Functional Consequences of Human Induced Pluripotent Stem Cell Therapy

Abstract

Background

The use of cells derived from human induced pluripotent stem cells as cellular therapy for myocardial injury has yet to be examined in a large-animal model.

Methods and Results

Immunosuppressed Yorkshire pigs were assigned to 1 of 3 groups: A myocardial infarction group (MI group; distal left anterior descending coronary artery ligation and reperfusion; n=13); a cell-treatment group (MI with 4×10⁶ vascular cells derived from human induced pluripotent stem cells administered via a fibrin patch; n=14); and a normal group (n=15). At 4 weeks, left ventricular structural and functional abnormalities were less pronounced in hearts in the cell-treated group than in MI hearts (P<0.05), and these improvements were accompanied by declines in scar size (10.4±1.6% versus 8.3±1.1%, MI versus cell-treatment group, P<0.05). The cell-treated group displayed a significant increase in vascular density and blood flow (0.83±0.11 and 1.05±0.13 mL·min⁻¹·g⁻¹, MI versus cell-treatment group, P<0.05) in the periscar border zone (BZ), which was accompanied by improvements in systolic thickening fractions (infarct zone, −10±7% versus 5±5%; BZ, 7±4% versus 23±6%; P<0.05). Transplantation of vascular cells derived from human induced pluripotent stem cells stimulated c-kit⁺ cell recruitment to BZ and the rate of bromodeoxyuridine incorporation in both c-kit⁺ cells and cardiomyocytes (P<0.05). Using a magnetic resonance spectroscopic saturation transfer technique, we found that the rate of ATP hydrolysis in BZ of MI hearts was severely reduced, and the severity of this reduction was linearly related to the severity of the elevations of wall stresses (r=0.82, P<0.05). This decline in BZ ATP utilization was markedly attenuated in the cell-treatment group.

Conclusions

Transplantation of vascular cells derived from human induced pluripotent stem cells mobilized endogenous progenitor cells into the BZ, attenuated regional wall stress, stimulated neovascularization, and improved BZ perfusion, which in turn resulted in marked increases in BZ contractile function and ATP turnover rate.

Heart failure is the end-stage clinical syndrome for a variety of cardiovascular diseases and affects more than 23 million patients worldwide, including 5.8 million in the United States.^1,2 The healthcare expenses associated with heart failure cost the US government more than $39 billion annually.^1,2 Heart failure often develops after acute myocardial infarction (MI) because the injured myocardial tissue fails to recover or regenerate.³ The heart has long been considered a postmitotic organ and incapable of self-renewal; however, recent reports have demonstrated that progenitor cells mediate cardiomyocyte turnover in normal adult hearts⁴ and that heart tissue can regenerate after injury through the differentiation of cardiac progenitor cells (CPCs) and through activation of the cell cycle in cardiomyocytes.⁵ The endogenous rate of cardiomyocyte regeneration is too slow to replace the cardiomyocytes that are lost during ischemia, but these mechanisms of cardiomyocyte turnover may be a key component of the therapeutic effects associated with cellular therapies,^6–8 which have recently been shown to benefit patients with postinfarction left ventricular (LV) remodeling.^9,10

Perhaps one of the most significant recent achievements in medical science has been the development of induced pluripotent stem cells (iPSCs).^11,12 These cells reproduce much of the regenerative potential possessed by pluripotent embryonic stem cells but are generated from a patient’s own somatic cells and consequently minimize the ethical concerns and potential immunogenic complications associated with embryonic stem cell therapies. The myocytes generated through cellular therapy, or from the activity of endogenous progenitor cells, are believed to reduce LV dilatation and bulging at the site of infarction, which subsequently decreases myocardial wall stress and improves myocardial function and metabolism^7,13,14; however, the use of iPSCs or cells derived from iPSCs for the treatment of ischemic myocardial injury has not yet been examined in a clinically relevant large-animal model.

The hypothesis that cardiac dysfunction in failing hearts develops because of energy starvation is an old one¹⁵; however, whether the decreased reserve of myocardial ATP turnover rate contributes to the progression of cardiac dysfunction is largely a topic for speculation, because accurate measurements of myocardial ATP turnover rates cannot be made in vivo. Although the magnetic resonance spectroscopy–magnetization saturation transfer (MRS-MST) technique has been used extensively to measure ATP flux via creatine kinase, the accurate evaluation of ATP turnover rates via mitochondrial and cytosolic enzymes (ADP+P_i↔ATP) has generally not been successful in vivo because the level of myocardial free inorganic phosphate (P_i) is too low to be measured directly.^16,17

Here, we introduce a novel MRS-MST method that enables us to calculate the ATP hydrolysis rate without measuring P_i levels (see online-only Data Supplement for mathematical demonstration), which overcomes the primary barrier in determining the ATP turnover rate in vivo. The accuracy of this method was examined rigorously and confirmed in both skeletal muscle and the heart under baseline and high cardiac workloads. Using an immunosuppressed swine model of postinfarction LV remodeling and this novel nuclear magnetic resonance technique, we examined the hypothesis that in vivo hearts with postinfarction LV remodeling have a decreased reserve of myocardial ATP turnover rate. We hypothesized that treatment with an epicardial fibrin patch–enhanced delivery of vascular cells derived from human iPSCs (hiPSC-VCs) would decrease the periscar border zone (BZ) myocardium overstretch and wall stresses, which in turn would result in an improvement in myocardial ATP turnover rate.

Methods

A more detailed description of the experimental procedures used in this investigation is provided in the online-only Data Supplement.

Generation of Vascular Cells From Human iPSCs

The characteristics of the human iPSCs (hiPSCs) used for these studies were provided in Knorr et al.¹⁸ The generation of iPSCs from neonatal human dermal fibroblasts was accomplished by lentiviral transduction of OCT4, SOX2, KLF4, and cMYC.¹⁹ The iPSC criteria include polymerase chain reaction and immunostaining of pluripotent markers including Oct3/4, Sox2, Nanog, SSEA-4, and Tra-1-81; ability to form teratomas; and normal karyotype. hiPSCs that expressed green fluorescent protein (GFP) were generated separately by the “Sleeping Beauty” transduction method (Amaxa, Gaithersburg, MD), as described previously.²⁰ The GFP⁺ hiPSCs were then subjected to vascular differentiation by a previously published protocol for human embryonic stem cells.²¹ The hiPSC-VCs consisted of 2 distinct cell types that corresponded to endothelial cells and smooth muscle cells (SMCs).

