Introduction

Unlike the adult mammalian heart, the neonatal mouse heart exhibits a remarkable regenerative capacity after myocardial infarction (MI) up to 7 days after birth¹ A similar regenerative capacity was shown in new-born humans². This regenerative potential is linked to the epicardium, a mesothelial layer covering the outer surface of the heart. During development, a subset of epicardial cells undergo epithelial to mesenchyme transition (EMT), generating epicardium-derived cells (EPDCs) that contribute to coronary vessel formation and cardiomyocyte proliferation and compaction Click or tap here to enter text..³ A number of transcription factors are expressed in the developing epicardium, including Wilms’ tumour 1 (WT1), which is instrumental in the regulation of EMT. Loss of WT1 impairs the ability of EPDCs to migrate into the underlying sub-epicardial region and mutant embryos die between E14.5-E16.5 due to cardiovascular failure⁴. EPDC plasticity gradually decreases during embryogenesis and after birth the epicardium becomes quiescent. Embryonic genes, such as Wt1, are downregulated and EMT ceases⁵.

Following injury, the epicardium reactivates, resembling an embryonic-like state. Epicardial cells re-express WT1⁶, proliferate, undergo EMT, and differentiate into fibroblasts to stimulate tissue repair⁷. Thymosin β4 (Tβ4) pre-treatment results in EPDC-mediated neovascularization after injury⁸ and induces limited de novo cardiomyocyte formation by differentiation from a progenitor population of epicardial origin⁹.

Hypoxia, or low oxygen levels, has been linked to tissue regeneration¹⁰, and hypoxiainducible factors (HIFs) play a central role in cellular responses to hypoxia¹¹. HIFs are heterodimeric transcription factors, consisting of α and β subunits¹², with the latter being constitutively expressed, whereas expression of the α subunit is oxygen-dependent (reviewed in¹³). Under well-oxygenated conditions (i.e. normoxia), prolyl hydroxylase domain proteins (PHD1-3) hydroxylate specific conserved proline residues on the α subunit, tagging it for proteasomal degradation. When oxygen levels drop, PHD activity is inhibited, enabling the α subunit to translocate to the nucleus and bind to hypoxia response elements (HRE) regulating transcription of a plethora of target genes¹⁴. In human, mouse and rat there are three isoforms of the α subunit, but the best characterized are HIF-1α and HIF-2α activating distinct, although partially overlapping, sets of genes¹⁵. Hypoxia and HIF activation are evident throughout the developing heart¹² and disruption of oxygen sensing pathways during development results in a wide range of cardiac abnormalities¹⁶. Notably, WT1 expression is upregulated in the heart and kidneys of rats exposed to hypoxia and an HRE binding site for HIF-1α has been identified within the Wt1 promoter¹⁷ suggesting a link between hypoxia, HIF signaling and epicardial activation. In this context, the role of epicardial HIF signaling in heart development and regeneration remains unclear. Studies have shown that stabilizing HIF-1α inhibits EPDC migration in avian embryos ¹⁸ but promotes their differentiation into vascular smooth muscle cells (VSMCs)¹⁹. Endothelium-specific depletion of Phd2/3 promotes cardiomyocyte proliferation and prevents left ventricular failure in a model of MI²⁰, and exposure to hypoxia following MI induces a regenerative response in adult mice by metabolic reprogramming of cardiomyocytes leading to cell cycle re-entry²¹.

The epicardium and subepicardial space are identified as hypoxic niches, housing a unique progenitor cell population^22,23. Given that hypoxia has been implicated in regenerative responses post-MI, and epicardial WT1 expression is HIF dependent, we hypothesised that HIF signalling contributes to epicardial cell activation during heart development and following injury and as such might represent a therapeutic target for extending the regenerative window²⁴. Here, we demonstrate that the developing epicardium is hypoxic and epicardialspecific deletion of Hif1 a disrupts epicardial EMT and heart vascularisation. In mice, we observe a gradual and concurrent decrease in WT1 expression and HIF-mediated signalling over the first week of life. Stabilisation of HIF signalling was sufficient to maintain activation of the epicardium beyond postnatal day (P) 7, improve cardiac remodelling and preserve function post-MI. These findings provide novel insight into the molecular mechanism regulating epicardial activation in the mouse heart and suggest that modulation of HIF signalling in the dormant epicardium may enhance cardiac repair and regeneration, representing a potential therapeutic strategy for the treatment of ischaemic heart disease and heart failure.

Materials & Methods

Mouse Lines

All animal experiments were carried out in accordance with the UK Home Office project license (PPL) 30/2987, PPL30/3155 and PDDE89C84 compliant with the UK animals (Scientific Procedures) Act 1986 and approved by the local Biological Services Ethical Review Process. The following mouse lines were used: Rosa^26+/CreERT2;Hif1 a^fl/fl;, Rosa^26+/CreERT2;Phd2^fl/fl. The Wt1^CreERT2/+;Hif1 a^fl/fl and Wt1^CreERT2/+;Phd2^fl/lf mouse line was generated by crossing Wt1^{CreERT2/+ 20} with Hif1 a^fl/fl animals or Phd2^fl/lf for two generations. Genetically modified mouse lines used were kept in a pure C57BL/6 background. Both males and females were used in the study. For timed-mating experiments, 8-12-week-old mice were set up overnight and females checked for vaginal plugs the following morning; the date of a vaginal plug was set as embryonic day (E) 0.5. For tamoxifen-dependent gene activation, 2 doses of 40 mg/kg of body weight of tamoxifen (Sigma) were administered to pregnant dams by oral gavage, at embryonic stages E9.5 and E10.5. For neonate studies, pups were injected intraperitoneally (i.p.) with a single 10 μl dose of 20 mg/ml tamoxifen, at postnatal day (P)2, using a 25-gauge needle.

