a three-dimensional process, so ideally, in the future, EPR will be able to produce three-dimensional data using more than one observed plane of rotation. With present methods, this can only be done by attaching spin labels at other locations on the myosin RLC. Therefore, there has been an effort to find a new residue location to spin label other than the native cysteine location of the gizzard RLC, residue 108 (Laconte, 1998). If another effective spin-label location were found, researchers would be able to compare the rotation of the myosin observed from this new angle to the data already gathered1 on residue 108 (Thomas, 5 Jun 1998).
Information from computer modeling can be used to decide which RLC positions of spin-label attachment should be attempted by EPR researchers. In a wet-bench laboratory, site-directed mutagenesis can theoretically be used to insert a cysteine residue at any desired site on the protein so that a cysteine-reactive spin label can be attached (Laconte). However, researchers need information on which positions are most appropriate for attachment.
Modeling this spin-labeled myosin on a computer is useful in another way because modeling can provide new information about how the spin-labeled myosin is configured. The arrangement of the atoms in the protein is clearly visible on a computer screen, so the impact of the spin label on the original structure can be observed. With that knowledge, the computer should be able to determine which positions would be best suited to accommodate a transferred cysteine residue with an attached 3-(5-fluoro-2,4-dinitroanilino)-2,2,5,5-tetramethyl-1-pyrrolidinyloxy (FDNASL) spin label (Fig. 2) (Laconte, Nov 1998).
By attaching a spin label to a cysteine residue on the myosin RLC (Fig. 1), the movements around the spin label site can be studied to determine the motion of the entire myosin head. In EPR, the orientation of the spin label relative to the magnetic field can be observed. It is important to be sure that the observed spin-label motion actually indicates the orientation of the object under study (the myosin muscle protein). A proper site for attachment immobilizes the spin label so part of the spin label is buried in the main structure, and a rigid part of the spin label is exposed to the outside. If the spin label is not rigidly attached and anchored to the myosin RLC, the indicated movement of the spin label will not meaningfully indicate the movement of the entire myosin head (Baker; Roopnarine; Thomas, 1980).
A problem that must be avoided when editing the structure of a protein is trapping the atoms of a spin label too closely to surrounding atoms. If the computer is unable to position the atoms of the main structure far enough away from the atoms of the spin label, the repelling forces of those particles will make the structure unstable. Some residue locations provide enough room to accommodate a spin label comfortably and some do not. The issue of seeking a physically stable structure and dealing with the intricacies of crude software to achieve this are the main points of focus dealt with in this project (Thomas, 5 Jun 1998).
The computer is a useful tool to subjectively observe the quality of a location by looking at a three-dimensional model, and to quantitatively determine which location is more stable (lower energy in kcal/mol). The computer's ability to automatically construct a complete myosin with a cysteine located anywhere on the RLC provides a unique opportunity to search for another attractive mounting site. The Appendix of this report contains detailed procedures created by this writer for the tasks outlined in the methods section. By using a computer to reliably and quickly simulate the various attachment sites for the spin label, a great deal of time and expense in the laboratory can be saved, since in the laboratory, each structure must be custom-built in an expensive trial-and-error process. I hypothesized that computer modeling will determine favorable spin-label sites for protein structural studies by considering all possible residue positions on the RLC for spin label attachment. At the time this project was begun, no reliable method existed for this type of prediction.
Methods and Materials:
Materials:
The software used in this project was Insight II ver. 97.0 from Molecular Simulations, Inc. Insight II consists of a base program that handles display and several modules for different tasks. Modules used include Builder, Biopolymer, Discover, and Analysis and they were run on a Silicon Graphics workstation with 320MB of RAM.
Preparation Methods:
The structure of the regulatory domain of scallop myosin, generated by x-ray crystallography, was imported from the Protein Data Bank (PDB) on the internet into the Insight II program. This structure was missing seven residues before the N-terminus of the RLC and eleven after the C-terminus. The Append:Residue command was used to append the missing residues after the C-terminus of the structure in a left-handed helix motif. The appended structure was less random and more unnatural when compared to the crystallized structure. When a molecular dynamics simulation was performed on this structure, using the Discover module, the results were not satisfying because no single position could be quantitatively proven to be optimal. It was determined that little benefit could be gained from trying to append a small "fake" structure onto a valid crystallized structure for this project. It was decided that work should continue on the natural stretch of RLC structure rather than on the artificially appended RLC. Consequently, the scallop amino residues of the RLC were replaced with those of a smooth muscle from a chicken gizzard. The chicken gizzard RLC sequence was retrieved from the SWISSPROT database, and its residues were arranged over the existing geometry to mimic the fabricated protein, which is commonly used for EPR study.
