Unique Angiogenesis From Cardiac Arterioles During Pericardial Adhesion Formation

Objectives The molecular mechanisms underlying post-operative pericardial adhesions remain poorly understood. We aimed to unveil the temporal molecular and cellular mechanisms underlying tissue dynamics during adhesion formation, including inflammation, angiogenesis, and fibrosis. Methods and Results We visualized cell-based tissue dynamics during pericardial adhesion using histological evaluations. To determine the molecular mechanism, RNA-seq was performed. Chemical inhibitors were administered to confirm the molecular mechanism underlying adhesion formation. A high degree of adhesion formation was observed during the stages in which collagen production was promoted. Histological analyses showed that arterioles excessively sprouted from pericardial tissues after the accumulation of neutrophils on the heart surface in mice as well as humans. The combination of RNA-seq and histological analyses revealed that hyperproliferative endothelial and smooth muscle cells with dedifferentiation appeared in cytokine-exposed sprouting vessels and adhesion tissue but not in quiescent vessels in the heart. SMAD2/3 and ERK activation was observed in sprouting vessels. The simultaneous abrogation of PI3K/ERK or TGF-β/MMP9 signaling significantly decreased angiogenic sprouting, followed by inhibition of adhesion formation. Depleting MMP9-positive neutrophils shortened mice survival and decreased angiogenic sprouting and fibrosis in the adhesion. Our data suggest that TGF-β/matrix metalloproteinase-dependent tissue remodeling and PI3K/ERK signaling activation might contribute to unique angiogenesis with dedifferentiation of vascular smooth muscle cells from the contractile to the synthetic phenotype for fibrosis in the pericardial cavity. Conclusions Our findings provide new insights in developing prevention strategies for pericardial adhesions by targeting the recruitment of vascular cells from heart tissues.


INTRODUCTION
Pericardial adhesions are frequently recognized in repeat open-heart surgeries and complicate the procedure (1,2). Growing evidence indicates that the risk of adverse events, such as intraoperative injury and bleeding, is increased during cardiac reoperations with adhesiotomy. Therefore, understanding the molecular mechanism underlying adhesion formation is necessary to prevent its occurrence (2).
The heart consists of various cell types, including cardiac myocytes, myofibroblasts, vascular smooth muscle cells (VSMCs), vascular endothelial cells, and mesothelial cells (3). Mesothelial cells typically surround myocardial tissues and maintain homeostasis of the pericardial cavity via serous fluid secretion (4). Of the various cell types in the pericardial cavity, mesothelial cells play a central role as the initiators of adhesion formation through multi-step reactions after pericardiotomy (5). Previous investigations using a bacterial toxin-induced pericardial adhesion model in sheep demonstrated that mesothelial cell transition from adhesion to free-floating type in the pericardial cavity induces inflammatory reactions (6)(7)(8). Following this, fibrin accumulation then occurs on the injured epicardial surface after the development of a scaffold for subsequent fibrosis (6). Finally, numerous inflammatory cells and fibroblasts invade the injured area, leading to collagen fiber production that eventually fills the pericardial space (7).
Although new blood vessels have also been observed in collagendeposited areas, the temporal cell behaviors of endothelial cells, myofibroblasts, and their source tissues during adhesion formation are poorly understood.
Blood vessels support the maintenance and survival of various tissues or organs by delivering nutrition and oxygen (9). Hypoxic and inflammatory signals induce the proliferation and migration of endothelial cells through activation of vascular endothelial growth factor (VEGF) receptors by ligands such as VEGF-A, -B, -C, and -D. The molecular mechanisms underlying the formation of new blood vessels in different developmental stages have been well characterized using a retinal angiogenesis murine model (10)(11)(12)(13). Excessive leading and following cell proliferation of endothelial cells frequently occur during the growth and extension stages. Immature vascular trees are subsequently covered by pericytes, including myofibroblasts, for stabilization or specification into arteries (13), followed by the formation of lumen structures for blood perfusion. Nevertheless, the molecular and cellular mechanisms underlying pathological angiogenesis observed in adult cardiovascular diseases and repair are yet to be elucidated.
Abbreviations: VSMC, vascular smooth muscle cell; VEGF, vascular endothelial growth factor; TGFB, transforming growth factor B; EGFP, enhanced green fluorescent protein; αSMA, alpha smooth muscle actin; PCNA, proliferating cell nuclear antigen; MMP, matrix metalloproteinase; GFP, green fluorescent protein; FGB, fibrinogen beta chain; HE, hematoxylin eosin; PBS, phosphatebuffered saline; HRP, horseradish peroxidase; DAB, diaminobenzidine; DAVID, Database for Annotation, Visualization and Integrated Discovery; KEGG, Kyoto Encyclopedia of Gene and Genomes; GO, Gene Ontology; Itgam, integrin subunit alpha M; Ptprc, protein tyrosine phosphatase receptor type C; Itgb2, integrin subunit beta 2; Anpep, alanyl aminopeptidase; Tlr2, toll-like receptor 2; Tlr8, tolllike receptor 8; Sorl1, sortilin-related receptor 1; Col1a2, collagen type 1 alpha Molecular-based analyses of inflammatory signaling in pericardial fibrosis have been previously reported. Transforming growth factor-beta (TGF-β) is reportedly a central player in the activation of inflammatory reactions and the production of the extracellular matrix for fibrosis (14). The activation of TGFβ signaling also contributes to the transition from epithelial or endothelial cells to mesenchymal cells, promoting tissue fibrosis (15,16). A previous human study reported elevated TGF-β levels in both serum and pericardial fluid of patients with pericarditis (17). Further, a pericardial adhesion model in pigs also demonstrated excessive upregulated expression of tumor necrosis factor-α, TGF-β, and phosphorylated SMAD, indicating that TGF-β is essential for inflammation in the pericardial cavity and adhesion formation (18). However, no studies have investigated the role of TGF-β in the early stages of pericardial adhesion.
We have previously established a simple model of pericardial adhesion in mice (19). In the current study, we conducted spatiotemporal visualization of histological changes using this model to elucidate the molecular and cellular mechanisms underlying tissue dynamics during adhesion formation, including inflammation, angiogenesis, and fibrosis.