5-ethynyl-2’-deoxyuridine (EdU, Thermofisher) was intraperitoneally injected at a dose of 50mg/kg at the day of surgery (P7) and every other day. Hearts were collected at 9 dpi and the assay was carried out according to manufacturer’s protocol.

Hypoxyprobe

For hypoxia studies, pregnant females were injected i.p. with 1.5 mg Hypoxyprobe (pimonidazole hydrochloride, Hypoxyprobe-1 Inc). After 2 hours, hearts from embryos were harvested and fixed. Pimonidazole is a 2-nitroimidazole that is reductively activated specifically in hypoxic cells and forms stable adducts with thiol groups in proteins, peptides, and amino acids at oxygen levels below 1.3%.

Cell lines

The mouse immortalised embryonic epicardial cell line MEC.1²⁶ was purchased from Merck/Millipore (SCC187) and cultured according to manufacturer’ recommendations.

Myocardial Infarction

All surgical and pharmacological procedures were performed in accordance with the Animals (Scientific Procedures) Act 1986 (Home Office, UK). MI was induced by permanent ligation of the left anterior descending (LAD) coronary artery in wild type mice at postnatal day (P) 7, as previously described²⁵. Neonates are not able to be sexed during the first week of life so mixed male and female cohorts were used throughout the protocol. Mice were anesthetized with 4% isoflurane for 1 minute and then transferred to an ice box for up to 6 minutes. Mice were recovered under 0.5% isoflurane.Animals received intraperitoneal injections of BAY 85-3934 (Molidustat, Selleck) (10 mg/kg) or vehicle (DMSO), upon recovery (day 0) and 7 days later. Mice were sacrificed by cervical dislocation for tissue collection at 4-, 9- and 21-days following ligation. Mice with excessive or minimal infarct sizes at early time points were excluded from further histological or functional assessment (see under Cine MRI).

Epicardial explants

Sterile 12-well plates (Fisher Scientific), containing sterile 13-mm diameter coverslips were coated with 0.1% gelatin (Millipore) and allowed to stand for 20 minutes. The gelatin was then replaced by Dulbecco’s Modified Eagle Medium (DMEM) (Sigma) containing 10% FBS (Sigma), 1% Penicillin/Streptomycin (Sigma). Hearts were isolated from E11.5 embryos and the outflow tract and atria were removed. Each ventricle was then cut in half and placed epicardial side down on the coverslip. Explants were kept at 37°C/5% CO₂. 24 hours later, 1 μM of tamoxifen was added to explants. After 48 hours of culture, the explant tissue was carefully peeled off, leaving only the epicardial sheet remaining on the coverslip. For explants prepared from C57BL/6 embryos, 50 μM of BAY 85-3934 (Molidustat, Selleck) or vehicle (DMSO) was added. After a total of 72 hours in culture, the media was removed, coverslips were washed briefly in ice-cold PBS and then fixed in 4% PFA at room temperature for 15 minutes. Coverslips were subsequently washed in PBS before proceeding with immunostaining.

Histological analysis

Following overnight (O/N) fixation in 2% PFA at 4°C, hearts were washed in PBS and dehydrated by passage through rising concentrations of ethanol. Samples were then washed in Butanol O/N, at RT, before being placed in molten 50:50 butanol: Histoplast paraffin wax (Fisher) at 56°C. After 1 hour the solution was replaced with 100% Histoplast paraffin wax. After several changes of 100% wax, hearts were oriented and embedded into a mould pre-loaded with paraffin. The wax was then rapidly cooled on a bed of ice and stored at 4°C. A microtome was used to cut 10μm-thick sections through the heart. Before staining, sections were first deparaffinised in Histoclear solution (Fisher), followed by rehydration through a decreasing concentration of ethanol. For Masson’s Trichrome Staining, the Masson’s Trichrome kit (Abcam) was used as per the manufacturer’s instructions. Briefly, slides were immersed in Bouin’s solution for 15 minutes at 56°C. Sections were then stained by serial immersion in the following solutions: Weigert’s Iron Haemotoxylin solution (5 minutes); Biebrich Scraletacid Fuchsin (5 minutes); Aniline Blue (5 minutes); Phosphotungstic/Phosphomolybdic acid solution (5 minutes) and finally 1% Acetic acid solution (2 minutes). Samples were then rinsed in distilled water and dehydrated. Finally, sections were cleared in Xylene and mounted using DPX mounting media.

