Wnt10b Gain-of-Function Improves Cardiac Repair by Arteriole Formation and Attenuation of Fibrosis

Abstract

Rationale

Myocardial infarction causes irreversible tissue damage, leading to heart failure (HF). We recently discovered that canonical Wnt signaling and the Wnt10b ligand are strongly induced in mouse hearts after infarction. Wnt10b regulates cell fate in various organs, but its role in the heart is unknown.

Objective

To investigate the effect of Wnt10b gain-of-function on cardiac repair mechanisms and assess its potential to improve ventricular function after injury.

Methods and Results

Histological and molecular analyses showed that Wnt10b is expressed in cardiomyocytes and localized in the intercalated discs of mouse and human hearts. After coronary artery ligation or cryoinjury in mice, Wnt10b is strongly and transiently induced in peri-infarct area cardiomyocytes during granulation tissue formation. To determine the effect of Wnt10b on neovascularization and fibrosis, we generated a mouse line to increase endogenous Wnt10b levels in cardiomyocytes. We found that gain of Wnt10b function orchestrated a recovery phenotype characterized by robust neovascularization of the injury zone, less myofibroblasts, reduced scar size, and improved ventricular function compared to wild-type mice. Wnt10b stimulated expression of Vascular Endothelial Growth Factor Receptor 2 (Vegfr-2) in endothelial cells and Angiopoietin-1 in vascular smooth muscle cells through NF-κB activation. These effects coordinated endothelial growth and smooth muscle cell recruitment, promoting robust formation of large, coronary-like blood vessels.

Conclusion

Wnt10b gain-of-function coordinates arterial formation and attenuates fibrosis in cardiac tissue after injury. Because generation of mature blood vessels is necessary for efficient perfusion, our findings could lead to novel strategies to optimize the inherent repair capacity of the heart and prevent the onset of HF.

INTRODUCTION

Coronary heart disease leading to myocardial infarction (MI) is the major cause of death in both men and women. Each year about 735,000 people in the U.S. have an MI and suffer tissue damage, often leading to ventricular remodeling, hypertrophy, dilatation and heart failure (HF).^1,2

Cardiomyocytes within the infarct area die within minutes after coronary artery occlusion.^3,4 Toxic products and alarm signals released from dying cells induce cell adhesion proteins, cytokines and chemokines to recruit inflammatory cells, which remove cellular debris and degrade extracellular matrix.^5–7 After debris is cleared, usually within 2–3 days after MI, the gap is filled with granulation tissue composed of proliferating myofibroblasts that secrete extracellular matrix proteins and endothelial cells that form new capillaries.^6–8 Two to three weeks after the original MI, the infarcted tissue begins to mature into a dense scar.⁸ Despite considerable progress to restore circulation to infarcted areas using thrombolytics and percutaneous interventions, we currently have no pharmaceutical agents to modulate endogenous cardiac repair processes with the goal of preserving left ventricular (LV) function and preventing HF.

We and others have demonstrated that canonical Wnt signaling is activated in myofibroblasts and endothelial cells during granulation tissue formation following experimental MI in mice, suggesting that the Wnt pathway plays a role in both angiogenesis and fibrosis.^9–13 Our results further showed that Wnt10b, a canonical Wnt ligand, is induced after MI.¹² Although previous studies in animal models have shown that modulation of Wnt antagonists such as Sfrp1 and intracellular downstream targets such as β-catenin can influence the wound healing process after MI, the effects of canonical Wnt ligands such as Wnt10b on cardiac tissue repair have not been investigated.^9,10,14,15

To determine the effects of Wnt10b on the cardiac repair process, we analyzed the spatial and temporal patterns of Wnt10b activation in the heart before and after injury. We discovered that in normal human and mouse hearts, Wnt10b is expressed in cardiomyocytes and stored in the intercalated discs. Following ventricular tissue injury in mice, Wnt10b expression is strongly and transiently up-regulated in cardiomyocytes around the injury, peaking during granulation tissue formation. Our results show that Wnt10b gain-of-function in cardiomyocytes leads to i) increased angiogenesis both directly by stimulation of endothelial cell proliferation and by induction of Vegfr2 expression, ii) formation of coronary-like vessels surrounded by vascular smooth muscle cells (vSMCs) via induction of Ang-1 in vSMCs, iii) cardiomyogenesis within scar tissue and iv) decreased numbers of myofibroblasts and infiltrating inflammatory cells. Furthermore, we show that the mechanisms underlying arteriole formation by Wnt10b depend on coordinated stimulation of the NF-κB signaling pathway in endothelial cells and vSMCs. These findings suggest Wnt10b has the potential to orchestrate and improve the natural reparative capacity of the heart.

METHODS

A complete Methods section is available in the Online Data Supplement.

RESULTS

Wnt10b is expressed in cardiomyocytes and localized in the intercalated discs

To identify the cell types expressing WNT10B protein in the adult human heart, we stained tissue sections from healthy heart donors with an antibody recognizing the N-terminal domain of WNT10B. Immunofluorescence (IF) analysis showed that WNT10B was localized in the intercalated discs of cardiomyocytes, co-localizing with β-catenin (CTNNB1) and Connexin 43 (GJA1; Figure 1A). In contrast, we detected abundant cytoplasmic staining and WNT10B accumulation along the lateral borders of cardiomyocytes in ventricular tissue from ischemic cardiomyopathy patients (Figure 1B). Using a distinct antibody against the C-terminal protein domain of Wnt10b,¹⁶ we found similar accumulation in the intercalated discs of the mouse heart (Fig 1C). Western blot analysis of protein samples prepared from freshly isolated cardiomyocytes confirmed myocyte-specific Wnt10b expression (Online Figure I).

