Neonatal Apex Resection Triggers Cardiomyocyte Proliferation, Neovascularization and Functional Recovery Despite Local Fibrosis

Establishment of Neonatal Apex Resection Injury Model

One-day-old (P1) C57BL/6 mice were subjected to apical resection by thoracotomy using the exposure of the LV lumen as a standardized method for resection size (Mahmoud et al., 2014, Porrello et al., 2011). Sham controls underwent the same surgical procedure without resection of the apex (no cardiac manipulation of any kind).

Immediately after the surgical procedure, the survival rate of apex-resected animals was equivalent to sham controls (∼87%). However, maternal cannibalization reduced survival at 24 hr post-surgery to 61% and 82% for injury and sham groups, respectively.

Hearts Are Not Fully Restored at the Histological Level

The ventricular surface area and heart length/body weight ratio were significantly reduced upon apical resection by ∼11% and ∼14%, respectively (Figures 1A–1C), similar to other reports (Andersen et al., 2014a, Porrello et al., 2011). Representative sampling of the heart (Figure S1A) and histological characterization by Masson's trichrome (MT) staining (Figures 1A and S1B) revealed that apex resection induced noticeable cardiac remodeling, with impact on heart morphology and local deposition of fibrotic tissue (Figures 1B–1D). From 7 to 21 days after surgery, apex regrowth was evident as myocardial tissue was observed below the resection site. Alongside, the myocardial tissue in the innermost part of the apex where the lumen was exposed, was surrounded by a subepicardial scar (Figures 1A and S1B). Higher magnification revealed that the scar undergoes a maturation process, which starts with a scattered deposition of collagenous material and matures to a fibrillar appearance, occupying most of the central apex wall from 60 days after resection onward (Figures 1A1–1A4, S1B1, S1B2, and S1C). In contrast, areas surrounding the central apex region displayed myocardial integrity.

Ventricular surface area was restored at 7 days after injury (Figure 1B), and resected hearts weighed the same or were even heavier than intact hearts (Figures S1D and S1E). However, injured hearts were consistently shorter throughout the study, indicating that heart growth was mainly achieved through short-axis expansion (Figure 1C). The extent of injury, estimated by determining the percentage of sections with cardiac damage (i.e., myocardial disruption or cardiac fibrosis), decreased over time, which correlates with an actual reduction in scar volume (Figures 1D and 1E). Overall, these results indicate that resected hearts remain shorter, have permanent scarring, and do not fully regenerate.

Resected Hearts Are Functionally Competent

Echocardiograms at 21, 60, 120, and 180 days following surgery confirmed that resected ventricles underwent severe morphological alterations (i.e., long-axis shortening and short-axis widening) (Figure 2A). However, in contrast to previous reports (Andersen et al., 2016), our data showed that short-axis expansion occurred due to a thickening of the anterior and posterior wall of the LV (Figures 2B and 2C) and left ventricular chamber dilation (Figure 2D). Because LV enlargement is concurrent to a reduction in heart length (Figure 2E), the eccentricity index of injured hearts was significantly increased (Figure 2F). End-diastolic volume of injured hearts was increased at 21 days, but normalized from 60 days onward (Figure 2G). Importantly, no differences were detected in the ejection fraction (Figure 2H), stroke volume (Figure 2I) or cardiac output (Figure 2J) across groups, thus showing that the systolic function of injured hearts was not significantly impaired. Regarding diastolic function, the ratio between LV early (E) to late (A) filling velocities was marginally decreased at 180 days, suggesting that incomplete relaxation and increased filling pressures might be present in resected hearts (Figure 2K). This can be explained by increased myocardial stiffness caused by the scarring, wall thickening, and/or alterations in ventricular chamber geometry. Accordingly, myocardial performance index (also known as Tei index), which comprises both systolic and diastolic parameters, showed a tendency to be increased in resected hearts (Figure 2L) at the expense of longer relaxation times (data not shown). Electrocardiogram (ECG) tracing of injured hearts was similar to control animals (Figure 2M), which was confirmed by the absence of any significant modification in the evaluated ECG parameters (Table S1).

These findings demonstrate that, despite consistent LV chamber enlargement and wall thickening, accompanied by a reduced heart length, resected hearts are functionally competent.