Immunofluorescence staining

Embryos were harvested at the required embryonic stage, placed in ice-cold PBS (Sigma) and the heart micro-dissected. Similarly, hearts from neonates were removed and washed in ice-cold PBS. Both embryonic and neonatal hearts were fixed for 6 hours in 2% paraformaldehyde (PFA; Santa Cruz Biotechnology) at 4°C and equilibrated in 30% sucrose overnight at 4°C. Hearts were then placed in 50:50 30% sucrose/PBS: Tissue-Tek OCT (VWR) for 30 minutes at room temperature (RT) and embedded in OCT. 10 pm-thick cryosections through the heart were cut. Before use, slides were left to dry for 10 minutes at RT and then washed in PBS for 5 minutes to remove the OCT. Samples were permeabilised with 0.5% (sections) or 0.1% (explants) Triton X-100 in PBS (PBTr) for 10 minutes at RT and subsequently rinsed twice in PBS. Samples were blocked in 10% Serum, 1% bovine serum albumin (BSA, Merck), 0.1% PBTr for 1 hour at RT prior to incubation with the primary antibodies overnight at 4°C. The following day, slides were washed three times for at least 5 min in 0.1% PBTr. Samples were incubated with Alexa Fluor®-conjugated secondary antibodies (1:200 dilution; Invitrogen) and 4’,6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI; 0.1 μg/ml, Invitrogen), for 1 hour at RT, protected from light. After final washes, slides were mounted in 50% glycerol in PBS. For wholemount staining, samples were washed in 0.3% PBTr and blocked in 1% BSA (Merck), 0.3% PBTr for at least 2 h. The samples were then incubated with primary antibodies in the blocking solution overnight at 4°C. On the second day, the samples were washed at least five times in 0.3% PBTr and then incubated with secondary antibodies and DAPI (Invitrogen) diluted in PBS overnight at 4°C. The samples were then washed with PBS at least five times the next day and mounted in 50% glycerol in PBS. The following primary antibodies were used: Dylight 549 Mab (Hypoxyprobe-1 Inc, 1:200) antibody to detect hypoxyprobe labelling, HIF-1α (1:100, Novus Biologicals), HIF-2α (1:100, R&D Systems), endomucin (1:50, Santa Cruz Biotech), podoplanin (1:200, Fitzgerald), WT1 (1:100, Abcam), Alexa Fluor™ 488 Phalloidin (1:250, ThermoFischer), actin alpha-smooth muscle Cy3 (1:100, Sigma), Anti-α-Actinin (Sarcomeric) (1:500, Sigma-Aldrich), CD31 (1:100, Abcam), smooth muscle Myosin heavy chain 11 (SM-MHC, 1:100, Abcam), SM22 alpha (1:100, Abcam), Wheat-germ agglutinin (WGA, 1:100, Thermofisher), prolyl hydroxylase domain-2 (PHD2, 1:100, Novus Biologicals).

RNA isolation and qRT-PCR analysis

Total RNA was isolated from frozen ventricles and epicardial explants with Trizol (Invitrogen) using a Teflon homogeniser followed by aspiration with a sterile 25-gauge needle and syringe (BD sciences). RNA was transcribed into cDNA utilizing random primers (Promega) with Superscript Reverse Transcriptase III (Life Technologies). Real-time quantitative PCR was performed on a ViiA™ 7 Real-Time PCR System (Applied Biosystems), using SYBR Green mix (Invitrogen). Gene expression was evaluated as DeltaCt relative to control (Atp5b, Sdha and B2M).

Heart dissociation and FACS sorting

Hearts from P1 and P7 mice were harvested and minced into a single cell suspension of FACS sorted live cells. Briefly, for each sample, 3 hearts collected from the same litter were pooled together and finely minced with a scalpel. Tissue was then digested by gentle agitation (180 rpm shaker) in Collagenase II (Gibco,) using a solution of 500 units/ml in HBSS at 37 °C for 45 minutes (P7 hearts) or 20 minutes (P1 hearts). Cell solutions were then passed through a 70 μm filter, washed and incubated in Red Blood Cell lysis buffer (Cell Signaling Technology) for 10 minutes at RT to remove red blood cells. Finally, isolated single cardiac cells were centrifuged, resuspended in 2% FBS in PBS, passed through the filtering cap of the FACS tubes and incubated with 1% 7-AAD viability stain (Invitrogen) for 10 minutes. Approximately 1×10⁵ live cells per sample were sorted using the BD FACSAria Fusion Sorter.

Single cell RNA-Sequencing

FACS sorted cardiac cells viability and concentration were assessed by using an automated cell counter and 1.5×10⁴ cells per sample were loaded onto the 10X Chromium system (10X Genomics) to obtain a target cell recovery of ~6000 cells. Single cell RNA-seq libraries were generated using Single Cell 3 Prime Reagent Kits v1.3 (10x Genomics) according to the manufacturer’s protocol. Sequencing was performed on an Illumina NovaSeq 6000 System operated by the Oxford Genomics Centre at the Wellcome Centre for Human Genetics. Raw sequence reads were aligned against the mouse mm10/GRCm38 reference transcriptome using the Cell Ranger 3.1.0 pipeline (10x Genomics) and processed further using the scRNA-seq analysis R package Seurat (v.3.2.3). Initial filtering removed cells expressing less than 200 genes and genes that were expressed in less than 3 cells. To exclude low quality cells and doublets we filtered out cells with a very high mitochondrial genes content and cells that expressed more than 8000 genes. Based on these criteria, 19211 genes across 5308 and 8734 cells for P1 and P7 samples respectively remained for downstream analysis. Data from the two samples were combined and scaled by regressing out the nUMIs, percentage of mitochondrial gene expression and cell cycle. To correct for batch effect, samples were integrated using the Harmony package (v.1.0). Uniform manifold approximation and projection (UMAP) was performed on the scRNAseq harmonised cell embeddings and unbiased clustering was obtained using the FindCluster function of the Seurat pipeline. Cluster cell types were annotated using a combination of differentially expressed markers, identified using the Seurat FindAllMarkers function and the expression of selected canonical markers for specific cell types. The Epicardial cell cluster was then identified and subset into a new Seurat object with raw reads counts. The standard Seurat pipeline described above was performed and the Model-based Analysis of Single-cell Transcriptomics test (MAST v.1.14.0) was used to analyse the differential gene expression between the P1 and P7 groups in the epicardial cells. Finally, differentially expressed genes were ranked based on both the fold change and p value (avg_logFC * −log10 of the p_val) and the ranked list was submitted to the Metascape platform (www.metascape.org) to identify enriched pathways in the P1 or P7 group, respectively.

