Computational design of custom therapeutic cells to correct failing human cardiomyocytes

Figures

• 

  FIGURE 1. Genetic algorithm-based approach to identify custom cells expressing non-excitable cell channels are unable to restore failing cardiomyocyte phenotype. Initial populations of 2500 custom cell models were generated for heterocellular coupling with non-fibrotic (A) and fibrotic (B) human heart failure (HF) cardiomyocytes with pseudo-random perturbations of non-excitable cell ion channel maximal conductivities, gap junction conductances, and number of coupled cells. Custom cell populations underwent 5 generations of genetic algorithm evolution. Custom cells were accepted if root mean square error was less than 50% relative to untreated HF cardiomyocytes for both action potential and calcium transient waveforms; in this set of simulations, all models were outside this range and were therefore rejected (red). Waveforms for healthy human cardiomyocytes (hCMs) and untreated HF cardiomyocytes are indicated by dashed and dotted lines, respectively.

• 

  FIGURE 2. Custom cells expressing excitable cell channels restore a healthy phenotype in fibrotic failing cardiomyocytes. Initial populations of 2500 custom cell models were generated for heterocellular coupling with non-fibrotic (A) and fibrotic (B) heart failure (HF) human cardiomyocytes (hCMs) with pseudo-random perturbations of gap junction conductances, number of coupled cells, and both cardiomyocyte and non-excitable cell ion channel maximal conductivities. Custom cell populations underwent 5 generations of genetic algorithm evolution, whereby custom cells were accepted if root mean square error was less than 50% for both action potential and calcium transient waveforms (blue); models outside this range were rejected (red). Waveforms for healthy and untreated HF cardiomyocytes are indicated by dashed and dotted lines, respectively. (C) Distribution of accepted custom cell model parameter fold changes (relative to baseline values) in the fibrotic HF condition shown as box and whisker plots. Note for certain parameters, a fold-change >10 is achieved due to the genetic algorithm mutation step. Abbreviations: hMSCs, human mesenchymal stem cells; hCIC, c-kit+ cardiac interstitial cells;…

• 

  FIGURE 3. Effects of accepted custom cell heterocellular coupling on fibrotic heart failure cardiomyocyte electrophysiology and calcium handling. Action potential and calcium transient metrics of fibrotic heart failure (HF) cardiomyocytes when coupled to accepted custom cells are shown as box and whisker plots. Action potential metrics include: (A) Action potential duration (APD) at 50% repolarization (APD50) and 90% repolarization (APD90); (B) peak voltage; (C) upstroke velocity; and (D) resting membrane potential. Calcium transient metrics include: (E) peak intracellular calcium and calcium transient amplitude; (F) diastolic calcium; and (G) calcium relaxation time constant at 50% decay (τ50) and 90% decay (τ90). Values for untreated healthy and fibrotic HF cardiomyocytes are shown as dashed and dotted lines, respectively.

• 

  FIGURE 4. Custom cells expressing cardiomyocyte L-type calcium and inward rectifier channels are capable of correcting fibrotic failing cardiomyocytes. Initial populations of 2500 custom cell models were generated expressing (A) L-type calcium (ICaL) and inward rectifier (IK1) channels; (C) IK1 only; and (E) ICaL only with pseudo-random perturbations to gap junction conductances, number of heterocellularly coupled custom cells to fibrotic heart failure (HF) cardiomyocytes, and custom cell ion channel maximal conductances. Custom cell populations underwent 5 generations of genetic algorithm evolution, whereby custom cells were accepted if root mean square error was less than 50% for both action potential and calcium transient waveforms (blue); models outside this range were rejected (red). Distributions of accepted custom cell model parameter fold changes (relative to baseline values) in the fibrotic HF condition are shown as box and whisker plots for (B) ICaL and IK1, (D) IK1, and (F) ICaL. Note for certain parameters, a fold-change >10 is achieved due to the genetic algorithm mutation step. Waveforms for healthy and untreated HF cardiomyocytes are indicated by dashed and dotted lines,…