Swine Model of Ischemia Reperfusion and Cell Transplantation

The experimental protocol was approved by the University of Minnesota Research Animal Resources Committee. All experimental and animal maintenance procedures were performed in accordance with the animal use guidelines of the University of Minnesota and were consistent with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals (NIH publication No. 85–23).

Female Yorkshire farm swine (weight ≈15 kg; Manthei Hog Farm, Elk River, MN) were randomly assigned to 1 of 3 experimental groups: A cell-treatment group (MI with 4×10⁶ hiPSC-VCs administered via a fibrin patch; n=14); an MI group (distal left anterior descending coronary artery ligation and reperfusion; n=13); and a normal group (n=15). Ischemia-reperfusion injury was surgically induced in animals from the cell-treatment and MI groups by temporary ligation of the coronary artery for 60 minutes as described previously^13,22 and as summarized in the online-only Data Supplement; animals in the normal group underwent the same surgical procedure except for the ligation step. A fibrin patch that contained 4 million hiPSC-VCs (hiPSC endothelial cells and hiPSC SMCs, 2 million each) was placed over the ligation site of hearts in the cell-treatment group.²² Because the transplanted cells were of human origin, the immune systems of animals in all 3 experimental groups were suppressed via cyclosporine injection (15 mg·kg⁻¹·day⁻¹, supplemented with food).⁷

Contractile Function and Infarct Size

Cardiac functional parameters (LV ejection fraction and systolic thickening fraction) and infarct size were evaluated via magnetic resonance imaging as summarized in the online-only Data Supplement. Measurements were performed with a 1.5-T clinical scanner (Siemens Sonata, Siemens Medical Systems, Iselin, NJ); functional parameters were determined from short-axis cine images, and infarct sizes were determined via delayed-enhancement magnetic resonance imaging.

Myocardial ATP Turnover Rate

In vivo measurements of the myocardial ATP turnover rate were obtained via an open-chest ³¹P MRS-MST protocol as summarized in the online-only Data Supplement. Measurements were performed on a 65-cm-bore 9.4-T magnet interfaced with a VnmrJ console (Varian, CA). Radiofrequency transmission and MRS signal detection were performed with a double-tuned (¹H and ³¹P) surface coil (28-mm diameter). BZ measurements were obtained by suturing the coil directly to the LV epicardium over the peri-infarct region in hearts from the MI and cell-treatment groups, and remote zone measurements were obtained by suturing the coil over a noninfarcted region; measurements in normal hearts were obtained with the coil positioned on the anterior wall. Measurements were performed both under baseline conditions and a high cardiac work state induced by catecholamine stimulation (dobutamine and dopamine, each 10 μg·kg⁻¹·min⁻¹ IV).

A complete theoretical derivation of our experimental approach is provided in the online-only Data Supplement. Briefly, the in vivo metabolism of ATP can be modeled as a chemical exchange network comprising 2 reactions among 3 components: The shuttling of inorganic phosphate (P_i) between phosphocreatine (PCr) and ATP (PCr↔ATP) and the hydrolysis/condensation of ATP (ATP↔ADP+ P_i). The kinetics of ATP metabolism can be summarized with 2 pseudo first-order rate constants: K_PCr→ATP for the forward creatine kinase reaction and K_ATP→Pi for ATP→ P_i hydrolysis reactions. The K_ATP→Pi includes contributions from (1) all of the contraction/relaxation-associated reactions that are energized by ATP and (2) additional substantial contributions from a cytosolic near-equilibrium enzyme complex composed of GAPDH and phosphoglycerate kinase. Because the PCr↔ATP reaction is at equilibrium, the rate of the ATP→ADP+P_i reaction can be calculated by subtracting the flux of the ATP→PCr reaction from the total ATP turnover rate (ATP→PCr plus ATP→ADP+P_i):

$$k_{\text{ATP}\to \text{Pi}}=\left(\frac{M_{0,\text{ATP}\gamma }-M_{\text{ss},\text{ATP}\gamma }}{M_{\text{ss},\text{ATP}\gamma }}\right)/T_{1,\text{ATP}\gamma }^{int}-\frac{M_{0,\text{PCr}}}{M_{0,\text{ATP}\gamma }}\left(\frac{M_{0,\text{PCr}}-M_{\text{ss},\text{PCr}}}{M_{\text{ss},\text{PCr}}}\right)/T_{1,\text{PCr}}^{int}$$

where K_ATP→Pi is the pseudo first-order rate constant of the ATP→ADP+P_i reaction; M₀ and T₁^int are the fully relaxed magnetization and intrinsic longitudinal relaxation time constants of spin, respectively, as determined via ³¹P MRS; M_ss,ATPγ is the steady-state magnetization of ATPγ when PCr and P_i are completely saturated; and M_ss,PCr is the steady-state magnetization of PCr when ATPγ is completely saturated. The measurement of K_ATP→Pi can be achieved from as few as 3 spectra, 1 taken with the ATPγ signal saturated to measure the steady-state magnetization of PCr (M_ss,PCr), 1 taken with both P_i and PCr saturated to measure the steady-state magnetization of ATPγ (M_ss,ATPγ), and 1 taken with no saturation to measure the magnetization of PCr and ATPγ (M_0,PCr and M_0,ATPγ). The validity of the methodology was rigorously examined on swine skeletal muscle (n=5), where P_i is reliably measureable; the results from the novel approach were compared with those from the conventional MST method. Details of validation experiments are provided in the online-only Data Supplement.