Cardiac cine-MRI

Cardiac cine-MRI was performed at 7T using a Varian DDR system. Briefly, mice were anaesthetised with 2% isoflurane in O₂ and positioned supine in a custom animal handling system with homeothermic control. Prospectively gated proton cardiac images were acquired with a partial Fourier accelerated spoiled gradient echo CINE sequence (TR 5.9 ms, TE 2.2 ms, 30 kHz bandwidth, 25° FA, approximately 20-30 frames gated to the R wave with a 4 ms post-label delay; 20% partial acquisition; 4 averages) with a 72 mm volume transmit/4 channel surface receive coil (Rapid Biomedical GmbH) in order to acquire two and four chamber long-axis views and a short axis stack for functional quantification (128×128 matrix; 25.6 mm^2 FOV; 0.7 mm slice thickness, 0.2 mm resolution in-plane). Non-acquired partial Fourier data was reconstructed via the method of projection onto convex sets prior to a simple, cartesian, DFT. Blinded image analysis was performed in ImageJ as described previously. Mice with either excessive or minimal infarct sizes were excluded from further analyses by MRI scanning at day 1 post-MI.

Wound healing/scratch assay

2 × 10⁴ MEC.1 cells were seeded on a 2 well Culture insert (Ibidi, Germany). The day after, the insert was removed and 50 μM of BAY 85-3934 (Molidustat, Selleck) or vehicle (DMSO) was added to the culture. Cells were imaged every 2 hours up to 6 hours. The cell-free space area was determined by using the ImageJ software (NIH, Rockville, USA).

Image Analysis

For quantification of immunofluorescence of cryosections, images were captured on an Olympus FluoView 3000 confocal microscope and analysed with Fiji (NIH) and AngioTool⁷. To quantify nuclear fluorescence, a macro was written to identify all DAPI-stained nuclei and to analyse nuclear fluorescence based on the DAPI mask. To localise WT1+ cells in the epicardium and myocardium, cells were manually counted. To quantify the cell size, three independent samples per group with six different fields and positions, two from left and right ventricles, and septum were captured at 40x magnification. ImageJ (National Institutes of Health) was used to quantify the size of each cell.

Statistical Analysis

All data are presented as mean ± standard error of the mean (SEM) or as median, inter-quartile range (IQR) and upper and lower limits. Statistical analysis was performed on GraphPad Prism 8 software. The statistical significance between two groups was determined using an unpaired two-tailed Student’s t-test, these included an F-test to confirm the two groups had equal variances. Among three or more groups, one-way analysis of variance (ANOVA) followed up by Dunnett’s or Tukey’s multiple comparison tests were used. A value of p ≤ 0.05 was considered statistically significant.

Results

The epicardium is hypoxic during development

To study the role of low oxygen tension in the developing heart, we initially investigated the physiological levels of hypoxia in situ from mid- to late gestation stages, when the epicardium forms and epicardial EMT takes place. Following in utero treatment with the marker Pimonidazole-HCl (Hypoxyprobe-1, HP1)²⁶, we probed embryonic hearts at various developmental time points, ranging from E12.5 to E18.5 using an anti-pimonidazole fluorescence-conjugated monoclonal antibody (HP1), alongside immunostaining for the spatiotemporal expression of HIF-1α and HIF-2α.

At E12.5, characterized by a well-established epicardial layer amidst highly trabeculated myocardium, we observed extensive regions of hypoxia, mostly in areas of dense myocardium, such as the compact wall of the atrioventricular (AV) groove (Figure 1a). Notably, HP1 staining co-localized with the epicardial marker WT1 (Figure 1a), and both HIF-1α (Supplementary Figure 1a) and HIF-2α at the apex of the heart. Additionally, strong HIF-1α signals were noted in the AV groove (Figure 1a), while HIF-2α expression extended into the myocardial compartment (Supplementary Figure 2a).

Figure 1.

By E14.5, extensive HP1 staining persisted in areas of thicker compact myocardium, such as the interventricular septum (IVS) (Figure 1b). Co-localization with WT1 and HIF-1α was observed predominantly in the epicardium, notably in the apex (Supplementary Figure 1b), and AV groove regions. While HIF-2α co-localization with HP1 was present, it was primarily within the myocardium (Figure 1b, Supplementary Figure 2b).

Figure 2.

Analysis of hearts at E16.5 and E18.5 revealed a progressive decrease in HP1 signal, becoming largely confined to WT1-expressing epicardial cells at later stages (Figure 1c, d). Similarly, HIF-1α localization was restricted in discrete regions of the outermost layer of the heart (Supplementary Figure 1c, d), whereas HIF-2α expression transitioned from the epicardium (notably at the apex of the heart) to some sporadic expression in underlying cardiomyocytes by E18.5. (Supplementary Figure 2c–d, Figure 1c–d). Collectively, these data suggest that WT1+ cells making up the epicardial layer of the developing heart become hypoxic from mid-gestation onwards, preferentially expressing HIF-1α whereas HIF-2α expression predominantly occurs within the forming myocardium.

