Multiple Scale Approaches for Complex Diseases


Extensive research in last a few decades provided large amount of data from animal and human studies related to many complex diseases. The data were collected at different length scales spanning genetics, protein interactions and structure, and cell, tissue and organ function. Most of the findings at any scale were directly connected to particular disease, neglecting processes at different scales that can significantly distort outcome at organ function. This partial approach led to expensive and frequently unsuccessful development of new drugs and therapies.

Recent development of computational hardware and resources provided huge suppository of collected heterogeneous data sources available to be used for more complete quantitative understanding of subject-specific progress of disease. To achieve this we are developing the support systems based on a complex integration of data from various sources combined with subject specific data. This complex integration can be achieved by recently developed or under development subject-specific computer models at multiple scales. Outcome of this system will provide essential information for decision support systems, more efficient and far less expensive new drug development and developing more effective new therapies and monitoring systems for evaluating progression or regression of disease for a prescribed therapy.

Although this approach is general we will focus here on heart disease. Specifically we have focused on a large number of mutations in sarcomeric proteins associated with human skeletal and cardiac myopathies. Cardiomyopathies are one of the most common causes of death among young adults. At the moment there is no cure and diagnosis and prediction of disease progression remains difficult. Yet one in 500 people carry a mutation associated with the disease. About 40% of these mutations occur in myosin and another 40% in myosin binding protein C and the rest are mostly in thin filament proteins troponin and tropomyosin. In myosin alone there are more than 300 known mutations associated with cardiomyopathies.

Multiple scale approach for integration of measurements at protein, subcellular, cellular, tissue and whole organ level, combined with state of the art measurement techniques provides strong informational data base for following evolution of cardiomyopathies from genotype to phenotype.

Despite the lack of understanding of the disease progression the evidence that the mutations are causally related to the disease is strong in many cases. There are many in vitro studies of the properties of proteins carrying the mutations and studies of muscle from trans-genetic mouse with the mutations. However, currently there is a large gap between understanding how the mutation alters the behavior of the protein and how that leads to the disease. Part of the complexity lies in the fact that the protein is present in the tissue from birth but in many cases there is no altered phenotype until adolescence of even later. Thus, the disease is seen as a result of the compensatory mechanism to deal with what may be a relatively minor problem in the protein leading to a long term degeneration of muscle performance.

To date there is no way to join up the in vitro studies of individual proteins carrying mutations and studies of intact systems - either from trans-genetic mice, human tissue or patient-specific data. Modelling provides a route to join up multiple studies including underlying genetics, protein structural and kinetics alternations, observed cell and tissue changes in disease to clinical outcomes. In order to achieve this we need to interlink:

  1. Bioinformatics to relate specific gene or genes to protein structural and functional changes and its link to modulated cell function and clinical markers of a specific disease. This part of the project will include data mining over accumulated personal medical data and data from relevant animal studies including genomic and epi-genomics in vivo and in vitro imaging data, and effect of administration of therapeutics and on nutrition/exposure to environmental factors.

  2. Protein and cell data. For simulations bridging modulated crossbridge kinetics and thin filament         regulation to fiber mechanics, caused by known mutations, it is necessary to collect subject specific data from left ventricle biopsies. From these biopsies the extracted key contractile proteins will be used for the experiments in solutions to provide kinetic data associated with myosin ATP-ase and thin filament regulation by calcium. Also from the biopsies, the extracted intact fibers will provide dynamic contractile characteristics of muscle cells and tissues (force-pCa relations and twitch transients). These data will be used in MUSICO simulations to provide the parameters for whole heart modelling by MUSICO meso-scale FE and Alya-red FE software.

  3. Models for linking data from multiple length scales. We have currently available at IIT the computational platform MUSICO (Muscle Simulation Code) that can link molecular interactions to the whole organ function. For example, MUSICO can predict the effect of a known mutation (defined from in vitro protein studies) on the intact system. It can also do the reverse, take the data from intact muscle fiber studies and predict what molecular properties of the protein components need to be altered to produce the phenotype. It can also predict where potential targets for drug therapy may be in the ensemble of protein.