Short-Term Response to Cardiac Injury Involves Extracellular Matrix Remodeling and Fibroblast Activation

Apex amputation was evident by the disruption of the expression pattern of sarcomeric-α-actinin (s-α-Actinin) and laminin (LN) (Figure S2). Forty-eight hours after resection, the apex region was marked by the deposition of fibronectin (FN) and tenascin-C (TN-C), and by an abundant CD45⁺ immune cell infiltrate (Figures 3A and S2, arrowheads). Myofibroblasts (α-SMA⁺ cells) also started to appear at the injury site at this point (Figure S2, arrows). Resolution of the inflammatory phase was evident at 7 days post-injury, whereas myofibroblasts became more frequent (Figures 3A and S2). At 21 days post-resection, α-SMA was restricted to smooth muscle cells in vessels, as observed in sham hearts. Of note, TN-C remained elevated in the border between the resection and the adjacent myocardium (Figure S2). Confocal microscopy analysis further showed that FN and TN-C were produced by CD45⁺ hematopoietic cells during the inflammatory phase (48 hr post-surgery) and by α-SMA⁺ myofibroblasts during inflammatory resolution (7 days post-surgery) (Figure 3A). Although largely overlapping, TN-C expression was restricted to the interface between the healthy and damaged myocardium, whereas FN displayed a more ubiquitous deposition, suggesting different biological roles (Figure 3A).

We profiled hematopoietic cells (CD45⁺) infiltrating the apex and remote regions by flow cytometry at 2, 5, and 7 days after surgery. Single-cell suspensions were stained with antibodies to distinguish total myeloid cells (CD11b⁺), macrophages (CD11b⁺F4/80⁺), monocytes (CD11b⁺F4/80⁻), neutrophils (CD11b⁺Gr-1⁺), T lymphocytes (CD3⁺), and B lymphocytes (CD45R⁺) (Figure S3A). At 2 days after surgery, an influx of CD45⁺ cells was observed in the apex, paralleled by a reduction in the remote LV (Figures S3B and S3C). The rate of hematopoietic cells in the myocardium of apex-resected hearts was comparable with those observed in sham-operated controls at 7 days post-surgery, corroborating our immunofluorescence data. The immune infiltrate was mainly composed of myeloid cells. At the initial stages of injury-response (i.e., 2 days) neutrophils, monocytes, and macrophages infiltrated the myocardium at the apex, although only macrophages remained elevated up to 5 days post-surgery. No differences were observed regarding the percentage of T lymphocytes between groups, whereas B lymphocytes were significantly decreased 2 days post-injury (Figures S3B and S3C). Additional staining of myeloid populations allowed discrimination of CD11b⁺F4/80⁺Ly6C^lo macrophages, CD11b⁺F4/80⁺Ly6C^hi macrophages, CD11b⁺F4/80⁻Ly6C^lo monocytes, and CD11b⁺F4/80⁻Ly6C^hi inflammatory monocytes, as reported previously (Aurora et al., 2014). Of note, CD11b⁺F4/80⁺Ly6C^hi macrophages and CD11b⁺F4/80⁻Ly6C^hi monocytes were predominantly increased at the injury site, but not in the remote myocardium, during early inflammatory response (Figure S3D). In turn, CD11b⁺F4/80⁺Ly6C^lo macrophages were increased at 5 days after injury, similar to what was reported following myocardial infarction at P1 (Aurora et al., 2014).

Cardiac fibroblasts are master mediators of myocardial remodeling following injury (reviewed in Souders et al., 2009). However, studies determining the cardiac fibroblast surface signature during the neonatal period are lacking. Considering the heterogeneity of this cell type, we discriminated fibroblast populations, from embryonic day 17 to adulthood (Figure S4), through the expression of the three well-reported markers of cardiac fibroblasts, i.e., CD90 (Thy-1), CD140a (PDGFRα), and SCA-1 (Furtado et al., 2014, Hudon-David et al., 2007, McQualter et al., 2009, Smith et al., 2011), following exclusion of non-viable cells (PI⁺), endothelial cells (CD31⁺), hematopoietic cells (CD45⁺), and erythrocytes (TER-119⁺). These results demonstrated that fibroblasts express the mesenchymal marker CD140a (≥85%) at all stages analyzed, either in the presence or absence of CD90 (Figure S4). Throughout time these fibroblasts also acquire the expression of SCA-1 and, thus, in the adult, the most abundant subsets are CD140a⁺CD90⁻SCA-1⁺ (Figure S4, green) and CD140a⁺CD90⁺SCA-1⁺ (Figure S5, gray), as reported previously (Furtado et al., 2014).