HIF-1α epicardial deletion reduces the number of Wt1+ cells and alters coronary vessel development in E16.5 embryos

Given that the epicardium is hypoxic during mid-late gestation, and we observed expression of HIF-1α by WT1⁺ cells within the epicardium, we next determined the functional implications of HIF-1α on epicardial development. To specifically target Hiflα expression in the epicardium, we crossed tamoxifen-inducible epicardial-specific CreERT2 driver, Wt1^CreERT2/+27 mice crossed with mice in which exon 2 of the Hiflα gene is flanked by LoxP sites (Hif1α^fl/fl). Importantly, there is no known phenotype attributed to hemizygosity for Wt1²⁸ as might be associated with the Wt1^CreERT2/+ mice²⁷ and as such any interpretation of downstream effect is attributed to epicardial-loss of Hiflα. Pregnant wt1^CreERT2+; Hiflα^fl/fl females were injected with tamoxifen at E9.5 and E10.5, to target epicardial development, and embryos were collected at E16.5. Notably, deletion of Hiflα did not affect HIF-2α expression in WT1+ cells (Supplementary Figure 3a). Immunostaining for WT1 and the epicardial marker podoplanin (PDPN) revealed a marked reduction in the number of WT1+ cells in Wt1^CreERT2/+;Hif1α^fl/fl mutant (KO) hearts as compared to littermate controls (CTR) (Figure 2a–b), affecting both the epicardial layer (Figure 2c; CTR: 0.72 ± 0.11; KO: 0.32 ± 0.010; p=0.0268) and the myocardium (Figure 2d; CTR: 3.96± 1.01; KO: 0.98 ± 0.25; p= 0.0461). Reduction in the number of Wt1+ cells in the myocardial compartment suggested an impairment in epicardial EMT. Hearts were also probed for the endothelial marker endomucin (EMCN), which also marks the endocardium and the outline of myocardial trabeculae. No significant alterations were observed in myocardial compaction (Figure 2e), myocardial trabeculation as assessed by fractal analysis (Figure 2f), or the proliferation rate of WT1⁺ cells between mutants and controls (Supplementary Figure 3c–d).

Given the essential role of the epicardium in coronary vessel development²⁹ we assessed coronary vessel formation using whole-mount immunofluorescence staining for EMCN associated with AngioTool™ software quantification³⁰. Mutant hearts exhibited significantly impaired coronary vasculature, as evidenced by decreased total vessel length (microns) (Figure 2h, CTR: 129229 ± 5509; KO: 81073 ± 10915; p= 0.0170), number of junctions (Figure 2i, Mean ± SEM; CTR: 2505 ± 63.74; KO: 1457 ± 282.1; p= 0.0223) and vessel end points (Figure 2j, Mean ± SEM; CTR: 3422 ± 249.5; KO: 2429 ± 218.6; p= 0.0402).

During development, EPDCs undergo EMT and migrate into the sub-epicardial space to support coronary vessel formation. Thus, we sought to investigate whether the vascular impairment observed in the wt1^CreERT2/+; Hiflα^fl/f1 mutant embryos was due to compromised epicardial EMT. Epicardial explants from Wt1^CreERT2+;Hif1α^fl/fl embryos treated with tamoxifen showed a reduced HIF-1α expression (Supplementary Figure 4b; CTR: 0.99 ± 0.16; KO: 0.46 ± 0.12; p= 0.0397) without affecting HIF-2α levels (Supplementary Figure 4c; CTR: 2505 ± 63.74; KO: 1457 ± 282.1; p= 0.223). Tamoxifen-treated epicardial explants from Wt1^CreERT2+; Hif1α^fl/fl hearts revealed an increased epithelial and reduced mesenchymal phenotype. This was evident by reduced stress fibre formation (Figure 2k), reduced alphasmooth muscle actin α-SMA) staining (Figure 2l), and altered sub-cellular localisation of the EMT marker and structural component of cell-cell adhesion complexes zonula occludens-1 (ZO-1) (Figure 2m) compared to controls. Specifically, mutant explants showed a predominant membrane α-SMA staining, demarcating their epithelial cell shape, whereas control cells displayed a more intense cytoplasmic labelling of the filamentous actin cytoskeleton (Figure 2l). Similarly, in control cells the localisation ZO-1 protein was predominantly cytoplasmic and/or nuclear, whereas in mutant cells ZO-1 staining was membranous, indicating a lesser migratory and more epithelial-like phenotype (Figure 2m). Consistent with our in vivo findings, explants derived from wt1^CreER2/+;Hif1α^fl/fl embryos exhibited decreased WT1 expression, indicating impaired epicardial EMT (Figure 2n–o; CTR: 725.7 ± 86.93; KO: 490.9 ± 37.65; p= 0.0479).

To elucidate the direct regulation of WT1 by HIF-1α, we interrogated a publicly available dataset of chromatin immunoprecipitation coupled with next generation sequencing (ChIP- seq) generated in the renal adenocarcinoma 786-O cell line¹⁵. ChIP-seq analysis confirmed direct HIF binding within intron 3 of the Wt1 gene (Supplementary Figure 5), consistent with previous findings¹⁷.

In summary, epicardial-specific deletion of Hif1 a in vivo and ex vivo significantly reduced the number of WT1+ cells, both in the epicardium and underlying myocardium, arising through EMT affecting coronary vessel development at E16.5. Taken together our data demonstrate that HIF regulates the expression of WT1 and associated epicardial EMT.

