Laboratory Exercise Molecular Mechanics and Dynamics Characterization of an In Silico Mutated Protein: A Stand-Alone Lab Module or Support Activity for In Vivo and In Vitro Analyses of Targeted Proteins
Harry Chiang† Lucy C. Robinson‡ Cynthia J. Brame§ Troy C. Messina†*
From the †Department of Physics, Centenary College of Louisiana, Shreveport, Louisiana 71104, ‡Department of Biochemistry and Molecular Biology, LSU Health Sciences Center, Shreveport, Louisiana 71103, §Center for Teaching, Vanderbilt University, Nashville, Tennessee 37212
Abstract Over the past 20 years, the biological sciences have increasingly incorporated chemistry, physics, computer science, and mathematics to aid in the development and use of mathematical models. Such combined approaches have been used to address problems from protein structure–function relationships to the workings of complex biological systems. Computer simulations of molecular events can now be accomplished quickly and with standard computer technology. Also, simulation software is freely available for most computing platforms, and online support for the novice user is ample. We have therefore created a molecular dynamics laboratory module to enhance undergraduate student understanding of molecular events underlying organismal phenotype. This module builds on a previously
described project in which students use site-directed mutagenesis to investigate functions of conserved sequence features in members of a eukaryotic protein kinase family. In this report, we detail the laboratory activities of a MD module that provide a complement to phenotypic outcomes by providing a hypothesis-driven and quantifiable measure of predicted structural changes caused by targeted mutations. We also present examples of analyses students may perform. These laboratory activities can be integrated with genetics or biochemistry experiments as described, but could also be used independently in any course that would benefit from a quantitative approach to protein structure– C 2013 by The International Union of function relationships. V Biochemistry and Molecular Biology, 41(6):402–408, 2013
Keywords: molecular dynamics; homology; modeling; genetics; computers in biology; physical biology; teaching
Introduction Teaching and research in the biological sciences have been undergoing vast changes over the past 2 decades [1]. The changes include incorporating a broadening number of scientific disciplines, especially chemistry, physics, computer science, and mathematics, to solve complex biological problems. The development and use of mathematical models based on fundamental principles of chemistry and physics have become the basis of systems biology, using knowledge gained from
*Address for correspondence to Department of Physics, Centenary College of Louisiana, 2911 Centenary Boulevard, Shreveport, Louisiana 71104, USA. E-mail:
[email protected] Received 9 May 2013; Accepted 22 August 2013 DOI 10.1002/bmb.20737 Published online 9 November 2013 in Wiley Online Library (wileyonlinelibrary.com)
402
biological inquiry to drive modeling that can aid in understanding the workings of biological systems from subcellular machines to cells and tissues. Visioning for future research and education along these lines has been detailed by the National Research Council (NRC) [1]; the American Association of Medical Colleges (AAMC) in collaboration with HHMI [2], and the National Science Foundation (NSF) and the American Association for the Advancement of Sciences (AAAS) [3]. Computer usage has increased dramatically in the physical sciences to assist with previously intractable problems. For example, protein folding and structure=function relationships can now be simulated, with results available within a few months for most systems. This makes computational results available on the same timescale as laboratory work. Furthermore, simulation software is freely available for most computing platforms (Windows, Mac, and Linux) with a substantial amount of technical support (see, e.g.
Biochemistry and Molecular Biology Education
http:==www.ks.uiuc.edu=Research=namd=). The increased use of computational tools in biology and readily available software induced us to use these approaches to enrich a research-based lab project that is pursued annually in a Genetics course. Specifically, we used online molecular modeling tools to allow students to generate data to complement their wet lab experiments. After observing the students’ use of the tools, we concluded that the computational exercises would also make excellent stand-alone pedagogical tools.
