The Essential Role for Endothelial Cell Sprouting in Coronary Collateral Growth

Abstract

Rationale –

Coronary collateral growth is a natural bypass for ischemic heart diseases. It offers tremendous therapeutic benefit, but the process of coronary collateral growth is incompletely understood due to limited preclinical murine models that would enable interrogation of its mechanisms and processes via genetic modification and lineage tracing. Understanding the processes by which coronary collaterals develop can unlock new therapeutic strategies for ischemic heart disease.

Objective –

To develop a murine model of coronary collateral growth by repetitive ischemia and investigate whether capillary endothelial cells could contribute to the coronary collateral formation in an adult mouse heart after repetitive ischemia by lineage tracing.

Methods and Results –

A murine model of coronary collateral growth was developed using short episodes of repetitive ischemia. Repetitive ischemia stimulation resulted in robust collateral growth in adult mouse hearts, validated by high-resolution micro-computed tomography. Repetitive ischemia-induced collateral formation compensated ischemia caused by occlusion of the left anterior descending artery. Cardiac function improved during ischemia after repetitive ischemia, suggesting the improvement of coronary blood flow. A capillary-specific Cre driver (Apln-CreER) was used for lineage tracing capillary endothelial cells. ROSA mT/mG reporter mice crossed with the Apln-CreER transgene mice underwent a 17 days’ repetitive ischemia protocol for coronary collateral growth. Two-photon and confocal microscopy imaging of heart slices revealed repetitive ischemia-induced coronary collateral growth initiated from sprouting Apelin⁺ endothelial cells. Newly formed capillaries in the collateral-dependent zone expanded in diameter upon repetitive ischemia stimulation and arterialized with smooth muscle cell recruitment, forming mature coronary arteries. Notably, pre-existing coronary arteries and arterioles were not Apelin⁺, and all Apelin⁺ collaterals arose from sprouting capillaries. Cxcr4, Vegfr2, Jag1, Mcp1, and Hif1α mRNA levels in the repetitive ischemia-induced hearts were also upregulated at the early stage of coronary collateral growth, suggesting angiogenic signaling pathways are activated for coronary collaterals formation during repetitive ischemia.

Conclusions –

We developed a murine model of coronary collateral growth induced by repetitive ischemia. Our lineage tracing study shows that sprouting endothelial cells contribute to coronary collateral growth in adult mouse hearts. For the first time, sprouting angiogenesis is shown to give rise to mature coronary arteries in response to repetitive ischemia in the adult mouse hearts.

Graphical Abstract

1. Introduction

Coronary heart disease is the leading cause of mortality in the United States. It will continue to be a dominant basis of mortality because about 21 million individuals over 20 years of age live with coronary heart disease [1]. In patients with coronary artery disease, the gradual occlusion of the coronary artery by an atheroma is one of the etiologies leading to ischemic heart disease (IHD). An adaptation to this pathological progression is the development of the coronary collateral circulation, which occurs in roughly 50 percent of all patients. However, a robust collateral circulation only develops in a small fraction of patients [2]. Coronary collaterals are anastomoses between two adjacent branches of the coronary arterial tree, allowing blood to bypass the atherosclerotic lesion and supply the under-perfused myocardium downstream of the stenosis, thereby restoring oxygen supply [3]. The development of the coronary collateral circulation, i.e., abluminal expansion of the collateral vessels, decreases the incidence of adverse cardiovascular outcomes in patients with CAD [4]. Stimulating coronary collateral growth (CCG) could be a potential treatment to increase the size and number of collaterals to accommodate an arterial occlusion [5, 6]. Currently, the postnatal process of CCG is not entirely understood in terms of the specific determinants such as duration, cell types and signaling pathways at different time points, and metabolic factors.

While studies of large animal models of CCG have a long history [7–10] and small animal (rat) models of CCG have been implemented for years [11–14], the interest in a mouse model of CCG has recently emerged to better study the process of CCG using genetically modified and lineage tracing murine models. Several mouse models of CCG have been reported [15–19], and these studies have yielded new insights into this incompletely understood process. Overall, the appearance of coronary collaterals has been attributed to the expansion of pre-existing arterial-arterial connections and de novo formation or artery re-assembly in the mouse myocardial infarction (MI) model [18–20]. However, details to resolve the possibilities are lacking even with current reports from murine models.

To investigate the specific determinants of postnatal CCG, such as duration and cell types, we developed a robust mouse model of CCG induced by repetitive ischemia (RI). Notably, high-resolution micro-computed tomographic (micro-CT) analysis validated such well-developed coronary collaterals. This mouse model enabled us to investigate the process of CCG and determine the origin of cells that populate coronary collaterals. We hypothesized that capillary ECs could contribute to the coronary collateral formation in an adult mouse heart after RI. To test this hypothesis, we used a capillary-specific Cre driver (Apln-CreER) to trace the fate of capillary ECs during RI. Apelin (Apln) regulates cell proliferation, migration, inflammation, apoptosis, and autophagy [21]. Apln-creER mice crossed with fluorescent reporter mice have been utilized to label sprouting ECs to elucidate vascularization processes in the mouse heart in a mouse MI model [18]. In this study, we used Apln-CreER mice crossed with Gt (ROSA)26Sor^{tm14(CAG-tdTomato)Hze}/J (mT/mG) reporter mice to study the role of capillary ECs in CCG in adult hearts by cell lineage tracing. We found that CCG induced by RI initiates from sprouting angiogenesis, which was followed by vascular maturation, leading to larger diameter arterial blood vessels. Such findings contradict the traditional dogma that all coronary collaterals expand from pre-existing vessels.

2. Materials and Methods

2.1. Animals

C57BL6/J (WT), mT/mG reporter (Gt (ROSA)26Sor^{tm14(CAG-tdTomato)Hze}/J), and Tie2-cre (B6.Cg-Tg[Tek-cre]1Ywa/J) mice were purchased from Jackson Laboratories (JAX). Apelin (Apln)-creER mice were generated by Dr. Bin Zhou [18, 22, 23]. To generate lineage-tracing mice that specifically express endogenous fluorescent proteins in ECs, we crossed mT/mG mice with Tie2-cre or Apln-creER.

2.2. RI Surgery and Protocol

A pneumatic occluder was placed over the LAD to induce CCG by RI. The RI protocol was introduced by inflation of the occluder using the following protocol: 6 minutes of occlusion every 3 hours, 4 times a day for 17 consecutive days. Echocardiography was performed to test the occluder placement at day 0 as described previously [11], and again at the end of RI to assess the cardiac function.

2.3. Retrograde Microfil® perfusions of mouse hearts and micro-CT analysis

The Microfil® polymer was perfused until the coronary arterial circuit was completed, and cleared hearts were imaged and scanned using a Bruker Skyscan 1172 micro-CT scanner and reconstructed using Avizo software. Analyze 14.0 was used for tree map analysis of micro-CT images.