Stabilisation of HIF signalling enhances Wt1 expression and epicardial EMT

We next sought to investigate whether stabilising HIF signalling under normoxic conditions is sufficient to enhance epicardial EMT. To this end, we established epicardial explants from Rosa26^+/CreERT2;Phd2^fl/fl mice, where Cre-mediated recombination induced loss of Phd2, thereby stabilising HIF signalling in all cell types but restricted to epicardium in explants for ex vivo analysis. Upon tamoxifen treatment, Rosa26^+/CreERT2;Phd2^fl/fl derived explants (KO) exhibited a significant induction of HIF-1α and HIF-2α expression as compared to controls (CTR) (Supplementary Figure 6a–c; HIF-1a; CTR: 0.26 ± 0.13; KO: 0.86 ± 0.09; p= 0.0061; HIF-2a; CTR: 0.37 ± 0.01; KO: 1.12 ± 0.15; p= 0.0027). Subsequent visualization of stress fibres via a-SMA staining revealed augmented mesenchymal morphology in the mutant- derived explants (Figure 3a). Consistent with this WT1 nuclear fluorescence was significantly increased in KO explants as compared to controls (CTR) (Figure 3b–c; CTR: 158.7 ± 48.34; KO: 330.5 ± 27.26; p = 0.0364). In addition, expression levels of the EMT transcription factor snail family transcriptional repressor 2 (Snai2) gene were increased in Phd2-KO explants relative to CTR (Figure 3d; CTR: 1 ± 0.025; KO: 1.166 ± 0.036; p = 0.0464).

Figure 3.

To complement our genetic studies, we undertook a more therapeutically relevant approach by using a pharmacological PHD inhibitor to stabilise HIF signalling. Epicardial explants were generated from C57BL/6 wildtype embryos and treated with DMSO control (CTR) or Molidustat (BAY 85-3934, Mol) a specific PHD inhibitor, used in clinical trials for renal anaemia³¹. Immunostaining for HIF-1α and HIF-2α confirmed the stabilisation of HIF signaling upon PHD inhibitor treatment (Supplementary Figure 6d–g). Staining with phalloidin (Figure 3e) and α-SMA (Figure 3f) showed an enhanced mesenchymal morphology with increased stress fibres in inhibitor-treated explants, suggestive of EMT induction. Furthermore, immunostaining demonstrated a significant increase in WT1 expression in treated explants as compared to control (Figure 3g–h; CTR: 1039 ± 134.9; Mol: 2500 ± 503.7; p = 0.03). Likewise, the PHD inhibitor enhanced the motility of cultured mouse embryonic ventricular epicardial cells (MEC.1 cell line)³² compared to control-treated cells (Figure 3i–j; CTR: 436869 ± 9359; Mol: 372482 ± 20880; p = 0.0477), as assessed by scratch assays. These findings collectively suggest that genetic perturbation and/or pharmacological inhibition of PHDs stabilise HIF signalling in the epicardium ex vivo, leading to increased WT1 expression and stimulation of epicardial EMT.

Reduced HIF signalling is associated with epicardial quiescence in the neonatal heart

To gain unbiased insight into hypoxia and HIF-related molecular pathways which are differentially regulated in P1 versus P7 mouse hearts, we performed single cell RNA-seq analysis using the 10x Genomics Chromium platform and next generation sequencing. Cell clusters were visualised through uniform manifold approximation and projection (UMAP) (Figure 4a). An epicardial cell cluster (Epi) was identified based on previously described specific epicardial gene expression signatures³³ (Figure 4b). As expected, a cardiomyocyte cell cluster (CM) was underrepresented due to both low survival and size incompatibility with the FACS cell sorting³⁴. Gene Ontology (GO) analysis for biological processes showed an enrichment of hypoxia-related pathways in P1 compared to P7 hearts (Supplementary Figure 7a), particularly in the epicardial cell population (Figure 4c). In contrast, hypoxia- related pathways were enriched in P7 hearts compared to P1 hearts in the fibroblast cluster (Supplementary Figure 7b–d), suggesting a role for hypoxia in the fibrotic response leading to scarring. Accordingly, expression of well-known HIF-induced genes such as Vegfa and pyruvate dehydrogenase kinase 3 (Pdk3) was increased in P1 derived epicardial cells (Figure 4d). Conversely, the expression of Phd2, encoding for the main suppressor of HIF signalling, was enriched in P7 cells (Figure 4e). These findings were further confirmed by a time-course analysis of mRNA levels using ventricle lysates and immunostaining which revealed a marked upregulation of Phd2 cardiac levels from P7 to adulthood (Figure 4f, Supplementary Figure 8a–b). Notably, this increase in Phd2 levels coincided with a reduction in WT1 expression in P7 versus P1 hearts (Figure 4g).

Figure 4.

Activation of HIF signalling improves the response to injury in non-regenerative P7 hearts

To establish a causative role in the regulation of epicardial WT1 activity, we extended HIF signaling beyond P7 using a genetic gain-of-function approach. Neonatal Wt1^CreERT2/+;Phd2^fl/fl pups received a single dose of tamoxifen at P2 to induce Cre-mediated loss of Phd2. MI was then induced by permanent ligation of the proximal left anterior descending (LAD) coronary artery in P7 mice (Figure 5a), a time-point when the heart’s regenerative capacity is lost. Hearts were collected at 21 dpi. Immunostaining analysis of the endothelial marker CD31 and a-SMA did not reveal a significant difference in the number of vessels between Phd2^fl/fl (CTR) and Wt1^CreERT2;Phd2^fl/fl (KO) hearts (Figure 5b–c, CTR 0.0030 ± 0.0004, KO 0.0034 ± 0.0002, p= 0.45), suggesting that activation of HIF signalling in WT1-expressing cells does not affect the neovascularisation response. To investigate whether epicardial activation of HIF signalling improved heart function following MI, we performed cine-MRI analysis at 3-weeks post-MI. We observed a substantial increase in the ejection fraction (EF) and decrease in the end-diastolic volume (EDV) in KO animals compared to controls (Figure 5d–f; EF: CTR 43.21 ± 6.25, KO 64.30 ± 6.82, p< 0.05. EDV; CTR 54.19 ± 3.50, KO 33.72 ± 2.67; p < 0.05). Furthermore, histological analysis by Masson’s trichrome staining revealed a significant reduction of the fibrotic scar in KO animals as compared to controls (Figure 5g–h, CTR 15.44 ± 1.088, KO 10.34 ± 1.374, p= 0.027) without changes in the size of cardiomyocytes, as assessed by wheat germ agglutinin staining (WGA) (Figure 5i–j, CTR 167.6 ±3.27, KO 179.3 ± 8.68, p= 0.32). Taken together these findings suggest that activated hypoxia signalling in the epicardium contributes to heart regeneration by reducing scar size and improving heart function.