Myocardial Perfusion, LV Hemodynamics, and Histological Assessments

Evaluations were performed after the chest had been opened for the ³¹P MRS assessments. Myocardial perfusion was measured via the injection of fluorescently labeled microspheres as described previously.^23–25 LV hemodynamics were monitored via a catheter inserted through the apical dimple and into the LV,²³ and histological assessments were performed according to standard protocols. More detailed descriptions of these experimental procedures are provided in the online-only Data Supplement.

Statistical Analysis

All data are reported as mean±SD. Statistical analyses were performed with SigmaStat version 3.5 (San Jose, CA). Comparisons among groups were analyzed for significance with 1-way ANOVA. Comparisons among different time points and between experimental groups were performed with 2-way repeated-measures ANOVA. P<0.05 was considered significant. When ANOVA demonstrated a significant effect, post hoc analysis was performed with the t test with Bonferroni correction. Comparisons of experimental condition versus baseline within a group were performed with a paired t test. Comparisons between groups were performed with an unpaired t test. Linear regression and correlation analyses were performed to analyze the variables of rate-pressure product, LV systolic thickening fraction, end-diastolic volume, and wall stress versus the myocardial flux rate of the ATP hydrolysis reaction.

Results

hiPSC-Derived Vascular Cells

The hiPSC-VCs consisted of 2 distinct cell types: hiPSC-derived endothelial cells and hiPSC-derived SMCs (Figure 1). The hiPSC endothelial cells expressed endothelial markers such as CD31, CD73, CD105, CD144, CD146, and von Willebrand factor and formed robust tubelike structures when cultured in Matrigel (Figures 1A through 1C). The hiPSC SMCs displayed a typical SMC phenotype and expressed the SMC markers smooth muscle actin (SMA), SM22, and calponin (Figure 1D). These results are consistent with those obtained when human embryonic stem cells were cultured with the same vascular differentiation protocol.²¹

Figure 1.

Characterization of human induced pluripotent stem cell (hiPSC)–derived vascular cells. A, Flow cytometric analysis of hiPSC-derived endothelial cells (ECs), demonstrating the expression of CD31, CD73, CD105, and CD146. Blue line shows samples stained with respective antibodies; red line, isotype control. B and D, Representative immunostained images of hiPSC-ECs and smooth muscle cells (SMCs), respectively. ECs expressed endothelium-specific markers such as CD31, von Willebrand factor (vWF), and CD144, whereas SMCs expressed smooth muscle actin (SMA), SM22, and calponin. C, hiPSC-ECs formed capillary tubelike structure after 24-hour culture in Matrigel. Max indicates maximum.

hiPSC-VC Transplantation Reduced LV Remodeling and Improved Cardiac Performance After MI

One week after MI injury, infarcts were ≈10% of the size of the LV in hearts in both the MI and cell-treatment groups (Figure 2A) and were accompanied by evidence of cardiac remodeling, such as cardiac hypertrophy (Figure 2B) and dilatation (Figure 2C); however, infarcts at week 4 were significantly smaller and remodeling was significantly less extensive in hearts in the cell-treatment group than in MI hearts. MI injury also led to a significant decline in LV ejection fraction (Figure 2D). However, the systolic thickening fraction at the BZ of the infarct (Figure 2E) was significantly greater in hearts in the cell-treatment group than in MI hearts, and the bulging of the infarct was also prevented by cell treatment (Figure 2E); these improvements were accompanied by declines in measurements of myocardial wall stress at both the BZ and infarct zone (Figure 2F). Cardiac hemodynamic parameters in hearts in the MI and cell-treatment groups were similar to measurements in normal hearts both at baseline and during high cardiac work states (Table 1), which suggests that the injured hearts were in the compensated phase of LV remodeling. Typical cardiac magnetic resonance images of hearts from each of the experimental groups at end diastole and end systole are shown in Figure 2G. Videos of typical hearts from each of the experimental groups are provided in the online-only Data Supplement.

Figure 2.

Structural and functional benefits of transplantation of human induced pluripotent stem cell–derived vascular cells (hiPSC-VCs). A, At week 4 after infarction, the cell-treatment group showed significantly smaller infarct size (delayed-enhancement magnetic resonance imaging) than the myocardial infarction (MI) group. B, MI induced left ventricular hypertrophy, as evidenced by significantly increased left ventricular weight over body weight (LVW/BW). hiPSC-VC transplantation attenuated the left ventricular hypertrophy. C, MI induced dilation of the left ventricular chamber compared with the normal group. hiPSC-VC transplantation attenuated the chamber dilation. D, hiPSC-VC transplantation significantly improved the ejection fraction of infarcted swine hearts. E, Border zone (BZ) myocardium in postinfarcted swine hearts showed significant contractile dysfunction (systolic thickening fraction), which was significantly improved in response to hiPSC-VC transplantation. hiPSC-VC treatment also prevented the infarct zone (IZ) from bulging out during systole (negative thickening fraction). F, hiPSC-VC transplantation alleviated the abnormal myocardial wall stress in the BZ and IZ myocardium. G, Representative cardiac magnetic resonance images from normal, MI, and cell-treatment groups at end systole (left) and end diastole (right). #P<0.05 vs normal group; *P<0.05 vs MI group. Quantification of cardiac functional parameters was based on normal hearts (n=10), MI hearts (n=9), and hearts treated with cell therapy (n=9). LVEDV indicates left ventricular end-diastolic volume; and RZ, remote zone.

Table 1.

Hemodynamic Data

	Baseline				High Cardiac Workload
	HR, bpm	LVSP, mm Hg	LVEDP, mm Hg	RPP (×103 mm Hg/bpm)	HR, bpm	LVSP, mm Hg	LVEDP, mm Hg	RPP (×103 mm Hg/bpm)
Normal (n=10)	92±11	98±7	5±3	9.0±1.1	186±29	166±33	8±3	25.5±1.9
MI (n=9)	96±12	94±5	7±2	9.0±1.2	194±40	136±22	7±5	25.8±4.2
Cell treatment (n=9)	98±13	91±8	6±2	9.0±1.6	224±43*	147±21	8±3	33.1±8.2

Values are mean±SD.

HR indicates heart rate; LVEDP, left ventricular end-diastolic pressure; LVSP, left ventricular systolic pressure; MI, myocardial infarction; and RPP, rate-pressure product (HR×LVSP).