Visualization of Tissue Structure and Fibrosis
Mouse chest cavities were excised and fixed using 10% neutral buffered formalin for 24 h. After decalcification with 10% ethylenediaminetetraacetic acid for 3 weeks, the tissues were embedded in paraffin and sliced into 5 µm sections. Paraffin-embedded heart tissue from the autopsy was used for comparative investigations of fibrosis and structure comparison between mice and humans. All clinical experiments were approved by the Ehime University Internal Review Board (protocol no.2105012), and informed consent was obtained based on the opt-out principle. Hematoxylin and eosin (H and E) and Masson's trichrome staining were performed to visualize tissue structures and collagen deposition, respectively. After the specimens were deparaffinized with xylene, samples were rehydrated and stained as previously described (19). For H and E staining, slides were stained with Meyer's hematoxylin solution for 15 min, rinsed with water for 10 min, and then stained with eosin, followed by washing with water. For Masson's trichrome staining, samples were treated with a mixture of 10% potassium dichromate and 10% trichloroacetic acid for 30 min, rinsed for 3 min, and then treated with Weigert's iron hematoxylin for 5 min. Samples were subsequently stained with 0.75% Orange G solution (Wako, Osaka, Japan) for 1 min and rinsed with acetic acid. Samples were then treated with xylidine ponceau acid for 20 min, 2.5% phosphotungstic acid for 10 min, and aniline blue for 5 min. Sagittal images of the chest cavity were obtained using a BZ-X800 microscope (Keyence, Kyoto, Japan). Highmagnification images were acquired using a BX-51 Olympus microscope (Olympus, Tokyo, Japan).
To investigate tissue co-localization, immunofluorescence assays were performed using multiple antibodies. After deparaffinization and antigen activation, samples were blocked using Dako REAL TM Antibody Diluent, followed by reaction with the following primary antibody mixtures overnight at 4

RNA Isolation and Sequencing
Pericardial adhesion tissues and the myocardium were separately harvested from euthanized mice, homogenized with the ISOGEN II reagent, and stored at −20 • C until further use. Total RNA was isolated according to the manufacturer's instructions for the ISOGEN II reagent. Purified total RNA was quantified using the NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific), and RNA integrity score was evaluated using the Bioanalyzer RNA 6000 Nano Assay Kit (Agilent Technologies, Santa Clara, CA, USA). RNA sequencing using the MiSeq system (Illumina Inc., San Diego, Cam USA) was performed as previously described (20). Signal pathway and gene enrichment analyses were performed using DAVID version 6.8 bioinformatics software (http://david.abcc.ncifcrf.gov).