Figure 5.

To further investigate our hypothesis that prolonged maintenance of HIF signalling has a beneficial effect on the epicardium and response to MI in the postnatal heart, we explored a complementary pharmacological approach recapitulating our gain of HIF-function studies using epicardial explants. We induced MI by LAD ligation in P7 mice and administered Molidustat or DMSO control (vehicle) by intraperitoneal injection, immediately after surgery and one week later, to stabilise HIF signalling in the postnatal heart (Figure 6a). Treatment with PHD inhibitor effectively stabilized HIF-1α and HIF-2α, as observed by immunostaining (Supplementary Figure 9a–b) in hearts collected at 9-days post-MI (9dpi). Moreover, this pharmacological approach revealed a significant increase in epicardial WT1 expression in both the infarct and remote zones (Figure 6b) in Molidustat-treated hearts (MI T) compared to controls (CTR) (Figure 6c–d, infarct zone: MI CTR 0.34 ± 0.08, MI T 0.50 ± 0.06, remote zone: CTR 0.075 ± 0.02, MI T 0.25 ± 0.05, p < 0.05). Similarly, an increase in WT1 expression in the myocardium was observed in the PHD-inhibitor treated group and was more pronounced proximal to the infarct region (Figure 6e–f, infarct zone: CTR 5.02 ± 1.34, MI T 7.09 ± 1.86, p=<0.05; remote zone: CTR 13.88± 4.96, MI T 18.61 ± 6.18; p < 0.05). We observed an extensive co-localization of WT1 and EMCN within the myocardial compartment at 9dpi (Supplementary Figure 9c), supporting an endothelial identity for these cells. Furthermore, an assessment of the proliferation of WT1+ cells and cardiomyocytes revealed no significant differences between control and Molidustat-treated animals (Supplementary Figure 9d, e, g).

Figure 6.

To analyse the long-term histological effects of HIF-stabilisation after MI, we collected hearts at 21dpi. Co-immunostaining and quantification of CD31 and a-SMA revealed no difference in the number of vessels between controls and the treatment group (Figure 6g–h). Cine-MRI analysis showed a marked reduction in left ventricular systolic function following MI in the vehicle-treated group (MI CTR) as compared to uninjured hearts (Sh) and, more importantly, a significant improvement in systolic function in the PHD inhibitor treated-groups (MI T). EF and EDV were significantly improved in the drug-treated groups (Figure 6i–j; EF: Sh 78.10 ± 3.22, MI CTR 67.70 ± 4.31, MI T 77.61 ± 1.76, p < 0.05; EDV: Sh 27.53 ± 3.10, MI CTR 36.11 ± 2.49 MI T 26.44 ± 0.93; p< 0.05). Histological analysis by Masson’s trichrome staining revealed the presence of a fibrotic scar in drug-treated animals, albeit this was reduced relative to controls (Figure 6k–l) and the at-risk myocardium was comparatively thicker with minimal dilatation or thinning of the left ventricle. The thickened myocardium was not due to an effect on cardiomyocyte hypertrophy, as determined by WGA staining (Supplementary Figure 9f–h) nor an effect on cardiomyocyte proliferation (Supplementary Figure 9e, g) suggesting potential cardio-protection and improved preservation of survived heart muscle. More generally these findings also indicate that improved remodelling and function following drug treatment are possible against a background of fibrosis.

Discussion

In this study, we characterised the levels of hypoxia in the heart during mid-to-late gestation when the epicardium is fully formed. At E12.5, hypoxia was mainly observed in areas of compact myocardium, most significantly in the atrioventricular (AV) groove, close to where the coronary plexus first develops³⁵. By E14.5, hypoxic regions localised to the interventricular septum (IVS), which is perfused later than the muscle of the free wall³⁶. Finally, by E16.5 and E18.5, when the heart is adequately perfused by the coronary vessels, hypoxia was primarily restricted to the epicardium. We further analysed the expression and distribution analysis of HIF-1 isoforms, finding a predominance of the HIF-1α isoform in WT1 + cells, whereas HIF-2α was more localized in the myocardium. Given hypoxia’s known role in promoting EMT³⁷ we confirmed a functional role for HIF-1α in promoting morphological changes associated with EMT by direct regulation of Wt1 expression (Figure 2 and 3, Supplementary Figure 5). Moreover, epicardial- specific deletion of Hif1a significantly reduced the number of WT1 expressing cells in both epicardial and myocardial compartments, along with impaired coronary vessel development in E16.5 embryos. Collectively, this suggests that HIF-1α induces epicardial EMT and plays an essential role in coronary vessel formation during development. While Cre-mediated deletion of HIF-1α, induced at E9.5 and E10.5, predominantly targeted epicardial cells, it is worth noting that expression of WT1 in the coronary endothelium has been reported as early as E11.5³⁸, although the biological significance of endothelial WT1 remains elusive. Thus, aberrant HIF signalling in WT1₊ endothelial cells cannot be completely excluded as contributing to the observed defects on the expansion of the coronary vasculature. However, stabilisation of HIF signalling in epicardial explant cultures, under physiological oxygen levels, either via genetic ablation of Phd2, or chemical inhibition of PHD enzymes with Molidustat, resulted in enhanced EMT and WT1 expression, further confirming a role for HIF signalling in epicardial activation.