We applied the same strategy to injured and sham-operated animals 7 days post-surgery (Figure 3B). Three main populations were identified: CD140a⁺CD90⁻SCA-1⁻ (blue; injury: 34.71% ± 2.24%; sham: 32.50% ± 2.72%), CD140a⁺CD90⁺SCA-1⁻ (purple; injury: 36.53% ± 0.95%; sham: 40.29% ± 1.85%), and CD140a⁺CD90⁺SCA-1⁺ (gray; injury: 21.26% ± 1.15%; sham: 17.77% ± 1.40%). A minor population was also found expressing CD90 only (red; injury: 3.79% ± 0.94%; sham: 6.34% ± 1.17%), whereas SCA-1 was mainly confined to cells also expressing the other surface markers (Figure 3B). No statistically significant differences were detected in the relative percentages of each population between injured animals and sham-operated controls (Figure 3B). This might be explained by the use of the whole heart and/or by variability created during the isolation procedure, which may mask small differences in cell numbers in the apex. Indeed, immunofluorescence results showed a fibroblast-mediated response to injury largely confined to the lesion site (Figures 3A and S2).

The activation of most abundant populations was assessed by real-time qPCR (sorting strategy detailed in Figure S5). The CD140a⁺CD90⁺SCA-1⁻ (purple, SP1) subpopulation, which is particularly frequent in earlier developmental periods (Figure S4), responded to apex resection by upregulating genes associated with scarring (Col1a1, Col3a1, Tgfb1, and Tgfb3), CM proliferation (Fn1, Tbx20, Igf1, Igf2, and Fstl1), and neovascularization (Vegfa) (Figure 3C). In contrast, CD90⁺CD140a⁺SCA-1⁺ (gray, SP3) cells, a subpopulation upsurging after birth (Figure S4), shows a repair-associated phenotype, with increased expression of Col1a1, Col3a1, Tgfb1, and Tgfb3 (Figure 3C). CD140a⁺CD90⁻SCA-1⁻ (blue, SP2) cells, more frequent in earlier developmental periods (Figure S4), were not responsive to injury, although they exhibited high expression levels of most evaluated transcripts (Figure 3C). These findings indicate that development-associated fibroblasts support the regenerative response, whereas fibroblast subsets abundant in adulthood signal mostly toward scar formation.

CM Proliferation Is Induced upon Cardiac Injury and Leads to Newly Formed CMs

The activation of CM proliferation after apex resection has been a matter of intense debate (Zebrowski et al., 2016). At 7 days post-injury, immunostaining for histone H3 phosphorylated at serine 10 (PH3) and s-α-Actinin highlighted cycling CMs that were distributed throughout the myocardium. Quantification of PH3⁺ CMs in the myocardium below the papillary muscle and in isolated CMs from the same region showed that injured hearts displayed increased rates of mitotic CMs (PH3⁺) in comparison with sham-operated hearts (Figures 4A and 4B). Moreover, CMs undergoing mitosis show sarcomere disruption, which is essential for effective cell division (Figures 4C and S6A; Movies S1, S2, and S3). These findings were further supported by a significant increase in CMs displaying Aurora B kinase in the nucleus and in the cleavage furrow, of the border zone and remote myocardium, of resected hearts, indicating efficient cytokinesis (Figures 4D and 4E). Close to the injury, mitotic CMs were found predominantly near FN, but not TN-C. In accordance, CD29 (β1 integrin), which forms an integral part of several integrins involved in cell-matrix interactions, was found in most CMs (Ieda et al., 2009) (Figure S6).

The loss of CM proliferative potential is commonly attributed to the progression to a terminally differentiated phenotype characterized by the presence of a mature sarcomeric apparatus and binucleation (Collesi et al., 2008). To assess whether increased rates of mitosis correlated with higher rates of CM proliferation rather than multinucleation, we assessed the number of nuclei and morphometric parameters of CMs isolated from the apex by imaging flow cytometry. Injured hearts showed less multinucleation and a significant increase of mononucleated and round-shaped CMs, attesting successful cell-cycle progression and cell division (Figures 4F and 4G).