2.4. Tissue slicing, sectioning, and imaging

Mouse hearts were embedded in agarose and sectioned with a vibratome. Images were acquired with confocal and multi-photon microscopy. Images were analyzed by Image J for quantitation.

2.5. RNA isolation and quantitative real-time RT-PCR (qRT-PCR)

RNA extraction and qRT-PCR were performed as previously described [24, 25]. The specificity of the primers was verified by Primer-BLAST.

2.6. Graphics

BioRender software was used for the lower panel of graphic abstract with license number DU238LPSHM.

2.7. Statistical analysis

Data have been presented as mean ± the standard deviation (SD). Data were analyzed using PRISM 7.0 statistical software. A one-way or two-way analysis of variance (ANOVA) was made between groups. Comparisons between two groups were made using unpaired Student’s t-test. The parametric test was used when the normal distribution test showed that the data distributions did not significantly differ from normal. Otherwise, a nonparametric test was used. Differences were considered statistically significant at a value of P < 0.05.

3. Results

3.1. RI stimulates CCG in adult mouse hearts

A robust mouse model of CCG induced by short episodes of RI was developed. In this model, the LAD is periodically occluded with a pneumatic snare to induce ischemia over 17 days (Figure 1A–B). A short period of inflation of the occluder induces the ischemia, but deflation of the occluder restores the blood flow; therefore, such a RI model does not cause MI (caused by ligation of LAD), which causes a permanent ischemic injury. Therefore, cardiomyocyte death does not occur in our RI model. Figure 1A shows the timeline of the RI protocol; echocardiography was performed to assess the cardiac function on day 0 and day 17 with and without the inflation of occluder. The position of the pneumatic snare over the LAD is shown in Figure 1B–C. Coronary arteries were visible in a Microfil® perfused heart after clearing under widefield microscopy (1C). In theory, after complete ligation of the LAD, retrograde Microfil® perfusion would fill the distal LAD below the ligation only if collaterals are present; lack of Microfil® perfusion of the distal LAD indicates the absence of collaterals. Retrograde Microfil® perfusion below the ligation (white circle) was absent in the naïve heart (1D), indicating there are no pre-existing coronary collaterals in the mouse heart, consistent with the literature [20]. However, after 17 days of RI stimulation, the vessels under the ligation point of the LAD in the RI hearts (1E) were perfused through the anastomoses developed during the RI, suggesting that a robust collateral circulation compensated the ischemia caused by the ligation of LAD.

Figure 1.

Mouse model of coronary collateral growth by repetitive ischemia (RI). (A) Timeline of the RI protocol. (B) Schematic showing surgical placement of pneumatic snare around left anterior descending (LAD) artery for RI protocol. Normal zone (NZ) denotes the area above the pneumatic snare and collateral-dependent zone (CZ), the area below the pneumatic snare (ischemic during snare inflation). During the occlusion of LAD, CZ blood flow is dependent on the degree of coronary collaterals. (C) Brightfield (BF) image of a Microfil® perfused heart showing pneumatic snare placement over the LAD artery. (D-E) BF image of a Microfil® perfused mouse hearts without (D) and with (E) RI. (F-G) Micro-computed tomographic (micro-CT) reconstruction of hearts without (F) and with (G) RI. D and F represent filling of the coronary arterial circulation upon acute LAD artery ligation. Note the large area (white oval) without filling, indicating a lack of collaterals or vessels too small to be visualized. E and G show extensive collateral growth (white oval) in a mouse heart subjected to RI. (H) Magnification of G. In F-H, the white filling represents the right coronary artery (RCA) perfusion territory, green is the septal artery, blue is the LAD, and yellow and purple vessels denote likely and definite collaterals, respectively.

Even though the collaterals are visible in a Microfil® perfused heart with traditional microscopy, it is hard to tell where the anastomoses are in such 2D images and how they form. To better understand this, we combined the retrograde perfusion of Microfil® with high-resolution micro-CT scans. A 3D reconstruction of micro-CT images of a Microfil® perfused heart made a direct visual confirmation of collateral growth possible. A 3D reconstruction of micro-CT images of the coronary vasculature (1F) confirms no collaterals in the distal LAD area (white circle) of the heart shown in 1D. Interestingly, a 3D reconstruction of the high-resolution micro-CT images of the coronary vasculature (1G-H) validated the formation of newly developed coronary collaterals in the RI heart (1E), where collaterals (shown in purple and yellow) were visualized connecting the arterial trees from the septum (green) and LAD (blue).

Moreover, individual collaterals can be seen connecting branches from LAD-septal and LAD-RCA arterial trees. The movies in the supplementary materials show the micro-CT images of coronary perfusion in WT and WT RI hearts with ligation of LAD and provide a 360° 3D view of F and G, showing where the anastomoses are. This is the first report showing postnatal mouse CCG by high-resolution micro-CT imaging, evidence that anastomoses were developed between LAD-RCA and LAD-septal arterial trees. Notably, the results show that well-developed coronary collaterals are sufficient to compensate for the ischemia caused by the acute occlusion of LAD, suggesting the therapeutic significance of CCG in IHD.

3.2. RI promotes global vascular branching in the mouse heart

Transluminal distribution and connectivity of coronary collaterals have been shown in 3D reconstruction images within human hearts [26], but there are no reports regarding the mouse heart. The high-resolution micro-CT image and a 3D reconstruction provide new insights into where anastomoses are in the mouse hearts. These exciting results spurred us to investigate the coronary vascular architecture and the connectivity of coronary collaterals in the mouse heart. Mouse hearts are widely used in the research of IHDs; therefore, it could be helpful to investigate their coronary vascular architecture.

We accomplished this using Analyze 14.0 tree analysis of Microfil® perfused WT mouse hearts without acute LAD ligation (Figure 2). Figures 2A and 2B show the tree map with arterial branches and branch-ends of a native mouse heart and a mouse heart undergone with 17-day RI. The coronary vascular tree starts from the aorta (shown in yellow); daughter branches are shown in blue, and the end branches are shown in red. The quantitation of total branches and end branches of mouse hearts with and without CCG in Figure 2E shows more coronary branching and connections in the mouse with CCG, suggesting RI promotes global branching in the whole heart, not only locally in the ischemic area or CZ.

Figure 2.

Coronary vascular architecture and vascular tree analysis of micro-computed tomographic reconstruction images of Microfil® perfused hearts. (A) Vascular tree from a naïve mouse heart. (B) Vascular tree from a mouse heart with RI. (C) Total arterial branches and branching ends in the heart from A. (D) Total arterial branches and branching ends in heart from B. Yellow (aortic root), red (end branches), blue (parent branches). (E) The numbers of total branches and branch ends of coronary vascular tree in mice with coronary collateral growth (CCG) and without CCG (n=3 mice per group, two-way AN OVA (Sidak’s multiple comparison test) was used for statistical analysis, *P <0.05).