Figure 1. Wnt10b is expressed in cardiomyocytes and induced in chronic and acute cardiac injury.

(A) Localization of WNT10B protein in the normal adult human ventricular tissue. IF for WNT10B (red), β-catenin (CTNNB1, top, green) and connexin-43 (GJA1, bottom, green) illustrates localization of WNT10B in the intercalated discs of cardiomyocytes. Scale bars, 20 μm. (B) Aberrant accumulation of WNT10B (red) in the cytoplasm of cardiomyocytes (arrows) in ischemic cardiomyopathy (ICM) patients (right), in addition to its localization in the intercalated discs (arrowheads). Normal human heart (left) shown as control. Collagen IV (COL4AI, green) antibody staining of basal membranes outlines borders between cardiomyocytes. Scale bars, 20 μm. (C) Localization of Wnt10b in cardiomyocytes of normal adult mouse ventricular tissue. IF for α-Actinin (Actn2, top, green) and plakoglobin (Jup, bottom, green) shows localization of Wnt10b (red) in the intercalated discs. Scale bars, 20 μm. (D) Wnt10b and Tgf-β1 mRNA expression by qPCR analysis at sequential time points post experimental myocardial infarction (MI) in mouse hearts. Wnt10b levels peak at day 7 after MI, during granulation tissue formation. * P < 0.05; ** P < 0.01; *** P < 0.001. One-way ANOVA with Dunnett’s multiple comparisons test. N≥3 for all time points. All data are means ± SEM. (E) Wnt10b remains associated with cardiomyocyte junctions in distal, non-infarcted, areas of mouse ventricle 7 days post MI. (F) Wnt10b expression (red) is induced and becomes pervasive in the cytoplasm of cardiomyocytes (stained in green for Actn2) in the border zone of mouse hearts 7 days post MI. Low (top) and high (bottom) magnification of cardiac tissue is shown. Left bottom panels depict boxed areas on top. Scale bars, 100 μm. BZ=border zone, INF=infarct tissue. All tissue sections were counter-stained with DAPI (blue) to mark cellular nuclei.

Wnt10b is induced in cardiomyocytes at the infarct border during granulation tissue formation

To determine the spatio-temporal pattern of Wnt10b expression in the heart after injury, we induced MI in C57BL/6 adult mice by permanent ligation of the left coronary artery and analyzed whole heart RNA samples at specific time points of the cardiac tissue repair process. Specifically, we quantified Wnt10b levels during the inflammatory response phase (day 1 to 3 after injury), granulation tissue formation (i.e., during neovascularization and fibrosis after day 4) and in mature scar at day 21 in comparison to baseline levels. Our results showed Wnt10b RNA levels began to rise around day 3 post MI, peaking at day 7 by 6–8 fold, but returning to baseline levels during scar maturation (Figure 1D). Wnt10b peak levels followed the induction of TGFβ1 that is known to promote granulation tissue formation and fibrosis (Figure 1D).¹⁷

To identify the location of Wnt10b induction in the heart after injury, we analyzed mouse cardiac tissue sections on day 2 and 7 post MI. While little to no change in Wnt10b protein localization was detected at day 2 post MI (Online Figure I), we observed strong induction of Wnt10b protein specifically in the myocytes of the infarct border zone at day 7 post MI (Figure 1E and 1F). In addition to the intercalated disc localization in normal hearts or in cardiomyocytes remote from the infarct, Wnt10b accumulated prominently in the cytoplasm of border zone cardiomyocytes (Figure 1E and 1F).

Taken together, our data show that Wnt10b protein localizes in the intercalated discs of cardiomyocytes in the normal adult heart. This pattern is disturbed in ischemic cardiomyopathy patients with cytoplasmic accumulation of WNT10B. Moreover, Wnt10b protein is strongly increased in peri-infarct cardiomyocytes during the neovascularization and fibrosis phase of cardiac repair after experimental MI.

Wnt10b gain-of-function promotes functional recovery after injury

To determine whether Wnt10b gain-of-function is beneficial or detrimental to cardiac repair, we genetically enhanced the natural Wnt10b expression in cardiomyocytes by generating a transgenic mouse line carrying the entire Wnt10b coding sequence under the adult cardiomyocyte-specific alpha Myosin Heavy chain (αMHC or Myh6) gene promoter (Figure 2A).¹⁸ Molecular analyses of ventricular tissue showed that αMHC-Wnt10b transgenic (TG) mice expressed higher levels of Wnt10b RNA and Wnt10b protein in the heart compared to wild-type (WT) controls; however, Wnt10b protein in TG was primarily localized in the intercalated discs, as in WT cardiomyocytes (Figure 2B and Online Figure II). Wnt10b overexpression did not cause changes in the ventricles of TG mice at the macroscopic or ultrastructural level, but it led to enlargement of both atria (Online Figure II). TG mice displayed normal ventricular dimensions and functional parameters compared to WT controls (Online Table I). Flow cytometry of primary non-myocyte cells isolated from adult WT and TG ventricles also showed comparable numbers of immune, endothelial and stromal cells (Online Figure III).

