Research ArticleCardiology

Platelet-derived growth factor-AB improves scar mechanics and vascularity after myocardial infarction

See allHide authors and affiliations

Science Translational Medicine  01 Jan 2020:
Vol. 12, Issue 524, eaay2140
DOI: 10.1126/scitranslmed.aay2140

Straightening up scars

As a result of myocardial infarction (MI; heart attack), scar tissue forms that negatively affects cardiac function. Thavapalachandran et al. investigated the potential therapeutic effects of intravenous delivery of recombinant human platelet-derived growth factor-AB (rhPDGF-AB) after MI in a randomized, double-blinded study in pigs. Cardiac function was improved 28 days after treatment was initiated and rhPDGF-AB enhanced angiogenesis; however, there was no difference in the size of scars between rhPDGF-AB–treated and vehicle-treated pigs. Scars in rhPDGF-AB–treated pigs showed less heterogeneity and greater fiber alignment, and rhPDGF-AB–treated pigs were less susceptible to arrhythmia after MI. Results support the translational potential of rhPDGF-AB to modulate scars and augment cardiac repair.


Therapies that target scar formation after myocardial infarction (MI) could prevent ensuing heart failure or death from ventricular arrhythmias. We have previously shown that recombinant human platelet-derived growth factor-AB (rhPDGF-AB) improves cardiac function in a rodent model of MI. To progress clinical translation, we evaluated rhPDGF-AB treatment in a clinically relevant porcine model of myocardial ischemia-reperfusion. Thirty-six pigs were randomized to sham procedure or balloon occlusion of the proximal left anterior descending coronary artery with 7-day intravenous infusion of rhPDGF-AB or vehicle. One month after MI, rhPDGF-AB improved survival by 40% compared with vehicle, and cardiac magnetic resonance imaging showed left ventricular (LV) ejection fraction improved by 11.5%, driven by reduced LV end-systolic volumes. Pressure volume loop analyses revealed improved myocardial contractility and energetics after rhPDGF-AB treatment with minimal effect on ventricular compliance. rhPDGF-AB enhanced angiogenesis and increased scar anisotropy (high fiber alignment) without affecting overall scar size or stiffness. rhPDGF-AB reduced inducible ventricular tachycardia by decreasing heterogeneity of the ventricular scar that provides a substrate for reentrant circuits. In summary, we demonstrated that rhPDGF-AB promotes post-MI cardiac wound repair by altering the mechanics of the infarct scar, resulting in robust cardiac functional improvement, decreased ventricular arrhythmias, and improved survival. Our findings suggest a strong translational potential for rhPDGF-AB as an adjunct to current MI treatment and possibly to modulate scar in other organs.


The current treatment paradigm for myocardial infarction (MI) focuses on expedient primary reperfusion to reestablish myocardial oxygen supply. Decreasing symptom-to-reperfusion time has been shown to improve clinical outcomes (1). However, despite timely reperfusion and optimal medical therapy, up to 25% of patients experiencing their first MI will develop heart failure (HF) within 1 year (2). Furthermore, sudden cardiac death (SCD) remains an important clinical problem accounting for up to 40% of total mortality, with SCD rates 10-fold higher in the first 30 days after MI (3). For these reasons, there is an urgent unmet need for novel therapeutics to enhance cardiac repair.

The postinfarction scar is often considered an inert tissue composed primarily of collagen with cross-linked fibers providing tensile strength and resistance to deformation (4). Therefore, conventional pharmacological therapies such as renin-angiotensin-aldosterone system inhibitors and adrenergic receptor blockade have focused on attenuating fibrosis and prevention of adverse remodeling (5). Recently, cardiac scar tissue has been shown to undergo rapid remodeling through fibroblast populations, stabilizing the tissue and preventing cardiac rupture (6, 7). Now, a concept of the infarct scar as a “dynamic” tissue is emerging. Cardiac scar tissue is increasingly being viewed as cellular, vascularized, metabolically active and contractile. In computational and animal models, interventions that increase anisotropy (the presence of highly aligned collagen fibers) improve tissue mechanics and preserve left ventricular (LV) function after MI (8, 9). However, translational efforts to anisotropically reinforce the infarct region to improve LV function have been limited (10, 11).

We recently demonstrated that systemic delivery of recombinant human platelet-derived growth factor-AB (rhPDGF-AB) via an osmotic mini-pump for 5 days after MI in a murine model led to functional cardiac improvement at 1 month (12). Functional improvement was accompanied by activation of fibroblasts, suggesting that the post-MI fibrotic response may be a viable therapeutic target. Angiogenesis was also enhanced, and resolution of the inflammatory response was accelerated. Our present study extends this work by assessing the effects of rhPDGF-AB delivered intravenously at the time of coronary artery reperfusion in a clinically relevant porcine model of myocardial ischemia-reperfusion.

We hypothesized that rhPDGF-AB administered in the acute phase of MI healing would alter fibroblast behavior and, in turn, scar matrix remodeling. To determine the putative therapeutic effects, we used methods similar to clinical trials to obtain appropriate preclinical large-animal data for translation, including clinical grade–randomized and double-blinded study design, gold standard cardiac functional assessment including cardiac magnetic resonance (cMR) imaging, myo-architectural assessment, and catheter hemodynamic and electrophysiology studies. Here, we demonstrate that rhPDGF-AB promotes post-MI wound repair by altering the properties of the infarct scar, resulting in robust cardiac functional improvement, reduced cardiac arrhythmias, and increased survival rates.


rhPDGF-AB improves cardiac function 28 days after MI

Details of the randomized and double-blinded study design are shown in Fig. 1 and described in the Materials and Methods section. Briefly, MI was produced in pigs by angioplasty balloon inflation to occlude the proximal left anterior descending coronary artery, with reperfusion initiated by balloon deflation after 90 min (fig. S1). This simulated the clinically relevant ischemia-reperfusion injury that occurs in patients with MI. Systemic rhPDGF-AB infusion was commenced immediately after coronary artery reperfusion and continued for 7 days via an osmotic mini-pump (ALZET; fig. S2A) loaded with either sterile water as a vehicle or rhPDGF-AB. An enzyme-linked immunosorbent assay (ELISA) at day 2 confirmed serum rhPDGF-AB concentrations were at least 10-fold higher in the rhPDGF-AB–treated group than in the vehicle group (fig. S2B). cMR was used to assess baseline infarct size using late gadolinium enhancement (LGE) and LV cardiac function. We undertook baseline cardiac assessment at an initial time point (day 2), when an infarct scar had formed but the total dose of rhPDGF-AB was minimal, using cMR to ensure there were no baseline differences between the groups. This was followed by imaging of the same animals at day 28 after MI, a time point when rhPDGF-AB had been ceased and there was therefore no potential for direct cardiac effects.