^*P<0.05 vs MI hearts.

hiPSC-VC Transplantation Improved Myocardial Energetics After MI

Recently, we used a novel ³¹P MRS 2-dimensional chemical shift imaging protocol to demonstrate that the effect of MI injury on myocardial bioenergetics is heterogeneous: PCr/ATP ratios declined near the border of the infarcted region but not (or to a much lesser extent) in remote (ie, noninfarcted) tissue.⁷ The results obtained in the present investigation with hiPSC-VCs are consistent with our previous report: The PCr/ATP ratio was significantly lower at the BZ of the infarct in MI hearts than in normal hearts under baseline workload conditions and was significantly improved by hiPSC-VC treatment (Table 2). For the study reported here, we further developed another ³¹P MRS protocol that enabled us to measure, in vivo, the myocardial ATP turnover rate in different regions of the heart.

Table 2.

³¹P MRS Data in the In Vivo Heart

	Normal (n=10)		MI-BZ (n=9)		CELL-BZ (n=9)
	Baseline	HWL	Baseline	HWL	Baseline	HWL
PCr/ATP	2.09±0.09	1.94±0.23	1.64±0.22*	1.40±0.22*‡	1.86±0.17*†	1.44±0.18*‡

Values are mean±SD.

CELL-BZ indicates border zone of cell-treated hearts; HWL, high cardiac work load; MI-BZ, border zone of hearts with myocardial infarction; and MRS, magnetic resonance spectroscopy.

^*P<0.05 vs normal hearts;

^†P<0.05 vs MI-BZ;

^‡P<0.05 vs baseline for same group.

In the heart, ATP turnover can be modeled as a chemical exchange network with 3 components (PCr↔ATP↔P_i), and the kinetics of ATP metabolism can be summarized with 2 pseudo first-order rate constants (K_PCr→_ATP for the creatine kinase reaction and K_ATP→_Pi for the ATP→P_i reaction). The corresponding chemical reaction flux can be calculated by multiplying the rate constant with metabolite concentration: Flux_PCr→_ATP=K_PCr→_ATP X [PCr] and Flux_ATP→_Pi=K_ATP→_Pi X [ATP]. Myocardial ATP turnover rates were determined in vivo as illustrated in Figure 3 and were rigorously validated with equivalent in vivo analyses performed on skeletal muscle (n=5; online-only Data Supplement Figure I). The intrinsic T₁ (T₁^int) of ATPγ and PCr was measured under baseline conditions via progressive saturation (Figure 3A; saturation of ATPγ) and inversion recovery (Figure 3B; double saturation of both PCr and P_i), respectively. At 9.4 Tesla, T₁^int in the heart was 1.1±0.1 seconds for ATPγ and 3.2±0.1 seconds for PCr, which is in good agreement with measurements in skeletal muscle (online-only Data Supplement Table I) and with previously reported values at the same magnetic field.²⁶ These observations indicate that T₁^int is determined by the microenvironment of the magnetic field that surrounds a given molecule and is not affected by physiological or pathological conditions or by the kinetics of the chemical exchange network.

Figure 3.

Measurement of in vivo myocardial ATP turnover rate with ³¹P magnetic resonance spectroscopy–magnetization saturation transfer experiments. A and B, Measurement of intrinsic T₁ (T₁^int) for phosphocreatine (PCr) and ATPγ using progressive saturation (A) and inversion recovery (B) experiments. Spectra a1 through a7 are from progressive saturation experiment with ATPγ saturated (arrow) for 0.44, 0.88, 1.32, 2.2, 3.52, 5.28, and 7.04 seconds, respectively. Spectra b1 through b6 are from inversion recovery experiment with both inorganic phosphate (P_i) and PCr saturated (arrows) for 0.44, 0.88, 1.32, 1.76, 2.64, and 3.96 seconds, respectively. The magnetizations of PCr and ATPγ were quantified and then subjected to exponential models to fit the intrinsic T₁s. The experiments yielded an average intrinsic T₁ of 3.2±0.1 seconds and 1.1±0.1 seconds for PCr and ATPγ, respectively. C, Three-spectrum magnetization saturation transfer experiment to measure myocardial ATP turnover rate based on the equation given in Methods. c1 indicates control spectrum without saturation to quantify magnetization of PCr (M_0,PCr) and ATPγ (M_0,ATPγ); c2, ATPγ-saturated spectrum to quantify steady-state magnetization of PCr (M_ss,PCr); and c3, P_i - and PCr-saturated spectrum to measure steady-state magnetization of ATPγ (M_ss,ATPγ). BISTRO-based frequency-selective saturation pulses are indicated by arrows. Detailed information of BISTRO (B1-insensitive train to obliterate signal) saturation pulse sequence is shown in online-only Data Supplement Figure IV. 2,3-DPG indicates 2,3-diphosphoglycerate. D, Scatter-plot of Flux_ATP→Pi vs rate-pressure product (RPP) for individual hearts from normal group, which indicates a tight coupling between ATP hydrolysis rate and cardiac workload (R²=0.89, P<0.05). E and F, Summarization of myocardial K_ATP→Pi and K_PCr→ATP for all groups at both baseline and high cardiac workload (HWL) conditions. K_ATP→Pi is significantly reduced in border zone of myocardial infarction (MI) hearts compared with normal hearts, whereas K_PCr→ATP was similar among all groups and all conditions. In response to transplantation of human induced pluripotent stem cell–derived vascular cells, the abnormality of ATP turnover rate in border zone was abolished. Flux_ATP→Pi was calculated from K_ATP→Pi×[ATP]. [ATP] value was estimated to be 5.3 μmol/g wet tissue weight for normal myocardium. #P<0.05 vs normal group; *P<0.05 vs MI group; ‡P<0.05 vs same group in baseline condition. Quantification of myocardial ATP turnover rate was based on normal hearts (n=10), MI hearts (n=8), and hearts treated with cell therapy (n=9).

