Multiscale Modeling in Biomedical Research
New approaches extend multiscale models to represent cellular mesoscales and bridge from molecular to cellular models
In an era of increasingly comprehensive molecular characterizations of living systems, computation has emerged as a key technology to facilitate integrative understanding of biological mechanisms. Computation can be integrative in biomedical science in several different ways. Perhaps the best recognized of these is the role of computing for information integration. This is a central goal of bioinformatics. A related but distinct type of integration is functional integration. Once bioinformatics has organized and annotated the molecular components of the biological system, we can build functional interaction networks whether genomic, transcriptional, signaling, metabolic or physiological. These reconstructed networks serve as a foundation for developing comprehensive functionally integrated systems models of the cell or living system. This kind of functional integration is at the core of systems biology. A third way that computation can integrate is structurally, across physical scales of biological organization, from molecule to organism. This type of computational biology is often now commonly referred to as multiscale modeling. Indeed the Inter-Agency Modeling and Analysis Group1 comprised of officers from nine federal agencies (plus the Canadian agency, MITACS) has formed the Multiscale Modeling Consortium (MSM) of over 100 funded investigators around this shared interest. Whereas systems biology modeling is very frequently data-driven and data-limited, multiscale modeling is more often physics-driven and compute-limited.
Of course these different types of integration are neither independent nor mutually exclusive. On the contrary they are interdependent. There is no single computational approach that can integrate comprehensively from molecule to organism, from genotype to phenotype or across interacting physiological subsystems. But generalizable paradigms are emerging that are making the prospect of models that span increasingly broad spatial and temporal scales of biology—blue sky pipedreams just a short time ago—seem much more feasible, at least for certain classes of problem.
Much of this progress has been driven by improvements in technology such as higher resolution three-dimensional measurements of biological structures made possible by improved microscopy and structural biology approaches. A recent example is the three-dimensional model of the cardiac myocyte calcium release unit shown below that was used to model the calcium “spark” that occurs when a single calcium release unit on the sarcoplasmic reticulum membrane opens.2 This model was made possible by improved electron tomography techniques3 and new methods for generating high-quality computational meshes.4 New probes for localizing receptors and macromolecular complexes within these microanatomic domains will further improve these models.5 Improved computational performance and novel algorithms are also allowing increasingly large-scale particle-based models to be implemented. The particle-based Monte Carlo modeling of vesicular release from pre-synaptic neurons by Nadkarni and colleagues6 is an exciting example. And improved algorithms together with higher resolution whole organ imaging are making possible large-scale organ models that can investigate the effects of fine spatial heterogeneities such as distributed scarring on clinical phenotypes such as cardiac arrhythmias7.
Another way to push the spatio-temporal boundaries of multiscale models is through emerging new strategies for course-graining molecular models and for bridging molecular models to cellular scales. Fedosov and colleagues8 recently demonstrated how coarse-grained techniques such as Dissipative Particle Dynamics can be used to model cell membrane dynamics and adhesive forces which are then used in multi-cellular models of red blood cell aggregation and in turn included in continuum models of whole blood non-Newtonian viscous properties. In 2009, Silva and Rudy9 introduced another strategy for bridging from all-atom molecular dynamics and molecular electrostatics simulations to whole cell models by using a Markov model of a delayed rectifier potassium channel as the intermediate and then exploring the mechanisms by which disease-causing mutations can affect clinical electrophysiological phenotypes.
I have focused on new approaches to extending multiscale models to represent details at cellular mesoscales and on bridging molecular to cellular models. But there are other opportunities too. The span of important temporal scales in biomedicine is even larger than that of the spatial scales. There is tremendous potential to extend multiscale models from the time-scale of physiological responses to those longer time-scales of growth, remodeling, and the natural history of disease and aging. The challenges will ensure an exciting and robust future for integrative multiscale modeling in biomedicine.
1 See http://www.imagwiki.nibib.nih.gov.
2 Hake J, Edwards AG, Yu Z, Kekenes-Huskey PM, Michailova AP, McCammon JA, Holst MJ, Hoshijima M, McCulloch AD. Modelling cardiac calcium sparks in a three-dimensional reconstruction of a calcium release unit. J Physiol (Lond). 2012 Sep 15;590(Pt 18):4403–22.
3 Hayashi T, Martone ME, Yu Z, Thor A, Doi M, Holst MJ, Ellisman MH, Hoshijima M. Three-dimensional electron microscopy reveals new details of membrane systems for Ca2+ signaling in the heart. J Cell Sci 2009 Apr 1;122(Pt 7):1005–13.
4 Yu Z, Holst MJ, McCammon JA. High-fidelity geometric modeling for biomedical applications. Finite Elem Anal Des 2008;44, 715–723.
5 Shu X, Lev-Ram V, Deerinck TJ, Qi Y, Ramko EB, Davidson MW, Jin Y, Ellisman MH, Tsien RY. A genetically encoded tag for correlated light and electron microscopy of intact cells, tissues, and organisms. PLoS Biol. 2011 Apr;9(4):e1001041.
6 Nadkarni S, Bartol TM, Sejnowski TJ, Levine H. Modelling vesicular release at hippocampal synapses. PLoS Comput Biol. 2010;6(11):e1000983.
7 Winslow RL, Trayanova N, Geman D, Miller MI. Computational medicine: translating models to clinical care. Sci Transl Med. 2012 Oct 31;4(158):158rv11.
8 Fedosov DA, Pan W, Caswell B, Gompper G, Karniadakis GE. Predicting human blood viscosity in silico. Proc Natl Acad Sci USA. 2011 Jul 19;108(29):11772–7.
9 Silva J, Pan H, Wu D, Nekouzadeh A, Decker K, Cui J, et al. A multiscale model linking ion-channel molecular dynamics and electrostatics to the cardiac action potential. Proc Natl Acad Sci USA 2009;106(27):11102.
Andrew McCulloch, PhD, is professor of bioengineering and Jacobs School Distinguished Scholar at the University of California San Diego, La Jolla, CA.