Flow Cytometry
To quantify the cell population of neutrophils in the peripheral blood and heart tissue of adhesion-induced mice, Ly6g and CD11b double-positive cells were selectively collected using a flow cytometer. TALC-injected and sham mice were anesthetized via mixed injection of ketamine (80 mg/kg) and xylazine (10 mg/kg), and blood was collected using heparin-treated 23G needles. After perfusion with PBS, heart tissues were excised and digested with PBS-based collagenase (Worthington Biochemical, Freehold, NJ, USA) solution for 3 h at 37 • C. After digestion, the samples were filtered through 40 µm cell strainers (BD Biosciences, Bedford, MA), and the cells were labeled with Alexa Fluor R 488 anti-mouse/human CD11b antibodies and APC anti-mouse Ly-6G antibody. Peripheral blood treated with the antibodies were hemolyzed using OptiLyse C Lysing Solution according to the manufacturer's instructions. The samples were then washed with 10% fetal bovine serum and analyzed using a FACS Calibur flow cytometer (Becton Dickinson, San Jose, CA).

Gelatin Zymography
The enzymatic activities of MMP9 were measured using a Gelatin Zymography Kit following the manufacturer's instructions. Forty micrograms of the extracted proteins from heart tissues were separated by SDS polyacrylamide gel electrophoresis, and the gel was incubated in the reaction buffer for 36 h at 37 • C. The bands were visualized by staining with coomassie blue staining solution.

Depletion of Ly6g-Positive Cells
To eliminate neutrophils, mice were injected intraperitoneally with 387 µg of Ly6g neutralizing antibodies or isotype control antibodies in 200 µl PBS, as previously reported (21). After 2 days, pericardial adhesions were induced by talc injections. Both antibodies were injected every 2 days. On the seventh day after talc injection, heart tissues were excised and adhesion scores measured, followed by histological analyses by Masson's trichrome staining and immunohistochemical staining for CD31 and αSMA.

Statistical Analysis
All data are presented as mean ± standard error (SE). GraphPad Prism Ver. 7 (GraphPad Software, San Diego, CA, USA) was used to conduct a one-way analysis of variance (ANOVA) for normal distributions and the Kruskal-Wallis test for nonnormal distributions. Statistical significances between two groups were calculated using Student's t-tests (parametric variables) and Mann-Whitney U-test (non-parametric variables). The normality of the obtained experimental data was assessed using the Shapiro-Wilk normality test. ImageJ software (NIH) was used for quantification histological data. ns.: not significant. ND.: not detected.

Visualization of Tissue Dynamics During Adhesion Formation
To determine the turning point in pericardial adhesion, identified as an intensive increase in the adhesion score of our model, we injected a talc solution into the pericardial cavities of mice and evaluated tissue structures at both the gross and histological levels. Adhesions around the myocardium were observed (Figure 1A), and adhesion scores were remarkably increased ( Figure 1B) on day 3 after talc injection. The highest adhesion scores, indicating complete adherence to the epicardium, were observed after day 4 ( Figure 1B). These histological findings were consistent with the adhesion scores, suggesting that crucial changes occurred at the cellular level around day 4 after talc injection. Next, we excised the chest cavity for histological analysis to avoid inflicting pericardial injury during the thoracotomy (Figure 1C, right). H and E-stained specimens displayed excessive cell invasion and fibrosis in the pericardial cavity from day 4 ( Figure 1C, arrowheads). Masson's trichrome staining results confirmed intense production of extracellular matrix proteins connecting the epicardium and pericardium, such as collagen, whose level increased significantly after day 4 ( Figure 1C, arrowheads). To investigate cell-based histological features, we comprehensively compared the histological images at each time point. Visualization of Masson's trichrome-and H and E-stained specimens at high-magnification displayed neutrophils in the early stages (day 2), extravasation of red blood cells into the pericardial space (day 3), and invasion of fibroblastic cells into the myocardial tissue (day 5) (Figures 1C,D).
The pathological findings in model mice regarding collagen depositions onto the surface of heart tissues were similar to those in human specimens with adhesions in the heart ( Figure 1E). The ratios of epicardial erosion defined as discontinuous epicardium significantly increased (p < 0.001) at the third hour after injection (Figure 1F), while intact epicardia stained by immunohistochemical staining with GPM6A, which is known as a marker of mesothelial cells (22), was observed in the preinjected or blood-injected mice using a previously established bleeding model (19) (Supplementary Figure 1). Hemorrhage was confirmed using immunohistochemical staining with antifibrinogen beta chain (FGβ) antibodies. Expectedly, the FGβ was located in the pericardial cavity on day 1 ( Figure 1G). These data indicated that our model reproduced the pathologies of epicardial injury in the early stage, including mesothelial erosion, bleeding, and inflammatory cell accumulation previously reported in humans and other animal models.

