A Heart Failure Knowledgebase Combining Experimental Data with Tools for Integrative Biological Modeling Raimond L. Winslow1, Patrick Helm1, William Baumgartner Jr.1, Paul Delmar1, Tilak Ratnanather2, Srinivas Peddi2, Elliot McVeigh3 and Michael I.

1
A Heart Failure Knowledgebase Combining
Experimental Data with Tools for Integrative
Biological Modeling Raimond L. Winslow1,
Patrick Helm1, William Baumgartner Jr.1, Paul
Delmar1, Tilak Ratnanather2, Srinivas Peddi2,
Elliot McVeigh3 and Michael I. Miller2 The
Whitaker Biomedical Engineering InstituteCenter
for Computational Medicine and Biology1 Center
for Imaging Sciences2, Johns Hopkins
UniversityThe NIH Laboratory of Cardiac
Energetics Medical Imaging Section3
Institute URL http//www.wbmei.jhu.eduCCMB
URL http//www.ccmb.jhu.eduCIS
URL http//www.cis.jhu.edu NIH-LCE
URL http//zeus.nhlbi.nih.gov/
2
NIH Specialized Center of Research in Sudden
Cardiac Death(NIH P50 HL52307)
Goal To understand the molecular basis of sudden
cardiac death in human heart failure
Experiments (Human and Canine)
Channel/ Transporter Function
Cell Electro- physiology
Gene Expression
Ventricular Remodeling
Ventricular Conduction
RPA RT- PCR Microarrays Protein Assays
Histological Analyses MR Diffusion Imaging
Recombinant Channels Somatic Gene Transfer
Ca2 V NADH, FADH, Vmito, Ca2mito
Electrode Arrays MAPs
Modeling Data Analysis
3
Topics

• Genomics of Heart Failure
• Modeling the Cellular Phenotype of Heart Failure
• Structural Remodeling in Heart Failure

4
Genomics of Human Heart Failure

• Goals
• To develop accurate statistical models of gene
  expression intensity in normal and failing human
  ventricular myocardium
• Use these models to identify changes in gene
  expression characteristic of end-stage human
  heart failure

5
Genomics of Human Heart FailurePatient
Demographics
NF1 Subarachnoid Hemorrage 34 M
NF2 Subarachnoid Hemorrage 57 F
NF3 Blunt Head Trauma 25 M
NF4 Unknown 47 M
Non-Failing Subjects
IDCM1 Amiodarone, Digoxin, Enalapril, Furosemide 34 M
IDCM2 Butemanide, Clonidine, Digoxin, Lisinopril, Norvasc, topical Nitroglycerine 57 F
IDCM3 Aldactone, butemanide, demodex,Omeprazole, Aspirin 25 M
Failing Subjects Idiopathic Dilated CardioMyopathy
6
Genomics of Human Heart Failure Hybridization
and Data Analysis

• Rosetta Flexjet Oligonucleotide Arrays
• 24,466 probes (60 mers) with 10,892 in Locus
  Link
• 68 hybridizations

• Maximum-Likelihood Estimation of gene-specific
  model parameters

7
Human Cardiac Gene Expression Knowledgebase
(CaGE)http//www.cage.wbmei.jhu.edu
8
Genomics of Human Heart FailureResults
9
Genomics of Human Heart FailureResults (cont.)
Up-Regulation
Down-Regulation

• Ion Channels
• inward-rectifying K channel
• Kv4.3 component of Ito1
• Ca2 Cycling
• serca2
• phospholamban
• Angiogenesis
• Ephrin-A1
• Cysteine-rich, angiogenic inducer, 61
• Cell Structure/Extracellular Matrix
• Spectrin, beta, non-erythrocytic 1
• Human extracellular matrix protein 1

• Transcription/Translation
• cAMP Response Element Binding Protein-1
• 11 ribosomal proteins
• Cell Structure
• Alpha L integrin
• Sarcoglycan, alpha DAG
• Contractile Proteins
• Tropomodulin
• Tropomyosin
• Myosin light chain
• Membrane Receptors
• Human G-protein coupled receptor
• G-protein beta 5 subunit
• Ca2 Cycling
• NaCa Exchanger

10
Genomics of Human Heart FailureNext Steps
Next Steps

• Increase of patient samples (10 normals, 40
  failing)
• Perform additional hybridizations using
  Affymetrix HG-U95Av2 arrays
• Time-series analyses of changes in gene
  expression during development of heart failure
  (canine tachycardia pacing model)
• Add data access/analysis capabilities to CaGE

11
Cellular Phenotype of End-Stage Heart Failure

• Goals
• To develop accurate, predictive computational
  models of normal and failing canine and human
  ventricular myocytes
• To relate changes in gene expression measured in
  end-stage human heart failure to mechanisms of
  arrhythmia using a combination of experimentation
  and modeling