Figure 2. Wnt10b gain-of-function improves ventricular function after injury.

(A) Schematic drawing of the αMHC-Wnt10b transgene (TG). (B) Western blot analysis of protein samples isolated from total cardiac tissue shows higher Wnt10b protein levels in TG hearts compared to wild types (WT). β-actin served as loading control. Molecular weights are indicated in kilodaltons (kDa). (C) Fractional shortening percentage changes at 1 day, 1 week and 3 weeks after cryoinjury show that cardiac function recovers over time in TG mice, whereas it deteriorates in WT controls. WT N=10, TG N=11. (D) Representative M-mode echocardiograms at baseline and 3 weeks after cryoinjury demonstrate improved ventricular function in TG mice compared to WT counterparts 3 weeks after injury. (E) Percentage changes in left ventricular internal dimension in systole (LVIDs) and diastole (LVIDd) 3 weeks post cryoinjury, relative to baseline values, indicate functional recovery in TG mice is primarily due to improved systolic function. WT N=10, TG N=11. (F) mRNA analysis by qPCR at baseline (arbitrarily set as value 1 in WT) and 7 days post cryoinjury shows that expression of heart failure predictor genes is attenuated in TG mouse hearts compared to WT. BL=baseline, dpc=days post cryoinjury. N=3–6 mice/group. (G) Masson’s trichrome staining of ventricle sections 4 days and 3 weeks after cryoinjury (left) and quantification of scar size as percentage of horizontal ventricle area (right). TG and WT mice display similar size of original injury. 3 weeks later, scar size in TG hearts is reduced by approximately 2-fold compared to WTs. N=4 mice per group. (H) Immunohistochemistry of Col1A in injured tissue of WT and TG ventricles 3 weeks post cryoinjury shows reduced collagen deposits in TG (left). Boxed areas are shown at higher magnification. Scale bars, 200 μm. Quantification of relative fibrotic density (arbitrarily set as value 1 in WT) based on Col1A staining (right). N=4 mice per group. (I) Comparison of Actn2 staining for cardiomyocytes in WT and TG ventricles 7 days post cryoinjury in low magnification shows the appearance of cardiomyocyte patches in TG. Injury areas between epicardium and healthy tissue are delineated with dashed lines. (J) Actn2 staining in WT and TG ventricles 7 days post cryoinjury in high magnification. Images from 3 distinct TG mice display abundant Actn2 staining within scar tissue. Inset (far right) shows sarcomeric structures in Actn2⁺ cells. Scale bar, 200 μm; inset, 50 μm. INJ=injured tissue, Non-INJ=non-injured tissue. * P < 0.05; ** P < 0.01; NS, not significant, unpaired t test. All data are means ± SEM.

Because the enlarged atria of TG mice hindered ligation of coronary vessels, we assessed the effects of Wnt10b overexpression on cardiac tissue repair using the cryoinjury model.^19,20 Histological and molecular analyses of Wnt10b expression after cryoinjury showed induction at day 7 after injury that was confined to cardiomyocytes in the injury border zone, as observed in the MI model, suggesting that both spatial and temporal Wnt10b expression patterns are comparable in the two injury models (Online Figure IV).

Cardiac functional parameters were measured by echocardiography at baseline before injury, and at 1 day, 1 week and 3 weeks after injury (Figure 2C–E). The results showed that the extent of the initial injury was similar in WT and TG mice as demonstrated by similar drops in fractional shortening (FS) 1 day after injury. However, whereas the function of WT hearts continued to deteriorate over time, left ventricular functional output of TG hearts improved 3 weeks after injury (Figure 2C). Functional recovery in TG mice was primarily due to preservation of systolic function, with preserved chamber size and contractility (Figure 2D and 2E). Improved cardiac function in TG mice matched attenuated induction of prognostic markers of HF, such as Matrix Metalloproteinase-9 (Mmp9)^21–23 and atrial and brain natriuretic peptides (Nppa, Nppb),^22,24,25 when compared to WT counterparts (Figure 2F).

Consistent with the echocardiography data, histological analysis demonstrated that the original injury size was comparable in WT and TG mice. By 3 weeks post injury, however, TG ventricles displayed reduced scar than WT counterparts, with attenuated deposition of collagen 1A (Figure 2G and 2H). The injured tissue in TG ventricles instead contained numerous patches of α-Actinin⁺ myocytes with developing sarcomeric structures (Figure 2I and 2J). Together, our data suggest Wnt10b has a favorable effect on cardiac repair, leading to reduced scar size, cardiomyocyte generation, improved systolic function and reduction in the expression of HF markers.

Wnt10b overexpression increases neovascularization of scar tissue

To test the effect of Wnt10b on cardiac tissue repair, we compared scar neovascularization and myofibroblast accumulation between WT and TG mice after cryoinjury. Histological and flow cytometry analyses showed that ventricles of uninjured WT and TG mouse hearts contained comparable numbers of endothelial cells (Online Figure V). After injury, IF analysis using the CD31 antibody to mark endothelial cells showed the presence of small and sparse microcapillaries within the injured tissue of WT hearts, as previously reported.²⁰ In contrast, the corresponding injured tissue in TG ventricles contained more vascular beds than WT, organized in dense vascular regions or as large-diameter blood vessels (Figure 3A). Vascular density of tissue away from the site of injury was comparable in WT and TG mice, whereas the injured tissue of TG ventricles was approximately 2-fold more densely vascularized than the corresponding area in WT counterparts (Figure 3A). Consistent with the histology data, flow cytometry analysis of dissociated total ventricular tissues showed TG hearts to contain approximately 30% more endothelial cells (CD31⁺αSMA⁻) than WT (Figure 3B, Online Figure V).