Overview of the Project The research-based lab project has been described previously [4]. In brief, students use bioinformatics tools to identify conserved regions of a protein of interest (in our case, a casein kinase 1 [CK1] protein kinase) and investigate the function of these conserved regions. In the “wet lab” component of the project, the students 1) use visualization software (Cn3D, supported by the National Center for Biotechnology (NCBI) [5] to visualize the protein structure and identify the locations of conserved regions; 2) form hypotheses about the function of the conserved region of their choice based on these observations; 3) design and generate mutations to test their hypotheses. The molecular modeling=molecular dynamics component described here complements the wet lab by allowing students to develop quantifiable predictions of structural changes caused by the mutations they designed. We incorporated the molecular dynamics module after students had identified conserved regions, developed hypotheses about their function, and planned mutations to test their hypotheses. The students were therefore able to enrich their ongoing wet lab experiments using molecular dynamics software to generate “their” mutations and observe the effects on enzyme structure in silico. These structural effects were quantified by calculating the distance between two sets of atoms purported to interact. The distribution of distances was compared between wild-type and mutated enzymes to determine whether the mutations had a structural impact. The computational module allows visualization of otherwise somewhat abstract concepts, provides experience in use of computational tools, and emphasizes the value of application of physical methods to biological questions; it can therefore serve as a valuable addition to an in vivo mutagenesis experiment to investigate protein function. In addition, the module can also serve as a stand-alone experiment, offering visualization of three-dimensional structure, Brownian and Newtonian motion, energy comparisons, distributional analysis, and many other physics concepts.
The Module
Technical Requirements The laboratory activities were designed to fit in a 3-h lab period with an additional 1-h period for analysis. Students worked in pairs, or occasionally, groups of three. We anticipate that working individually or in larger groups would not affect the outcomes. All activities were performed using Dell TM laptops running WindowsV 7 with Intel Core i5 processors. We have also had students run simulations using UbuntuV R
R
Chiang et al.
R
(versions 9, 10, 11, and 12) Linux and WindowsV XP on single processor desktops with success and no issues with computational time. Furthermore, most modeling software can be run on a centralized server if one is available. This would expose students to secure shell clients (SSH) and file transfer protocols (SFTP) while using institutional or personal computers. The modeling software we used is NAMD (Not [just] Another Molecular Dynamics program) [6]. We chose NAMD because it is free, multiplatform, simple to install, has extensive documentation, and has a graphical user interface (GUI), VMD, from which all the modeling and analysis is possible. One point of emphasis in this exercise is that VMD allows users to go from start to finish without extensive programming or command line interfacing. However, VMD and NAMD also offer experienced users a great amount of flexibility and opportunity for customization. VMD has been used for its visualization and alignment capabilities previously [7]. However, all the activities we describe could be performed with other modeling packages [8–11] and viewers [12–14] if there are reasons to use other software packages. For example, the use of gromacs for similar modeling analysis of structural homologies has been reported [15]; however, this report did not include related wet lab work for tying simulation to experiment. The activities described in our report and its supplemental information assume no previous experience with NAMD, VMD, or other modeling=visualization software. We offer a step-by-step approach to any science student or faculty member interested in becoming acquainted with molecular modeling and analysis. It should be noted that this exercise may require 1–3 days of instructor preparation depending on the level of previous experience. Potential quantifiable, experimentally driven questions that could be addressed using our approach extend far beyond our application presented here. All techniques described are easily expandable through the tcl scripting available in the VMD interface with many scripts available online [16]. Given 1) the cross-discipline integration that has become the norm within the biological sciences, 2) the importance of visualization in understanding biochemical and molecular biology concepts [17, 18], and 3) the importance of using multiple approaches to obtain reliable understanding of biological systems in a research setting, we believe that the combination of approaches described here provides students with a particularly rich and valuable classroom-based research experience.
