Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: James Fonseca Project Name: dynamics Division: MCS Project title: Dynamics of biophysical systems. Associated funding: DOE Other Systems: MCS workstations, Beagle at Computation Institute Science: Determine key characteristics of ion selectivity of the Na channel using a reduced model combined with Monte Carlo simulations. Materials' Science, polymer physics, electrophysiology, molecular biology Project description: List of funding agencies that support this project. DOE Description of the science objective(s) for the project. Determine key characteristics of ion selectivity of the Na channel using a reduced model combined with Monte Carlo simulations. PROJECT SUMMARY This project focuses on a range of computational studies of electrolyte and electrolyte-protein solutions. We investigate the fundamental behavior of *confined* electrolytes outside of equilibrium, which is of paramount importance to the understanding of the functioning of many biological and physical systems. The primary mode of investigation is Monte-Carlo (equilibrium) simulations of protein-electrolyte mixtures (in the primitive or restricted primitive model of the electrolyte). This project scope is twofold. The first project continues work performed on LCRC systems over the last two years. The goal this year is to extend simulations of the reduced model of the sodium channel as described below. A second request is an extension of an existing project and is included with this allocation request. This project is similar to the first in that it focuses on reduced models of ion channels and the software core is the same. But the goal of these simulations will be provide verification of a recently developed method which uses an energetic variational approach applied to the primitive (implicit solvent) model of ionic solutions. This project is called PNP (Poisson-Nernst-Planck) simulations with steric effects and is discussed in a following section. PROJECT TEAM A new team member, Janhavi Giri, PhD, has been brought on to the project to help with manuscript preparation, simulation execution, and project direction. Janhavi performed her doctoral work with Prof. Eisenberg at Rush University and is experienced in both experimental and theoretical research of the Na and related ion channels. Janhavi joins the team with Gary Leaf, Jim Fonseca, and Bob Eisenberg. REDUCED MODEL OF THE SODIUM CHANNEL Overview Sodium vs. Potassium selectivity is one of the most important properties of sodium channels. Any failure in this selectivity means quick death. Small nerve terminals in the brain quickly swell and burst, doing irreversible damage to the nervous system stopping respiration and then a wide range of other functions. Potassium leakage from nerve terminals of the brain is the immediate cause of brain death in almost all cases. Indeed, a substantial fraction of all the ATP used in the body (e.g., some 30 to 50%) is spent pumping out potassium (and sodium) that leaks out of or into cells. There is hardly a more important physiological property than sodium vs. potassium selectivity. The system representative of our investigation is a membrane channel protein operating under a concentration and voltage gradient. The primary function of channel proteins is to pass ionic current in and out of the cell *selectively* (i.e., favoring some types of ions over others). This selectivity mechanism is crucial to the functioning of the cell and the overall organism, but it is still relatively poorly understood. In particular, since a channel can be sensitive to trace amounts of the selected ion, the permeation and selectivity mechanisms are subtly related to the atomic structure of the protein. At the same time, since the ionic current is largely independent of the thermal fluctuations of the protein (in the open configuration), the channel function is rather robust. Thus, there is a need to isolate the relevant features of the channel geometry and, most importantly, permanent charge distribution, that are responsible for this sensitive yet robust selective beha vior. Simple models of channels have been able to account for essentially all the selectivity properties of two types of calcium channels, with distinctly different functions, properties, and structures, and the classical voltage activated sodium channel DEKA (so-named for the amino acid residues implicated in ion selectivity as determined by electrophysiological and biochemical experiments). The DEKA channel selects sodium over calcium in a complex manner not yet understood. The DEKA channel excludes potassium from the channel in a number of conditions and thus does not conduct this dangerous ion at all. Past Work on LCRC systems We have nearly completed a survey of the Na channel model we employ to determine its behavior. The survey has included a wide variety of bath concentrations and model parameters such as pore radius (structural effects) and channel dielectric (polarization effects) constant. So far we have collected selectivity data on a two parameter grid for a variety of common mono and divalent cations at various physiological concentrations. Work thus far has included the following sets of simulations (most of which include dozens of simulations): 1. Size selectivity: Na+ vs. monovalent ions (Li+, K+, Rb+, Cs+) at various concentrations 2. Effect of pore radius on Na+ vs. K+ and Na+ vs. Li+ selectivity 3. Charge selectivity: Na+ vs. divalent ions (Mg2+, Ba2+, Sr2+) at various concentrations 4. Effect of relative difference in bulk ionic concentrations on Na+ vs. K+, Na+ vs. Li+ selectivity 5. Anomalous Mole Fraction Effect in Na channel: Na+ vs. K+, Na+ vs. Li+ Proposed Work This approach has provided several unexpected results that we wish to investigate further. They include a strong dependence of conductance dielectric constant, as well as a possible selectivity mechanism that may help explain the selectivity of Na and Li in the Na channel. Future work will focus on determining the robustness of this phenomenon under different conditions. We will also continue with planned studies to investigate the role of the model’s permanent structural ions. Although the model works robustly, the mechanism by which DEKA selects sodium over potassium is not known. The mechanism by which potassium is excluded is not known. We propose here to study the role of the following aspects of the model. We have estimated the number of simulations we believe will be required for each aspect. Each set of simulations will investigate the model under a set of conditions, such as different bath salt concentrations, including physiological scenarios as well bath concentrations to match to experiment. 