Figure 3. Wnt10b gain-of-function promotes neovascularization and attenuates myofibroblast expansion during cardiac repair after injury.

(A) Panoramic IF images of CD31 antibody staining (green) in the injury (INJ) and peri-injury (Non-INJ) zones of WT and TG mouse cardiac tissue sections 3 weeks after cryoinjury (left panels). Injury zones between epicardium and non-injured tissue are delineated with dashed lines. The injury zone in WT mice contains mostly micro-capillaries, whereas in TG mice it contains numerous large blood vessels (arrowheads). Scale bars, 200 μm. Quantification of vascular density in non-injury and injury cardiac sections (right) shows scar tissue in TG mice is approximately 2-fold more densely vascularized than WT controls (right). N=3 mice per group. (B) Number of CD31⁺αSMA⁻ endothelial cells, (C) CD140a⁺CD31⁻ fibroblasts, (D) αSMA⁺CD31⁻ myofibroblasts, and (E) αSMA⁺Col1A⁺ myofibroblasts among non-cardiomyocyte CD45⁻ primary cells isolated from WT and TG ventricles, determined by flow cytometry. Cell population numbers from WT ventricles at baseline shown as reference. TG hearts contain higher endothelial cell, equal fibroblast and lower myofibroblast numbers compared to WT 5 days after cryoinjury. BL=baseline, dpc=days post cryoinjury. N=6 mice per group. (F) IF analysis with antibodies recognizing Ki-67 (red) and CD31 (green) shows higher numbers of proliferating endothelial cells (arrows) in the injured tissue of TG ventricles compared to WT, 7 days after cryoinjury. Right panels show magnified images from boxed regions on the left and middle panels. Sections were counter-stained with DAPI (blue) to mark cellular nuclei. Scale bar, 100 μm; insets 50 μm. (G) Ratio of Ki-67⁺CD31⁺ proliferating endothelial cells to Ki-67⁺CD31⁻ proliferating stromal cells in WT and TG ventricles 5 days post cryoinjury, determined by flow cytometry. N=3 mice per group. ** P < 0.01; *** P < 0.001; NS, not significant, unpaired t test. All data are means ± SEM.

We found Wnt10b overexpression had the opposite effect on the number of myofibroblasts generated after injury (Figure 3C–E). Flow cytometry showed ventricles of uninjured WT and TG mice contained comparable numbers of CD31⁻CD140a⁺ fibroblasts (Online Figure V). After injury, the number of total fibroblasts in WT and TG mice remained comparable. We observed a significant increase in the number of myofibroblasts in both WT and TG ventricles as expected, though TG ventricles contained approximately half the number of αSMA⁺, collagen-producing myofibroblasts compared to WT (Figure 3C–E, Online Figure V).

IF analysis with a nuclear proliferation marker Ki-67 showed the majority of proliferating cells within the injured tissue of TG ventricles were endothelial cells (Figure 3F). In contrast, most Ki-67⁺ cells within the injured tissue in WT ventricles were non-endothelial cells (Figure 3F, Online Figure V). Flow cytometry quantified that the ratio of endothelial to non-endothelial Ki-67⁺ cells was 2.5 higher in TG mice compared to WT counterparts (Figure 3G). These data indicate that Wnt10b promotes proliferation of endothelial cells at the expense of non-endothelial cells, which in the case of the injured tissue are mostly myofibroblasts.⁸

Wnt10b gain-of-function in cardiomyocytes induces canonical Wnt signaling in adjacent endothelial cells, increasing their proliferative capacity

Analysis of RNA from TG ventricles prior to injury showed Wnt10b gain-of-function led to up-regulation of canonical Wnt signaling target genes Axin2, Lef1 and Tcf7 (Figure 4A).^26–28 To identify the cell types with canonical Wnt activation in TG hearts, we stained cardiac tissue sections with an antibody recognizing the canonical Wnt signaling molecule β-catenin. IF analysis showed accumulation of β-catenin in endothelial cells, but not in cardiomyocytes, indicating Wnt10b overexpression activated canonical Wnt signaling in endothelial cells in a paracrine fashion (Figure 4B). Furthermore, transfection of freshly isolated cardiac endothelial cells with canonical Wnt activity luciferase reporter constructs confirmed higher canonical (i.e., β-catenin-mediated) Wnt signaling in cells from TG mice compared to those from WT mice (Figure 4C).

Figure 4. Wnt10b gain-of-function activates canonical Wnt signaling and increases proliferative capacity in resident cardiac endothelial cells.