Software Installation The software should be installed as described on each machine to be used. VMD is installed through a typical Windows installer. By default, it will be located in the directory C:nProgram FilesnUniversity of IllinoisnVMD. It will also appear in the Start menu programs. NAMD extracts as a folder with the primary executable,
403
Biochemistry and Molecular Biology Education namd2.exe. There are other executables, dynamic-link libraries (.dll), and a folder of libraries for performing various types of molecular dynamics. The entire NAMD folder can be placed in the University of Illinois subdirectory. NAMD runs via a command prompt or via VMD. The system PATH variable can be edited to make the namd2.exe executable from a command prompt or from within VMD without typing the full path to namd2.exe. To do this in WindowsV 7, right-click on “Computer” and select “Properties”; click “Advanced system settings”; click “Environment variables…”; Select “PATH” and click “Edit.” At the end of the variable value, add the path to the namd2.exe, which should be similar to “;C:nProgram FilesnUniversity of IllinoisnNAMD_2.8_Win32-multicoren.” The preceding semicolon is necessary for separating a previous path in the variable. Earlier versions of WindowsV have a slightly different method, and Linux and Mac OS should update the PATH variable during the installation process. R
R
Setup and Simulation The procedure the students should follow is detailed in the companion Powerpoint presentation supplemental material, slides from which will be cited hereafter [19]. As a first step, students download a structure of the protein of interest. Protein crystal and NMR structures are available for free download from the Research Collaboratory for Structural Bioinformatics (RCSB) Protein Data Bank (PDB) at the website www.pdb.org as well as from the Molecular Modeling Database (MMDB) maintained by the NCBI at www.ncbi.nlm.nih.gov [20]. In our project, we used data for the CK1 casein kinase 1 (CK1) protein kinase (pdb file 1CSN) from the fission yeast Schizosaccharomyces pombe. These data serve as a model for the enzyme used in our wet lab experiments, the budding yeast CK1 protein kinase Yck2, for which NMR and crystal structures are unavailable [21]. The e-value for the BLASTP alignment of CK1 with Yck2 is 1E-150, and all the residues the students focused on were conserved between these two orthologs. They differ in total amino acid length, and the residues of interest in 1CSN are shifted by 64 amino acids as Yck2 is smaller. Here, we refer to the 1CSN residue numbers as they are the ones relevant to the molecular modeling. Approximately 15 min were spent presenting the use of www.pdb.org as a structure database and demonstrating both its online tools for viewing and its links to literature corresponding to the structures. Next, students were introduced to the anatomy of the downloaded pdb file in a text editor such as Word or Wordpad (slides 4–12). The 1CSN structure was opened in VMD. About 1 h was spent learning to use VMD and some of its capabilities (slides 13–27). We created various graphical representations using “Drawing Methods” (CPK, ribbons, cartoon, etc.), “Coloring Methods” (Name, Type, ColorID, etc.), and “Selected Atoms” (all, protein, residue, resname, etc.). This exercise allowed students to become familiar with the three
404
FIG 1
The crystal structure of 1CSN.pdb shown with ATP, Mg, and SO4 included. The red region is the highly conserved activation loop. Two conserved loop anchor amino acids, serine 179 and arginine 130, are shown with a distance of 8.35 Å between their C-alpha atoms [21].
dimensional structure, relating this structure to both the primary sequence and the functions of the protein, and determining the locations of highly conserved elements, such as the ATP-binding pocket. If desired, it also would be possible at this point to label and measure the distance between a set of atoms. Figure 1 shows a VMD-generated representation of CK1 with the activation loop colored red and two conserved amino acids on the loop labeled. The distance between these two residues is indicated. Our major goals for these activities were to introduce students to the physical approach they would be using with added visualization of their mutation, the visualization offering insight as to how the protein structure might shift. Initial student responses suggest that these goals were met. The crystal structure of CK1 includes water molecules, sulfate ions, a magnesium atom, and the ATP molecule. All these non-protein molecules should be removed for performing the most basic simulations (slides 28–33). They can be dealt with in NAMD but would require a more advanced usage of the software. A protein-only structure can be saved using the TkConsole found in the Extensions menu. Commands entered in TkConsole for this exercise will be shown in the courier font and should be entered exactly as written. The commands for saving a protein-only structure are set foo [atomselect top protein] $foo writepdb filename.pdb
Molecular Mechanics and Dynamics Characterization
The first command creates and sets the variable foo (foo is a common name used in computer programming for a variable that will only be used momentarily). Foo stores the pdb file data. Atomselect is a command for selecting all atoms meeting a specification, in this case, the protein without ATP, Mg, SO4, and water. Top makes the selection from the topmost structure, indicated by the “T” on the VMD Main window. Assuming only one structure is open in VMD, top will use that structure. If more than one structure is open, care must be exercised to select the structure of interest. Closing structures in VMD is detailed on slide 33. The structures in VMD are given ID numbers that can be seen in the VMD Main window. In the second command, the $ in front of foo designates the data stored in the variable foo, and writepdb is a command for creating a new pdb file with the file name given. The desired protein structure data must be stored in a protein structure file (.psf) to inform NAMD about charges, bonds, angles, and so forth. The psf creation process also will add hydrogen atoms to the structure if not already present. The autopsf generator can be used to generate these files (slides 34–40). Manual generation of pdb and psf files is possible if necessary or desired (see the NAMD tutorial [22]). The procedure for using the autopsf generator and verifying the output is detailed in the supplemental PowerPoint [19]. The structure is now ready for mutation. NAMD has a built-in mutator extension that works well for the 20 standard amino acids. Patches in supplemental topology files can be used to make changes to non-standard amino acids or phosphorylated natural amino acids [19]. The mutation process will create new pdb and psf files (slides 41–44). The final preparation before simulation is to solvate the protein in water (slides 45–47). This is not strictly necessary, since simulations may be performed in vacuum with much faster computation. However, the results may reflect the less realistic approach as solvation in water puts the protein in a more realistic environment. For solvation, either a water box or a water sphere may be used. Solvating in a water box is a simpler process because it can be performed using the VMD user interface. The disadvantage of a water box is that long simulations could result in the water solvation adopting a spherical shape due to the surface tension of water. Solvating in a water sphere requires running a script. The script for solvating in a water sphere is available with the NAMD tutorials and also with our supplemental information. Scripts are run in VMD through a built-in console (TkConsole). In either solvation case, boundary conditions must be used during simulation to ensure that NAMD appropriately handles water molecules at the outer boundaries, for example, reflective or periodic boundaries and the size of the water-bound structure. Reflective boundaries must be used with a spherical solvation and cause solvent molecules to reflect from the outer
Chiang et al.