Fig. 1 Study flowchart and randomization.

*Seven days of rhPDGF-AB (65 μg/kg) or vehicle infusion via mini-pump was started immediately after ischemia-reperfusion.

Infarct size, assessed by LGE, was comparable between the vehicle- and rhPDGF-AB–treated groups at day 2 and, if anything, slightly larger in the rhPDGF-AB–treated group (rhPDGF-AB: 20.16 ± 2.24 g versus vehicle: 13.90 ± 2.85 g; P = 0.108) (Fig. 2A). Similarly, LV ejection fraction (LVEF) at day 2 was similar between the groups (rhPDGF-AB: 39.64 ± 2.15% versus vehicle: 40.78 ± 1.2% P = 0.705) (Fig. 2B). However, we found a significant improvement in the LVEF of rhPDGF-AB–treated animals at day 28 compared with vehicle-treated animals (rhPDGF-AB: 51.17 ± 2.31% versus vehicle: 37.33 ± 2.49%; P = 0.002), with an absolute increase in LVEF from baseline of 11.53 ± 3.28% compared with a decrease of 3.46 ± 1.47% in the vehicle group (P = 0.005; Fig. 2, B and C). Sham LVEF was unchanged between time points (day 2: 54.76 ± 2.6% versus day 28: 55.38 ± 3.19%; P = 0.88).

Fig. 2 rhPDGF-AB improves cardiac function 28 days after MI.

(A) Infarct size, as assessed by LGE, day 2 after MI. (B) LVEF of sham, vehicle-, and rhPDGF-AB–treated pigs at days 2 and 28 after MI. (C) Change in LVEF in comparison to baseline (ΔLVEF). (D) Representative magnetic resonance short-axis end-diastolic (top) and end-systolic (bottom) images; red circle, endocardial border. (E) EDV and ESV on day 2 and day 28. n = 5 in sham group; n = 10 in rhPDGF-AB group; n = 6 in vehicle group. All data are presented as mean ± SEM. *P < 0.05 (unpaired t test); ns, nonsignificant.

We evaluated LV dimensions to assess maladaptive LV remodeling. Representative images and quantitative analysis of end-diastolic volume (EDV) and end-systolic volume (ESV) from the two groups are shown in Fig. 2D and table S1. On day 2, LV volumes were comparable between the groups, but at the day 28 end point, ESV was significantly lower in the rhPDGF-AB group (rhPDGF-AB: 1.26 ± 0.1 ml/kg versus vehicle: 1.77 ± 0.15 ml/kg; P = 0.01), whereas EDV was similar between the groups (rhPDGF-AB: 2.56 ± 0.12 ml/kg versus vehicle: 2.83 ± 0.21 ml/kg; P = 0.26) (Fig. 2E). This suggests that rhPDGF-AB administered after MI increases LV contractility, rather than affecting LV dilation during remodeling.

rhPDGF-AB improves myocardial contractility and energetics with minimal effect on ventricular compliance

Despite cMR being the gold standard for cardiac functional assessment, the derived measures remain load dependent. Therefore, we performed pressure-volume (PV) loop studies at the day 28 post-MI end point to determine intrinsic (load independent) cardiac contractility (Fig. 3 and table S2). Steady-state index of systolic function, the maximal rate of pressure change during systole (dP/dtmax), increased with rhPDGF-AB treatment at day 28 (rhPDGF-AB: 1352 ± 48 mmHg/s versus vehicle: 1107 ± 83.3 mmHg/s; P = 0.015) (Fig. 3A), indicating improved cardiac contractility. Contractility assessed during inferior vena cava (IVC) occlusion showed greater contractile function in the rhPDGF-AB group at the day 28 end point, including a steeper slope of the end-systolic PV relationships (ESPVR) (rhPDGF-AB: 2.48 ± 0.2 mmHg/ml versus vehicle: 1.64 ± 0.3 mmHg/ml; P = 0.03) (Fig. 3, B and C) and significantly higher preload recruitable stroke work (PRSW) slope (rhPDGF-AB: 72 ± 5 mmHg versus vehicle: 46 ± 4.9 mmHg; P = 0.004) (Fig. 3D), confirming improved myocardial load-independent contractility with rhPDGF-AB treatment after MI.

Fig. 3 rhPDGF-AB improves myocardial contractility and energetics with little effect on ventricular compliance.

Systolic function in rhPDGF-AB–treated pigs versus vehicle as measured by (A) maximal rate of pressure change during systole (dP/dtmax) at steady state. (B) Representative PV recordings with corresponding end-systolic PV relationships (ESPVR) during IVC occlusion. Slope of the regression curve for (C) ESPVR and (D) preload recruitable stroke work (PRSW) index, metrics of myocardial load-independent contractility. PV diastolic function as assessed by the maximal rate of pressure decay (E) dP/dtmin, (F) Slope of the regression curve for EDPVR, and (G) τ, the isovolumic relaxation time. n = 10 in the rhPDGF-AB group; n = 6 in the vehicle group. All data are presented as means ± SEM. *P < 0.05 (unpaired t test); **P < 0.005 (unpaired t test); ns, nonsignificant.