(A) qPCR analysis using ventricular RNA samples shows levels of canonical Wnt target genes Axin2, Lef1 and Tcf7 are increased in TG hearts compared to WT controls. N=3-4 mice per group. (B) IF images of CD31 (green), β-catenin (red), and DAPI staining (blue) in cardiac tissue sections. β-catenin is present in the intercalated discs of WT and TG cardiomyocytes (arrows), but shows additional staining in endothelial cells (arrowheads) of TG mice. Scale bar, 20 μm. (C) Luciferase reporter assay using the canonical Wnt reporter SuperTOPFlash in primary cardiac endothelial cells isolated from WT and TG ventricles shows TG endothelial cells have activated canonical Wnt signaling. Results obtained in independent cell isolation experiments from 3 mice per group. (D) Primary cardiac endothelial cells (ECs) were isolated from WT and TG ventricles, seeded, grown in culture for 6 days and counted. Endothelial cells isolated from TG ventricles show higher growth than WT. Results obtained in independent cell isolation experiments from 3 mice per group. (E) Images of isolated endothelial cells from WT and TG hearts at day 5 in culture. Endothelial cell cultures from TG mice are denser than those from WT controls. (F) Doubling time of cardiac primary endothelial cells isolated from TG mice is 2-fold higher than the corresponding WT controls. N=3 mice per group. (G) Tube formation assay on GFR Matrigel shows cardiac endothelial cells from TG ventricles form more tubes than WT controls. (H) Cardiac endothelial cells from TG ventricles were cultured with canonical Wnt inhibitor (IWR-1-endo, 1 μM, top right) and its negative control counterpart (IWR-1-exo, 1 μM, top left) as well as with JNK inhibitor II (1 μM, bottom right) and its negative control counterpart (1 μM, bottom left) during the entire duration of culture. Canonical Wnt inhibition blocks endothelial cell growth, but JNK inhibition has no effect. (I) Quantification of endothelial cells for each group (in panel G) counted at day 5 of culture. Cells isolated from WT ventricles served as baseline control. Results obtained in independent cell isolation experiments from 3 mice per group. * P < 0.05; ** P < 0.01; NS, not significant, unpaired t test. All data are means ± SEM.

To test whether canonical Wnt activity promotes the proliferative capacity of endothelial cells, we purified and cultured freshly isolated endothelial cells from WT and TG ventricles (Online Figure VI). Our data show endothelial cells from TG ventricles possessed higher growth potential in culture and branching morphogenesis on Growth Factor Reduced (GFR) Matrigel than endothelial cells isolated from WT ventricles (Fig 4D–G). This effect was abolished when endothelial cells were treated with the small molecule canonical Wnt inhibitor IWR-1-endo,²⁹ but not the JNK inhibitor SP600125,³⁰ demonstrating that the high growth potential of cardiac endothelial cells isolated from TG ventricles was due to elevated canonical Wnt activity (Figure 4H and 4I).

Wnt10b enhances endothelial cell growth both directly and through induction of Vegfr-2 expression

To test whether the high growth potential of endothelial cells isolated from TG ventricles was a direct effect of Wnt10b on endothelial cells, we isolated cardiac microvascular endothelial cells from WT mice and exposed them to increasing concentrations of recombinant Wnt10b protein, which induced canonical Wnt activity (Online Figure VII). Wnt10b protein directly promoted endothelial cell growth in culture, suggesting canonical Wnt activity induction and stimulation of endothelial cell proliferation was a direct effect of Wnt10b (Figure 5A, Online Figure VII).

Figure 5. Wnt10b stimulates endothelial cell growth directly and in synergy with VEGF.

(A) Growth of primary cardiac endothelial cells, CD31⁺CD140a⁻, isolated from 4 week-old WT ventricles in presence of PBS as vehicle control or Wnt10b protein (300 ng/ml). At day 6, cultures treated with Wnt10b exhibit increased numbers of CD31⁺CD140a⁻ cells counted by flow cytometry. Results obtained in independent cell isolation experiments from 3 mice per group. (B) qPCR analysis shows that Wnt10b induces Axin2 and Vegfr-2 in microvascular cardiac endothelial cells (MCEC-1), but it has no effect on Vegfr-1 and Vegf-a expression. (C) Wnt10b promotes branching morphogenesis and synergizes with VEGF165 (10 ng/ml) in MCEC-1 cells on GFR Matrigel (left panels). Total number of branches per 1 mm² field (right). Results obtained in independent cell isolation experiments from 3 mice per group. * P < 0.05; ** P < 0.01; *** P < 0.001; NS, not significant, unpaired t test in panel A and B, one-way ANOVA with Bonferroni’s multiple comparisons test in panel C. All data are means ± SEM.

The Vegf-a/Vegfr-2 ligand/receptor axis specifically regulates endothelial cell proliferation.^31,32 As shown in Figure 5B, Wnt10b induced Vegfr-2 in addition to the canonical Wnt gene target Axin2 in cultured endothelial cells, indicating that Wnt10b can further enhance the response of endothelial cells to Vegf-a by increasing the levels of its receptor. To test this possibility, we analyzed the effect of Vegf-a in vascular tube formation in the presence or absence of Wnt10b. We found Wnt10b and Vegf-a cooperated in stimulating endothelial tube formation in vitro (Figure 5C). These findings indicate Wnt10b both directly stimulates endothelial cell growth and amplifies the pro-angiogenic effects of Vegf-a by induction of Vegfr-2.

Wnt10b promotes formation of mature coronary-like blood vessels

In addition to greater density, the vasculature within injured tissue of TG hearts consisted of large-diameter blood vessels, in contrast to small-diameter capillaries present in WT hearts (Figure 3A). Co-staining with antibodies recognizing endothelial and vSMCs confirmed scar tissue in TG mice contained greater number of blood vessels surrounded by a prominent smooth muscle cell layer, as compared to that in WT (Figure 6A).