The natural scallop myosin structure was modified to represent the hybrid scallop/gizzard structure used at the laboratory of Dr. David Thomas at the University of Minnesota. The RLC section of the original scallop myosin was replaced by a chicken gizzard RLC, which included the vital cysteine residue. The cysteine served as the anchor for proper spin-label attachment. Although a cysteine could not be added to the original scallop regulatory light chain because that would change the balance of atomic charges in the protein, the cysteine of a gizzard RLC could be moved to different positions with little adverse effect. The Biosym Homology module on the computer was used to extract the sequence of the scallop RLC structure and visually line up the residues with the corresponding residues on the imported gizzard sequence. A structure similar to the scallop RLC was built for the gizzard RLC sequence, which was appropriate for spin-label attachment.
The resulting hybrid structure closely approximated the hybrid structure used for EPR work in the lab. There were two main difficulties in preparing this virtual structure properly. First, the entire original RLC was deleted from underneath the newly generated RLC model. This left gaps in the gizzard structure, which had to be repaired manually paying careful attention to atomic bonds. Secondly, this program corrupted certain bond types (in the PDB file format, some double bonds turn into single bonds after the Homology module is used). Since that problem prevented potential charges from being fixed, which was necessary before the computer could begin calculations, the bonding problems were repaired using Biopolymer. This model of the gizzard/chicken hybrid was the basis for the subsequent work.
Simulated Annealing:
Proteins like myosin and even molecules like the FDNASL spin label have certain rest configurations, toward which their atoms naturally gravitate. It was important to be sure that the computer files for the basic myosin and the spin label were well balanced before any combination took place. To stabilize the protein structures physically, they were allowed to settle in their preferred, most relaxed physical configuration by employing extended energy minimization calculations. This procedure was used to finalize the vacuum structure, water-soaked structure, and saltwater structure with potassium and propionate ions. Simulated annealing was also used to finalize the structures of the gizzard hybrid structure and the FDNASL spin label. Molecular dynamics were performed on this structure to avoid traps in minimizing energy. The program identified trends in energy fluctuations as it moved the structure.
Sometimes, instead of finding the true natural state of a protein, the structure became "trapped" at a local minimum energy reading where moving it in any direction increased the total energy. The possibility always existed that another position might be even lower in total energy. The program was not able to use imagination to find another position. Dramatic random agitation of the protein was done using the dynamics calculations.
In Vacuo Molecular Dynamics:
Molecular-dynamics runs were done on a computer-simulated hybrid gizzard structure in a theoretical vacuum. By default, the computer thinks of each atom of the structure individually and does not assume that any particular matter surrounds the structure. Some dynamics runs were performed at 300 K on the structure in a vacuum, but no movement trends were observable due to a slow calculation speed. Extended dynamics runs were done at 900 K to speed up the movement process. Since living tissue is not normally subjected to 900 K, a temperature of 300 K was used for further simulations, using the Distributed Queuing System (DQS) for job submissions to achieve better calculation speed.
Solvated Molecular Dynamics:
Insight II was used to add a six-angstrom layer of water to the structures in the hopes of producing a more realistic structure. The water layer included just over 2500 molecules. The dramatic stabilization benefits of soaking the structures in water and the uniformity between soaked structures supported a decision to do all primary quantitative comparisons in this project using solvated molecules.
Auxiliary comparative calculations were done:
• Water-soaked vs. vacuum minimization energy
• 2000 iterations vs. 50 iterations dynamics relaxing ability.
• Water vs. vacuum movement freedom (animation playback)
Saltwater Molecular Dynamics:
To more accurately simulate the chemical makeup of myosin in a living organism, potassium and propionate ions were added. This was done to determine the most natural environment possible in which to test delicate structural variables, such as spin-label location. Only seven sets of potassium ions and propionate salts were used, simulating a concentration of approximately 200 mmol/L. Although slightly more stable than pure-water soaking, saltwater was not used for primary quantitative comparisons for reasons cited in the Discussion section.
Finding Attractive Mounting Sites for the FDNASL Spin Label:
Since previous laboratory work has shown that residue number 108, the native cysteine location on the gizzard regulatory light chain, is particularly stable dynamically and is harmonious in terms of latent energy when fitted with a spin label, the primary method of searching for a good mounting site centered around comparison to that benchmark. Since residue 95 is known as an unstable residue, initial tests were run to determine whether it would reproduce this fact.