• 

  FIGURE 5. Effects of accepted custom cell heterocellular coupling on fibrotic heart failure cardiomyocyte electrophysiology and calcium handling. Action potential and calcium transient metrics are shown as box and whisker plots for coupling fibrotic heart failure (HF) cardiomyocytes to accepted custom cells with cardiomyocyte L-type calcium channel (ICaL), inward rectifier channel (IK1), or both (ICaL + IK1). Action potential metrics include: (A) Action potential duration (APD) at 50% repolarization (APD50) and 90% repolarization (APD90); (B) peak voltage; (C) upstroke velocity; and (D) resting membrane potential. Calcium transient metrics include: (E) peak intracellular calcium and calcium transient amplitude; (F) diastolic calcium; and (G) calcium relaxation time constant at 50% decay (τ50) and 90% decay (τ90). Values for healthy and untreated fibrotic HF cardiomyocytes are shown as dashed and dotted lines, respectively.

• 

  FIGURE 6. Ionic currents of fibrotic heart failure cardiomyocytes during heterocellular coupling with accepted custom cells. Plots of the following ionic currents of fibrotic heart failure (HF) cardiomyocytes when coupled to 25 randomly selected accepted custom cells (blue) that express cardiomyocyte inward rectifier channel: (A) inward rectifier current (IK1); (B) rapid delayed rectifier (IKr); (C) slow delayed rectifier (IKs); (D) L-type calcium (ICaL); (E) sodium-calcium exchanger (INCX); (F) ryanodine receptor. Zoomed in views of currents for ICaL and ryanodine receptor are inset. Healthy and untreated fibrotic HF cardiomyocytes are shown as dashed and dotted lines, respectively.

• 

  FIGURE 7. Custom engineered human mesenchymal stem cells and c-kit+ cardiac interstitial cells expressing cardiomyocyte inward rectifier channels reduce root mean square error. Initial populations of 2500 (A) control human mesenchymal stem cells (hMSCs), (B) hMSCs expressing cardiomyocyte inward rectifier channel (IK1), (C) control c-kit+ cardiac interstitial cells (hCICs), and (D) hCICs expressing IK1 were generated for heterocellular coupling with fibrotic heart failure (HF) cardiomyocytes with pseudo-random perturbations of gap junction conductances, number of coupled cells, and IK1 maximal conductance. Each population underwent 5 generations of genetic algorithm evolution, whereby cells were accepted if root mean square error (RMSE) was less than 50% for both action potential (AP) and calcium transient (CaT) waveforms; all models outside this range were rejected (red). The minimum total RMSEAP + RMSECaT achieved for each group is shown in pink. Waveforms for healthy and untreated HF cardiomyocytes are indicated by dashed and dotted lines, respectively.

• 

  FIGURE 8. Effects of custom engineered human mesenchymal stem cell and c-kit+ cardiac interstitial cell heterocellular coupling on fibrotic heart failure cardiomyocyte electrophysiology and calcium handling. Action potential and calcium transient metrics are shown for coupling fibrotic heart failure (HF) cardiomyocytes to human mesenchymal stem cells (hMSCs) and human c-kit+ cardiac interstitial cells (hCICs) (red and blue circles, respectively) with and without cardiomyocyte inward rectifier channel (IK1) expression (filled and unfilled, respectively) that achieved minimum total root mean square error of action potential and calcium transient waveforms (see pink waveforms in Figure 7). Action potential metrics include: (A) Action potential duration (APD) at 50% repolarization (APD50) and 90% repolarization (APD90); (B) peak membrane voltage (Vm); (C) upstroke velocity; and (D) resting Vm. Calcium transient metrics include: (E) peak intracellular calcium and calcium transient amplitude; (F) diastolic calcium; and (G) calcium relaxation time constant at 50% decay (τ50) and 90% decay (τ90). Values for healthy and untreated fibrotic HF cardiomyocytes are shown as dashed and dotted lines,…