Next, we applied stereology to establish whether the proliferative upsurge observed upon apex resection impacts on total CM numbers. The mice were treated with the thymidine analog 5-ethynyl-2′-deoxyuridine (EdU), and the frequency of EdU incorporation in thick tissue sections was determined by co-labeling with PCM-1, a marker of CM nuclei (Figure 4H). The number of EdU⁺ CMs was greater in resected hearts relative to sham-operated controls (Figures 4H–4J), further supporting an increment in CM proliferation after injury. Although CM proliferation was observed throughout the LV, the frequency of EdU⁺ CMs was more pronounced in regions proximal to the apex, compared with remote regions (Figure 4J). In accordance, the density of myocyte nuclei was found to be significantly increased in the apex regions (Figure 4K). Combination of collagen type IV (COLL IV), to delineate cell boundaries, with PCM-1 immunolabelling, showed that CM binucleation in injured (95.7% ± 0.93%) and sham-operated hearts (95.98% ± 0.86%) was similar at 14 days after surgery (Figure 4L).

Taking multinucleation into account, we determined the density of CMs in the tissue. The volume of the LV was calculated based on morphometric analysis of longitudinal tissue sections (Figure S1A) and corrected for tissue shrinkage during fixation and processing for paraffin embedding. The total number of LV CMs in resected hearts, calculated based on CM density in the remote region, was significantly higher relative to sham-operated controls (Figure 4M). These findings show that the injury-activated CM proliferation burst, although more abundant in the apex, is also extended to remote regions, overcompensating the amount of CM loss after resection.

Resected Hearts Display Increased Capillary Density and No Signs of CM Hypertrophy or Edema

To evaluate the status of the capillary network, endothelial cells (CD31⁺) were quantified 60 days after surgery. The number of CD31⁺ endothelial cells was significantly increased in injured hearts in comparison with sham-operated animals both in the apex and in the remote myocardium (Figures 5A and 5B).

To further establish whether hypertrophy was also contributing for LV wall expansion, CM size was assessed in tissue sections and isolated CMs. No significant alterations were detected in the cross-sectional area of CMs over time and in different ventricular regions, apart from the vicinity of fibrotic tissue, where CMs had a larger cross-sectional area at 180 days post-surgery (Figures 5C and 5D). Corroborating these findings, no significant differences were found in the area and volume of isolated CMs, suggesting that this hypertrophic behavior is residual and mainly localized in the apex (Figures 5E–5H). Furthermore, as edema could also explain the increase in myocardial volume, through the accumulation of fluid in the extracellular interstitial space, we examined the myocardial water content at 180 days post-surgery, but no differences were detected between the two groups (Figure 5I). Overall, these results further support that the LV wall thickening in resected hearts results from an increase in the number of CMs and is thus a consequence of injury-activated proliferation.

Animals and Apex Resection Injury Model

All animal work was approved by the IBMC-INEB (Instituto de Biologia Molecular e Celular–Instituto de Engenharia Biomédica) Animal Ethics Committee, and by the Direcção Geral de Veterinária (permit 022793), and is in conformity with Directive 2010/63/EU. Humane endpoints were followed in accordance with the OECD Guidance Document on the Recognition, Assessment, and Use of Clinical Signs as Humane Endpoints for Experimental Animals Used in Safety Evaluation.

P1 C57BL/6 mice were anesthetized by hypothermia for ∼3 min. Animals were laid in lateral decubitus position, the skin cut, the thorax was opened in the fourth intercostal space and the ventricular apex was resected (removing a minimal amount of tissue but ensuring lumen exposure). Subsequently, the ribs and skin were sutured and animals were exposed to heat produced by an infrared lamp to revert anesthesia and returned the progenitor. Sham-operated animals underwent the same procedure with the exception of apex resection.

Histological Characterization

At 0, 7, 14, 21, 60, 120, and 180 days after surgery, apex-resected and sham-operated hearts were harvested and processed for paraffin-embedding (see Supplemental Experimental Procedures for detailed information). Hearts were sectioned longitudinally in 3 μm sections and sampled according to Figure S1A. Subsequent histological MT staining was used to assess apex re-growth, fibrosis deposition, and injury extension.

Immunofluorescence

Immunofluorescence was performed in cryosections (0 day, 48 hr, 5 days, 7 days, 14 days, and 21 days after surgery), in histological sections (60 days post-surgery), and in cytospins of isolated CMs; as detailed in the Supplemental Experimental Procedures. Assembly, merging, and contrast/brightness adjustments were performed in Fiji v2.0.0-rc-54/.51h. software.