A macro, called test10a, written by this researcher and found in Appendix X, instructed the computer to build a myosin with a gizzard RLC with any of its 137 native residues changed to a cysteine. The new structure or structures were automatically fitted with the FDNASL spin label and had calculation files created. This streamlined the task of eventually comparing large numbers of structure choices.
Discussion:
Computer simulations of spin-labeled myosin showed much promise by supporting existing laboratory data as well as beginning to predict new data. Molecular dynamics runs shown in Table 2 showed that better bonding sites, such as residue 108, tended to have lower total energy statistics in kcal/mol after dynamics runs compared to known unstable sites such as residue 95. The other residue locations tested, numbers 72 and 89, displayed far higher minimized total energies than 108 with post-dynamics minization. Residue 72 has indeed proven difficult to spin-label (Laconte, Nov 1998). This means that molecular energy readings could be used to meaningfully compare the differences in stability among competing structures. However, data from raw structure minimization, shown in Table 1, did not produce minimized energy readings that correspond with known laboratory data. This suggested that dynamics is a necessary intermediate step for the computer to produce a protein in a natural pose for accurate comparative minimization. With post-dynamics structures, a poor residue location, such as number 95, placed some atoms on the spin label too close to original myosin atoms, so their repelling forces generated both structural instability and an energy reading much higher than the optimal residue location. This suggested an increased level of movement and instability between the spin label and the protein, which is precisely what EPR researchers try to avoid.
The secondary aim of this work, finding the most meaningful way to simulate protein dynamics on the computer, yielded results related to how the surroundings of a protein affect its total energy. When simulated dynamics calculations were performed in a vacuum, the structure was prone to fall apart and completely disintegrate, no matter what bonding site was used for the FDNASL spin label. Water contained this movement very effectively and kept the protein from disintegrating. Water stabilized the structure so much that total energy fell dramatically compared to identical structures in a vacuum, as seen in Table 3. The addition of salt, in order to more closely simulate biological conditions, furnished a further drop in total energy. This showed that, for simulating the movements of a living protein on the computer, soaking in saltwater stabilizes the system and makes movements more realistic. However, uncertainties related to the effect of relative positioning of so few salt molecules around the molecule prompted a decision to use pure water soaking for all primary calculations in Tables 1 and 2.
An extra finding of this project dealt with the best way of performing molecular dynamics without wasting too much computer time. While minimization jobs stop automatically once they are unable to improve the molecule, dynamics jobs sometimes run for many hours with no real change in energy. The myosin proteins and other molecules tended to stabilize in total energy immediately after they began to move. Often the computer created the structure in an unsettled position. However, after as few as 10 iterations of dynamics at 300 K, the molecule reached a stable plateau of energy from which it could be more easily minimized. In the output file scm72_2.out, it seemed that during the 18 hours of calculations on a 195 Mhz R10000 processor only 20 minutes of minimization and 30 minutes of dynamics produced any changes. In future research of this type, dynamics should be limited to 500 iterations.
Conclusion:
Computers can be used in a number of ways, both qualitatively and quantitatively, to assist biochemistry researchers in examining myosin. Now that adequate methods exist for this type of exploration and more is known about the best way to simulate myosin, large-scale computer experiments are on the verge of producing useful data for EPR researchers to consider. The best mounting sites for spin labels can probably be identified visually on the screen using the script written by the author for this project (see Appendix). The files, created assembly-line style by this script, contain the three-dimensional structures of hybrid myosin with the chosen RLC residue converted to cysteine and spin-labeled. These files can also be used quantitatively when surveying for a good spin-label site, as was done in this project for residues 95, 72, and 89. The combination of these methods should provide a solid basis for future researchers to make decisions about spin-label attachment on myosin and other proteins.