Providing Myofibroblasts and Endothelial Cells From the Beating Heart
To clarify the cellular mechanism of fibrosis observed in the pericardial cavity after day 4, we investigated the cellular behavior of αSMA-positive myofibroblasts responsible for collagen production during adhesion formation. As expected, few myofibroblasts were observed in the pericardial cavity on day 4, and the cell population was significantly increased (p = 0.0001) on day 7 in proportion to collagen deposition (Figure 2A). Interestingly, myofibroblasts were also located around the pericardial area, and the cell population tended to increase, suggesting that adhesion-related myofibroblasts might originate from the pericardial tissues. Based on our hypothesis, we quantified myofibroblasts migrating from the myocardial tissues to the pericardial cavity. As shown in the whole heart view illustrated in Figure 2B, myofibroblasts were more densely located around heart tissues than on the pericardial side.
Notably, the abundance of vascular-like bridging myofibroblasts, which form lumen structures between cardiomyocytes and adhesion tissues, was significantly increased (p = 0.0005) on day 7 (Figure 2B). Based on our findings that the vascular-like structures were composed of myofibroblasts, we also investigated the localization of CD31-positive endothelial cells. A large number of endothelial cells were observed in adhesion and myocardial tissues on day 7 ( Figure 2C). Notably, the abundance of CD31-positive sprouting vessels was remarkably increased from day 4 ( Figure 2C), suggesting that angiogenesis and fibrosis might occur concomitantly during adhesion formation (Figures 2A,C). To validate our hypothesis, we conducted double immunofluorescence experiments using anti-CD31 and -αSMA antibodies. As shown in areas 1 of Figures 2D,E, several arterioles were composed of CD31-positive endothelial cells and αSMApositive myofibroblasts as the inner and outer layers, respectively. Histological analysis of heart tissues on day 7 also revealed initial stages of sprouting (area 2), elongating vessels from myocardial tissues to adhesion tissues (area 3), and elongating vessels in adhesion tissues (area 4), which maintained the association of endothelial cells with αSMA-positive cells, and lumen structures with red blood cells (Figures 2D,E). A few myofibroblastuncovered capillaries were observed in adhesion tissues (area 5 of Figure 2E). Next, we investigated whether arterioles and capillaries from the heart contributed to the development of fibrosis and vascular formation in the pericardial cavity. Using Rosa26 EGFP ; Cdh5 CreERT2 mice that harbored EGFP-labeled endothelial cells, we traced endothelial cell behavior during adhesion. Arterioles and metarterioles were observed in the inner heart areas, and numerous capillaries were formed throughout heart tissues ( Figure 3A). Immunofluorescence co-staining results also clearly indicated direct sprouting from arterioles and metarterioles (myofibroblasts located at the leading edge of sprouting vessels) (areas 1 and 2 in Figure 3B). Additionally, new capillary formation from arterioles was observed (area 3 of Figure 3B). However, no double-positive cells labeled with EGFP-and αSMA-derived fluorescence were present in heart tissues ( Figure 3C), suggesting that endothelial cells could not be considered the origin of the myofibroblasts responsible for the development of fibrosis within 7 days of adhesion induction. In addition, three-dimensional tissue imaging based on confocal microscopy with z-stacking demonstrated that several EGFP-positive neoformed vasculatures sprouted in adhesion tissue from the vascular network in heart tissues ( Figure 3D). To validate our findings obtained via the murine model, we evaluated the pathology of an autopsied patient who underwent cardiac surgery 6 months prior to the autopsy. Masson's trichrome staining results revealed the robust formation of adhesion tissues (indicated in blue) around layers of cardiomyocytes, while adipose tissues were partially observed within myocardial and adhesion tissues ( Figure 3E). Interestingly, sprouting vessels from heart tissues were observed, composed of CD31-positive endothelial cells in the inner regions and αSMA-positive smooth muscle cells in the outer regions ( Figure 3E). Tissue localization of CD31and αSMA-positive cells during sprouting was confirmed by double immunofluorescence staining ( Figure 3F). These angiogenic sprouting from the heart to adhesions were observed in all cases who underwent cardiac reoperation (Figure 3G). Taken together, the results indicated that endothelial cells and myofibroblasts originating from myocardial tissues might contribute to the development of post-surgical cardiac adhesions in humans.