When ascertaining the effects on relaxation kinetics, we found that the maximal rate of pressure decay (dP/dtmin) was significantly increased in the rhPDGF-AB–treated animals (rhPDGF-AB: −1787 ± 112 mmHg/s versus vehicle: −1328 ± 96 mmHg/s; P = 0.011) (Fig. 3E). However, linear end-diastolic PV relationship (EDPVR) slopes (rhPDGF-AB: 0.39 ± 0.06 mL/mmHg versus vehicle: 0.39 ± 0.04 mL/mmHg; P = 0.93) and tau (time constant of isovolumic LV pressure fall) were comparable between the groups (rhPDGF-AB: 41 ± 1.9 ms versus vehicle: 51 ± 6.5 ms; P = 0.08) (Fig. 3, F and G). Together, these data suggest that the rhPDGF-AB group had improved relaxation without altered LV distensibility at the day 28 end point.

rhPDGF-AB promotes angiogenesis and arteriolar collateralization after MI

Since PDGF-BB (13) treatment after MI and PDGF-AB (14) treatment before MI have been reported to be proangiogenic, we sought to determine the effects of rhPDGF-AB on the post-MI vasculature. Myocardial sections (5 μm thick) immunostained with von Willebrand factor (vWF) and α-smooth muscle actin (α-SMA) were assessed by investigators blinded to the experimental groups. Although the peri-infarct capillary number (vWF+) was significantly increased in the rhPDGF-AB–treated hearts (rhPDGF-AB: 107 ± 6/mm2 versus vehicle: 56 ± 4/mm2; P = 0.001) (Fig. 4A), there was no difference in peri-infarct arteriolar density (α-SMA+) between the groups (rhPDGF-AB: 5 ± 3/mm2 versus vehicle: 5 ± 2/mm2; P = 0.37) (Fig. 4B). Standard two-dimensional (2D) microscopy vessel quantification may be confounded by inadvertently missing vessels out of the plane of section. We therefore used 3D light sheet microscopy to better understand the effects of rhPDGF-AB on arteriolar vasculature. 3D tissue (5 to 10 mm thick) from the peri-infarct zone of experimental animals was cleared using the CUBIC (clear, unobstructed brain/body imaging cocktails and computational analysis) R1a protocol (Fig. 4C), immunostained with α-SMA, and then imaged using light sheet microscopy (Fig. 4D). Individual filament tracing using the IMARIS software was used to identify the origin, branching, and terminal points of arterioles. Quantitative analyses showed that rhPDGF-AB enhanced arteriogenesis (rhPDGF-AB: 1919 ± 231 arterioles per field versus vehicle: 500 ± 321 arterioles per field; P = 0.03) (Fig. 4E). It also promoted arteriolar branching (rhPDGF-AB: 220 ± 42 junctions per field versus vehicle: 23 ± 12 junctions per field; P = 0.03) (Fig. 4F), with reduction in mean arteriolar diameter (15 ± 0.1 μm versus 18 ± 0.4 μm; P = 0.02) (Fig. 4G), suggesting budding and de novo synthesis of microvessels.

Fig. 4 rhPDGF-AB promotes angiogenesis and arteriolar collateralization after MI.

Representative images and quantification of (A) vWF immunostaining, analyzing capillary density in the peri-infarct zone, and (B) arteriolar density (α-SMA+). Scale bar, 50 μm. n = 10 in the rhPDGF-AB group; n = 6 in the vehicle group. (C) Pig heart tissue sections (5 to 10 mm thick) cleared using CUBIC R1a protocol. (D) Representative images of α-SMA–stained vessels using 3D light sheet microscopy and quantification of (E) arteriolar density, (F) branching points, and (G) arteriolar diameter. n = 3 per group. All data are presented as means ± SEM. *P < 0.05 (unpaired t test); **P < 0.005 (unpaired t test); ns, nonsignificant.

rhPDGF-AB promotes collagen fiber alignment and wound healing without reduction in overall infarct scar size or noncardiac fibrosis

Unexpectedly, despite significant improvement in LV function, we observed that treatment with rhPDGF-AB did not reduce scar size as assessed by LGE (Figs. 2, 3, and 5A). We confirmed cMR findings histologically by quantifying the collagen content of the infarct, border, and remote zones using Gӧmӧri trichrome staining (Fig. 5, B and C). As such, no histological difference in scar size was observed between the two groups. The organization of collagen fibers in the scar may be an important determinant of the anisotropic mechanical properties of the scar (9, 11). Hence, to determine collagen fiber organization within the peri-infarct zones of experimental animals, we used second-harmonic generation (SHG) two-photon microscopy (a label-free collagen imaging technique), Gӧmӧri trichrome staining, and polarized light microscopy. We found that rhPDGF-AB treatment resulted in directional organization of collagen fibers (Fig. 5D and figs. S3 to S5). To quantify this observation, the orientation of single collagen fibers relative to one another from representative SHG images was measured. In the rhPDGF-AB group, there were more aligned fibers (P = 0.0094; Fig. 5, E and F), demonstrating increased scar anisotropy.

Fig. 5 rhPDGF-AB promotes collagen fiber alignment and wound healing without reduction in overall scar size.

(A) Representative cine cMR 4 chamber LGE views (scar, white; viable myocardium, black) and quantification. (B) Macroscopic pig hearts (scar, white; viable myocardium, brown). Scale bars, 10 mm. (C) Collagen quantification of scar by Gӧmӧri trichrome staining at day 28 (scar, purple; viable myocardium, pink). Scale bars, 100 μm. n = 6 vehicle and n = 10 rhPDGF-AB. (D) Representative peri-infarct images using second-harmonic generation (SHG) microscopy (red, transmitted SHG; green, reflected SHG) and polarized light imaging to analyze collagen fiber alignment. Scale bars, 50 μm. (E) Quantification of collagen fiber alignment assessed by SHG. Number of collagen fibers within a given angle orientation is represented with the maximum normalized to 0° ± 5°. n = 4 vehicle and n = 6 rhPDGF-AB. (F) Number of collagen fibers oriented to 0° (normalized maximum). n = 4 vehicle and n = 6 rhPDGF-AB. (G) Quantification of Col1a1 and Col3a1 assessed by quantitative real-time polymerase chain reaction (qPCR). Relative gene expression to glyceraldehyde 3-phosphate dehydrogenase (Gapdh) housekeeping gene and normalized against the vehicle group. Log base twofold change data are presented. n = 5 per group. (H) Endogenous tension generation in human mesenchymal cell microtissues with and without rhPDGF-AB treatment. n = 15 to 18 per condition. IZ, infarct zone; BZ, border zone; RZ, remote zone. LGE and collagen data are presented as means ± SEM. Microtissue data are presented as means and 95% confidence intervals. **P < 0.005 (unpaired t test); ***P < 0.0005 (unpaired t test); ns, nonsignificant.