Data Tables:
|
Table 1 - Minimization Energy of Raw Structures, 500 iterations at 300K, in kcal/mol
Residue 108 Initial: 1647138799849.981 Final: -16335.925
Residue 95 Initial: 503474795.080 Final: -16494.587
Residue 72 Intial: 29777619839.633 Final: -11328.733
Residue 89 Initial: (above maximum readable) Final: -17213.865
|
Table 2 - Minimization Energy of Post-Dynamics Structures, 500 iterations at 300K, in kcal/mol
Residue 108 Initial: -16335.925 Final: -25602.049
Residue 95 Initial: -16494.587 Final: -24078.812
Residue 72 Initial: -11328.733 Final: -22943.558
Residue 89 Initial: -17213.865 Final: -23361.865
|
Table 3 - CYS 108 Structure Energy Levels With Various Surroundings, in kcal/mol
Vacuum
Initial: 17188562280.698 Post-Min.: 4727.709 Dynamics after 100: 11781.981 after 1000: 10802.923
Water
Initial: 17188549625.573 Post-Min.: -18339.085 Dynamics after 100: -823.682 after 1000: -2219.820
Salt-Water
Initial: (above maximum readable) Post-Min.: -19481.231 Dynamics after 100: -2098.229 after 1000: -3686.905
|
Table 1 & 2 Notes:
Each molecule soaked in pure water 6 angstroms thick. Dynamics runs done in Discover at 300K for 3000 iterations.
Table 3 Notes:
Vacuum structure created by BCL script "test10a.bcl". Water structure derived from Vacuum structure by soaking in 6 angstrom layer of water. Salt water structure created by adding 8 phos-potassium salts. [Apparently the structure is more stable in water than a vacuum by a wide margin, and yet more stable when salt is added].
Structures subjected to 100 iterations of minimization, then 1100 iterations of dynamics at 300K. CPU time for vacuum = 2.5 hrs., water = 11.5 hrs., saltwater = 18.5 hrs.
––––––––––––––
Figures:
[click to enlarge]
Figure 1: Special Myosin Hybrid
Figure 2: FDNASL Spin Label
Figure 3: Ball & Stick at Res. 72
Figure 4: CPK at Rez. 72
Appendix:
I. Opening a Molecule File
From the UNIX window, once the .cshrc file is set up with permanent user preferences, type "InsightII," and the program suite launches automatically. A row of graphical menus and a large blank visual window comes up. To load a pre-made structure, pull down the "Molecule" window and select "Get" (let's say Molecule:Get). The "Get" window has many choices for different file formats for the structures, like PDB and Mol2. Specify what format the file is, then find it in the column of file names to the right. Once a file is "Gotten", or opened, it is displayed on the screen.
II. Viewing The Molecule Onscreen
Use the buttons on the mouse to get any view of the molecule, or even a small zoomed-in portion of it. Hold the left button to make a selection box for selecting individual atoms. Hold the middle button to translate the molecule up, down, or side to side on the screen. Hold the right-side button to twirl the molecule horizontally or tumble it vertically, depending on the direction of the mouse movement. This is useful because it allows a view of the molecule at most any angle. Hold the right and left buttons simultaneously to produce a Z-axis spin, which is also very useful. To zoom in and out, hold together the middle and right-side buttons.
III. Advanced Viewing of a Single Molecule
To change the color of the molecule, or parts of it, or make a 3D rendering for high-quality printouts use the following commands. To change the color of the structure, go to Molecule:Color. The "Color" window can specify either the whole molecule or individual parts of it for color change. This is how the program is directed to an atom or chain of atoms within a structure to perform an action. If the object name is "MOL" and it has three chains, "A", "B", and "C", type "MOL:A23" to specify the 23rd residue on the A-chain. To specify the whole chain, type "MOL:A*" ("*" means anything so it will select all atoms on the chain). In the little menu in the color window marked "Color Method", click to choose other color methods such as "By Atom" (which colors the atoms according to their makeup [green = carbon, red = oxygen, blue = nitrogen, white = hydrogen, yellow = sulfur, etc]).
Alternatively the menu can be left at "Specification" and the "Color" blank underneath used to choose the color. For 3D rendering go to "Molecule:Render". First, specify an object or parts of an object as above. Different Render Styles, like "CPK" which displays each atom as a large sphere or "Ball and Stick" can be chosen. Select "Ribbon" from the Molecule menu to spin a 3D ribbon along the backbone of the structure. Use the molecule movement procedures from Appendix section II to get a proper view of whatever part of the molecule holds interest. Use parts I-III to highlight points of interest using different colors and to spin and zoom into that critical areas to render for viewing in three dimensions.
To print the molecule view or export it as a graphic from Insight, the background should be switched to white to avoid massive black ink or toner waste. Go to Session:Environment and click the Background button. Use "File: Export_Image" to make a TIFF file or SGI RGB file of the view for printing. For best results when viewing a molecule, not more than ~20 residues, show the 3D structure rendered on-the-fly as the molecule is spun or tumbled on screen. Instead of spinning the lines-only structure and rendering it in 3D once molecule movement stops, continuously render the object to spin it. Go to Session: Environment and toggle "Render_in_Motion" to "on".