Kinetic measurements (K_PCr→_ATP and K_ATP→_Pi) were performed via a 3-spectrum MST experiment (Figure 3C). In normal hearts, Flux_ATP→_Pi measurements were linearly related to the rate-pressure product (LV pressure×heart rate; Figure 3D), which indicates that the ATP→P_i turnover rate and cardiac workload are tightly coupled. MI injury was associated with significantly lower K_ATP→_Pi measurements at the BZ of the infarct under both baseline and high-workload conditions (Figure 3E). However, K_ATP→_Pi measurements in the remote zone were unchanged (K_ATP→_Pi from remote zone of MI hearts [n=5] was 0.15±0.03 s⁻¹, P=NS versus normal hearts), which indicates that the effect of infarction on ATP turnover rate is localized to the BZ at this phase of the LV remodeling. Although the rate-pressure product in response to catecholamine stimulation in MI and normal hearts was similar (Table 1), the K_ATP→_Pi of BZ from MI hearts was less responsive. The decline in K_ATP→_Pi measurements observed at the BZ of MI hearts was significantly improved by hiPSC-VC treatment (Figure 3E). The BZ Flux_PCr→_ATP was significantly lower in MI hearts than in normal hearts. K_PCr→_ATP measurements in hearts in the normal, MI, and cell-treatment groups were similar under both baseline and high-workload conditions (Figure 3F and 4A).

Figure 4.

Correlation of myocardial ATP turnover rate with severity of postinfarction left ventricular remodeling. A, Energy transfer flux of ATP→PCr and ATP→P_i reactions from groups of normal, myocardial infarction (MI), and cell-treatment hearts. #P<0.05 vs normal; *P<0.05 vs MI. B–D, Relationship between myocardial Flux_ATP→Pi and border zone myocardium systolic thickening fraction, left ventricular end-diastolic volume (LVEDV), and border zone myocardial wall stress, respectively. PCr indicates phosphocreatine; and P_i, inorganic phosphate.

Myocardial flux of the PCr→ATP and ATP→P_i reactions are summarized in Figure 4A. The concentration of ATP in the myocardium of swine hearts (5.3 μmol/g wet weight) has been determined previously.^27,28 Flux_PCr→ATP and Flux_ATP→Pi were significantly lower in the BZ of hearts in both the MI and cell-treatment groups than in the corresponding region of normal hearts, and Flux_ATP→Pi but not Flux_PCr→ATP was improved by hiPSC-VC treatment (Figure 4A). Flux_ATP→Pi was also linearly correlated with measurements of the myocardial systolic thickening fraction, BZ wall stress, and LV end-diastolic volume (Figures 4B through 4D). These findings are consistent with existing understanding that regional wall stress is a key determinant of metabolic activity and demand and contractile element response.²⁹

Survival and Engraftment of hiPSC-VCs in Infarcted Swine Heart

The transplanted hiPSC-VCs had been engineered to express GFP; thus, the engraftment of hiPSC-VCs into the myocardial vasculature was evaluated by staining for the expression of GFP, CD31, and SMA, and proliferation was evaluated by staining for bromodeoxyuridine (BrdU) incorporation in the hearts of animals that had been infused with BrdU (10 mg·kg⁻¹·d⁻¹, administered via an insulin pump) for 7 days. GFP⁺ cell counts indicated that the engraftment rate was 7.1±0.8% of the total number of cells transplanted per animal at week 1 and 2.3±0.9% at week 4 (Figure 5A). The transplanted hiPSC-VCs appeared to integrate into preexisting vessels (Figure 5B) and to generate new vessel growth in the area of the fibrin patch (Figure 5C), and abundant evidence of BrdU incorporation by GFP⁺ cells confirmed that the engrafted hiPSC-VCs were highly proliferative (online-only Data Supplement Figure II).

Figure 5.

Functional engraftment and structural benefit of transplantation of human induced pluripotent stem cell–derived vascular cells (hiPSC-VCs). A, Engraftment rate of hiPSC-VCs at both week 1 and week 4 after transplantation. Calculation was based on green fluorescent protein (GFP)–positive cell counts in histology analysis and an initial cell number of 4 million. B, Transplanted hiPSC-VCs contributed to neovascularization through engraftment onto vascular structures. Inset i–iv, Costaining of GFP (green) and smooth muscle actin (SMA; red) demonstrated hiPSC-VCs grafting onto a host artery (arrowhead). C–F, Representative vascular staining from (C) a cell-treatment heart at patch/myocardium border, (D) a normal heart, (E) a heart in the myocardial infarction (MI) group at border zone (MI-BZ), and (F) a cell-treatment heart at border zone (CELL-BZ). Vessels were visualized by CD31 staining (green) and SMA staining (red), whereas myocardium was visualized by cardiac troponin I (cTnI) staining (blue). G, Summarization of quantification of arteriolar density (SMA⁺ vessels with diameter >25 μm) and total vascular density (all CD31⁺ vessels). H and I, Summarization of microsphere-based myocardial blood flow (MBF) data examined at 4 weeks after infarction. MBF measurements were performed at both baseline (H) and during hyperemia (I; adenosine 0.5 mg·kg⁻¹·min⁻¹ IV) conditions. Myocardium samples (≈1 g) from either border zone (BZ) or remote zone (RZ) were examined. #P<0.05 vs normal; *P<0.05 vs MI. All quantifications were performed based on data from 4 hearts from each group.

hiPSC-VC Transplantation Improved Myocardial Vascularity and Perfusion After MI

MI injury led to a severe loss of vascularity, as evidenced by the minimal amount of CD31 and SMA expression observed at the BZ of the infarct in hearts from the MI group; however, both CD31⁺ and SMA⁺ vascular density was significantly greater in cell hearts than in MI hearts (Figures 5D through 5G). To determine whether the increase in vascular density was accompanied by a corresponding increase in blood flow, regional myocardial blood flow was measured via cardiac magnetic resonance imaging and microsphere injection^30,31 (n=4 per group), and coronary reserve was examined by use of adenosine-induced myocardial hyperemia (0.5 mg·kg⁻¹·min⁻¹ IV).^23,32 These data demonstrate that cell treatment led to significant increases in BZ myocardial perfusion (Figure 5H), which were accompanied by significant improvements in systolic thickening fractions and ATP turnover rate in this region of the myocardium (Figure 4). Myocardial blood flow also tended to be higher in hearts from the cell-treatment group during hyperemia (Figure 5I), but the difference between hearts from the cell-treatment and MI groups was not significant (Figure 5I), and calculation of the coronary flow reserve did not identify significant differences among experimental groups or between different regions of myocardium (Figure 5I). Collectively, these observations demonstrate that the transplanted hiPSC-VCs were able to integrate into the growing vasculature and to generate new vessels, which suggests vasculogenesis. However, fewer than 5% of vessels expressed GFP, so the enhanced vascularity and perfusion associated with cell treatment likely evolved primarily through the sprouting of preexisting vessels.