Collagen subtypes are known to fluctuate within the healing infarcted heart. Collagen III expression is increased in the early reparative phase, whereas collagen I is expressed later during the maturation phase (15, 16). We performed quantitative real-time polymerase chain reaction (qPCR) analysis to assess collagen subtypes within the infarct scar core region. We found significantly lower Col1a1 expression in rhPDGF-AB–treated animals (P = 0.038). There was a trend toward lower Col3a1 after the rhPDGF-AB treatment group; however, this was not statistically significant (P = 0.18) (Fig. 5G).

Given the cardiac functional and collagen distribution results detailed above, we hypothesized that rhPDGF-AB may alter the scar environment through effects on mesenchymal stromal cells. To test this, we generated microtissues from human cardiac mesenchymal cells sorted for the PDGFRα+/CD90+/CD31 fraction (fig. S6) (17, 18). rhPDGF-AB treatment increased the rate of tissue condensation and tensile force generation (Fig. 4H). These effects occurred mainly at days 0 to 3, which is consistent with the short duration of rhPDGF-AB treatment (days 0 to 7) after MI in pigs having a sustained benefit at day 28. Together, these data suggest that rhPDGF-AB promotes post-MI wound healing by accelerating scar maturation and contractile force transmission without increased stiffness of the scar matrix.

We performed a thorough necropsy of all animals and did not find evidence of neoplastic or fibrotic “off-target” effects (fig. S7). However, longer-term animal studies focused on safety, together with thorough pharmacodynamic studies, will be required before commencing early-phase clinical trials.

rhPDGF-AB promotes early survival and reduces arrhythmogenicity

In our study, 9 of 36 (25%) animals died during balloon occlusion or reperfusion (before randomization). After randomization, there were six deaths (one in the rhPDGF-AB group and five in the vehicle group) over the course of the study. Postmortem examination revealed that a femoral bleed was the cause of death in two animals (one in the rhPDGF-AB group and one in the vehicle group). All four remaining deaths occurred in the vehicle-treated group, within 1 week of MI. These deaths were classified as SCDs after exclusion at necropsy of other causes such as HF, cardiac bleed, or myocardial wall rupture, suggesting that rhPDGF-AB improved early survival from arrhythmia. The Kaplan-Meier curve shows a 40% improvement in early survival from presumed arrhythmic deaths in the rhPDGF-AB–treated group (log rank, P = 0.04) compared with vehicle (Fig. 6A).

Fig. 6 rhPDGF-AB promotes early survival and reduces arrhythmogenicity.

(A) Kaplan-Meier curve shows that rhPDGF-AB treatment after MI is associated with reduced mortality from SCD at 1 month. (B) Quantification of inducibility of ventricular tachycardia (VT) and number of extrastimuli required to induce VT. (C) Map of myofiber heterogeneity within the scar core and scar border regions. Heterogeneity map: Representative images of myocardial scar tissue heterogeneity (color blue to red indicates increased myofiber clustering) in vehicle- (top) and rhPDGF-AB–treated pigs (bottom). Heterogeneity index: Small values indicate multiple small clusters of myocytes separated by small distances, thus facilitating an increased propensity for cell-cell coupling; large values indicate the opposite, where decreased coupling may facilitate electrophysiological conduction slowing or blockade. n = 6 vehicle and n = 10 rhPDGF-AB. (D) Gӧmӧri trichrome stain of infarcted myocardium in vehicle showing “islands” of preserved myocardium (red) surrounded by scar tissue (blue) compared with a scar in the rhPDGF-AB–treated group. A schematic representation of a hypothetical VT circuit (white dashed line) and exit (arrow) is depicted. Scale bars, 1 mm. Scar heterogeneity data are presented as means ± SEM. *P < 0.05 (unpaired t test).

Programmed electrical stimulation (PES) has been shown to strongly correlate with spontaneous arrhythmias (19). PES was performed in a subgroup of animals at the day 28 end point to identify animals at risk of SCD after MI. PES revealed a significant reduction in inducible ventricular tachycardia (VT) [rhPDGF-AB: 1 of 5 (20%) versus vehicle: 3 of 3 (100%); P = 0.04] associated with a reduced number of extrastimuli required for VT induction (Fig. 6B) in the rhPDGF-AB group. This suggests abatement of late arrhythmogenicity after rhPDGF-AB treatment. The high rates of arrhythmogenicity seen with vehicle-treated animals are consistent with previous observations in porcine models (20, 21).

We further explored these arrhythmic findings by investigating whether differences in the anatomic substrate for ventricular arrhythmias were present between the groups. In rhPDGF-AB–treated animals, the architectural organization of myofibers interspersed with collagen was less heterogeneous (lower heterogeneity index) than in vehicle-treated animals in both scar core (rhPDGF-AB: 0.49 ± 0.01/mm2 versus vehicle: 0.56 ± 0.01/mm2; P = 0.048) and scar border regions (rhPDGF-AB: 0.49 ± 0.01/mm2 versus vehicle: 0.51 ± 0.01/mm2; P = 0.03) (Fig. 6C). Conversely, vehicle-treated hearts showed considerable heterogeneity of putative conducting myofibers colocalized to collagen, which provides the substrate for a reentry circuit and a milieu for ventricular arrhythmias and subsequent SCD (Fig. 6D). Together, these results suggest that rhPDGF-AB decreases ventricular arrhythmias and SCD by decreasing the heterogeneity of post-MI scar composition.


Here, we demonstrate that intravenous infusion of rhPDGF-AB after reperfusion of the occluded coronary artery that caused MI (i) significantly improves LVEF with reduced LV ESV, (ii) increases scar anisotropy (high fiber alignment) without affecting overall scar size or stiffness, (iii) promotes angiogenesis and arteriolar collateralization, (iv) improves cumulative survival by reducing early SCD, and (v) decreases inducible ventricular arrhythmias by attenuating myocardial scar heterogeneity in pigs. LGE on cMR imaging is a robust and clinically validated method for assessment of postinfarct scar size (22). We found that pigs treated with rhPDGF-AB had no reduction in scar size. This effect was observed despite both cMR (functional and volumetric parameters) and micromanometric PV studies indicating significant restoration of cardiac function. Although these results initially seemed contradictory, we hypothesize that better resultant biomechanics of the altered scar and increased myocardial perfusion created by rhPDGF-AB treatment is responsible for the observed cardiac functional improvement.