Gene Expression Profiling of Pericardial Adhesions in Mice
To elucidate the molecular mechanism underlying pericardial adhesion, RNA sequencing was performed on samples obtained from heart and adhesion tissues. The scatter plot indicated that drastic changes did not occur in the expression levels of detected genes in heart tissues between days 3/0 and days 7/0 ( Figure 4A). In heart tissues, 736 and 737 upregulated genes (>three-fold) and 231 and 220 downregulated genes (three-fold) were detected on days 3 and 7, respectively. To characterize the obtained gene sets, we determined Kyoto Encyclopedia of Gene and Genomes (KEGG) enrichment using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) bioinformatics tool. KEGG pathway results revealed that inflammation-related genes played crucial roles in pathways for chemokine signaling, cytokine-cytokine receptor interaction, and PI3K-Akt signaling in both heart and adhesion tissues ( Figure 4A). In contrast, 708 and 874 upregulated genes (>three-fold) and 220 and 788 downregulated genes (<three-fold) were found in adhesion tissues on days 3 and 7, respectively. Cytokine-and PI3k-Akt signaling-related genes were also enriched in both adhesion and heart tissues ( Figure 4B), while the enriched gene patterns responsible for cytokine signaling differed considerably between heart and adhesion tissues (Supplementary Figure 2). Notably, Gene Ontology (GO) term analysis indicated genes associated with cell division-, cell cycle-, and proteolysis-related processes were enriched in both heart and adhesion tissues (Figures 4A,B), indicating that these genes might contribute to the formation of new blood vessels during adhesion. Downregulation of MYOCD signaling was seen as a common pathway between heart and adhesion tissues on day 3 using Ingenuity Pathway Analysis (IPA) (Supplementary Figure 3). IPA analysis also revealed upregulations of Leukocyte Extravasation and MMP9 signaling in the heart on day 3. Notably, upregulation of angiogenic signaling in the heart tissue on day 7 was also detected (Supplementary Figure 3).

Cell Proliferation of Heart-Derived Myofibroblasts and Endothelial Cells
Based on the RNA sequencing results regarding cell cycle regulators, to visualize proliferative cells during adhesion formation, we performed immunohistochemical staining using anti-PCNA antibodies. A few PCNA-positive cells were observed around the injured surface of the heart on day 1, and the cell population gradually increased for 7 days (Figure 5A). From days 4-7, the abundance of proliferative cells was also markedly increased in adhesion tissues ( Figure 5A). We next identified the types of PCNA-positive proliferative cells in heart and adhesion tissues using double immunofluorescence staining for endothelial cells and myofibroblasts. PCNA-derived fluorescence signals were not detected in EGFP-positive endothelial cells and αSMA-positive smooth muscle cells of heart tissues ( Figure 5B, upper). However, PCNA was expressed in the nuclei of endothelial cells and myofibroblasts during vascular sprouting (Figure 5B, middle). Further, the populations of PCNA-positive proliferative endothelial cells and myofibroblasts were increased in adhesion tissues compared to those in sprouting vessels (Figure 5B, lower), with significantly different ratios of PCNA-positive endothelial cells (p < 0.0001) and myofibroblasts (p < 0.0001) between sprouting vessels and adhesion tissues (Figure 5C). Cell proliferation of both cells in sprouting vessels was confirmed by the immunofluorescence staining for ki67 (Supplementary Figure 4), suggesting that dedifferentiation of VSMCs from the contractile to the synthetic phenotype might have been induced. To test our hypothesis, immunohistochemical co-staining of TAGLN, a marker of smooth muscle cells, and αSMA was performed. We also found that while cardiac arterioles and sprouting vessels surrounded by TAGLN-stained smooth muscle cells were also observed, the fluorescence intensities of TAGLN were significantly decreased in sprouting vessels (p < 0.0001), rather than in cardiac arterioles. Our data indicate that proliferative dedifferentiated smooth muscle cells in arterioles of the heart are one of the origins of fibrosis in the pericardial cavity.