Figure 3 shows the architecture of the main branches of coronary arteries and arterial branches and branch-ends of the left coronary artery (LCA), right coronary artery (RCA), and the septum. Compared to hearts without CCG (3A), the hearts with CCG (3B) clearly show more parent (blue) and end branches (red) in LCA, RCA, and septal arterial trees. Interestingly, the LCA had many more branches and branch ends than RCA and septum (3C), suggesting the mouse has a left dominant coronary system which might be an excellent model to study the disorders of the left dominant system [27, 28]. Interestingly, the total branches/end branches ratio is around 0.5 in all arterial trees (3C). These data indicate that RI stimulates CCG and greatly enhances cardiac perfusion by expanding the coronary circulation, which is consistent with a previous report employing a rat CCG model [13]. The tree map analysis of 3D high-resolution micro-CT images of coronary architecture provides an objective global view of the vasculature in the mouse heart compared to traditional histology of several microns tissue section or microscopy, especially regarding the 3D view of vasculature in different regions, branching, and connections between LCA, RCA, and septal branches. So high-resolution micro-CT is a potential alternative to histology in studying the development of coronary arteries and coronary collaterals.

Figure 3.

Coronary vascular architecture and vascular tree analysis of micro-computed tomographic reconstruction images of main branches of Microfil® perfused hearts. (A) Vascular trees from a mouse without coronary collateral growth (CCG). (B) Vascular trees from a mouse with CCG. Maps of arterial branches and branching ends for trees from the left coronary artery (LCA), the right coronary artery (RCA) and septum are shown, respectively. Yellow (aortic root), red (end branches), blue (parent branches). (C) The numbers of branches and branch ends in LCA, RCA and Septum of A and B.

3.3. Apln⁺ GFP⁺ cells specifically label sprouting ECs compared to Tie-2⁺GFP⁺ cells

With a robust mouse model of CCG, we were positioned well to study which cell types are involved in CCG and at what sequence. The traditional dogma of CCG is that collaterals are from the expansion of the pre-existing vessels. Though de novo mouse coronary collateral formation has been induced by MI, there was no evidence of whether the vessels were pre-existing without lineage tracing. Moreover, the mechanism of CCG in the MI model might be completely different from the RI model. We hypothesized that capillary ECs contribute to the CCG in our RI model. Therefore, two endothelial cell-specific cre mice were used to trace endothelial cell fate during RI. Tie2 is a pan-EC marker used to label capillary and arterial ECs [29, 30], and Apln is a marker that labels sprouting ECs [18, 22, 23]. We included the Tie-2 cre to serve as a positive control for visualizing the coronary collaterals, uncertain of whether capillary ECs contribute to the coronary collateral formation, a potential possibility of no GFP⁺ (Apln⁺) signaling at all.

To label sprouting ECs in adult mouse hearts with or without RI stimulation, we induced Apln-creER activation in mT/mG x Apln-creER mice using tamoxifen and mapped capillary sprouting in a specific temporal window. For comparison, Figure 4 shows stitched two-photon images of native heart slices from mT/mG x Tie2-cre (4A) and mT/mG x Apln-creER mice (4C). Figures 4B and 4D are the closeups of ROI from Figures 4A and 4C, respectively. As expected, pan-endothelial marker Tie2 labeled EC in blood vessels of all sizes throughout the heart (Figures 4A–B). It was impossible to distinguish whether Tie2⁺ cells were pre-existing before RI or proliferated during RI. Figure 4E, a closeup of 4B, shows that non-endothelial cells were red because of no occurrence of cre excision, while endothelial cells were GFP⁺ (green) because Tie2-cre excision occurred in all endothelial cells. In contrast, Alpn labeled more specifically only the sprouting endothelial cells (Figure 4C–D). Figure 4F, a closeup of 4D, shows that all other cell types were red except for sprouting endothelial cells that were GFP⁺ (green) because Apln-cre excision occurred in newly sprouting endothelial cells. These results confirmed the specificity of Apln-CreER for newly sprouting endothelial cells as reported [24, 37, 38] and showed that Apln signaling is detectable and visible in our model, indicating capillary endothelial sprouting occurs. Therefore, only Apln-CreER was used for the lineage tracing of sprouting endothelial cells for collateral formation in the rest of our studies.

Figure 4.

Two-photon microscopic images of native heart slices (1mm) from pan-endothelial (Tie2-cre⁺) and actively sprouting endothelial (Apln-cre⁺) mT/mG mice. (A) Native Tie2-cre⁺ heart with cells expressing Tie2 (GFP⁺). Regions of interests (ROIs; pink boxes) in A are shown in greater detail in B and E, where some Tie2⁻cells are also Td-tomato⁺. (C) Native Apln⁺ heart with cells expressing Apelin (GFP⁺). ROIs in C are shown in greater detail in D and F, where apelin⁻ cells are Td-tomato⁺. Apln⁺ GFP⁺ cells represent endothelial sprouting in the native heart.

Additionally, two-photon microscopy revealed that EC sprouting occurs in the left and right ventricles as well as the septum (Figure 4C). Short capillaries were labeled with GFP (GFP⁺) throughout the heart. Such an observation of global EC sprouting supports the previous results (Figure 2–3) that RI promotes global branching in the whole heart, not just in the CZ.

3.4. RI stimulates endothelial sprouting and coronary collateral formation in the adult mouse heart

With a capillary endothelial cell-specific tracing marker, we were able to test our hypothesis of whether capillary EC sprouting contributes to coronary collaterals. mT/mG x Apln-creER mice were subjected to 17 days’ RI after 5 days standard tamoxifen induction, but a lower tamoxifen dosage was used during RI. Thus, all sprouting endothelial cells during collateral growth were labeled. To better identify the newly-formed collaterals, we defined the areas of the mouse hearts with RI according to the position of the snare: normal zone (NZ) denotes the area above the pneumatic snare where no ischemia occurs with or without the inflation of the occluder, and collateral dependent zone (CZ) denotes the area below, which is ischemic during snare inflation if there are no collaterals present (Figure 1B). In theory, the ischemia or shear stress in the CZ during RI will lead to more blood vessel growth [43]. Accordingly, we defined the areas of non-surgical control hearts as apex and base relative to the CZ and NZ in the RI hearts. Supplementary Figure I shows that RI stimulates endothelial sprouting and capillary formation in the adult mouse heart. Stitched two-photon images show sprouting ECs in sham hearts (with surgery and occluder, but without RI stimulation, Figure IA) and RI hearts (Figure IB). More Apln⁺ sprouting ECs can be observed in RI hearts than in sham hearts. Since all Apln⁺ cells expressed GFP, we used GFP⁺Apln⁺ to refer to these cells. Interestingly, a greater abundance of GFP⁺Apln⁺ ECs are visible in the RI heart, and these GFP⁺Apln⁺ ECs constitute longer capillary segments (Figure IE–H) than the sham groups (C-D). We also found that GFP⁺Apln⁺ sprouting ECs make up a more significant fraction of capillaries in hearts with more prolonged RI stimulation (17 days, Figure I H) than in hearts with shorter RI duration (8 days, Figure IF), suggesting a longer time of RI is essential for larger collateral formation.