  4. Imaging: Mesoscale (DSI), CT scans, MRI, ultrasound etc. For patient specific simulations this projects, currently under development, will also include mesh generation based on mesoscale DSI tracks for inclusion of fiber direction in FE elements.

  5. Finite element programs linking (2) and (3). These simulations will be adapted to subject specific data including genomics, protein structure and kinetics, in vitro cellular data, mesoscale architecture, and the ventricular anatomy to decipher the effect of the microscale structure on the overall organ function. These computer model simulations are designed to predict the personalized physiology, functional disorders and other diseases.

  6. Standardization of collecting and reporting/ saving in depositories of newly collected data. This is important for the smooth use of the specialized software available for this purpose.

  7. Clinical critical analysis of collected data and data analyzed by multiple computer programs and data from MUSICO and FE simulations.

Schematic view of our multi-scale DSS approach into the influence of sarcomere protein mutation on cardiomyopathy. Transferring information from Genotype to Phenotype

Linking data from nano- to micro-scale

Systems biology has introduced new paradigms in science by switching to a more integrative approach toward the study of complex systems. Over the past years, researchers have produced an extraordinary wealth of knowledge on human physiology. The aim of the lab research is integrating this knowledge to reveal the relationships between the different components and scales that form the balance in cardiac systems. In order to integrate this knowledge, a number of mathematical models and ICT (Information and Communication Technologies) are to be developed and used.

Unifying Concepts of the Generalized Sliding Filament Model (MUSICO)

Computational platform MUSICO (MUscle SImulation COde) for modelling realistic sarcomeric system has been developed with the aim to simulate a wide variety of experimental muscle behavior. The platform offers a modular program structure that allows extension and replacement of any part of sarcomeric system (calcium activation, cross-bridge cycle, sarcomere geometry, etc.). The current version of the MUSICO involves a number of sarcomere geometry models including the three-dimensional spatial models of multi-sarcomere geometry. Furthermore, multiple actomyosin cycle models and calcium regulatory models are also incorporated. Nonlinear mechanical behavior of extensible filaments and crossbridges is addressed using iterative finite element scheme. Moreover, in order to speed-up simulations, the platform is provided with parallelized computational algorithm.

MUSICO physiological model integrates the kinetics of actomyosin interactions and the action of regulatory proteins with extensible sliding filament models.

Molecular models of contractility in striated muscle require an integrated description of the action of myosin motors, firstly in the filament lattice of the half-sarcomere and then over the whole fiber. Existing models do not adequately reflect the biochemistry of the myosin motor and its sarcomeric environment. An integrated model is defined by applying knowledge of the actin-myosin-ATP cycle subject to the constraint that: (a) strain-dependent actin-myosin kinetics are derived from reaction-energy landscapes which are applied to dimeric myosin, (b) actin-myosin interactions in the half-sarcomere and whole-sarcomere are defined in the context of 3D sarcomere geometry with discretely-positioned heads on the myosin filament and target zones on actin filaments, and  (c) the myosin and actin filaments are treated as elastically extensible. We specifically develop two kinds of sliding filament models: (1) a 3-D comprehensive molecular model of a sarcomere contraction which includes explicit position of each myosin head and corresponding site(s) and its association with thin filament regulatory proteins; (2) probabilistic models which include most of the essential features of the 3-D molecular model and are computationally more feasible for multiscale formulations. The development of these models is supported by molecular dynamic (MD) simulations and molecular experiments. We demonstrated in several preliminary studies the necessity for the revisions of currently accepted models based on fundamental understanding of intra- and inter-molecular interactions.  The novel approach provides a method for studying the effect of contractile protein mutations on mechanochemistry, mechanotransduction, and organ dysfunction and thus will constitute a tool to study physiologically relevant disease models.