Infarct healing is regulated by a well-orchestrated inflammatory response that ultimately leads to the formation of scar tissue (23, 24). The initial inflammatory phase of healing is followed by a proliferative phase in which activated myofibroblasts accumulate in the infarct granulation tissue and synthesize extracellular matrix (ECM) proteins (25). The orientation of collagen fibers in the myocardial infarct scar has been shown to greatly influence LV function (811). Furthermore, longitudinal fiber alignment limits scar deformation during systole, with increased circumferential compliance enabling greater diastolic filling and systolic contractility through the Frank-Starling mechanism (9). We acknowledge that as shown by Fomovsky et al. (8), the plane of sectioning and location of sample within the ventricular wall may influence collagen alignment interpretation and in situ mechanics of the scar. Most studies (8, 26, 27) have focused on the circumferential-longitudinal plane, compared with the circumferential-radial plane analyzed in our study. Nevertheless, our results suggest that improved collagen fiber alignment enhances contractile function, since SHG two-photon microscopy, light microscopy, and polarized birefringence revealed collagen disarray in the vehicle group compared with organized collagen fiber alignment after rhPDGF-AB treatment.

Therefore, we propose rhPDGF-AB as a treatment to directly affect regional scar organization and mechanical force transmission through the scar after MI, thus improving LV performance. This possibility is supported by previous studies, which suggest that PDGF signaling regulates fibrous tissue deposition by stimulating fibroblast migration, proliferation, and activation (12, 2830).

Our results show that rhPDGF-AB enhances maturation of collagen III fibers to collagen I fibers in both the infarct and border zones, implying that rhPDGF-AB alters the intrinsic composition and orientation of ECM. We have shown through microconductance PV studies that scar “stiffness” did not significantly change in the rhPDGF-AB–treated pigs. Furthermore, rhPDGF-AB increased tissue condensation and endogenous tension generation in engineered 3D human cardiac mesenchymal cell microtissues. These findings suggest that collagen alignment and cellular tension in the direction of forces acting on the myocardium may enhance LV systolic function without increasing scar tensile strength to a degree that leads to impaired LV diastolic filling.

The formation of infarct neovessels is a critical component of healing. Despite reperfusion of an occluded epicardial artery, microvascular rarefaction may prevent efficient reperfusion of the entire LV (31). Hence, effective wound healing and resultant myocardial functional improvements rely on the formation of a microvascular network capable of supplying the healing infarct with oxygen and nutrients (32). A proangiogenic response was evident using immunofluorescence staining with vWF in both the infarct and border zones of pigs treated with rhPDGF-AB. However, a proarteriogenic response was not evident with 2D immunofluorescence staining. This led us to pursue light sheet microscopy as a powerful imaging modality enabling 3D analysis of arterioles in infarcted tissue. 3D spatial mapping using this tool revealed a notable increase in arteriolar density and branching after rhPDGF-AB treatment. Moreover, the significant decrease in mean arteriolar diameter, coupled with an eightfold increase in branching points in rhPDGF-AB–treated pig hearts, suggests neovascularization through increased vessel sprouting. Our results are in line with previously described effects of PDGF isoforms (pretreatment with PDGF-AB and PDGF-BB) in small-animal models showing increased vascular density postadministration after MI (13, 14). Activation of PDGF receptor–β (PDGFR-β) is also associated with recruitment of mural cells by neovessels, promoting resolution of inflammation and enhancing scar maturation (33). Together, results from our study and others suggest that rhPDGF-AB enhances de novo angiogenesis and arteriogenesis, thereby limiting defective scar formation and preserving cardiac function. This raises the possibility that rhPDGF-AB therapy could efficiently increase perfusion to recovering viable cardiac tissue, thereby salvaging further ischemic myocardium, limiting infarct expansion, and reversing early myocardial hibernation after MI (34).

Unexpectedly, we observed greater early cumulative survival in animals receiving rhPDGF-AB treatment compared with controls. Although animals did not receive ambulatory electrocardiographic monitoring, rigorous necropsies suggested an arrhythmic cause of sudden death in vehicle-treated animals by exclusion of other causes of death (including acute HF, bleeding, and myocardial wall rupture). The mechanism of such early arrhythmic deaths (at 1 to 4 days after MI) is unlikely to be due to the realignment of collagen matrix discussed above, as this likely occurs later during cardiac remodeling. Rather, we hypothesize that the effect of rhPDGF-AB on early ischemia through angiogenesis and clearance of necrotic tissue might be responsible. Polymorphic VT and ventricular fibrillation are more prevalent in the acute-to-subacute phase of MI due to automaticity or triggered electrical activity as a result of these mechanisms (35). Furthermore, there is increasing evidence that fibroblasts may also contribute to cardiac electrophysiology and arrhythmogenesis through connexin-mediated fibroblast-to-myocyte coupling (36). Given the known effects of PDGF signaling on fibroblast activation, this might be another possible mechanism.

In contrast to early arrhythmias, the mechanism for decreased inducible VT at day 28 after MI in the rhPDGF-AB group is likely due to attenuation of myocardial fiber heterogeneity within the scar tissue and organized collagen fiber alignment. This may contribute to improved mechanoelectrical feedback within scar tissue, thus preserving coordinated excitation-contraction coupling (37). Previous studies have shown that heterogenous myocyte resorption and collagen deposition within infarct scars facilitate reentrant circuits for monomorphic VT via this mechanism (35). Furthermore, cell disarray from collagen misalignment can functionally alter conduction velocity of electrical signals, with conduction occurring transverse to the orientation of the myocytes, in addition to the effects imparted by structurally nonconducting barriers (38). Additional detailed electrophysiological mapping studies are required to compare the conduction characteristics of the scar substrate between the treatment groups and to further elucidate the mechanism attributing to arrested development of reentrant ventricular arrhythmias after rhPDGF-AB treatment.