Proteolysis-Dependent Tissue Remodeling in the Pre-angiogenic Stage
GO term and IPA analyses based on the RNA-seq data revealed enrichment of proteolysis-related genes, including MMP9 (Figure 4). Since MMP9 is a proangiogenic factor that induces angiogenesis by subjecting the extracellular matrix to proteolysis (23), we investigated the tissue distribution of MMP9. MMP9-positive cells were observed on the surface of myocardial tissues on day 1 (Figure 6A). Although few MMP9-positive cells existed in the blood (area 1 of Figure 6A), these cells were distinctly accumulated on the surface of myocardial tissues on day 2 (area 2 in Figure 6A). Immunofluorescence co-staining of MMP9 and MPO, a marker of neutrophils, and flow cytometry for CD11b and Ly6g double-positive cells confirmed that the MMP9-positive cells were neutrophils (Figures 6B,C). Indeed, the expression levels of activated neutrophil marker genes, such as itgam, ptprc, itgb2, mmp9, anpep, tlr2, tlr8, and sorl1, were markedly upregulated in heart tissues on days 3 and 7. Conversely, these genes were not upregulated in adhesion tissues, characterized by overexpression of proteins encoded by extracellular matrix-associated genes, including col1α2, col1α1, and co3α1 ( Figure 6D and Supplementary Figure 5). Temporal histological analyses revealed that the accumulation of MMP9-positive neutrophils on the surface of myocardial tissues gradually decreased after day 5 ( Figure 6E). Gelatin zymography and western blotting of tissue lysates from the heart also confirmed that MMP9 protein with protease activities was enriched after day 1, although protein levels of TAGLN (a marker of VSMCs) decreased with the progression of adhesion formation (Figure 6F). Immunofluorescence analysis also revealed the presence of MMP9-positive neutrophils around CD31-and CD34-double-positive blood vessels at the injured surface of myocardial tissues on day 3. After the occurrence of bleeding and the initial stages of collagen production in the pericardial cavity, the neutrophils gradually migrated into the adhesion tissues. The angiogenic sprouting of CD34/CD31 double-positive vessels covered by myofibroblasts occurred on the adhesion side ( Figure 6G). To investigate the roles of MMP9positive neutrophils, talc was administered to mice treated with isotype control antibodies-(cAb) or Ly6g neutralizing antibodies (nAb) (Figure 6H). Depletion of neutrophils was confirmed by flow cytometry for CD11b and Ly6g (Figure 6I). Fifty-five percentage of neutrophil-depleted mice could not survive talc injection, although control mice were not affected ( Figure 6J). Further investigation of the surviving nAb-treated mice revealed that adhesion score, collagen deposition, and the number of angiogenic sprouting were significantly suppressed by depletion of neutrophils (Figures 6K,L). These findings indicate that MMP-9-positive neutrophils contribute to survival and adhesion formation through neovascularization by enabling remodeling of the collagen scaffold in myocardial tissues.

MMP, PI3K, and SMAD Signaling-Dependent Angiogenic Sprouting Is Essential for Adhesion Formation
To identify the critical cell signaling responsible for angiogenic sprouting, we focused on the PI3k/ERK and TGF-β/SMAD signaling, respectively known as master regulators of cell growth and fibrosis. Phosphorylated ERK-and SMAD2/3-positive cells were observed in the heart tissues of talc-treated mice, but not sham groups ( Figure 7A). The fluorescence signals derived from phospho-ERK and SMADs were specifically detected in the CD31 or αSMA-positive sprouting vessels ( Figure 7B). Compared to sham groups, the numbers of phospho-ERK and SMAD-positive sprouting vessels were significantly increased in talc-treated mice compared to sham groups ( Figure 7C). To elucidate whether ERK or SMAD signaling is critical for angiogenic sprouting, talc-injected mice were treated with MEK/ERK inhibitor (PD98059) or TGF-β/SMAD inhibitor (SB431542). However, there was no difference between inhibitor and vehicle groups (Supplementary Figure 6). Our RNA-seq data also showed activation of PI3K/Akt, a master regulator  of positive regulation of cell cycle and cell survival, in cardiac tissues ( Figure 4A). In addition, MMP9, a powerful activator of TGF-β/SMAD signaling, was produced during angiogenic sprouting (Figure 6). Based on these data, we hypothesized that not only PI3k/ERK and TGF-β/SMAD signals but also MMP and PIk3/Akt signals might be essential for angiogenic sprouting during adhesions ( Figure 7D). To address our idea, we treated the adhesion-induced mice with inhibitor mixtures (PD98059, SB431542, LY294002, and Marimastat) every 2 days ( Figure 7E). As shown in Figure 7F, inhibitor mixture-treated mice exhibited a noticeable reduction of adhesion formation on the surface of heart tissues in the gross levels, although a dramatic change was not seen in groups of mice treated with a single inhibitor (Supplementary Figure 6). Adhesion score also significantly decreased when treated with the inhibitor mixture (p = 0.0006) ( Figure 7G). Additionally, collagen productions on the heart surface and the number of αSMA and CD31positive sprouting vessels were significantly decreased by the inhibitor mixture treatment (Figures 7H,I). Finally, to clarify the importance of PI3k/ERK and TGF-β/SMAD/MMP axes, adhesion-induced mice were treated with two-inhibitor cocktails (LY294002+ PD98059 and Marimastat+SB431542) every 2 days. The mixture of LY294002 and PD98059 significantly decreased adhesion scores, while treatment with a mixture of Marimastat and SB431542 tended to decrease the adhesion score ( Figure 7J). Histological investigations revealed that both cocktails inhibited collagen production and angiogenic sprouting ( Figure 7K). These data indicate that MMP9/TGF-β/SMAD and PI3K/ERK signaling activations are critical for angiogenic sprouting, followed by pericardial adhesions.