hiPSC-VC Transplantation Stimulated the Activation and Proliferation of Endogenous CPCs and Cardiomyocytes

To determine whether cell therapy enhanced the activation and proliferation of endogenous CPCs, sections were stained for the presence of α-sarcomeric actin (to identify cardiomyocytes), for expression of the progenitor cell marker c-kit, and for BrdU incorporation (Figures 6A and 6B; online-only Data Supplement Figure III). Hearts from animals in the normal group displayed little evidence of c-kit expression, but c-kit⁺ cells were common in the injury site of hearts from the MI group, and the proportion of c-kit⁺ cells that were also positive for BrdU incorporation was significantly higher in MI hearts than in normal hearts (Figures 6C and 6D). c-kit⁺ cells often coexpressed the cardiac cell marker GATA-4, but cells that coexpressed c-kit and the hematopoietic cell marker CD45 were rare (Figures 6E and 6F), which suggests that many of the progenitor cells were committed to the cardiac lineage and that inflammatory cells were not a significant source of c-kit expression. Furthermore, both the number of c-kit⁺ cells and the incorporation rate of BrdU by c-kit⁺ cells were significantly greater in hearts from the cell-treatment group than in MI hearts at week 1 but not at week 4. Thus, the transplanted hiPSC-VCs appeared to enhance the endogenous response to MI by transiently increasing both the number and proliferation of CPCs in the infarcted region. The proportion of cardiomyocytes that were positive for BrdU incorporation was also significantly higher in the BZ of hearts from the cell-treatment group than in the corresponding region of hearts from MI animals (Figures 7A through E), and BrdU⁺ cardiomyocytes were significantly smaller than BrdU⁻ cardiomyocytes from the same region (Figure 7F), which suggests that the enhanced activation of CPCs was accompanied by increases in cardiomyocyte turnover.

Figure 6.

Activation of c-kit⁺ progenitor cells in response to myocardial injury and transplantation of human induced pluripotent stem cell–derived vascular cells. A and B, Representative immunostained images from (A) normal and (B) infarct area of cell-treatment hearts at 1 week after surgery. Cardiac progenitor cells were identified by c-kit staining (green), cardiomyocytes by α-sarcomeric actin (αSA) staining (yellow), and cycling cells by bromodeoxyuridine (BrdU) staining (red). Arrow in Figure inset a1–a4 indicates c-kit⁺/BrdU⁻ cell; arrowheads in Figure inset b1–b4 indicate c-kit⁺/BrdU⁺ cells. C and D, Quantification of BrdU incorporation rate of c-kit⁺ cells (BrdU⁺-c-kit⁺/c-kit⁺ %) and c-kit⁺ cell density, respectively. All quantification was performed based on 5 slides per heart and 3 hearts per group. #P<0.05 vs normal; *P<0.05 vs myocardial infarction (MI); ‡P<0.05 vs same group at week 1. E1–E4, Costaining of CD45 and c-kit demonstrated that hematopoietic cells did not contribute significantly to the c-kit⁺ cells identified at 1 week after surgery. Arrow indicates 1 c-kit⁺/CD45⁺ cell. F1–F4, Costaining of c-kit and GATA4 indicated the cardiac commitment of these progenitor cells. Arrows highlight 2 GATA4⁺/c-kit⁺ cells.

Figure 7.

Transplantation of human induced pluripotent stem cell–derived vascular cells stimulated cardiac turnover. A–D, Cardiomyocyte turnover is evidenced by costaining of α-sarcomeric actin (αSA; green) and bromodeoxyuridine (BrdU; red). Representative BrdU⁺ myocytes with higher magnification are shown in Figure insets i–viii and denoted by arrowheads. BZ indicates border zone; MI, myocardial infarction; and RZ, remote zone. E, Quantification of myocardial regeneration level in terms of percentage of BrdU⁺ myocytes over total number of myocytes (BrdU⁺ myocytes %). The quantification was performed based on 10 slides per location (BZ and RZ) per time point (week 1 and week 4) and 3 hearts per group (MI and cell-treatment [CELL]). An average of >10 000 cardiomyocytes were analyzed per group. *P<0.05 vs RZ from same time point within same group; ^#P<0.05 vs MI group from same region and same time point; ‡P<0.05 vs same group at week 1. F, Quantification of myocyte size (cross-sectional diameter) from BZ myocardium of cell-treatment heart at week 4. BrdU⁺ myocytes were significantly smaller than BrdU⁻ myocytes. CM indicates cardiomyocytes.