3.5. Endothelial cell sprouting and proliferation during early and late stages of CCG during RI

Our data shown so far supported our hypothesis that sprouting ECs contribute to collateral growth. Next, we characterized endothelial sprouting within temporal windows. We first asked whether these GFP⁺Apln⁺ ECs proliferated after sprouting and how long the proliferation lasted. We compared the GFP⁺Apln⁺ cells and proliferating (EdU⁺) GFP⁺Apln⁺ cells in RI hearts subjected to different durations of RI to controls. We checked the EC proliferation at the early stage (3-5 days of RI) versus the late stage of CCG (8-10 days of RI). Figure 5 shows that sprouting ECs (GFP⁺Apln⁺ cells) had been actively proliferating at the early stages of CCG during RI, and they are visible in both NZ and CZ during the early stage of CCG. Notably, however, more GFP⁺Apln⁺ cells were present in the CZ (Figure 5B) than in the NZ, where relatively few GFP⁺Apln⁺ cells are present (Figure 5A), suggesting RI stimulated more capillary EC sprouting in CZ than NZ.

Figure 5.

Sprouting and proliferating endothelial cells (ECs) in the mouse heart during early stages of coronary collateral growth (CCG; induced by 5 days of repetitive ischemia [RI]). Images of normal zone (NZ) (A) and collateral dependent zone (CZ) (B) of mT/mG x Apln-cre mice after 5-day RI, showing more sprouting ECs in the CZ than ECs in NZ. Apln⁺ cells are GFP⁺. (C) 5-ethynyl-2’-deoxyuridine (EdU) labeling scheme of proliferating cells. EdU staining in the NZ (D) and CZ (E) of mT/mG x Apln-cre heart after 2-day’s EdU injections. Apln⁺ cells are labeled with anti-GFP antibody. (F) High magnification of EdU⁺ staining images in control (con) apex and RI CZ with red arrows marking co-localization of EdU⁺ and GFP⁺. (G-H) Quantitation of EdU⁺ nuclei (proliferating cells) and EdU⁺Apln⁺ (sprouting and proliferating EC) shows hearts with RI have more EdU⁺ or EdU⁺Apln⁺ cells (n= 12 fields, 4 fields per animals, unpaired t-test with Welch’ correction was used for statistical analysis, *P<0.05)

To further confirm the proliferation of GFP⁺Apln⁺ cells, EdU was used to label the proliferating cells. Figure 5C shows the scheme of EdU injection: EdU was injected 3 times a day (every 8 hours) in the RI mice on days 3-4 of RI or in the control mice. Mice were sacrificed 24 hours after the last injection of EdU, and heart tissue was imaged.

Few cells incorporated EdU (pseudo-colored orange) in the NZ (Figure 5D), while many more EdU⁺ cells were observed in the CZ (Figure 5E). High magnification confocal images reveal a greater abundance of EdU⁺ cells in the RI heart’s CZ than the apex of the control heart (Figure 5F). Quantification of EdU staining (Figure 5G) shows more EdU⁺ cells in the RI hearts than the control hearts. Also, the number of GFP⁺Apln⁺/EdU⁺ cells was higher in RI hearts than control hearts (Figure 5H), indicating that sprouting ECs were actively proliferating at the early stages of CCG in CZ in response to RI stimulation. An anti-GFP antibody was used in Figure 5D–E to detect GFP⁺Apln⁺ because endogenous GFP faded during EdU staining. These data show that capillary ECs actively sprouted and proliferated in the CZ during the early stages of CCG induced by RI.

To investigate how long the sprouting ECs continue to proliferate in the process of CCG, we used EdU and BrdU double labeling to trace the sprouting ECs in the RI hearts. The dual-tracing scheme is shown in Figure 6A. EdU was injected every 8 hours between 3-5 days of RI, and BrdU was injected every 8 hours between 8-10 days of RI. EdU/BrdU double staining was performed to compare the numbers of dividing cells during the two temporal windows. Figures 6B and 6C show EdU/BrdU double staining in the NZ and CZ in the RI hearts. Consistent with the previous results in Figure 5, there was a greater abundance of GFP⁺Apln⁺EdU⁺ cells (co-localization of yellow and green) in the CZ than in the NZ. In contrast, there were fewer GFP⁺Apln⁺BrdU⁺ cells (colocalization of blue and green) than GFP⁺Apln⁺EdU⁺ cells in both NZ and CZ regions (Figure 6B and C). High magnification images of Figures 6B and 6D revealed a scarcity of BrdU⁺ cells in CZ and NZ (Figure 6D and E). Quantification of GFP⁺Apln⁺BrdU⁺ staining (Figure 6F) shows that the number of GFP⁺Apln⁺BrdU⁺ cells is lower than that of GFP⁺Apln⁺EdU⁺ cells and supports the observation in Figure 6B–E, suggesting the EC proliferation is less active in the late stage of CCG (8-10 days of RI) than that in the early stage of CCG (3-5 days of RI). These results are interesting, and while we do not know why the proliferation of ECs decreased, but a recent report showing that arterialization requires the timely suppression of endothelial growth during embryo development [33] might imply that less active proliferation of ECs is required for arterialization even though our collateral formation is postnatal which might be different from embryonic development.

Figure 6.

Proliferating endothelial cells (ECs) in the heart during late stages of coronary collateral growth (CCG; induced by 8-10 days of repetitive ischemia [RI]). (A) Labeling scheme of proliferating cells with 5-ethynyl-2’-deoxyuridine (EdU) and 5-bromo-2’-deoxyuridine (BrdU). Images of normal zone (NZ) (B) and collateral dependent zone (CZ) (C) of mT/mG x Apln-cre mouse hearts after 17-day RI, showing more proliferating ECs in the CZ than NZ. Apln⁺ cells are labeled with anti-GFP antibody. (D and E) High magnification of EdU/BrdU staining in NZ (D) and CZ (E). (F) Quantitation of Apln⁺EdU⁺ nuclei (GFP⁺ proliferating ECs at the early stage of RI) and Apln⁺BrdU⁺ nuclei (GFP⁺ proliferating ECs at the late stage of RI) in NZ and CZ (n= 12 fields, 4 fields per animals, two-way ANOVA (Sidak’s multiple comparison test) was used for statistical analysis, *P <0.05).