1. Entropy and energy of the side chains DEKA (20 simulations) 2. The entropy and energy of the sodium ion (30 simulations) 3. The entropy and energy of the calcium ion (30 simulations) 4. The entropy and energy of the potassium ion (30 simulations) 5. Explore the selectivity of DEKA to lithium, rubidium and cesium, investigating the entropy and energy in interesting cases (30 simulations) 6. Study the effect of dielectric coefficient and diameter in cases (1)- (5) seeking the biological adaptation. (~200 simulations) It is clear from earlier work that dielectric coefficient and diameter have orthogonal actions on Na vs. K selectivity in DEKA. Diameter determines the RATIO of occupancy, i.e., selectivity, with virtually no effect on the amount of ions (which determines conductance) in the channel. Dielectric coefficient determines the actual OCCUPANCY with virtually no effect on the ratios. We are eager to see how the components of free energy of the system, listed above manage to make this happen. We are eager to see if other control systems are at play for other important ions listed above. PNP (POISSON-NERNST-PLANCK) SIMULATIONS WITH STERIC EFFECTS Overview The flow of current through an ionic channel is studied using the energetic variational approach, applied to the primitive (implicit solvent) model of ionic solutions. This approach allows the derivation of self-consistent (Euler Lagrange) equations to describe the flow of spheres through channels. The partial differential equations derived involve the global interactions of the spheres and are replaced here with a local approximation1, we call steric PNP (Poisson-Nernst-Planck). Kong combining rules are used and a range of values of steric interaction parameters are studied. These parameters change the energetics of steric interaction but have no effect on diffusion coefficients in model and simulations. Calculations are done for the calcium (EEEE, EEEA) and sodium channel (DEKA) previously studied in Monte Carlo simulations with comparable results. Biological function is quite sensitive to the steric interaction parameters and we speculate that a wide range of the function of channels and transporters, even enzymes, might depend on such terms. We point out that classical theories of channels, transporters, and enzymes depend on ideal representations of ionic solutions in which nothing interacts with nothing, even in the enormous concentrations found near and in these proteins, or near electrodes in electrochemical cells, for that matter. We suggest that a theory designed to handle interactions might be more appropriate. We show that one such theory is feasible and computable: steric PNP allows direct comparison with experiments measuring flows as well as equilibrium properties. Steric PNP combines atomic and macroscales in a computable formulation that allows calculation of the macroscopic effects of changes in atomic scale structures (size ≅ 10-10 meters) studied so extensively in channelology and molecular biology. Proposed Work Since the work is essentially a verification of an analytic approach (the results of which will be performed by other collaborators), there are many short simulations that will need to be performed. This parameter matrix will include filter radius, filter length, dielectric constants of the protein and the channel regions, boundary conditions in the form of applied voltages, and finally, bath concentration for various physiologically relevant salts. REQUEST FOR REDUCED MODEL OF THE SODIUM CHANNEL These calculations and similar less-intensive published results fit a large amount of important experimental data which is not fit by any other model. It should be noted that at these very low physiological concentrations, such computations, if based on molecular dynamics, would not be feasible. Nevertheless, even with this primitive model, the computations are extremely CPU intensive. The desired allocation of the Reduced Model of the Sodium Channel project is 90,000 core-hours per quarter. This allocation is slightly more than our current allocation. With a new team member and more experience running simulations, we feel we will be able to productively use the additional hours. Each simulation under a particular set of conditions requires in the neighborhood of 1,000 core-hours of computational time. We have found that the code and simulation setup and analysis is most efficient simulations are performed in parallel on 24 cores. With 340 proposed simulations, this request provides some leeway for issues like testing and post processing. Longer walltimes are required for advanced analysis and publication figures. A particular set of conditions means, for example, a particular concentration of NaCl and particular concentration of LiCl with given pore geometry and dielectric constant. REQUEST FOR PNP (POISSON-NERNST-PLANCK) SIMULATIONS WITH STERIC EFFECTS The desired allocation of the PNP (Poisson-Nernst-Planck) Simulations with Steric Effects project is 5,000 core-hours per quarter. Each simulation requires between 5 and 100 core-hours to perform. These jobs will run on single-cores and therefore will make up less than 6% of the total request. Related, previous, simulations have been performed on the Ohio Supercomputer Center and we would like to consolidate our work a single system for efficiency and ease of use. Project URL: http://www.phys.rush.edu/RSEisenberg/physioeis.html Current FY Hours Used: undetermined amount New FY Requested allocation: 380000 Q1: 95000 Q2: 95000 Q3: 95000 Q4: 95000 Justification: Thank You, The LCRC Accounts System