IV. Editing The Molecule: Step-By-Step, By Module
Insight's various modules can perform many tasks which must be classified according to which module can perform them. Click the the square "MSI" icon at the top left corner and hold; a menu of all of the modules appears, inlcuding Builder, Biopolymer, Discover, Homology, and Analysis. Use Builder and Biopolymer to make heavy structural modifications to the molecule including changing bonds, connecting common fragments or appending more amino acids, altering charges, or generating a new structure based on an outside amino sequence. Find the most energy-efficient state of a molecule's current configuration using Discover. Also use Discover for molecular dynamics, random motion, to try to find a new configuration which might be even more energy efficient than the original one.
V. Builder
The Fragment:Get command can be used to produce fragments and put them near a structure. Choose Amino Acids, Common Fragments, etc. in the Fragment Window. Residues can be added to the end of the peptide chain using the Residue:Append command in Builder to construct motifs for the appended residues, including left-or right-hand helix configurations. Check to see which direction the original structure is heading in the appended region. To symbolically complete a molecule that did not completely crystallize use this command and refer to the complete sequence information related to the incomplete structure. To execute a short minimization run on the structure to improve its energy efficiency, use the "Optimize" menu and command in Builder.
A note about fixing potentials in Insight... When structural changes have been made to the molecule and Insight is used to enter a new module such as Discover, the program requires that the potentials be fixed. It takes time to fix the potential charges for a big molecule. If it does not work, the computer will give an an error message such as "Error: MOL:C45:CB has undefined valences. Unable to assign potentials to MOL." If this happens, zoom in on MOL:C45 and set Molecule: Bond_Order to On for the object MOL. Now, look at the beta carbon on MOL:C45 to see what the problem is. Are there too many bonds to that atom? Too few? What other obvious problems can be seen? Fix this type of problem probably using Biopolymer.
VI. Biopolymer
To make more delicate changes to the structure, use Biopolymer. To perform operations such as fixing an undefined valence on an atom, merging with another protein, or fixing the potentials of the new combined structure: first, go to Molecule:Get and load the two molecules for repair and merge. Once the molecules are on screen, determine if there are undefined valence problems with the structure. Go to Forcefield:Potentials, and set the choices to "fix" everything. Click Execute. If this works, the structure is correct enough so that minimizations or molecular dynamics can be done on it using Discover. If not, repair the structure using Biopolymer.
A. Repairing Undefined Valences with Biopolymer
If a problem of undefined valences surfaces when trying to fix potential charges as a result of saving a file in a different format from its original format,2 examine the error message from the Potentials function. Scroll up in the Insight message window to look at the whole message. It will mention all problems, one offending atom on each line. To isolate a particular atom, go to the Molecule:Color menu, and first color the entire structure a soft shade like magenta. Then, to isolate the problematic residue, name it in the window and choose "By Atom" for the color method. Hit Execute. Now the problem residue is highlighted in light green. If bond order cannot be seen, set Molecule:Bond_Order to ON. Now, using molecule movement skills, zoom in on the problem residue. Use scientific knowledge: what is wrong with the atom in question? Are there too many bonds? Go to Modify:Bond and choose Modify Order to create, break, or change single, partial double, or double bonds. To merge this molecule with the other one on the screen in order to instruct the computer to treat them as one, go to Modify:Merge. To fix the potentials on the structure now that it has been repaired, go to Forcefield:Potentials again and choose to fix everything. If this does not work, repeat the steps just described. Once repairs have been completed and the potentials are correctly set, save immediately. Go to Molecule:Put, and save in Biosym format in your directory. Now the structure is ready for simulations in Discover.
B. Appending Residues To Your Structure
If the structure did not crystallize completely, showing residues in the known sequence that do not show up on the structure or producing gaps, use Homology to fill the gap as described below. Howver, ff the missing residues are either at the beginning or end of a chain, use Biopolymer's Residue:Append function. Choose the preferred motif in which the new residues should be arranged. Pick the molecule to which add residues will be added using Residue blank.Refine the structure using Discover.
Note: I t might be necessary to use Modify:Unmerge to declare a chain of the rotein a different molecule for the purpose of appending residues to that particular chain..