3.6. Newly-formed capillaries expand into larger diameter coronary collaterals in response to RI

Capillaries are small vessels with diameters of less than 10 μM [34], and the diameter of collaterals induced by myocardial infarction ranges from 15 μM to less than 20 μM [16]. To further confirm the collateral formation in our CCG model, we investigate the diameter and size of collaterals induced by RI. As shown in Supplementary Figure I, at the end of the CCG (17 days of RI), sprouting ECs expanded and formed longer vessels in the CZ of mT/mG x Apln-creER mouse hearts compared to the sham hearts. To better characterize these newly formed GFP⁺Apln⁺ blood vessels, we measured the size of these vessels in RI hearts. Figure 7 shows newly formed vessels in the CZ expanded in response to RI and GFP⁺Apln⁺ vessels in the CZ (Figure 7D) were more abundant, of larger diameter, and longer than ones in the NZ (Figure 7C). However, sprouting capillary expansion did not occur in sham hearts (Figure 7A and B). An extensive network of these wide GFP⁺Apln⁺ vessels expanding from sprouting capillary were visible from the epicardial surface to about 1 mm deep in the myocardium (Figure 7E). In the CZ of sham hearts, the widest capillaries were less than 10 μM in diameter (Figure 7B). However, the most prominent vessels in RI-treated hearts were 20-30 μM wide (Figure 7D and E), larger than the collaterals induced by MI as reported [16]. Capillaries in the NZ in sham and RI hearts were approximately 5 microns in diameter. These results show for the first time that newly formed vessels developed from sprouting capillaries can expand four-fold in diameter in the adult mouse heart in response to RI stimulation, suggesting a robust development of collaterals with a larger diameter and longer length after RI stimulation.

Figure 7.

Newly formed capillary expansion in response to repetitive ischemia (RI). (A-B) Images of the normal zone (NZ) and collateral dependent zone (CZ) from mT/mG x Apln-cre sham heart. (C-D) Images of the NZ and CZ from mT/mG x Apln-cre heart after 17-day RI. Capillary expansion occurs more in the CZ than NZ of RI heart in response to RI or more in the NZ or the CZ of the sham heart. GFP⁺ vessels in the CZ from the RI heart are more abundant, of larger diameter and more prolonged than in the NZ. An extensive network of these remodeled capillaries can be seen in E and newly formed, and expanded capillaries can be seen up to a depth of 1000 μM from epicardial surface. Approximate diameters of these large vessels (up to 30 μM) are shown.

3.7. The arterialization of newly-formed blood vessels

We have seen that the coronary collaterals developed during RI are larger than capillaries, and robust collateral circulation is shown by Microfil® perfusion, suggesting these vessels are functional collaterals; however, collaterals, by definition, are arterial anastomoses. Therefore, we investigated whether smooth muscle cells were invested around these newly-formed vessels after RI stimulation. We further determined if vascular smooth muscle cells (VSMCs) presented in the expanded GFP⁺Apln⁺ vessels in the RI hearts. Figure 8 shows in detail newly-formed vessels after 17 days of RI. Interestingly, Figure 8B shows tortuous GFP⁺Apln⁺ vessels in the CZ and a tdTomato⁺ outer layer accompaning the GFP⁺Apln⁺ endothelium of these newly-formed vessels.

Figure 8.

Arterialization of newly formed blood vessels. (A-D) Images of mT/mG x Apln-cre mouse heart with 17-day repetitive ischemia (RI). (B) Tortuous, newly-formed blood vessels with GFP⁺ endothelial layer and tdTomato⁺ outer layer in the collateral-dependent zone (CZ), indicating vessel maturation with a layer of mural cells. (C) Single z-slices from the z-stack in B revealed that the outer layer marked with yellow arrows is continuous at every z-plane. The Smooth muscle alpha-actin (αSMA) layer encompassing the GFP⁺ vessel is present in the CZ (E) but not in the normal zone (NZ) (D). Single z-slices (F) from the z-stack in E showed that the α-SMA layer is present outside the endothelial monolayer at every z-plane marked with yellow arrows, suggesting the maturation of sprouting capillaries with smooth muscle cell recruitment.

Single z-slices from the z-stack in B (Figure 8C) revealed that this tdTomato⁺ outer layer surrounding the GFP⁺Apln⁺ endothelium was continuous at every z-plane, suggesting the cells that covered GFP⁺Apln⁺ endothelium of newly formed vessels were pre-existing (labeled in red), not originating from sprouting ECs (labeled in green). This is consistent with the previous data (Figure 5) showing EC sprouting on days 3-5 of RI, where sprouting capillaries were not enveloped by other cell types (labeled in tdTomato⁺), suggesting that VSMCs recruitment did not occur at the early stage of CCG.

To determine the cell types of the tdTomato⁺ outer layer, immunostaining of these vessels with anti-α-SMA antibody was performed. We found α-SMA was expressed in the outer layer encompassing the newly formed GFP⁺Apln⁺ vessels in the CZ (Figure 8E) but not in the NZ (Figure 8D). Single z-slices from the z-stack in 8E (Figure 8F) show that the α-SMA layer exists outside the GFP⁺Apln⁺ endothelium at every z-plane. The presence of VSMCs in the outer layer of the newly formed vessels at the end of RI suggests VSMCs recruitment occurred when GFP⁺Apln⁺ vessels expanded to a larger diameter in the CZ. VSMCs coverage outside the GFP⁺Apln⁺ vessels also indicates maturation and arterialization of the sprouting capillaries. Therefore, VSMCs were recruited to expand vessels during the late stage of RI as part of the maturation of collaterals. These results suggest arterialization, not arteriogenesis, as the process behind RI-induced CCG in mouse hearts.

3.8. Hypoxia during early RI

Sprouting angiogenesis is reported to be initiated in response to specific physiological triggers, such as hypoxia [35]. To test whether hypoxia is involved in our CCG model, we injected Hypoxyprobe® i.v. into mice after three days of RI and harvested the hearts for immunohistochemical analyses. Anti-pimonidazole staining was abundant in the RI hearts compared to controls (Figure 9). The CZ of RI hearts (Figure 9F) had noticeably more positive staining than the apex of controls (Figure 9C). This result shows that hypoxia occurs in the CZ of the RI-treated heart, suggesting hypoxia might be involved in the sprouting at the early stage of CCG. This finding is further corroborated by a recent report in which mice subjected to reduced F1O₂ developed coronary collaterals [17].

Figure 9.

Hypoxyprobe® staining of wild-type (WT) mouse cardiac tissue. (A-B) Base and apex of non-surgical control heart. (C) High magnification of B. (D-E) Normal zone (NZ) and collateral-dependent zone (CZ) from 3-day repetitive ischemia (RI) hearts (F) High magnification of E. (G) The quantitation of green fluorescence of hypoxia probe in the control hearts and RI hearts. 4’,6-diamidino-2-phenylindole (DAPI) (blue), Hypoxyprobe® (green), Wheat Germ Agglutinin; WGA (red) (n= 12 fields in control and n=16 fields in RI group, 4 fields per animal, unpaired Mann-Whitney test was used for statistical analysis, *P <0.05).