DISCUSSION
Several prevention methods have been developed for postoperative pericardial adhesions based on the observations in humans and animal models that bleeding, inflammation, and fibrosis are involved in the adhesion process (24). However, time-course histological assessments have not been conducted with quantification and molecular profiling during adhesion formation. Our murine model enabled the visualization of the adhesion process in the pericardial cavity (Figure 8) and helped elucidate its molecular mechanism using genetic modification technology. We observed the recruitment of neutrophils to the epicardial surface after the induction of pericardial injury, followed by remarkable vascular malformation with myofibroblast coverage from myocardial tissues during adhesion formation, contributing to fibrosis and blood supply into the pericardial space via the new blood vessels generated therein. The adhesions, separated by a fibrous layer from myocardium tissues, remained for at least 6 months in the pericardial regions (Supplementary Figure 7). The adhesions shared similar features with those of the autopsied patient in terms of the existence of rich blood vessels, the disappearance of myofibroblasts in the adhesion tissues, and adipose tissues formation around myocardial tissues (Figures 3E-G). Our murine model exhibited signs of multiple inflammatory reactions, such as epicardial injury, bleeding, and fibrosis, similar to those reported in previous studies conducted with humans or other animal models (24,25). Thus, we demonstrated that the murine one-shot injection model was a powerful tool for investigating the molecular mechanism underlying pericardial adhesion formation. Angiogenic development has been well characterized using a retinal model of angiogenesis in mice (10)(11)(12)(13). These studies have shown that VEGF-A, a prime inducer of angiogenesis, guides endothelial cell migration toward peripheral regions, followed by vascular remodeling, such as arterialization. During the maturation and vein remodeling process, VSMCs are intensively recruited while the abundance of proliferative endothelial cells is subsequently decreased (13). Nevertheless, angiogenic processes during pericardial adhesion significantly differ from those observed during the developmental stage. Proliferative endothelial cells observed during sprouting were constantly wrapped by αSMA-positive cells, and red blood cells were contained in the vessels (Figure 2). Additionally, myofibroblasts located at the forefront of sprouting vessels played a crucial role as the leading cells of angiogenic sprouting (Figure 3B). Evidence of the generation of new blood vessels with αSMApositive cells coverage in myocardial tissues has been recently reported, and αSMA-positive arterioles were discovered in newly FIGURE 8 | Diagram depicting the molecular mechanism of adhesion formation in the pericardial cavity. After infliction of epicardial injury, several types of immune cells (including neutrophils) accumulate on the surface of heart tissues to enable tissue remodeling via secretion of proteolytic enzymes. Arterioles and capillaries in the heart are excessively sprouted into the pericardial space. Then, smooth muscle cells with a contractile phenotype transition to a hyperproliferative and synthetic phenotype, which potentially contribute to fibrosis via production of extracellular matrix (ECM) proteins. The endothelial cells in the arterioles also exhibit a hyperproliferative phenotype and promote neovascularization in adhesion tissues (whole heart view). Arterial angiogenic sprouting occurs by maintaining vascular coverage with myofibroblasts (heart tissue view).
formed vascular networks (26). αSMA-positive arterioles were also observed in skeletal muscles and heart tissues of mice with adeno-associated virus-mediated VEGF overexpression (27). The authors reported that VEGF-induced formation of new arterioles in the heart occurred to a lesser extent than in skeletal muscle. However, whether myofibroblasts covered the newly formed blood vessels following angiogenesis or whether vessel formation occurred via association between endothelial cells and myofibroblasts remains to be determined. Based on time-course experiments, our histological data demonstrated the occurrence of αSMA-positive cells-guided vascular formation along with red blood cell contents (Figure 2). Tracing analyses using endothelial cell-specific EGFP-expressing mice also revealed that endothelial-mesenchymal transition rarely occurred within 7 days, and each proliferated cell migrated into adhesion tissues (Figure 3). These results suggest that maintaining the vascular lumen structures of the arterioles may be essential for pathological neovascularization to avoid severe hypoxia during tissue repair, and direct recruitment of αSMA-positive cells from arterioles in myocardial tissues may be efficient for immediate adhesion formation.