12
Cellular Phenotype of End-Stage Heart FailureA
Minimal Model of Heart Failure
Normal, KCND3 66, KCNJ2 32, ATP2A2 62,
NCX1 75
Model
Model
ATP2A2 Down-Regulation
20
30
70
Experiment
Model
Winslow et al (1999). Circ. Res. 84 571
13
Cellular Phenotype of End-Stage Heart Failure
Results Next Steps
Major Results

• Altered expression of genes ATP2A2 and NCX1
  encoding Ca2 cycling proteins is primary
  contributor to AP prolongation in HF
• Restoration of normal JSR Ca2 levels may reduce
  risk of arrhythmia in HF patients
• Ca2-mediated inactivation of ICa,L may be the
  primary mode of inactivation during the AP

Next Steps

• Develop ventricular myocyte model describing
  local-control of Ca2 release
• Use model to understand the functional
  consequences of altered ICa,L b-subunit
  expression, RyR phosphorylation/dephosphorylation,
  diadic microstructure in HF
• Add new subsystems to myocyte model based on
  microarray results
• Cortassa et al (2001). An integrated model of
  cardiac mitochondrial energy metabolism and
  calcium dynamics.

14
Structural Remodeling in End-Stage Heart Failure

• Goals
• To develop rapid, accurate means for
  reconstructing cardiac geometry and fiber
  organization
• Using these techniques, reconstruct populations
  of normal and failing hearts (canine and human)
• Develop a transformation-based probabilistic
  atlas describing variation of ventricular
  geometry and fiber organization in normal human
  and canine ventricles
• Identify statistically significant changes in the
  anatomical structure of normal versus failing
  hearts
• Test the functional significance of these changes
  on cardiac conduction using both models and
  experimentation

15
Structural Remodeling in End-Stage Heart Failure
Imaging Heart Geometry and Fiber Structure
DTMRI Fiber Angles In Cross Section
DTMRI vs HISTO Fiber Angles
Holmes, A. et al (2000). Magn. Res. Med., 44157
Scollan et al (2000). Ann. Biomed. Eng., 28(8)
934-944.
16
Structural Remodeling in End-Stage Heart
FailureDTMR Imaging Results (Canine Model)
Fiber Anisotropy
Fiber Inclination Angle
Normal Canine Heart
Failing Canine Heart
17
Structural Remodeling in End-Stage Heart
FailureLarge-Deformation Transformations for
Computational Anatomy
18
Structural Remodeling in End-Stage Heart
Failure3-D Volume Mapping Results
19
Structural Remodeling in End-Stage Heart
FailureFinite Element Models of Cardiac Anatomy

• As described in Nielsen et al AJP 260(4 Pt
  2)H1365-78
• User selects number of volume elements/nodes
• Matlab GUI for visual control of the fitting
  process
• All imaging datasets, FE models, and FEM software
  will be available at www.cmbl.jhu.edu

Epicardial Fibers FEM Model
Endocardial Fibers FEM Model
20
Structural Remodeling in End-Stage Heart Failure
Functional Impact of Remodeling

• Electrically mapped and DTMR imaged 4 normal and
  3 failing canine hearts
• 128-electrode sock array, 7mm electrode spacing
• Complete anatomical and electrical reconstruction
  performed on one normal canine heart

21
Generalized Anatomic Database Interface and
Analysis Platform
22
Structural Remodeling in End-Stage Heart Failure
Results Next Steps
Major Results

• DTMRI is a rapid, accurate method for
  reconstruction of cardiac ventricular geometry
  and fiber structure
• Methods of Computational Anatomy, applied
  previously in the Brain Mapping Project, offer an
  approach to quantitative determination of
  structural differences in normal versus diseased
  hearts
• Electrical mapping, DTMR imaging and
  computational modeling based on data collected in
  the same heart will likely provide insights into
  the possible structural bases of arrhythmia in
  heart failure

Next Steps

• Increase number of electrically mapped, imaged,
  reconstructed and nmodeled canine hearts
• Reconstruction and analysis of human normal and
  failing human hearts
• Use of CA to identify statistically significant
  structural changes in normal and failing canine
  and human hearts
• Simulation of possible structural basis of
  arrhythmia in HF

23
Acknowledgements
Modeling and Data Analysis
Experiment
Bill Baumgartner Paul Delmar Pat Helm Alex
Holmes Saleet Jafri Jeremy Rice David
Scollan Christina Yung Jiangyang Zhang
Mike Bober Mark Boguski Ion Hobai Eduardo
Marban Brad Nuss Brian ORourke Suzanne
Szak Gordon Tomaselli Jody White Dave Yue Mike
Ziman
Supported by the NIH, the Whitaker Foundation,
the Falk Medical Trust, Physiome Sciences,
Research Genetics Inc., Rosetta Inpharmatics and
IBM Corporation