3.9. Signaling pathway(s) involved at an early stage of CCG

The initiation of collateral growth is essential, and understanding the key factors will help us better regulate CCG. To investigate the critical signaling pathway(s) contributing to the initiation of CCG, we harvested cardiac tissue from non-surgical control, sham control, and RI-treated mice at 3 days of RI and ran qRT-PCR for mRNA expression. Figure 10 suggests that C-X-C motif chemokine receptor 4 (Cxcr4), Vascular Endothelial Growth Factor Receptor 2 (Vegfr2), jagged canonical Notch ligand 1 (Jag1), Hypoxia Inducible Factor subunit alpha (Hif1α), and Monocyte Chemoattractant Protein-1 (Mcp-1) signaling is involved in CCG. Compared to the sham control, Cxcr4, Vegfr2, Jag1, and Hif1α mRNA expression levels were significantly increased in RI mice, suggesting these genes are critical to the collateral formation. Compared to the non-surgical control, Cxcr4, Vegfr2, Jag1, and Mcp-1 mRNA levels were significantly increased in RI mice, suggesting an inflammatory response was triggered during RI surgery. Interestingly, Hif1α expression in the CZ was markedly higher in the RI group vs. non-surgical and sham controls, indicating that the HIFα signaling cascade was specifically activated by the RI-induced ischemia and supporting the previous results in Figure 9 showing hypoxia present in the CZ during RI by Hypoxyprobe® staining. These data suggested hypoxia might be the critical factor in inducing CCG during RI, consistent with a recent report showing that hypoxia alone can induce collateral growth [17].

Figure 10.

mRNA expression in mouse hearts at early stages (3 days) of repetitive ischemia (RI) by qPCR. (A) Compared to non-surgical controls, C-X-C motif chemokine receptor 4 (Cxcr4), Vascular Endothelial Growth Factor Receptor 2 (Vegfr2), jagged canonical Notch ligand 1 (Jag1), expression in the mouse hearts with RI were significantly increased (**P<0.01); compared to the sham control, Cxcr4, Vegfr2, Jag1 and Hypoxia Inducible Factor subunit alpha (Hif1α) were significantly increased (n=7 mice/group, one-way ANOVA (Holm-Sidak’s multiple comparison test) was used for statistical analysis, *P < 0.05, **P < 0.01). (B) Monocyte Chemoattractant Protein-1 (Mcp-1) was upregulated in mouse hearts with RI compared to sham controls (n=9 in the control group and n=4 in the RI groups, unpaired Mann-Whitney U test was used for statistical analysis, *P < 0.05). (C) Hif1α was upregulated in the collateral-dependent zone (CZ) of RI mouse hearts compared to non-surgical controls and sham controls (n=5 in the control group, n=4 in the sham groups, and n=3 in the RI groups, one-way ANOVA (Turkey multiple comparison test) was used for statistical analysis, *P < 0.05, **P < 0.01). (D) Functional protein association networks via STRING show the signaling pathway linking genes with differences in A-C. (E-G) Functional protein association networks via STRING show signaling pathways of CXCR4, JAG1 and VEGFR2.

To understand how CXCR4, VEGFR2, JAG1, HIF1α, and MCP-1 are orchestrated in regulating CCG induced by RI, we used functional protein association networks to link CXCR4, VEGFR2, JAG1, HIF1α, and MCP-1 together. Figure 10D shows the signaling pathway functional networks of CXCR4, VEGFR2, JAG1, HIF1α, and MCP-1 via STRING (https://string-db.org). CXCR4 interacts with MCP-1 (CCL2). JAG1, HIF1α, and VEGFR2 (KDR). JAG1 interacts with CXCR4, HIF1α, and VEGFR2 (KDR). VEGFR2 interacts with MCP-1 (CCL2). JAG1, HIF1α, and CXCR4. MCP-1 (CCL2) interacts with CXCR4, HIF1α, and VEGFR2 (KDR). Figures 10E–G show the major signaling pathways of CXCR4, JAG1, and VEGFR2 by functional protein association networks, respectively. CXCR4 interacts with C-X-C Motif Chemokine Ligand 12 (CXCL12), which plays a fundamental role in cardiovascular development, cell trafficking, and myocardial repair, and CXCL12 induces artery assembly in MI-induced CCG [19] [36]. JAG1 interacts with Notch homolog 1, translocation-associated (Notch), which is activated by shear stress [37]. VEGFR interacts with Vascular Endothelial Growth Factors (VEGF), which is involved in cerebral and coronary collateral growth. These signaling pathways are all essential for blood vessel growth. Further studies with genetic knockout mice will warrant the detailed mechanism of regulating the CCG after RI.

3.10. Functional estimation of CCG

For mice undergoing the RI protocol, we evaluated CCG by measuring the cardiac function with total LAD occlusion by inflation of the occluder at days 0 and 17 as previously described [11, 13]. The rationale for this procedure was that if collaterals were developed, occlusion would not induce functional disturbances, and cardiac function would not change significantly with the occlusion of LAD by inflation of the snare. Alternatively, if collaterals were not mature, then occlusion of LAD would cause hemodynamic disturbances, and cardiac function would decrease. Suppl Fig III shows that after 17 days’ RI, changes in ejection fraction (EF) and fractional shortening (FS) when the LAD was occluded were significantly less than at day 0, suggesting that cardiac function improved during ischemia with CCG.

4. Discussion

Coronary collaterals are a natural coronary bypass in patients with IHD. Patients with well-developed coronary collaterals have a better prognosis in recovering from a MI than those with poorly-developed collaterals [5, 38]. Risk factors such as metabolic syndrome increase the risk of IHD while seemingly impairing CCG. Specifically, 30-40% of these patients show little or no CCG [39–41]. Before therapeutic implementation, it is imperative to understand the underlying processes of CCG for amplification in a treatment. Accordingly, we describe a mouse model of CCG stimulated by episodic RI that mimics the pathological presentation of patients with angina. Such a mouse model can open the doors for studying the regulation of coronary collateral formation with emphasis on the cellular and molecular contributors employing lineage tracing and genetic modification, which previous studies are lacking.

4.1. Significance of the studies

We developed a robust mouse model of CCG induced by short episodes of RI, which was validated by high-resolution micro-CT to show arterial-arterial anastomotic connections. Significantly, a well-developed collateral circulation functionally compensates for the deficiency of blood flow and significantly improves cardiac function during the occlusion of the LAD because coronary blood flow and cardiac function are directly correlated.