We now have compelling data in two species for the efficacy of rhPDGF-AB to improve cardiac outcomes after MI (12). However, there was a relatively small number of animals in the vehicle group that survived to end point, which was due, in part, to the high mortality rate associated with arrhythmias and the cost associated with large-animal studies.

Using a preclinical porcine model, we showed that rhPDGF-AB improves cardiac function and survival after MI. We also provide new hypothesis-generating data supporting further studies of infarct scar contractility and electrical mechanics. Our findings suggest that there may be strong translational potential for rhPDGF-AB as an adjunctive approach to the current MI standard of care. Because it is possible to target cardiac scar with delivery of rhPDGF-AB, there is also the potential to extend this research into modulation of scar tissue in other organ systems. This raises the exciting possibility for rhPDGF-AB to be used in other diseases in which organ scarring occurs.


Study design

Using a porcine model of ischemia-reperfusion, we used a randomized and double-blinded study design to determine the effect of rhPDGF-AB. All procedures in this study were approved by the Royal North Shore Hospital Animal Ethics Committee (protocol ID: RESP/15/104). Studies were conducted in adult, female Landrace swine (2 to 4 months, 25 to 30 kg). Briefly, animals underwent experimental MI followed by insertion of a mini-pump into the right external jugular vein and randomization to intravenous rhPDGF-AB (n = 11) or vehicle (n = 11) for 7 days. cMR and micromanometer conductance catheter PV loop analyses were conducted at days 2 and day 28 to assess structural and functional changes. Animals were euthanized at day 28. The experimental model is represented schematically in Fig. 1. For all procedures, animals were premedicated with intramuscular ketamine (10 mg/kg), methadone (0.3 mg/kg), and midazolam (0.3 mg/kg). All animals were intubated and mechanically ventilated (Anestar-S, Datascope) at 10 ml/kg tidal volume with supplemental oxygen to maintain an end-tidal CO2 of 35 to 45 mmHg. General anesthesia was induced with intravenous propofol (2 to 5 mg/kg) and maintained with 1 to 2% inhaled isoflurane. All animals were instrumented with a peripheral 24-gauge intravenous catheter, surface electrocardiography electrodes, and a right femoral artery (7F) and vein (8F) vascular access sheaths. Unfractionated heparin was given as a bolus followed by infusion with an activated clotting time of 300 to 400 s.

Experimental model of MI

The left coronary artery was engaged percutaneously via the right femoral artery using a 6F Hockeystick guiding catheter (Medtronic). A 0.36-mm coronary guidewire (Asahi Sion Blue) was delivered into the left anterior descending artery (LAD). MI was induced by inflation of the mid LAD distal to the first diagonal with a 3.0-mm angioplasty balloon (Boston Scientific) for 90 min and angiographic confirmation of balloon occlusion. Coronary angiography performed after reperfusion confirmed vessel patency and resolution of ST elevation.

Mini-pump insertion and randomization to treatment

After balloon deflation and at the time of reperfusion, animals were randomized to receive vehicle (sterile water) or rhPDGF-AB treatment via an intravenous osmotic mini-pump. rhPDGF-AB protein was purchased from Peprotech (cat no. 120-28, lot# 1111S396). rhPDGF-AB or vehicle was delivered at a fixed infusion rate of 65 μg/kg for 7 days via a 2-ml ALZET (Durect) osmotic mini-pump inserted into the right external jugular vein. rhPDGF-AB was measured in the plasma on day 2 using human rhPDGF-AB ELISA kit (Abcam).

Cardiac magnetic resonance imaging

cMR was performed using a 3-T Philips magnetic resonance imaging (MRI) scanner (Ingenia). A short-axis cine stack was acquired using a balanced turbo field echo sequence with the following parameters: time to echo, 1.5 ms; repetition time, 3.0 ms; flip angle, 45°; slice thickness, 8 mm; 0.08-mm overlap; and 30 phases per cardiac cycle with 125% sampling. Endocardial contours were drawn in a semiautomated fashion using cvi42 version 5.2.2 (Circle Cardiovascular Imaging Inc.) according to the Society of Cardiac Magnetic Resonance guidelines (39). Papillary muscles were excluded from the LV volume.

Micromanometer conductance catheterization

Hemodynamic measurements were performed at end point (day 28) before euthanasia. The right femoral artery was cannulated and a 5F micromanometer conductance catheter (Ventri-cath-510S, Millar) that was advanced across the aortic valve and into the LV. Baseline end-systolic and end-diastolic pressures and volumes were recorded. Transient LV preload reduction was achieved by inflation of an 18- to 20-mm balloon catheter (CRE Pulmonary balloon, Boston Scientific) in the IVC. Load-independent measures of cardiac function including slopes of ESPVR, PRSW, and EDPVR were obtained during IVC occlusion. Ves and Vw were calculated as volume-axis intercepts of the ESPVR and PRSW relationships, respectively. The time constant of fall in isovolumic LV pressure (τ) was studied in isovolumic beats after ejection.

PES protocol

PES was performed at 1 month to identify animals at risk of SCD, based on a previously validated PES protocol (40). PES was performed at the right ventricular apex at twice diastolic threshold using programmed stimulation. A drive train (S1S1) of eight beats at 400 ms was followed by up to four extrastimuli delivered one at a time. Initial extrastimuli were delivered at a coupling interval of 300 ms, which was then decreased by 10 ms until ventricular refractoriness. There was no set lower limit for the shortest permissible extrastimulus-coupling interval. Sustained monomorphic VT, defined as cycle length ≥200 ms lasting 10 s (or resulting in hemodynamic instability) induced by ≤4 extrastimuli, was considered a positive result.

Gross examination and histopathological analysis

On day 28, hearts were arrested with potassium chloride (75 to 150 mg/kg) and excised for subsequent analysis. For histological studies, the hearts and tissue samples from the lungs, liver, spleen, kidneys, and ovaries were harvested and fixed in 10% neutral-buffered formalin followed by paraffin embedding.