Flow Cytometry and Fluorescence-Activated Cell Sorting

Single-cell suspensions were prepared following digestion of cardiac tissue with crude collagenase (C2139, Sigma-Aldrich). Cells were incubated with fluorescence-conjugated antibodies or with the respective isotype controls for 30 min on ice and protected from light. Nonviable cells were excluded by adding 0.5% propidium iodide (P4170, Sigma-Aldrich) to cell suspension prior to analysis in a FACSCanto II cytometer (BD Biosciences). Data analysis was performed in FlowJo VX software. The three most frequent fibroblast subsets were isolated by FACSAria (BD Biosciences), as described in Supplemental Experimental Procedures and in Figure S5. See also Figure S3 for detailed characterization of the inflammatory infiltrate.

Real-Time qPCR

Total RNA was isolated from FACS-sorted cells using the RNeasy Mini Kit (QIAGEN). cDNA was obtained resorting to PrimeScript RT reagent Kit (Takara) and pre-amplified using SsoAdvanced PreAmp Supermix (Bio-Rad). Real-time qPCR were performed by using iQ SYBR Green Supermix (Bio-Rad) and according to the iQ5 Real-Time PCR Detection System (Bio-Rad). Primer sequences and temperature cycles are described in Supplemental Experimental Procedures. mRNA expression was defined as primer efficiency to the power of the difference in threshold cycle values between the reference (Gapdh) and the genes of interest.

EdU Pulse-Chase Assay

Neonates were subcutaneously injected with 25 μg/g of animal of EdU (C10640, Sigma-Aldrich) immediately after surgery and on days 5 and 7 after surgery. Hearts were harvested at 14 days post-surgery and LV subdivided in apex and remote myocardium, each sampled in 2 mm fragments before cryosectioning (40 μm). EdU incorporation was detected using the Click-iT reaction cocktail in accordance with manufacturer instructions (C10640, Sigma-Aldrich) followed by PCM-1 immunolabelling and DAPI counterstain. Images were acquired in a Leica TCS SP5 laser scanning confocal microscope. Tissue fragments (8–10) from both regions were quantified per animal with assistance of FIJI v2.0.0-rc-54/.51h and IMARIS 8.4.1. The total number of CMs was estimated by adapting the two-step N_V × V_REF method (Alkass et al., 2015), where N_V is an estimate of PCM-1 density and VREF is the volume of the myocardium at 14 days post-surgery. Myocardial volume was determined by multiplying the area of the myocardium in each histological longitudinal section by the distance separating adjacent sections in 14 days post-surgery hearts. This value was corrected for tissue shrinkage during fixation and processing for paraffin embedding.

CM Isolation

CMs were isolated at 7, 60, and 180 days post-surgery. Neonatal CMs were isolated as described previously (Mollova et al., 2013), with the following alterations: CMs were fixed in 4% paraformaldehyde at room temperature (RT) for 2 hr and digested with 3 mg/mL collagenase type II (Worthington). Adult CMs were isolated by reperfusion in a Langendorff system (∼80 mmHg and at 3 mL/min for 25–30 min) with Liberase TM (05401127001, Roche; 13.3 μg/mL) and Trypsin (27250018, Gibco; 13.8 μg/mL), and fixed in 4% paraformaldehyde in PBS.

Imaging Flow Cytometry

Cells were permeabilized with 1× BD Perm/Wash buffer during 15 min at RT. Incubations with primary and secondary antibodies lasted 2 and 0.5 hr on ice, respectively. Antibodies were diluted in 1× BD Perm/Wash. Immediately before acquisition in ImageStreamX cells were filtered and nuclei stained with 200 μM DRAQ5 (BioStatus).

Statistical Analysis

IBM SPSS Statistics 24 was used for statistical analysis. Shapiro-Wilk test was used to evaluate normal distribution of data. Outliers were excluded from the statistical analysis. Homoscedasticity of the sample was tested by Levene's test, which defined the statistical test(s) applied. Normally distributed and homocedastic data were tested with independent sample Student's t test and one-way ANOVA for two or three or more groups, respectively. Tukey's post-hoc test for correction of multiple comparisons was performed whenever significant differences were found between three or more groups.

Non-normally distributed and/or heterocedastic data were tested with Mann-Whitney U test and Kruskal-Wallis one-way ANOVA for two or three or more groups, respectively. Again, multiple comparisons were performed when three or more groups were considered statistically different. The statistical significance level chosen for all statistical tests was p < 0.05.