The current study integrated different techniques to evaluate the coronary collaterals such as Microfil® perfusion, high resolution of micro-CT and 3D reconstruction, stitched two-photon imaging, confocal microscopy, and collateral circulation functional assay. For the first time, we show mouse coronary collaterals using high resolution of micro-CT imaging to give a 3D view of the coronary circulation. These experiments revealed that collateral growth occurs between LAD-RCA and LAD-septum arterial trees, which have only been reported in other animal species [19, 42], not in the mouse hearts. High-resolution micro-CT and 3D reconstruction provide a different view of the development of coronary collaterals and will advance our study into the factors that control CCG. The software Analyze 14.0 was used to show a detailed mouse coronary architecture with quantitation of vascular branches and branch ends for the first time. Both micro-CT and two-photon microscopy are potential alternatives to classic histology for vascular imaging and analysis, which is laborious and prone to artifacts.

4.2. Experimental models of CCG

While RI protocols have previously been used for large animals [16, 18, 43],[44–46], these animals are not amenable to genetic manipulation or lineage tracing techniques; our mouse model fills in this gap. In addition, murine models are highly accessible tools for further exploring various determinants of CCG, including genetics, metabolism, shear stress, etc., in the future. Many patients with metabolic derangements, i.e., diabetes and metabolic syndrome, lack a functional collateral network and/or growth response, contributing to their poor prognosis during cardiac events [12, 47–49]. While the cause(s) of poor CCG remains elusive, our model presents a platform that can help address these biological questions and make progress in developing treatment strategies for stimulating CCG in patients with IHDs.

Our mouse model of CCG has several advantages. While an MI mouse model of CCG can recapitulate CCG post-MI, no other experimental mouse models produce a chronic short episode of myocardial ischemia under physiological conditions without wound and healing processes. This specific area of study is essential because these repetitive ischemic events, commonly presented as angina pectoris, can influence collateral development before an MI. In that way, a mouse model of short RI episodes that mimic the pathological presentation of patients with angina is clinically relevant and translational.

In our RI model, coronary collaterals developed over 17 days, while in the MI model, angiogenic remodeling peaks at 14 days post-surgery [16]. However, the diameter of the collaterals developed in our RI model is up to 30 μM, while the diameter of collaterals developed in the MI model is less than 20 μM. Also, in our RI model, the collateral circulation developed from 17 days of RI was sufficient to significantly improve cardiac function during ischemia in adult mice, suggesting the improvement of coronary blood flow.

We used lineage tracing in genetically modified mice to determine the process of RI-induced CCG. Our application of this technique revealed that CCG in the adult mouse heart initiated from capillary EC sprouting. Sprouting capillary vessels, a product of angiogenesis, can be unstable and subject to regression without proper smooth muscle or pericyte stabilization [50, 51], but we were also able to show for the first time that sprouting angiogenesis gives rise to mature blood vessels of larger diameter in response to RI in adult hearts by lineage tracing. These findings challenge the existing paradigm of CCG in animals, namely that coronary collateral vessels exclusively develop from pre-existing arteries (arteriogenesis) without any contribution from arterialization of capillaries. It also supports the report that mouse coronary collaterals formed de novo in an MI model [16]. Our new model is also different from the “artery reassembly” model in which arterial ECs migrated away from arteries and reassembled into collaterals in neonates [19].

Our results diverge from existing literature on how coronary collaterals grow [18, 19] though this could be explained mainly by the intrinsic differences between the models. Our model might produce different results from an MI model, considering that in a mouse model of acute MI alone, the expression of cytokines differs vastly depending on if the MI was produced during an acute surgical procedure versus after healing from the surgical preparation (thoracotomy) [43]. Repetitive, short episodes of ischemia over a long duration in our RI protocol appear to activate an alternative capillary arterialization pathway to CCG, in contrast to the acute ischemia caused by the conventional LAD ligation model, which evokes a different program for vascular growth [16, 18, 19]. Likely, RI triggers sprouting angiogenesis, and then increases blood flow in the CZ causes these newly formed capillaries to expand outward through a combination of signaling of HIF1“, CXCR4, VEGFR2, and JAG1 as well as vascular remodeling and capillary arterialization by shear stress.

Further, our RI model does not involve substantial loss of myocardium and alterations in the mechanical properties of the left ventricle, which are unavoidable confounding factors in studies using the MI model for CCG. Therefore, the findings of this study represent a previously unreported path to CCG. Notably, this study provides evidence of abluminal expansion and arterialization of coronary capillaries in response to RI.

4.3. The role of Apln+ EC sprouting

We found Apln to be a reliable marker for sprouting ECs. Apln signaling is involved in adaptations following injury [21], and a previous study using inducible Apln-creER lineage tracing in a MI model reported that the sprouting Apln⁺ ECs minimally incorporated into large arteries following angiogenic responses to MI [18, 23]. Our study contradicted this report even when using the same lineage-tracing mice. This again highlights a critical distinction between approaches to studying myocardial adaptations to ischemia, which may vary depending on whether there is overt tissue necrosis instead of brief episodes of ischemia. While the fate of Apln⁺ ECs under physiological conditions requires further investigation, Apln⁺ ECs appear unequivocally important for RI-induced CCG.

The other highlight of our lineage tracing study is that we induced the Apln-creER expression with standard tamoxifen for 5 days before RI, followed by a lower dosage of tamoxifen during the17 day RI protocol. Thus, all sprouting ECs during the entire process of collateral growth were labeled. This protocol was different than the traditional 5 days of tamoxifen induction which may underestimate the contribution of capillary ECs to the coronary collateral formation.

Interestingly, in our study, RI induced not only capillary sprouting and coronary collateral formation in the CZ, but also stimulated capillary sprouting in the NZ, though the sprouting in the NZ is significantly less than CZ and the sprouting capillaries did not develop into collaterals. Our micro-CT analysis also confirms more vascular branching in the whole RI heart, suggesting RI produced benefits for vasculature other than just collateral growth. This is consistent with a previous report of significant changes in capillary density (rising during an early phase and waning to control levels after growth stabilized) during collateral growth in a canine model [9]. We do not know the exact mechanism by which the RI stimulated capillary sprouting or vascular branching in the NZ, but it might be related to the remote ischemia condition, which provides a cardiac protective effect from factor(s) activated, produced, or transported throughout the heart during brief ischemia/reperfusion [52–55].