boundaries. This requires an external force to be applied to the system for the reflection to occur. This force can lead to unreal results if too large or too small. Furthermore, spherically solvated systems must be simulated through manual scripting because the GUI is not currently capable of handling spherical systems. Periodic boundary conditions for a water box allow water molecules to exit one side of the box, which then enter the opposite side. This eliminates possible issues with the boundary forces. In this exercise, we used a built-in extension for solvating in a water box. We used the molecular dimensions of our protein to create a water box with a 2.4 Å layer of water extending beyond the outermost regions of the protein. This corresponds to a single layer of water in these regions. A larger box can be created; however, the computational time increases very quickly when including water molecules that are freer to diffuse than the protein. New pdb and psf files are created by the solvate protocol. These will be automatically loaded into VMD. For simplicity, previously loaded structures should be deleted. The boundaries, center, and size of the water box structure must be defined for the boundary conditions input of the simulation (slides 48 and 49). The following two commands will provide the necessary information. set everyone [atomselect top all] measure minmax $everyone The first command selects everything in the structure and stores the data in the variable everyone. The second command returns the minimum and maximum coordinates in the xyz directions for the entire solvated structure. These can then be used to calculate the center and total size in each dimension. Crystal structures are obtained at very low temperatures that may result in somewhat different structures than those that exist at physiological temperatures. Introducing mutations and solvent further adds features that may not be structurally accurate. Thus, energy minimization and equilibration processes should be performed for all structures before beginning molecular dynamics simulations. A proper minimization and equilibration will slowly ramp the temperature from 0 K to the simulation temperature so as not to cause artifacts in the protein structure or “bubbles” in the water. This is not easily performed through VMD without scripting one’s own configuration files. The scripting for NAMD is not terribly complex. With more time or a course dedicated to teaching molecular dynamics simulations, some scripting could easily be taught as has been done with gromacs [15]. To avoid students getting lost in scripting details that we did not feel were crucial to the exercise at this introductory level, we simply performed minimization and equilibration at the simulation temperature of 310 K using the NAMDgui (an interface for NAMD via VMD) available in the Extensions menu (slides 50–61). The
405
Biochemistry and Molecular Biology Education
FIG 2
Total energy of the solvated CK1 structure as a function of energy minimization step. The asymptotic behavior shows that energy minimization occurred within 5,000 steps. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
minimization step offers a place to discuss forces and potential energy, which most students will have encountered in introductory physics. The simulations in this exercise used NVT (constant number, volume, and temperature) ensemble thermodynamics, because this closely mimics wet lab experimental conditions and allows for the specification of the water box boundaries. We performed a 10,000 step minimization. The results of the minimization were plotted to obtain a graphical representation of energy minimization, which can help to determine if the chosen number of steps is adequate for the user’s system of interest. Shown in Fig. 2 is such a plot that shows that a fully minimized structure of CK1 was obtained within 5,000 steps, as indicated by the asymptotic behavior of the total energy over time. The data for this graph was obtained as follows: the namdstats.tcl script was downloaded from the supplemental information and saved to the working directory of simulations. In TkConsole, the directory was changed to the working directory and the following commands were used. source namdstats.tcl data_time TOTAL filename.out first last The first command brings in the available analysis functions from namdstats.tcl. The second command searches the minimization output for the total energy at each time step for the range of times steps first to last. Ensure at this step that students use the appropriate file name. For example, csn_r130a_wb_eq.log is the output log from our R130A mutant. A file will be created called total.dat. This file has two columns of data, corresponding to time step and total energy in kcal=mol. The data may be plotted in Excel or other analysis software of choice.