To gain insight into the post-natal molecular pathways regulating the epicardium, we performed single cell RNA sequencing analysis comparing P1 and P7 stages, focusing on the specific gene expression signature of the epicardial cell cluster. GO terms analysis showed an enrichment of hypoxia-related pathways in P1 hearts, consistent with the expression of well-known HIF target genes. Importantly, Phd2 expression was enriched in P7 hearts, potentially contributing to epicardial quiescence during the first week of life, concurrent with a decrease in HIF signalling. Complete regeneration following MI requires both the replacement of lost cardiomyocytes and the formation of new blood vessels. Given the role of the epicardium during development in promoting coronary vessel formation and myocardial growth, the observed epicardial quiescence after birth likely contributes to the loss of regenerative capacity. Genetic and pharmacological stabilisation of HIF signalling beyond P7 proved to be effective in extending the regenerative window after LAD surgery, resulting in reduced pathological remodelling and preserved cardiac function against a background of fibrotic repair. Interestingly, genetic stabilisation of HIF signalling specifically in WT1-expressing cells significantly reduced scar size post-injury (Figure 5). Reduced fibrosis was also observed in PHD-inhibitor treated P7 mice (Figure 6). This is important since necrotic cardiomyocytes can trigger the activation of fibroblasts and promote fibrosis³⁹. The epicardium is a key source of mitogens in the embryonic heart³ and provides paracrine signals targeting cardiomyocytes during regeneration^40,41. In our study, we did not observe any change in either cardiomyocyte proliferation or hypertrophy, strengthening the hypothesis of an epicardium-mediated role in cardiomyocyte protection from apoptosis via paracrine signalling⁷.

It is noteworthy that despite persistent scarring (albeit reduced) following either genetic or pharmacological perturbation of PHDs, significant functional improvements were observed, in contrast to the prevailing view that fibrotic repair after MI is a barrier to effective tissue regeneration^42–44. Previous studies have suggested the involvement of HIF-1α in myocardial remodelling after injury. Mice with constitutive overexpression of HIF-1α in the myocardium showed enhanced angiogenesis, attenuation of infarct size, and improved cardiac performance after MI⁴⁵. Furthermore, cardiac-specific PHD2 inactivation and consequent HIF activation play a causal role in the pathogenesis of ischemic cardiomyopathy⁴⁶. In our study, we focused on the stabilisation of HIF signaling as a therapeutic approach to improve outcomes after heart injury. The mean half-life of Molidustat ranges from 4 to 10 hours⁴⁷ ensuring prolonged stabilisation of HIF signaling to enable therapeutic benefit. Molidustat is an orally administered small molecule used for the treatment of anaemia in patients with non-dialysis-dependent chronic kidney disease (CKD)³¹ and could potentially be repurposed to treat ischemic heart disease.

In summary, we show that the epicardium is hypoxic at later stages of development, with HIF-1α playing a crucial role in epicardial activation and EMT, necessary for supporting coronary vessel development. The epicardium becomes quiescent after birth, coinciding with decreased HIF-signalling. However, maintaining HIF signaling after birth can extend epicardial activation, leading to a significant improvement in cardiac remodelling and function after injury, thus representing a potential novel therapeutic target to improve cardiac regeneration following MI.

Supplementary Figure legends

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.

Supplementary Figure 5.

Supplementary Figure 6.

Supplementary Figure 7.

Supplementary Figure 8.

Supplementary Figure 9.

Acknowledgements

We are very grateful to Prof. Sir Peter J Ratcliffe for technical advice and highly informative scientific discussions throughout the project. We would like to thank Biomedical Services Unit for animal husbandry and Micron Oxford Advanced Bioimaging Unit (supported by Wellcome Strategic Awards 091911/B/10/Z and 107457/Z/15/Z) for access to and training in the use of confocal microscopy.

Additional information

Funding

This work was supported by the British Heart Foundation (BHF chair award to PRR: CH/11/1/28798; BHF programme grant to PRR: RG/18/33532; BHF Intermediate Basic Science Research Fellowship to JMV: FS/19/31/34158) and supported by the BHF Oxbridge Regenerative Medicine Centre (RM/17/2/33380).

Author Contribution Statement

PRR conceived the original project and sourced funding. PRR and JMV oversaw the project direction and data analyses. EG and ELP carried out experiments and analysed the data. CDV and MGR preformed surgeries. DP carried out the scRNA-seq experiment and bioinformatic analyses. RPC supervised the scRNA-Seq experiment and computational analyses. RS and DRM, assisted with the analysis of HIF-iβ ChlP-seq datasets. TB and CWP provided Rosa26+/CreERT2;Hif1αfl/fl and Rosa26+/CreERT2;Phd2fl/fl mouse strains, and provided advice and technical assistance. CAC performed blinded MRI scanning and analyses. EG wrote the main draft of the manuscript. PRR and JMV contributed to editing of the final manuscript.