4.4. The mechanism of CCG induced by RI

In terms of the mechanism of CCG stimulated by RI, we focused more on the early stages of CCG and investigated how the CCG was initiated by RI stimulation. Interestingly, we found some common pathways are essential for both RI- and MI-induced CCG. First was CXCR4; our gene expression data show that Cxcr4 mRNA expression is increased in RI hearts compared to non-surgical controls and sham controls, suggesting CXCR4 signaling was activated during CCG stimulated by RI. Interestingly, CXCR4 induced artery assembly in an MI model of CCG in neonatal mice [19], though our data indicate a different role for CXCR4 in our CCG model from its role in the MI model. The second was hypoxia; our data show increased hypoxia in the CZ of 3-day RI hearts, indicating that a hypoxia signaling cascade was activated at an early stage of CCG. In the MI model of CCG, hypoxia could solely induce coronary collaterals in adult hearts even though collaterals regressed after 28 days’ recovery of normoxia, suggesting the essential role of hypoxia signaling [17]. Moreover, during RI, mRNA upregulation of Hif1α, an upstream regulator of vascular response to ischemia and angiogenesis [56], supports the hypoxia probe staining data showing hypoxia during RI at the early stage of RI. Hypoxia-Hif1α signaling is further supported by activation of the VEGF-VEGFR2 signaling pathway with the upregulation of mRNA expression of Vegfr2 at days 3-5 of RI in our CCG model, indicated by a dramatic increase in endothelial sprouting in RI hearts. Moreover, the mRNA expression of Jag1 was upregulated at the early stage of RI, suggesting increased endothelial sprouting as JAG1 is triggered by myocardial ischemia and regulates sprouting angiogenesis [57]. In our RI model, Mcp-1 mRNA expression was also increased in the cardiac tissues with RI compared with non-surgical controls, suggesting that MCP-1 signaling is essential in CCG induced by RI [58–60].

4.5. Clinical relevance

Various clinical trials have attempted to promote the growth of new vessels in the ischemic hearts using cytokines like VEGF, bFGF, G-CSF, or GM-CSF, but the results have been modest [44–46, 61–65]. In that regard, our RI model could help screen potential therapeutic agents that might promote CCG and elucidate the mechanisms governing the process to benefit patients who lack appropriate coronary collateral functionality. CCG is also an alternative therapeutic approach for patients with intractable angina pectoris who are not indicated for percutaneous coronary intervention and/or coronary artery bypass grafting [66–68]. Furthering our understanding of CCG remains clinically relevant, as there are instances in diabetes and metabolic syndrome where patients fail to undergo adequate CCG and succumb to ischemic damage [47, 69, 70].

4.6. Considerations and limitations of the experimental design and approach

Scientific advancements have given rise to the ability to genetically manipulate our experimental mouse models to allow researchers to perform in-depth investigations into the mechanisms surrounding physiological processes. The current study utilizes ROSA mT/mG x Apln-CreER mice to uncover novel cellular contributions during CCG via lineage tracing. However, we would like to address several limitations of our study that pertain to the model used and experimental design.

4.6.1.

Aging is a key risk factor for developing many diseases, including IHD, which the current study did not address because of open chest RI surgery challenges. The angiogenic and regenerative potential is reduced across organs and tissues in older patients and experimental animals [71]. A key point of inquiry would be a comparison of RI-induced sprouting in older vs. younger mice. EdU and BrdU staining of capillaries during the early and later stages of CCG could reveal possible differences in EC proliferation between the two groups. As there are multiple steps in our RI-induced CCG process, each or all of them could be affected in aged mice. Future studies are warranted to clearly define whether and how aging could impact the results presented here.

4.6.2.

Sex is another key biological risk factor for certain diseases. Sex-based differences in clinical presentations of heart disease have been well known for decades; for example, women are more likely to develop microvascular dysfunction [72]. Gender differences have been shown in blood flow recovery have been shown in a mouse model of hindlimb ischemia where the female mice demonstrate impaired blood flow recovery [73 ]. Potential divergences in the extent of collateralization, endothelial sprouting, and/or rescue of cardiac function between males and females would be of profound clinical importance. Age and sex as biological variables could be combined to produce even more nuanced experiments. Excluding neonate and adult models of CCG, currently, there is no information on differences between young vs. old males or young vs. old females. To this end, two cohorts of mice – male and female – were used in our experiments and we did not observe any sex differences in this study. However, sex differences could be explicitly addressed in future studies.

4.6.3.

Humans and mice differ in their coronary collateral circulation in that humans possess native collaterals while mice do not. There is also the fundamental disparity in vessel size when using mice instead of a larger vertebrate such as a pig. However, these are true of many, if not all, mouse models. Overall, we acknowledge the limitations and weaknesses of the current study, but these do not diminish the novelty and usefulness of the findings.

4.7. Conclusion and future directions

This study successfully developed a mouse model of CCG induced by RI, and the RI-induced coronary collateral circulation compensated for blood flow deficiencies and improved cardiac function during ischemia. Our lineage-tracing studies found that CCG induced by RI involves capillary EC sprouting, capillary expansion, and smooth muscle cell recruitment, also known as arterialization, suggesting capillary angiogenesis gave rise to coronary arteries in adult mouse hearts.

To better understand the process of CCG, the roles played by inflammation, hypoxia, immune cells, and angiogenic signaling must be further studied in greater detail using conventional genetic and lineage-tracing approaches. Further molecular analyses are required to discover which genes are differentially expressed in a cell-type-specific and time-dependent fashion between RI and non-RI hearts. Validation of the genetic regulation of CCG using transgenic or knockout mice would yield greater insight into the molecular mechanism. Knockdown or knockout approaches will be used to confirm our results. Additionally, single-cell RNA sequencing or spatial RNA sequencing could be used as an unbiased screen to reveal essential genes and pathways in Apln⁺ ECs at different stages of CCG. For studying transcriptomic changes across cell types in the heart, single-cell ATAC sequencing could also be employed. Moreover, using such a robust CCG mouse model in pathological conditions like diabetes will also provide more insight into the therapeutic potentials for treating IHDs.

Highlights.

• Development of a murine model of coronary collateral growth by repetitive ischemia

• 3D micro-CT imaging shows well-developed coronary collaterals in adult mouse hearts

• Sprouting endothelial cells contribute to coronary collateral growth in adult hearts

• Sprouting angiogenesis gives rise to mature coronary arteries

Nonstandard Abbreviations and Acronyms:

Apln

Apelin

Apln-creER

Apelin-cre-recombinase

BrdU

5-bromo-2’-deoxyuridine

CCG

Coronary collateral growth

CZ

Collateral-dependent zone

EdU

5-ethynyl-2’-deoxyuridine

ECs

Endothelial cells

IHDs

Ischemic heart diseases

LAD

Left anterior descending

LCA

Left coronary artery

MI

Myocardial infarction

micro-CT

Micro-computed tomography

mT/mG

Gt (ROSA)26Sor^{tm14(CAG-tdTomato)Hze}/J

mT/mG x Apln-cre

Apelin-cre-recombinase with mTmG

MI

Myocardial infarction

NZ

Normal Zone

RCA

Right coronary artery

RI

Repetitive Ischemia

RI heart

RI-treated heart

RI mouse

RI-treated mouse