Figure 6. Wnt10b enhances blood vessel stabilization by induction of Angiopoietin-1 and Tie2 signaling.

(A) IF images of CD31 (green) and αSMA (red) antibody staining in the scar areas of WT and TG mouse hearts 3 weeks post cryoinjury show Wnt10b promotes formation of stable blood vessels with recruitment of αSMA⁺ smooth muscle cells (left). Scale bar, 100 μm. Quantification of the number of αSMA⁺ vessels shows that scar areas in TG hearts have 2-fold more arterioles than WT controls. N=3 mice per group. (B) qPCR analysis shows Ang-1 expression 1 week after cryoinjury is enhanced by 2-fold in TG ventricles, as compared to WT controls. In contrast, Ang-2 induction levels are, whereas Tie2 receptor expression remains unchanged after injury in both groups. N=3–4 mice per group. (C) IF images of CD31 (green, left), αSMA (green, right), and Ang-1 (red) antibody staining in the injury areas of WT and TG mouse hearts 1 week after cryoinjury shows higher levels of Ang-1 expression in perivascular cells of TG cardiac tissue sections as compared to WT controls. (D) Co-culture assay of cardiac endothelial cells (ECs) stained with DiD (red) and vascular smooth muscle cells (vSMCs) stained with DiO (green) on GFR Matrigel to monitor migration and ensheathment of endothelial tubes by vSMCs in the absence of presence of Wnt10b (300 ng/ml) and/or Tie2 Kinase Inhibitor (1 μM). Number of vSMCs per ring counts are shown on the right. N=3–5 wells counted per group. * P < 0.05; ** P < 0.01; NS, not significant, unpaired t test. All data are means ± SEM.

To identify the molecular basis of the observed vascular phenotype, we measured expression of Ang-1 and its receptor Tie2, which regulate vSMCs recruitment and blood vessel stability.^33,34 Analysis of ventricle tissue RNA of WT and TG mouse hearts showed comparable baseline Ang-1, Ang-2 and Tie2 expression levels prior to injury. However, there was a strong (15-fold) induction of Ang-1 expression 1 week after injury in WT ventricular tissue, which was further increased by 2-fold in TG ventricles (Figure 6B). In contrast, there were no major differences in the expression of Tie2 and the Ang-1 antagonist Ang-2 between WT and TG tissue (Figure 6B). Histological analysis pointed to higher Ang-1 levels in αSMA⁺ cells surrounding blood vessels within TG scar tissue compared to WT controls (Figure 6C).

To test whether Wnt10b-induced arteriole morphogenesis is due to activation of the Ang/Tie2 signaling axis, we co-cultured endothelial and vSMCs with or without Wnt10b and a Tie2 kinase-specific inhibitor.³⁵ We found Wnt10b induced association of endothelial and smooth muscle cells in GFR Matrigel. Moreover, Tie2 inhibition abolished Wnt10b-induced endothelial/smooth muscle cell interactions (Figure 6D). Taken together, our results indicate Wnt10b promotes vSMCs recruitment and blood vessel stability by activating the Ang-1/Tie2 pathway.

Wnt10b orchestrates vessel stabilization through coordinate induction of NF-κB signaling in endothelial and smooth muscle cells

Previous studies have identified NF-κB and Notch signaling as critical pathways in arteriogenesis, collateral vessel formation and vascular network size during development, or in the adult through coordinate promotion of angiogenesis and vSMC recruitment.^36–38 We observed no significant differences in Notch signaling targets such as Hey1 and Hey2 after injury between WT and TG mice (Online Figure VIII). To test whether Wnt10b acts upstream of the NF-κB signaling pathway, we investigated its effect on NF-κB signaling activation. Our data show Wnt10b treatment of endothelial and smooth muscle cells, transfected with NF-κB luciferase reporter constructs, led to induction of NF-κB signaling in both cell types (Figure 7A and 7B). The NF-κB signaling activation effect was Wnt10b-specific, since the non-canonical Wnt11 ligand, which is also strongly induced in the heart after injury,¹² had only a moderate effect. No synergistic interaction was observed between Wnt10b and TNFα-induced NF-κB signaling in endothelial cells even at low TNFα concentrations (Online Figure VIII).

Figure 7. Wnt10b induction of angiogenic and arteriogenic factors depends on activation of NF-κB signaling in endothelial and smooth muscle cells.

(A,B) Luciferase activity assay in pNF-κB-luc-plasmid-transfected endothelial cells MCEC-1 (A) and smooth muscle cells MOVAS (B) treated with vehicle (PBS), Wnt10b (300 ng/ml) or Wnt11 (300 ng/ml). Wnt10b activates NF-κB signaling in both cell types. (C,D) Wnt10b induction of Vegfr-2 and Pdgf-b in MCEC-1 (C) and Ang-1 in MOVAS (D) is blocked by the NF-κB inhibitor SN-50 (50 μg/ml). (E,F) Flow cytometry based quantification of CD45⁺ immune cells (E) among viable non-cardiomyocytes and CD11b⁺Ly6G⁺ neutrophils (F) among CD45⁺ cells isolated from WT and TG ventricles 5 days post cryoinjury (dpc) shows immune cell infiltration is attenuated in TG hearts. Cell populations from WT ventricles at baseline used as reference. N=4–6 mice per group. * P < 0.05; ** P < 0.01; *** P < 0.001; NS, not significant, unpaired t test. All data are means ± SEM.