C. Renaming Residues Quickly And Easily
If the names assigned to some residues do not match the names used for the rest of the protein after appending residues to a structure, go to Protein:Rename and type in the full extent of the molecules to be renamed, e.g. MOL:1-13. The computer will automatically name the new residues in order like B68, B69, etc.
Note: If the computer named the new residues 1B, 2B, or similar with the etters after the numbers, there is no known quick way to rename them all at once. If the computer named them this way ,they need to be renamed invividually.
VII. Discover
To move the structure to discover (sic) other possible orientations or find the lowest-energy position, first, enter the Discover module through the MSI menu. Go to Parameters:Dynamics and choose how many iterations the computer should do. For better results, choose more iterations, however, it takes a very long time to do thousands of iterations on a complex molecule containing thousands of atoms and the bonds between them. Consider either raising the temperature to speed up the motion or constraining the bond lengths using Discover 3 (not covered here; ask someone for a Discover 3 input file). Find choices for preparation of the dynamics run in the Run:Run window. If the run will be a simple one of under ~250 iterations or with a small molecule of under ~100 atoms, choose the Interactive choice within this menu. (This is slower but you can watch it happen and there is no need to exit the program at all.) Check the appropriate boxes for Minimize, Dynamics, or both, and hit Execute. The computer will create the appropriate input files and begin running the job on the screen, updating the picture with each new update. Make sure there is enough space in the home directory (at least 10MB available for EACH run-the files are not deleted automatically when done either). To change the assigned directory to a more accommodating one, such as /wrk, go to Session:Change_Directory. If the job is running interactively, little statistics in white letters should appear saying what iteration it is being run, how long it will take, and the current total energy of the molecule. This is especially useful for minimizing something quickly and for determining how fast the energy is falling. For a longer, more complex job involving a large molecule over a long time period, submit the job to the DQS (Distributed Queuing System). Once the minimize or dynamics parameters are set correctly in Discover, go to the Run:Run window and choose Command File. When this executes, three files are created in the home directory: if the object is "mol", then mol.inp (the input file with instructions), mol.car (the Biosym-format molecule) and mol.mdf (the molecular data file) are created. The common prefix for these files is thus "mol". Go to the UNIX window. Make sure to change the UNIX directory to the one with the files Discover created, like this : cd /wrk/johnson/discover/. Now the window will assume any files mentioned are located here. Type "discover". This launches the text-based version of the program. Type in the prefix, e.g. "mol". Use forcefield #1 and any number (it doesn't matter). When it asks if you want to run Discover now, type no. It has now created mol.csh. Now, to get the DQS to perform the job, a ".dqs" file must be accessed. If there is none, go to Bob Milius' DQS help page at http://www.msi.umn.edu/bscl/info/dqs/ and use his format to make your own ".dqs" file. Open the file and make sure it refers to mol.csh in the last line of the file. Once the mol.dqs file is made and saved it in the same directory as the other files with the "mol" prefix, go to the UNIX window again. Make sure the directory for the window is set to the one you are working with. Now type "qsub3 mol.dqs" to submit the job. Check the status of the jobs in the DQS by typing "qstat3" at any time.
VIII. Homology
Use Homology to deal directly with the sequences of physical structures. To view the sequence of the structure, enter Homology, and go to Sequence:Extract. Select the object and the Sequence Alignment Window should come up. It displays the residues of the structure in single-letter form, in capital letters. Use this window to generate a new structure according to a different but related sequence. A tutorial describing how to do this is below:
**** HOMOLOGY TUTORIAL ****
This is a set of clear, step-by-step, informal language instructions for using Biosym Homology to generate a 3D structure for a sequence based on the existing 3D structure of another related sequence. A meaningful structure can only be generated if a related structure exists, so find one before proceeding.
I. LOADING THE BASE STRUCTURE
(1) Launch Insight II.
(2) In the square-icon "MSI" menu, select Homology.
(3) You are now in the Homology module. Load the base structure by going to the Molecule menu and selecting "get".
(4) A dialog box appears. Choose your file format from the list in the box, and point it to the directory. Once the file is loaded, click on it and the computer should show its Object in the blank. Click execute.
(5) The 3D structure should be visible in the graphical window. If needed, adjust the slab thickness by clicking on "Slab Position" and dragging the mouse to the right.
(6) To load the sequence directly from the 3D structure you are displaying, go to the Sequences menu and select "Extract". Choose the object and hit Execute.
(7) The Sequence Alignment Window appears. You have now loaded your base structure into the program.