Discussion

One of the most unsettled questions in cardiovascular physiology is how the rate of ATP production and utilization is regulated in the heart and how the limitation of this rate may contribute to contractile dysfunction in hearts that become hypertrophic in response to MI-induced LV remodeling, pressure overload, or volume overload. The heart has the highest ATP turnover rate per gram of tissue of any organ in the body³³; however, the ATP reserve in myocardial tissue is so low that it would be exhausted after only a few dozen beats if the ATP production machinery stopped functioning.³³ In vivo methods for accurately measuring the ATP utilization rate in the heart have not been established because multiple mechanisms for ATP turnover happen simultaneously,³⁴ and the conventional method requires quantification of the intracellular P_i level, which is intrinsically low and overlaps with 2,3-diphosphoglycerate resonance peaks. In addition, whether intracellular levels of myocardial free P_i are completely visible on nuclear magnetic resonance is somewhat controversial.^16,17 In the present report, we introduce a novel, double-saturation ³¹P MST protocol that measures the ATP utilization rate without measuring P_i, thereby circumventing the barrier for in vivo applications (online-only Data Supplement). By applying this nuclear magnetic resonance method, we have observed abnormalities in the rate of ATP utilization at the BZ of infarction; furthermore, the BZ bioenergetic abnormality was significantly corrected in response to cell therapy, and this improvement was associated with significantly increased BZ contractile function (Figures 2 and 4). The observation that ATP hydrolysis/synthesis rates increased in the BZ of cell-treated hearts (Figure 4) is not surprising given the significant volume of literature that suggests that the factors responsible for myocyte overstretch and elevations in wall stress also lead to a decline in contractile function²⁹ and that correcting these factors would contribute to increases in the BZ systolic thickening fraction.¹³ However, in the in vivo heart, the mechanisms responsible for this simple observation likely involve multiple factors throughout the network of ATP machinery, which functions seamlessly to maintain the balance between energy delivery and demand for the contractile apparatus.^34,35 A more comprehensive and inclusive discussion of these mechanisms is beyond the scope of the present investigation, but future studies combining measurements of in vivo ATP turnover rates and intermediary metabolism³⁶ with measurements of changes in protein activity, the expression levels of creatine kinase isozymes,²⁷ and ATPase subunits^28,37 and isolated mitochondrial function could advance our understandings of the molecular pathways that are activated by myocyte hypertrophy, overstretching, and regional LV wall stress.^7,13,36

Validation of ATP Utilization Rate Measurements in Skeletal Muscle

Methods for accurately measuring the total rates of myocardial ATP synthesis (or utilization) in the in vivo heart have not been established because of difficulties in accurate measurement of myocardial free P_i level.³⁸ In the present report, we have introduced and rigorously validated a novel ³¹P MST protocol that measures the ATP utilization rate without the need for a P_i measurement. The validation experiments were performed in swine skeletal muscle, in which adequate P_i resonance definition ensured that measurements could be made with both the new and the classic MST techniques (online-only Data Supplement Figure I). The equivalence of the results obtained with the novel and classic methods validates our new method for measuring the K_ATP→Pi and Flux_ATP→Pi in the in vivo heart, in which the myocardial free P_i level is too low to be quantified accurately.

Myocardial ATP Turnover Rate in the In Vivo Heart

In response to acute MI, molecular, structural, and functional abnormalities first occur in the BZ. The surviving BZ myocytes are overstretched and then spread to adjacent regions of healthy myocardium until the entire LV is affected and signs of heart failure appear.^28,39 The present findings demonstrate that the Flux_ATP→Pi rate in MI hearts is highly heterogeneous, with a pronounced reduction in the BZ but not in the remote zone. Although the underlying mechanisms responsible for the decline in BZ Flux_ATP→Pi and its failure to increase in response to catecholamine stimulation are beyond the scope of the present investigation, these mechanisms are likely similar to those present in myocardium of failing hearts, which show downregulation to catecholamine stimulation.

The decline in BZ Flux_ATP→Pi was accompanied by contractile dysfunction, a loss of vascular density, and elevated wall stress. Previous reports using both a localized ³¹P MRS protocol and, more recently, a 2-dimensional chemical shift imaging technique^7,28 have shown that the PCr/ATP ratio and, by extension, the bioenergetic efficiency of the BZ are impaired. This abnormal bioenergetic status is related to elevations in BZ wall stress, which could contribute to the progression of the disease state from compensated MI to congestive heart failure. The decline in BZ Flux_ATP→Pi is also consistent with previous observations that the number of F₁F₀-ATPase subunits is severely reduced in the BZs of compensated swine hearts with LV remodeling.^28,37 The present observation that the ATP hydrolysis rate is severely reduced in BZ myocardium that was recovered in response to the patch-enhanced cell treatment suggests that cellular therapy targeting the BZ of hearts in the compensated phase of remodeling may prevent progression to heart failure (Figures 4C and 4D).

The transplantation of hiPSC-VCs corrected the regional abnormalities in ATP hydrolysis rate that were observed in MI hearts (Figure 4). This beneficial effect on ATP hydrolysis supports the previous observation that cell transplantation into infarcted hearts improved myocardial bioenergetics, as indicated by the PCr/ATP ratio,^7,13,14 and that these bioenergetic benefits were accompanied by significant improvements in BZ myocardial contractile function, wall stress, vascular density, and myocardial blood flow. Future investigations are warranted to identify the molecular pathways that are induced by myocyte overstretching and support the beneficial effects of cardiac cell therapy.

The hypothesis that failing hearts are energy starved is an old one.¹⁵ On the basis of studies that used animal models of heart failure and patients with LV hypertrophy, it is a rather consistent finding that myocardial HEP levels and the forward flux rate of creatine kinase are reduced significantly, which is linearly related to the severity of LV dysfunction.^23,27 Myocardial HEP levels^23,32 and creatine kinase kinetics have been studied extensively²⁷; however, rates of ATP hydrolysis have been largely overlooked in in vivo investigations because of the intrinsically low myocardial P_i level. In the in vivo heart, myocardial P_i produces a weak signal that overlaps with the resonance peaks of 2,3-diphosphoglycerate from erythrocytes in LV chamber blood. Our novel double-saturation ³¹P MRS protocol enables the ATP hydrolysis rate to be calculated without quantification of the P_i signal, thereby providing a new approach for the study of myocardial ATP turnover rate in vivo (online-only Data Supplement). The ATP hydrolysis rate is important because it reflects the total ATP utilization rate, combining the contractile apparatus utilization and all other ATPase pumps that maintain the structural integrity.