The transition of VSMCs from the contractile to the synthetic phenotype, which exhibits high proliferative/migrative activities and overexpression of extracellular matrix-associated proteins, has been well characterized in neointimal formation during atherosclerosis (28). Multiple studies using lineage-tracing and gene expression analyses have demonstrated that plaque cells originating from dedifferentiated VSMCs are characterized by downregulated expression of MYOCD, ACTA2, Tpm1, Tagln, Cnn1, ltga7, Des, and MYH11, and upregulated expression of Vim, Myh9, CD44, MMP3, and collagen, which contribute to the development of vascular fibrosis via promotion of conversion to the synthetic phenotype (28)(29)(30)(31)(32). Additionally, TGF-β signaling is an essential regulator of VSMC dedifferentiation (33). Both our histological and RNA sequencing analyses demonstrated the conversion of VSMCs in the arteriole to proliferative myofibroblasts (Figure 5), downregulated expression of contractile marker genes, and upregulated expression of synthetic marker genes, as well as TGFB1 and TGFB-induced (TGFBI) expression (Supplementary Figures 8A,B) and SMAD2/3 activation (Figure 7) in heart tissues during adhesion formation. Western blotting also confirmed that activation of SMAD2 in murine endothelial cells was induced by exposure to lysates from adhesion tissue (Supplementary Figure 8C). Proliferative αSMA-positive mural cells stained with anti-PCNA and -Ki67 antibodies were observed in the sprouting vessels from heart tissues to adhesion. These results suggest that analogical signaling and pathology exist between atherosclerosis and pericardial adhesions, and TGF-β-induced acquisition of the VSMC synthetic phenotype may also be essential for fibrosis in the pericardial cavity.
Comprehensive gene expression analysis also revealed that multiple cytokine signaling pathway-related genes that exhibit potential activities for enabling transition into the synthetic type, such as CCL, CXCL, IL, and the TNF family, were markedly enriched in the adhesions. Additionally, MMPs, including MMP9 and MMP2, known as a potent inducer of angiogenesis (34,35) and activator of TGF-β (36), were overexpressed in heart tissues at the pre-angiogenic stages (Figure 6 and Supplementary Figure 9). However, the expression of VEGF families remained unchanged in both myocardial and adhesion tissues within 7 days of induction (Supplementary Figure 10). We successfully identified ERK and SMAD signaling as the driving force for cardiac angiogenic sprouting (Figure 7). The rescue experiments using chemical inhibitors confirmed that MMP/TGF-β/SMAD and PI3K signaling are essential for unique angiogenic sprouting. Indeed, several studies have demonstrated that four chemical inhibitors effectively abrogate ectopic overgrowth and transition into the synthetic type of VSMCs (37)(38)(39)(40)(41)(42)(43)(44)(45)(46). Additionally, we found that the accumulation of MMP9-positive neutrophils at early stages was required for mouse survival following angiogenesis and fibrosis, suggesting that MMP9-positive cells may play dual roles as tissue protectors against acute pericarditis as well as angiogenic inducers. Our data suggest that atherosclerosis-related proliferation and recruitment of VSMCs and endothelial cells derived from arterioles in heart tissue play crucial roles in the adhesion formation on the injured epicardium.
The current study has several limitations. Solid live-imaging data of the angiogenic process during adhesion formation using multiphoton excitation fluorescence microscopy will be needed to establish a complete story for angiogenesis from the heart to adhesion tissues. Further investigations focusing on signal transduction of sprouting vessels are warranted to identify critical factors that regulate adhesion-related unique angiogenesis and the role of neutrophils in mouse survival during adhesions. We will further demonstrate the importance of angiogenesis in pericardial adhesion using angiogenic activitydeficient mice or smooth muscle cell-deletion mice. Our novel findings regarding unique vascular formations from heart tissues will help develop therapeutic options for pericardial adhesions, as well as for treatments of ischemic heart diseases, such as myocardial regeneration, which require the provision of oxygen and nutrition via new blood vessels.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI [accession: GSE182735].

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ehime University Internal Review Board (protocol no.2105012). Informed consent for participation was obtained based on the opt-out principle, as approved by the Ehime University Internal Review Board. The animal study was reviewed and approved by Ehime University Animal Care Committee (Project number: 05-RO-7-1 and 05-RO-42).