II. BUILDING A NEW STRUCTURE FOR A LONE SEQUENCE
(1) Click out of the Sequence Alignment Window (you cannot close it).
(2) Go back to the Sequences menu and select "Get".
(3) Select SWISSPROT for your file format, assuming that is where you got your sequence. Now, type in the exact file name of the sequence, including suffix. You cannot browse your directory in this window. If your sequence does not have an object name, assign one.
(4) If that worked, you will see your lone sequence written right below your structure's original sequence in the Sequence Alignment Window. Sequences with an associated structure are displayed in capital letters in this window, while lone sequences are in lower case.
(5) You can use the middle mouse button to align the two sequences if necessary. Just drag either row of letters and they should slide freely. Perhaps the computer can do this automatically but it is easy and fun to do it yourself.
(6) At the bottom of the Alignment window change Mode to "Box". Make sure that you have accounted for gaps in either of the sequences. If you matched the sequences based on the first few letters, go to the end of the region you plan to manifest and make sure the sequence is still aligned properly so your new structure keeps the common points stable.
(7) Using the left mouse button, drag and make a box holding both rows of a section of uninterrupted, aligned sequences. Make more boxes around the other parts you want to manifest, being careful to avoid gaps we will deal with later.
(8) Go to the Boxes menu and choose "Freeze". Enter "*" for box number and Execute. All boxes will be solidified and turn red.
(9) Go to the Sequences menu and select "AssignCoords". In the Alignment window, click on each red box in order. This is when the new structure is made.
(10) If there were gaps in your sequence that you avoided when making boxes, the computer has not created a structure for them yet. Go to the Loops menu and choose "Generate". Click on any residue in the box preceding the gap and another residue in the box succeeding it. The computer will adjust its structure as well as it can, often giving you ten choices which you must sort through. Pick your favorite.
(11) It is likely that there are various objects on the screen that you no longer want. Go to the Object menu, choosing Delete. There may be leftover things the computer created for you in the Generate Loops command. Delete those except for the one you want to use for each gap. Also, you may want to delete the original structure that you used as a basis for the new one. When everything is cleaned up, go to the Molecule window and choose Put to save you new object. If you want to save multiple objects in their relative positions, go to the File menu and select Save_Folder.
IX. Analysis
To look at the results, open mol.out, the output file that is created by Discover. It tells what was done, how long the run took, and some basic statistics about the molecule during the run. For a visual replay of the job, enter the Analysis module of InsightII. First, go to Molecule:Get and get mol.car, the file used as input for the run. Notice that mol.cor now exists. This is the final image of the molecule, after the dynamics or minimizations have been done to it. Do not use that file for this task. Once the molecule is onscreen, go to Analysis and pick Trajectory:Get. This window needs mol.his, the history file of the DQS run, and MOL, the object already loaded. The bottom three blanks deal with how many frames to display. The maximum number of frames the computer can load is about 200 if the molecule is like a scallop myosin. Do the math, considering that the history file was propbably set at default, every 10 iterations. If 2000 iterations were assigned, that may be too many frames to display. Choose "2" for Step so you only get 100 frames. In general, divide the number of frames you think were created by the Step number so it gets below 200. Otherwise the computer will run out of RAM with a big molecule. Once the trajectory is captured, you will want to animate it. Go to Trajectory:Animate and animate all of the images.
X. Macros
Use Biosym Command Language (BCL) to streamline any repetitive operations. Notice that when operations are done in Insight, the text window shows what command was just completed. This text output is saved in the format of a log file. In the home directory, there should be a file called WBLOGFILE (the current-session log) and WBLOGFILE.save (log of the last session). The text from these files can be used as commands to do the same thing all over again. Simply select the text of the commands performed in the past and paste it into the command window. The program will execute the commands again, provided the existing structures are the same. This gets even more useful when a ForEach... parameter is used, which counts through a certain set of numbers and lets the program create things with successive numbers, assembly-line style. Any letter in front of a colon in this macro was added here to identify areas of interest. To use the macro "test10a" displayed below, delete letters and the colons at the beginning of each line.
A:Define_Macro test10a int rfrom int rto
int res
B:Get Molecule Sybyl_Mol2 /home/meland/732scm.mol2 SCM -Reference_Object
C:Biopolymer
D:Rename Protein SCM:A1-A60 SCM:A777
Rename Protein SCM:C61-C209 SCM:C4
Rename Protein SCM:B210-B347 SCM:B25
E:Replace Residue SCM:B108 ALA L
F:Potentials Forcefield Fix -Print_Potentials Fix -Print_Part_Chargs Fix
-Print_Form_Chargs SCM
G:Print "Source Protein Ready. Starting Homology..."