To test whether the effect of Wnt10b on the induction of genes regulating endothelial cell growth (Vegfr-2) and vSMC recruitment (Ang-1) genes was due to activation of NF-κB signaling, we incubated endothelial and smooth muscle cells with Wnt10b in the presence or absence of the NF-κB signaling inhibitor SN50.³⁹ The results show Wnt10b-mediated induction of Vegfr-2 in endothelial cells and Ang-1 in smooth muscle cells was abolished by SN-50 (Figure 7C and 7D). Furthermore, Wnt10b also induced Pdgf-b expression in endothelial cells in a NF-κB-dependent manner. Pdgf-b is known to act on Pdgfr-β receptor, expressed in vSMCs, to promote recruitment to newly formed blood vessels.⁴⁰

Previous studies have shown that leaky blood vessels hinder tissue recovery because they allow infiltration of immune cells resulting in unresolved inflammation, which in turn may aggravate ventricular remodeling.⁶ This raised the possibility that Wnt10b-induced blood vessel stability attenuates inflammation. Consistent with this possibility, flow cytometry analysis 5 days after cryoinjury showed lower levels of inflammatory cells within ventricular tissue of TG mice compared to WT (Figure 7E and 7F, Online Figure IX).

Consequently, our data indicate Wnt10b promotes angiogenesis and blood vessel stabilization through parallel stimulation of NF-κB signaling in endothelial and smooth muscle cells, a pathway that is known to regulate induction of pro-angiogenic and arteriogenic factors in endothelial cells (Figure 8).^36,37,39,41 The coordinated formation of stable blood vessels improves perfusion of injured tissue and prevents prolonged infiltration of inflammatory cells, creating an optimized microenvironment for cardiomyocyte growth and enhanced repair of injured cardiac tissue.

Figure 8. Model of Wnt10b gain-of-function effects on arteriole formation in the injured heart.

Wnt10b coordinates blood vessel growth and stability by a) activation of canonical Wnt/β-catenin signaling to stimulate endothelial cell proliferation, and b) NF-κB-dependent induction of Vegfr-2 and Pdgf-b in endothelial cells (EC) and Ang-1 in vascular smooth muscle cells (vSMC) to further promote vascular growth, as well as vSMC recruitment.

DISCUSSION

Here we show canonical Wnt signaling activating Wnt10b ligand is expressed in cardiomyocytes and stored in intercalated discs. After injury, Wnt10b expression is induced in cardiomyocytes in border zone during the granulation tissue formation phase of cardiac repair, when neovascularization and fibrosis take place. Our data show that Wnt10b gain-of-function stimulates new blood vessel growth in two separate ways. First, Wnt10b activates canonical Wnt signaling in endothelial cells, a process that is known to promote angiogenesis after myocardial injury.⁴² Second, by stimulating Vegfr-2 expression in endothelial cells, Wnt10b further enhances Vegf-a-mediated endothelial cell proliferation.

Furthermore, Wnt10b mediates stable vessel formation via induction of Pdgf-b expression in endothelial cells and Ang-1 in smooth muscle cells, two growth factors that promote vascular stabilization, reduce permeability and enhance blood flow.^40,43,44 Previous studies have established that simultaneous supply of VEGF and Ang-1 leads to superior functional neovascularization of ischemic tissues while administration of VEGF alone generates leaky vessels and hemorrhages.^45–47 Although such secreted factors regulating individual aspects of blood vessel formation have been identified, knowledge of upstream regulators that coordinate the entire process was missing. To our knowledge, Wnt10b is the first morphogen that coordinates coronary-like vessel formation, pointing to new approaches to effectively re-vascularize poorly perfused areas of the heart. However, future studied will need to address the mechanisms leading to atria enlargement as a potential negative effect caused by Wnt10b overexpression.

Formation of large blood vessels with robust support structures involves endothelial cell growth, branching and recruitment of smooth muscle cells.⁴³ In support of our data on the role of Wnt10b in adult cardiac repair, it was recently shown that endothelial transcription factor ERG promotes vascular stability and growth through canonical Wnt activation during development.⁴⁸ Previous studies have also identified the NF-κB pathway, Notch signaling, RAF-1/ERK and the adaptor protein Shc as intracellular components that coordinate arteriogenesis, collateral vessel formation and vascular network size during development or in the adult.^36–38 Although we did not detect changes in Notch signaling in TG mouse hearts, we discovered that Wnt10b activated NF-κB signaling in both cardiac endothelial and smooth muscle cells. It is thus likely that coordinated activation of NF-κB signaling in both cell types mediating vessel growth and stability is responsible for the Wnt10b effects. In support of this notion, we found inhibition of NF-κB signaling abolished Wnt10b-induced expression of Vegfr-2 and Ang-1 in endothelial and smooth muscle cells, respectively. Links between Wnt and NF-κB signaling have also been identified in osteosarcoma cells and hair follicle formation,^49,50 suggesting the cross-talk between the two pathways has broader roles in development and disease.