2D vascular density

Neoangiogenic effects were assessed by immunohistochemical staining to determine capillary density. Briefly, paraffin-embedded sections were deparaffinized, rehydrated, and antigen retrieved in 10 mM sodium citrate buffer with 0.05% Tween-20. Sections were blocked with 5% goat serum in phosphate-buffered saline (PBS)/Tween-20 0.05% (blocking buffer) for 1 hour. Primary antibodies for vWF (1:500, Dako, #A0082) and α-SMA (1:500, Dako, #M0851) were then applied to the sections and incubated for 1 hour at room temperature. After this, sections were washed and stained with appropriate secondary antibodies—goat anti-mouse Alexa Fluor 488 (1:500, Life Technologies, A11029) or goat anti-rabbit Alexa Fluor 594 (1:500, Life Technologies, A11037). After counterstaining with 4′,6-diamidino-2-phenylindole (DAPI; 1 μg/ml), sections were mounted and examined. Each tissue section was digitally scanned (Nanozoomer, Hamamatsu) at 20× objective magnification and imported into ImageJ analysis software. Arteriolar and capillary density were represented as frequency per millimeter squared. Capillaries were defined as vessels stained with vWF; ≥4 DAPI-positive cells in the endothelium and arterioles were considered to be structures containing vWF colocalized with α-SMA.

3D vascular density

Light sheet microscopy was used for a 3D assessment of arteriogenesis. Tissue samples (5 to 10 mm) from the infarct zones were optically cleared using the CUBIC R1a method (41). Briefly, tissue sections were fixed overnight in 4% paraformaldehyde at 4°C and immersed in CUBIC Reagent 1A at 37°C for 3 days, with fresh Reagent 1A replaced daily. Samples were then washed in PBS and blocked in PBS-containing 0.5% Triton X-100 and 10% goat serum overnight at 4°C and then incubated in anti–α-SMA, diluted 1:250 in blocking buffer at 4°C for 5 days with gentle rocking. Tissue was washed three times for 1 hour in PBS and incubated in Alexa Fluor 488 secondary antibody (1:500) for 2 days, washed in PBS for 1 hour, and incubated with DAPI (10 μM) for 1 hour. Samples were then transferred to CUBIC Reagent 2 at 37°C for 24 hours. Samples were mounted on metal pins and imaged using the light sheet microscope (Z.1, Zeiss) while immersed in CUBIC Reagent 2 and imported into IMARIS 9.2.0 for analysis. A total of 10 images per pig were analyzed for arteriolar density, branching points, and diameter using IMARIS quantification. Initially, background noise was reduced through the use of the volume blend tool and applied equally across all samples. After this, the filament trace mode was used using a manual trace and autopath (no loop) setting. The draw tool was set to analyze the cone with an adaptive diameter setting. Correction parameters were enabled to account for vessel diameter and vessel center discrepancies by checking both “automatic center” and “automatic diameter.” Full traces were conducted for each image ensuring that all vessels were accurately traced across the x, y, and z axes. Excel data generated by the program allowed a range of comparisons and statistical analyses. In our case, we analyzed mean diameter, branching points, and vessel numbers.

Histological analysis

To quantify fibrosis, myocardial sections were stained with Gӧmӧri trichrome for collagen, digitally scanned (Nanozoomer, Hamamatsu, Japan) at 20× objective magnification, and imported into an in-house histology analysis software previously described (42). The software was able to differentiate between viable myocardium and collagen within the scar, based on a threshold algorithm using the red and blue coloring of pixels. The quantity of viable myocardium and collagen was calculated as the percentage of red color– or blue color–stained pixels, respectively, expressed per unit area of the myocardial section. Scar heterogeneity, a well-recognized arrhythmogenic substrate for reentrant ventricular arrhythmias, was assessed using a previously validated heterogeneity index (43). Small values in the heterogeneity index indicate multiple small clusters of tissue separated by small distances, and large values indicate the opposite. LV cross sections (fig. S3) were divided into scar core, border, and remote zones. Heterogeneity index results are displayed in the scar core and scar border regions (n = 6 vehicle and n = 10 rhPDGF-AB). A heterogeneity map of local myofiber clustering portrays a visual representation of the intermingling of viable myocytes and collagen, with color blue to red indicating increased clustering, and yellow showing intermediate clustering.

Polarized light microscopy

To assess the collagen fiber alignment of myocardial scar, 10 infarcted hearts were sectioned and stained with 0.1% picrosirius red (44), and their relative scar regions were imaged using polarized light microscopy. This was carried out on an Olympus VS120 Slide Scanner fitted with a polarized light filter.

Imaging of collagen fibers and collagen fiber angle analyses

Paraffin-embedded LV samples from infarcted regions of vehicle-treated and rhPDGF-AB–treated pigs were sectioned (4 μm) and then stained with picrosirius red to identify the myocardial scar. Collagen fiber alignment was assessed in the scar core region of the mid LV wall by a microscopist blinded to the sample treatment. All samples used for analyses were obtained approximately 20 mm from the apex and within the anterior LV scar core (fig. S3). This region was then imaged using a two-photon microscope (Leica TCS SP8 MP), and collagen was detected using SHG. The degree of orientation for collagen fibers from the SHG microscopy fields of view (FOVs) was analyzed using the angle tool on the ImageJ software. Assessors were blinded to sample treatments. Forty collagen fibers were measured per FOV, and three FOVs were analyzed per sample. Angle measurements were then binned into 5° bins [number of fibers between 0° and 5°, number of fibers between 5° and 10°, etc. (fig. S5A)] to produce histograms representing collagen fiber orientation (fig. S5B). Data were normalized by taking the maximum peak and shifting it to the 0° orientation while simultaneously shifting all other values by the equivalent amount (fig. S5C). This normalization was performed to account for variations in orientation of samples/slide. Technical replicates (120 fibers from three FOVs) were then combined followed by analysis of the arithmetic mean and standard deviation of all biological replicates (Fig. 5E; vehicle n = 4 and rhPDGF-AB n = 6). Unpaired two-tailed t tests were performed to compare the number of fibers oriented to 0°, representing the maximum number of fibers oriented in one direction for all biological replicates (Fig. 5F).

qPCR analysis

To quantify Col1a1 and Col3a1 collagen subtypes, qPCR was performed on frozen heart sections. Total RNA was collected from 25-μg scar core using the Isolate II RNA mini kit (Bioline). Complementary DNA was synthesized from 500 ng of RNA using reverse transcriptase enzyme and primers from Promega (table S3). qPCR was performed using SensiFAST SYBR No-ROX Mix (Bioline) run on the cycler CFX-384 (Bio-Rad). PCR conditions were 95°C for 2 min and 40 cycles of 95°C for 5 s, 60°C for 10 s, and 72°C for 15 s, followed by melt curve analysis. Analysis was performed with Bio-Rad CFX Manager 3.1 software. Relative gene expression was calculated using the ΔΔCt method, relative to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) housekeeping gene and normalized against vehicle group. Log base twofold change data are presented with error bars representing SEM. Statistical analysis was performed on the ΔΔCt raw data (45).