Transplanted hiPSC-VCs Enhanced Activation and Proliferation of Endogenous CPCs for Cardiac Repair

The paradigm of the heart as a postmitotic organ incapable of regeneration has been challenged by the discovery that cardiomyocytes can reenter the cell cycle and begin dividing and by the presence of CPCs in the adult heart that are capable of generating new cardiomyocytes.^{4,5,7,8,40,41} Here, we show that MI led to the activation of c-kit⁺ cells at the injury site, which is consistent with previous reports.^8,42 The higher c-kit⁺ cell density in cell hearts was accompanied by a significantly higher rate of BrdU incorporation (Figure 6C), which suggests that hiPSC-VC transplantation also increased c-kit⁺ cell proliferation. The present data also show that hiPSC-VC transplantation increased the BrdU incorporation rate of BZ cardiomyocytes at week 4 after injury. Collectively, these observations suggest that hiPSC-VC transplantation activates endogenous progenitor cells, which may contribute to the increased myocyte turnover.

In conclusion, the present study introduces a novel MRS-MST approach for obtaining in vivo measurements of the myocardial ATP hydrolysis rate. The ATP hydrolysis rate was linearly related to the rate-pressure product but was significantly reduced in the BZ of infarction, perhaps because of increases in wall stress in this region. Transplantation of hiPSC-VCs significantly increased the activation of endogenous CPCs in response to MI injury, which was accompanied by significant improvements in myocardial wall stress, vascular density, infarct size, and contractile function, as well as the ATP utilization rate. These beneficial effects may impede progression from the compensated phase of myocardial hypertrophy to heart failure.

Supplementary Material

Supplemental Video 1

Supplemental Table 1. ³¹P MST-MRS measurement data on swine skeletal muscle

Supplemental Figure 1. Magnetization saturation transfer (MST) experiments. Panel A, pulse sequence of BISTRO-based magnetization saturation. s_tat and t_ack represent the time of saturation and acquisition, respectively. Panels B–C, RF pulses used for single (red) and double (blue) saturation in the MST experiments. Panels B and C showed the relative amplitude and the phase of the hyperbolic secant-based RF pulses. Panel D showed the response of spin magnetization to RF pulses excitation with various power levels. The single saturation RF pulse (red) resulted in a single excitation frequency (a) whereas the double saturation RF pulse (blue) resulted in two excitation frequencies (a and b). Note that the excitation band b (for Pi saturation) has a wider bandwidth than a in order to compensate the possible change of Pi chemical shift due to change of pH values.

Supplemental Figure 2. ³¹P MRS-MST experiments on swine skeletal muscle. Panels A–C, measurements of apparent T₁s for PCr and Pi (panel A, with saturation on ATPγ) and ATPγ (panel B, with saturation on both PCr and Pi) using inversion-recovery method. The magnetizations were plotted against inversion time and fit with an exponential model to retrieve the corresponding apparent T₁s (panel C). Panel D, steady-state magnetization saturation transfer experiments to measure the rate constants of k_PCr→ATP, k_Pi→ATP and k_ATP,tot based on equations [8] and [17], respectively. d1: ATPγ-saturated spectrum to measure M_ss,PCr and M_ss,Pi. d2: control spectrum without saturation to measure the M_0.PCr, M_0,ATPγ and M_0,Pi. d3: both Pi- and PCr-saturated spectrum to measure the M_ss,ATPγ. Panel E, plot of normalized flux (flux/[ATP]) for total ATP production and utilization. The total ATP production consists of two reactions: PCr→ATP and Pi→ATP, and are measured by conventional MST approach with saturation on ATPγ. The total ATP utilization flux is measured by double saturation on both PCr and Pi. The results from 5 independent measurements showed no statistical difference between the two measurements.

Supplemental Figure 3. Panel A, numerous hiPSC-VCs were detected at 1 WK post surgery by GFP staining (green). Co-staining with BrdU (red) indicated active proliferation of these transplanted cells in vivo. Panel B, transplanted hiPSC-VCs contributed to neo-vascularization by generating new vessels. Arrows indicate vessels stained positive for GFP (green) and human specific CD31 (hCD31, red) inside the fibrin patch.

Supplemental Figure 4. Activation of c-kit⁺ progenitor cells in response to myocardial injury and hiPSC-VCs transplantation. Panels A–F, Representative immuno-staining images from Normal (A and D), infarct area of MI (B and E) and CELL (C and F) hearts at both 1 week (A–C) and 4 weeks (D–F) post surgery. Cardiac progenitor cells were identified by c-kit staining (green), cardiomyocytes were identified by alpha sacromeric actin staining (αSA, yellow), and cycling cells were identified by BrdU incorporation (red). Arrowheads indicate c-kit⁺/BrdU⁺ cells whereas arrows indicate c-kit⁺/BrdU⁻ cells.

Supplemental Video 1. Illustration of an epicardial fibrin patch to deliver the mixture of 4 million hiPSC-VCs on the surface of the LV infarct.

Supplemental Video 2. Supplemental Video 2.

Representative cardiac cine imaging of an aged-matched NORMAL heart.

Supplemental Video 3. Supplemental Video 3.

Representative cardiac cine imaging of an MI heart at 4 weeks post infarction.

Supplemental Video 4. Supplemental Video 4.

Representative cardiac cine imaging of a CELL heart at 4 weeks post infarction.

CLINICAL PERSPECTIVE.

The present study introduces a novel magnetic resonance spectroscopy–magnetization saturation transfer approach for obtaining in vivo measurements of the myocardial ATP hydrolysis rate. The myocardial ATP hydrolysis rate was linearly related to the left ventricular rate-pressure product but was significantly reduced in the border zone of infarction, perhaps because of increases in wall stress in this region. Transplantation of human induced pluripotent stem cell–derived vascular cells significantly increased the activation of endogenous cardiac progenitor cells in response to myocardial infarction injury and was associated with significant improvements in vascular density, infarct size, left ventricular wall stress, and contractile function. Cell therapy also restored the border zone ATP hydrolysis rate to near-normal levels, which may impede progression from the compensated phase of heart disease to heart failure. Because the severity of the reduction in myocardial ATP turnover rate is linearly related to the severity of left ventricular dilatation, and the improvement of myocardial ATP turnover rate is linearly related to the improvement in myocardial contractile performance, the new nuclear magnetic resonance methodology has the potential to add another dimension to our ability to evaluate the prognosis of patients with postinfarction left ventricular remodeling in response to different therapeutic interventions.