The effects of Wnt10b on neovascularization of the injured tissue led to significant restoration of ventricular function after injury, as evidenced by improved systolic parameters, reduced infarct size and attenuated expression of HF biomarkers in TG mice. This recovery phenotype mimics that of mice with sustained Ang-1 treatment, which promoted vessel size expansion and increased the diameter of arterioles, suggesting Wnt10b acts upstream of the Angiopoietin/Tie2 signaling axis.⁴⁶ Furthermore, Wnt10b promoted formation of cardiac tissue within scar, which may further account for the improved ventricular function. This result would be consistent with findings showing that canonical Wnt signaling activation promotes cardiomyocyte proliferation in the embryonic heart.⁵¹ Finally, Wnt10b gain-of-function led to decreased numbers of collagen-producing myofibroblasts and infiltrating pro-inflammatory cells, although further studies are needed to test whether this is a direct Wnt10b effect on these cells types. At present it is not clear if these responses reflect direct effects of Wnt10b on myofibroblast or cardiomyocyte proliferation, expansion and differentiation of cardiac progenitor cells, or regulation of cell fates during endothelial-to-mesenchymal (EndMT) or mesenchymal-to-endothelial (MEndT) transitions^7,12,52–54

Interestingly, Wnt pathway antagonists such as sFRPs also have positive effects in cardiac repair because they stimulate early cell survival through Akt-1 activation.^9,10,14,55 In contrast, activation of canonical Wnt signaling in myofibroblasts accelerated scar formation, benefiting recovery as shown in mice with constitutive fibroblast expression of β-catenin.^13,56 It would be important to test whether stage-, ligand- and cell-specific manipulation of Wnt signaling could further optimize the recovery process.

A limitation of the Wnt10b gain-of-function approach in our study is that the observed phenotypes may not only recapitulate the endogenous role of Wnt10b, but also reflect the collective overexpression effects of structurally related Wnt ligands in the heart. This problem is particularly acute in the Wnt signaling pathway, which consists of 19 ligands and 10 frizzled receptors with overlapping, often redundant expression and function.^57,58 However, it is noteworthy that the observed phenotypes are distinct from those described in αMHC-Wnt11 transgenic mice, which displayed cardiomyopathy due to hypertrophic ventricular myocytes,⁵⁹ suggesting that, at least part of the results are specific to the Wnt10b ligand. Thus, development of Wnt10b administration protocols, or compounds designed to mimic Wnt10b, will likely help orchestrate stable blood vessel formation after injury, improve cardiac tissue recovery, and potentially decrease ventricular remodeling and heart failure.

Novelty and Significance.

What Is Known?

• Angiogenesis and fibrosis are launched simultaneously during scar tissue formation after myocardial infarction.

• Canonical Wnt signaling is activated within the infarct zone in blood-vessel-forming endothelial cells and collagen-producing myofibroblasts.

• Wnt proteins are induced after myocardial infarction, but their roles in cardiac repair processes are not known.

What New Information Does This Article Contribute?

• The canonical Wnt signaling ligand Wnt10b is stored in the intercalated discs of adult cardiomyocytes and induced in the border zone of the infarct.

• Wnt10b regulates canonical Wnt and NF-κB signaling activation in multiple cell types that take part in the injury repair process.

• Overexpression of Wnt10b in the heart promotes formation of blood vessels and cardiomyocytes, attenuates fibrosis, and preserves left ventricular function after injury.

Despite considerable progress in thrombolytics and percutaneous interventions to restore circulation after myocardial infarction, we have no means to improve the natural healing processes of the heart with the goal of preserving ventricular function and prevent heart failure. We discovered that the canonical Wnt ligand Wnt10b is highly induced in peri-infarct cardiomyocytes after injury, at the onset of scar tissue formation. To study the effects of Wnt10b on the cardiac tissue repair process, we generated a transgenic αMHC-Wnt10b mouse that overexpresses Wnt10b in cardiomyocytes. We found that αMHC-Wnt10b mice show robust formation of blood vessels and cardiomyocytes, less fibrosis, and better preserved ventricular output after acute injury as compared to wild-types. Wnt10b induces endothelial cell proliferation via canonical Wnt signaling activation and stimulates arteriole formation in NF-κB-dependent manner. Our findings suggest Wnt10b has the potential to enhance cardiac tissue repair and regeneration and may lead to novel therapeutic strategies to improve ventricular function after injury.

Nonstandard Abbreviations and Acronyms

αMHC

alpha Myosin Heavy Chain (Myh6)

αSMA

alpha Smooth Muscle Actin

Ang-1

2, Angiopoietin-1, Angiopoietin-2

CD31

Cluster of differentiation 31, or Platelet Endothelial Cell Adhesion Molecule-1

FS

Fractional Shortening

GAPDH

Glyceraldehyde-3-phosphate dehydrogenase

GFR

Growth Factor Reduced (Matrigel)

HF

Heart failure

IF

Immunofluorescence

JNK

c-Jun N-terminal Kinase

Lef1

Lymphoid Enhancer-binding Factor 1

MI

Myocardial infarction

MMP

Matrix metalloproteinase

NF-κB

Nuclear Factor kappa-light-chain-enhancer of activated B cells

Pdgf-b

Platelet-Derived Growth Factor subunit B

Pdgfrβ

Platelet-Derived Growth Factor Receptor beta

Tie2

Tyrosine kinase with immunoglobin-like and EGF-like domains 2

Tcf7

Transcription Factor 7

Vegf-a

Vascular Endothelial Growth Factor-A

Vegfr-2

Vascular Endothelial Growth Factor Receptor 2 (Flk-1 or KDR)