3D human cardiac mesenchymal cell microtissues

Human cardiac PDGFRα+ cells were used to make 3D cardiac mesenchymal cell microtissues. Heart-Dyno (a 96-well device used for functional screening of cardiac organoids) culture inserts were fabricated using standard SU-8 photolithography and polydimethylsiloxane (PDMS) molding practices (46). For each tissue, a 3.2-μl mixture, containing 32,000 cells, collagen I (3.3 mg/ml), and 22% (v/v) Matrigel (growth factor reduced), was prepared on ice and pipetted into the cell culture inserts. The mixture was then gelled at 37°C for 75 min before the addition of medium (150 μl per tissue) with a complete media change 2 days later. rhPDGF-AB (100 ng/ml) or vehicle was added to the wells daily from day 0 to day3. The Heart-Dyno design facilitates the self-formation of tissues around in-built PDMS exercise poles (designed to deform ~0.07 μm/μN). The tissues were imaged on days 0, 1, 2, 3, and 5 using a Nikon confocal microscope at 37°C, and the tissue width and distance between the poles were measured using ImageJ. The endogenous tension was calculated by the difference in the distance between the poles relative to day 0 in micrometers multiplied by 14 μN/μm.

Statistical analysis

Continuous data are presented as mean ± SEM. Organoid data are presented as mean and 95% confidence intervals. Normality was assessed using the Shapiro-Wilk test. Statistical comparisons were performed by unpaired Student’s t test or one-way analysis of variance (ANOVA) followed by the Holm-Sidak post hoc test to adjust for multiple comparisons. Survival analyses were performed using the Kaplan-Meier method, and the log-rank test was applied to determine significance between overall survival between the two groups. P values <0.05 were considered statistically significant. All analyses were performed using the SigmaPlot 12.5 software. All individual subject-level data are reported in data file S1.


Fig. S1. Fluoroscopy and electrocardiograph during balloon inflation.

Fig. S2. Mini-pump insertion and rhPDGF-AB ELISA.

Fig. S3. Schematic representation of cardiac tissue source for collagen alignment studies.

Fig. S4. Representative images of full-thickness LV wall.

Fig. S5. Method of histogram normalization of collagen fiber angle data.

Fig. S6. Engineered 3D human cardiac mesenchymal cell microtissues.

Fig. S7. Necropsy histology.

Table S1. Cardiac magnetic resonance imaging data.

Table S2. PV studies data.

Table S3. Primer sequences used for qPCR analysis.

Data file S1. Individual subject-level data.


Acknowledgments: We are grateful to the Kearns animal facility and radiographers from Royal North Shore Hospital, S. Igoor (Center for Heart Research, Westmead), and O. Wroth for the assistance. We acknowledge the facilities and the scientific and technical assistance of the Westmead Institute for Medical Research Core Facilities Hub (histology and microscopy) and Microscopy Australia at the Australian Centre for Microscopy and Microanalysis at the University of Sydney. We thank the Children’s Medical Research Institute for providing access to the light sheet microscopy equipment. This work was performed in part at the University of Queensland node of the Australian National Fabrication Facility, a company established under the National Collaborative Research Infrastructure Strategy to provide nano- and microfabrication facilities for Australia’s researchers. Funding: The study was funded by the National Heart Foundation Australia (NHF) Cardiovascular Research Network Grant (100711) and by National Health and Medical Research Council (NHMRC) project grants (1126277 and 1100046). J.J.H.C. was supported by a Future Leader Fellowship (100463) from the NHF Australia and a Sydney Medical School Foundation Fellowship. S.T. was supported by a Westmead Medical Research Foundation Scholarship (The Stephen and Barbara Penfold PhD Scholarship) and a cofunded postgraduate scholarship from the NHF Australia (101108) and NHMRC (1114472). S.M.G. acknowledges the support of the Parker Hughes Bequest. J.E.H. acknowledges funding from QIMR Berghofer Medical Research Institute, project support from NHF Australia, and fellowships and grants from the NHMRC. S.C. was supported by a scholarship from Heart Research Australia. R.P.H. received funding from the Australian Research Council Strategic Initiative in Stem Cell Science (110001002), NHMRC Program Grant (1074386), NHMRC Australia Fellowship (573705), NHMRC Senior Principal Research Fellowship (1118576), Foundation Leducq Transatlantic Network of Excellence in Cardiovascular Research (13CVD01, 15CVD03), and the Australia India Strategic Research Fund (BF020084, BF050024). Author contributions: J.J.H.C. conceived and designed the experiments. S.T., R.P.D., A.M.B., P.B., J.L., and T.Y.L.L. performed large-animal experiments and data analysis. S.M.G., G.A.F., and S.C. performed cardiac MRI studies and analyzed the data. R.D.H., J.P., K.R., F.N.R., S.T., and T.Y.L.L. performed the histological analyses. J.E.H., R.J.M., and E.T. performed the cardiac organoid experiments and data analysis. M.P.F. and S.H.K. analyzed the PV data. S.T., J.J.H.C., R.P.H., R.M.G., S.M.G., J.E.H., M.P.F., E.K., T.Y.L.L., P.K., J.P., L.T., A.R.D., N.S.A., and M.X. contributed to the interpretation of the results and writing of the manuscript. Competing interests: J.J.H.C. and R.P.H. are inventors on PCT 2019/050617 filed by the University of Sydney that covers “cardiac treatment.” J.E.H. and R.P.M. are inventors on PCT patent 2018/0335574A1 filed by the University of Queensland for the 3D tissue culture device. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.

Stay Connected to Science Translational Medicine

Navigate This Article