406
We then determined whether structural variations around low energy structures of wild-type and variant proteins differ using molecular dynamics. Molecular dynamics allows the protein to structurally fluctuate around the lowest energy structure. The results of a molecular dynamics simulation can be used to generate a distribution of structures fluctuating around this energy minimum. As an example, we asked students to compute the distance between a conserved phosphorylatable amino acid and a putative interacting site that is positively charged. The measurements were compared between wild-type protein and variants with amino acid substitutions at the conserved phosphorylatable residue that mimic or inhibit phosphorylation, for example, Serine (S) to Alanine (A) and Glutamate (D). For a more global picture of structural change, one could also calculate the RMS distance that a portion of the protein backbone moves. VMD has an interface to perform this calculation, which may make it easier for students to perform. To avoid loading the minimized structure files into VMD, we used the NAMDgui, setting the molecular dynamics to begin from the minimization output structural coordinates (slides 62–65). The minimization output is a .coor or restart.coor file. There is also a restart.xsc file, which should contain the boundary conditions. The default time step size is 1 fs. Larger time steps may be used to reduce computation time; however, accuracy of the results is reduced with larger step size. The number of steps corresponds to a total number of femtoseconds. Our students ran molecular dynamics for 1 ns (1,000,000 steps at 1 fs per step). The length of our simulation was chosen to provide as long a simulation time as possible during a 3-h lab, while allowing time to both set up the simulation and analyze the results.
Data Analysis A script for calculating the distance between the centers of mass of two selected positions is included in the supplemental materials. This tcl file can be sourced into the TkConsole. The calculation is performed using the command: source distance.tcl distance seltext1 seltext2 N_d f_r_out f_d_out where seltext1 and seltext2 are the two selected positions (here, residues 179 and 130, respectively). N_d is the number of bins to use for calculating a histogram of the distances. f_r_out is a file to be created that stores the distances calculated at each time step. f_d_out is a file to be created that stores the distance histogram. An example histogram is shown in Fig. 3. One conclusion to be drawn from this figure is that the distances between the phosphorylation site and its potential interaction partner Arginine in the wild-type enzyme fluctuate across a wide range, whereas the distance in the charged Glutamate mutant is confined to a smaller separation, presumably due to the charge interaction with the arginine. These results support the
Molecular Mechanics and Dynamics Characterization
FIG 3
Histogram of the distance between the ATP binding pocket center of mass and the center of mass of the mutated amino acid (S179) in CK1. Histograms for the wild-type and mutant, S179D, enzymes are shown. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
hypothesis that phosphorylation at Ser179 plays a role in CK1 deactivation by limiting structural freedom. More procedural details of this analysis are given on slides 66–70. Students used the computational results in conjunction with results from their experiments in yeast to support or refute their hypotheses. This semester, students investigated the hypothesis that a positively charged pocket on the surface of CK1 protein kinases (termed the RD pocket) is part of the reason for the reduced activity seen when phosphorylation at a particular site (Ser243 in Yck2) was mimicked by substitution of the negatively charged amino acid aspartate (S243D). Students were able to discuss that the selected residues in the S179D mutant CK1 protein exhibited a smaller range of motion in the computer simulations than did these residues in the wild-type enzyme. Furthermore, they were able to consider the effect of a mutation within the RD pocket, by considering both the different ranges of motion exhibited in the computer simulations and the viability of yeast expressing the mutant protein. Two examples of analyses in students’ final lab reports are reproduced here: “Though the data concerning mYck2:K240Q1S243D do not support our hypothesis, they are consistent across assays. We predicted that partially neutralizing the RD pocket should return the S243D mutant, which mimics constitutive phosphorylation and has reduced activity, to a wild-type activity level. Both the drop growth test and the microscopy assay, however, counter this prediction. The mYck2:K240Q1S243D mutant instead exhibits an activity level lower than mYck2:S243D (see Figure 9). Similarly, the microscopy data reveal that the number of budding mutants is increased in the mYck2:K240Q1S243D double
Chiang et al.