Homology
H:Foreach $res From $rfrom To $rto
Copy Object SCM ("SCM"//$res)
Replace Residue ("SCM"//$res//":B"//$res) CYS L
End
I:Change_directory /wrk/meland/
Print "Cloned Objects Created. Ready To Attach Spin Labels..."
Print "Entering Biopolymer..."
Biopolymer
Get Molecule Sybyl_Mol2 /home/meland/H.mol2 H -Reference_Object
J:Foreach $res From $rfrom To $rto
Copy Object H ("H"//$res)
Bond Create -Fragment_Window Single ("SCM"//$res//":B"//$res//":HG")
("H"//$res//":4:H4") -Assimilate
Potentials Forcefield Fix -Print_Potentials Fix -Print_Part_Chargs Fix
-Print_Form_Chargs ("SCM"//$res)
Print ("SCM"//$res//":B"//$res//" now has attached spin label.")
End
Print "Spin Labels Attached. Saving Command Files..."
Discover
K:Foreach $res From $rfrom To $rto
Rename Object ("SCM"//$res) ("SCM"//$res//"_")
Run Local -List ("SCM"//$res//"_") "Command File" -Strategy Add_Auto
Run_Minimization Run_Dynamics -PBC Reduce_Output
End
Change_directory /home/meland/
End_Macro
Explanations: (A) gives the title of the macro. Name the text file the same as the title in the macro. In this case the title is "test10a.bcl". Thus, in the "File:Source_File" window, pick the test10a.bcl file. Then type "test10a.bcl" into the command line and the program activates this macro. (B) tells the program to get a molecule in Mol2 format of the specified name. (C) switches InsightII to the Biopolymer module. (D) renames all of the residues because the mol2 format tends to corrupt the atom names. (E) since the script is built to add a cysteine to any specified residue number on the light chain, the native cysteine location is changed to an ala nine. If position 108 is used often, the script will change 108 back to cysteine. (F) The potentials are fixed on the new structure here and then fixed after the new cysteine is added later in the script. (G) Anything you tell it to "Print" is displayed in the command window before it does the next task. This is useful to see if the script is proceeding well or if it got stuck earlier in the job. These lines to not affect the calculations. (H) This is one of those ForEach commands that tells the computer to create a copy of the SCM structure, such as SCM65, and replace its B65 residue with a cysteine. (I) This changes the directory used by the program for writing and reading files to the specified one. (J) For each of the numbered objects, an FDNASL spin label named "H" is created and bonded to the appropriate cysteine residue. (K) This is another ForEach command which sets up files for identical molecular dynamics runs.
References:
Baker, J. E., I. Brust-Mascher, S. Ramachandran, L. E. W. LaConte, and D. D. Thomas. 1998. A large and distinct rotation of the myoin light chain domain occurs upon muscle contraction. Proceedings of the National Academy of Science, 95:2944-2949.
Holmes, K. C. 1997. The Swinging Lever-Arm Hypothesis of Muscle Contraction. Current Biology, 7:R112-R118.
Houdusse, A. and C. Cohen. 1996. Structure of the Regulatory Domain of Scallop Myosin at 2Å resolution: implications for regulation. Structure, 4:21-32.
Laconte, L. E. W. 1998. Site-Directed Spin Labeling of Myosin: Conformational Changes During Muscle Contraction.
Laconte, L. E. W. 1998. Letter to author. 10 Nov 1998.
Roopnarine, O. and D. D. Thomas. 1994. A spin label that binds to myosin heads in muscle fibers with its principal axis parallel to the fiber axis. Biophys. J. 67:1634-1645.
Thomas, D. D. Letter to author. 5 Jun 1998.
Thomas, D. D. Letter to author. 19 Jun 1998.
Thomas, D. D. and R. Cooke. 1980. Orientation of Spin-Labeled Myosin Heads in Glycerated Muscle Fibers. Biophys. J. 32: 891-906.
Xie, X., D. H. Harrison, I. Schlichting, R. M. Sweet, V. N. Kalabokis, A. G. Szent-Györgyl, and C. Cohen. 1994. Structure of the Regulatory Domain of Scallop Myosin at 2.8 Å resolution. Nature, 364: 306-312.