mutant relative to WT-Yck2 and mYck2:S243D (see Figure 10). According to our histogram data, the mYck2:K240Q1S243D double mutant, favors two energy minima—one in which the ATP binding pocket is 10.4 Å wide and another conformation, which occurs more than twice as often, in which the ATP binding pocket is 11.9 Å wide (see figure 6). Though these modeling data suggest that mYck2:K240Q1S243D should more often favor the relatively-accessible conformation, our growth test and microscopy data suggest that it instead favors the more restricted conformation, thus conferring reduced activity.” “Based on data from Figure 10, we can conclude that wild-type Yck2 ATP binding pocket has three stable conformations. The stable conformations fluctuate between the measurements represented by the highest points in the histogram which are 11.3, 11.9, and 12.6 A . From data in Figure 11, we can conclude the mutant S243D Yck2 has only one favored stable conformation of around 11.8 A . This demonstrates that the mutant causes a smaller, more constrained ATP binding pocket supporting our hypothesis of decreased activity in this mutant. In Figure 12, we can see the ATP binding pocket in mutant K223Q Yck2 has only one stable conformation of 10.4 A . This demonstrates that when the RD pocket is neutral the ATP binding pocket only has one stable conformation compared to the three seen in the wild-type Yck2 which has a positively charged RD pocket. Figure 12 indicates that the ATP binding pocket possibly collapses on itself when the RD pocket is neutrally charged because of the smaller distance across the ATP binding pocket. We can predict that this mutant will show no activity during phosphorylation because the ATP binding pocket has possibly collapsed. Figure 13 shows the double mutant K223Q=S243D ATP binding pocket which has one stable conformation fluctuating around 12.5 A . This corresponds to the conformation in the wild-type ATP binding pocket (Figure 10) at approximately 12.5 A . The double mutant conformation means that the RD pocket is neutralized by the substitution of lysine (K223Q) and that phosphorylation is being mimicked with the S243D. This double mutant conformation suggests that the neutral RD pocket makes the stable conformation of ATP binding pocket slightly larger and less restricted then when seen in the S243D mutant. Thus if the double K223Q mutant in conjunction with the S243D mutant has a larger site for ATP to bind, this supports our hypothesis that we will see more activity in this mutant compared to the S243D mutant alone.” Collectively, the final lab reports illustrated that the module illustrated the utility of combining multiple approaches to answer a question. It may also have promoted students’ visualization of the physical events they were investigating, although this internal event is harder to measure using the blunt tool of the lab report. In addition, students indicated some enthusiasm for the module. Representative student responses to a question about the project include:
407
Biochemistry and Molecular Biology Education “The computer analyses were great for understanding…just how complex proteins are. For me, I often find a disconnect between mixing the chemicals and remembering what they do…but watching and understanding it in the computer really helped me grasp the concept of [our experiment].” “Finding the actual energy minima of the structure and being able to quantify the physical parameters of [the protein] allowed us to form new hypotheses for why our genetics assays resulted how they did, and enabled us to make better decisions on what further research could be done. The genetic assays could stand alone as tools for our inquiry, but would be much less guided without the physical modeling.” “The modeling data allowed us to connect large-scale phenotypic effects of our mutations with the molecularlevel effects on protein structure (and presumably efficiency of function). Neither approach alone would have allowed us to draw conclusions about the relationship of [mutated] Yck2 structure and function.”
Summary In summary, this module is an effective tool for demonstrating the utility of multiple approaches to answer scientific questions. Students report having a stronger understanding of the structural implications arising from amino acid changes. Visualization by molecular simulation is widely recognized as an important tool for clarifying events at the molecular level (see, e.g. Jenkinson and McGill [17], and Tibell and Rundgren [18]). In fact, in its 2012 report, the National Research Council recognized a vital role for visualization tools and spatial thinking in developing students’ understanding of concepts (NRC 2012) [23]. The exercises presented here build on previous work implementing molecular visualization and simulation. The activities are modular to provide some flexibility to how they are implemented (e.g. studio formatted courses or more traditional 3-h laboratory courses). Simple approaches to analysis have been presented that are amenable to all levels of biology, chemistry, and physics; however, there are many ways to make the exercises more advanced or to incorporate them into independent research projects. Interactive molecular dynamics, steered molecular dynamics, and replica exchange modeling are three advanced techniques that are reasonably straightforward to perform using NAMD.
Acknowledgements The authors wish to thank Centenary College of Louisiana for support through Student-Faculty Summer Research Awards and the Gus S. Wortham Endowed Chair of Engineering.
References [1] National Research Council (2009) A New Biology for the 21st Century, National Academies Press, Washington, D.C.
408
[2] American Association of Medical Colleges, H. H. M. I. (2009) Scientific Foundations for Future Physicians, Washington, D.C. https:==www. aamc.org=download=271072=data=scientificfoundationsforfuturephysicians. pdf. [3] American Association for the Advancement of Science and the National Science Foundation (2011) Vision and Change in Undergraduate Biology Education: A Call to Action, American Association for the Advancement of Science and the National Science Foundation. www.visionandchange.org. [4] Brame, C. J., Pruitt, W. M., and Robinson, L. C. (2008) A molecular genetics laboratory course applying bioinformatics and cell biology in the context of original research. CBE Life Sci. Educ. 7, 410–421. [5] Hogue, C. W. (1997) Cn3D: A new generation of three-dimensional molecular structure viewer. Trends Biochem. Sci. 22, 314–316. [6] Phillips, J. C., Braun, R., Wang, W., Gumbart, J., Tajkhorshid, E., Villa, , L., and Schulten, K. (2005) Scalable E., Chipot, C., Skeel, R. D., Kale molecular dynamics with NAMD. J. Comp. Chem. 26, 1781–1802. [7] Floriano, W. B. (2008) A portable bioinformatics course for upperdivision undergraduate curriculum in sciences. Biochem. Mol. Biol. Educ. 36, 325–335. [8] Case, D. A. (2012) AMBER 12. www.ambermd.org. [9] Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S., and Karplus, M. (1983) CHARMM: A program for macromolecular energy, minimization, and dynamics calculations. J. Comput. Chem. 4, 187–217. [10] Lindahl, E., Hess, B., and van der Spoel, D. (2001) GROMACS 3.0: A package for molecular simulation and trajectory analysis. J. Mol. Model. 7, 306– 317. [11] Arnold, K., Bordoli, L., Kopp, J., and Schwede, T. (2006) The SWISSMODEL workspace: A web-based environment for protein structure homology modelling. Bioinformatics 22, 195–201. [12] Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch G. S., Greenblatt, D. M., Meng, E. C., and Ferrin, T. E. (2004) UCSF Chimera—A visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612. [13] DeLano, W. L. (2002) The PyMol Molecular Graphics System, Schrodinger, LLC. www.pymol.org. [14] Guex, N. and Peitsch, M. C. (1997) SWISS-MODEL and the SwissPdbViewer: An environment for comparative protein modeling. Electrophoresis 18, 2714–2723. [15] Elmore, D. E., Guayasamin, R. C., and Kieffer, M. E. (2010) A series of molecular dynamics and homology modeling computer labs for an undergraduate molecular modeling course. Biochem. Mol. Biol. Educ. 38, 216–223. [16] Schulten, K. VMD online script library. VMD Script Library at http:==www.ks.uiuc.edu=Research=vmd=script_library=. [17] Jenkinson, J. and McGill, G. (2012) Visualizing protein interactions and dynamics: Evolving a visual language for molecular animation. CBE Life Sci. Educ. 11, 103–110. [18] Tibell, L. A. and Rundgren, C. (2010) Educational challenges of molecular life sciences: Characteristics and implications for education and research. CBE Life Sci. Educ. 9, 25–33. [19] Cynthia, J. Brame and Troy, C. Messina molecular modeling of CK1. Supplemental Materials for Molecular Modeling at http:==www.centenary.edu=physics=tmessina=NAMD-Lab. [20] Madej, T., Addess, K. J., Fong, J. H., Geer, L. Y., Geer, R. C., Lanczycki, C. J., Liu, C., Lu, S., Marchler-Bauer, A., Panchenko, A. R., Chen, J., Thiessen, P. A., Wang, Y., Zhang, D., and Bryant, S. H. (2012) MMDB: 3D structures and macromolecular interactions. Nucleic Acids Res. 40, D461–D464. [21] Xu, R. M., Carmel, G., Sweet, R. M., Kuret, J., and Cheng, X. (1995) Crystal structure of casein kinase-1, a phosphate-directed protein kinase. EMBO J. 14, 1015–1023. [22] Schulten, K. NAMD Tutorials. NAMD Tutorials at http:==www.ks. uiuc.edu=Training=Tutorials=. [23] National Research Council (2012) Discipline-Based Education Research: Understanding and Improving Learning in Undergraduate Science and Engineering, National Academies Press, Washington, D.C.
Molecular Mechanics and Dynamics Characterization
