Functional recovery from LV ischemic heart injury involves a prominent adaptive response of the RV. In order to create consistent, ischemic heart lesions in neonatal mice, we used a small cauterizer to ablate the root of the LCA in P1 (P1 MI) mice (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.176281DS1). This produced a large, highly reproducible ischemic injury in the LV, with recovery rates above 90%. At 2 days post surgery (dps), large parts of the LV of P1 MI mice were injured, as shown by lack of troponin I (TNNI3) staining (Figure 1A). The lesion typically affected the middle layer of the myocardium, leaving uninjured inner and outer layers of myocardium and intact endocardial and epicardial layers (Figure 1A and Supplemental Figure 1B). The injury size was very large at 2 dps (38.99% ± 1.18% of the LV, n = 5), decreased significantly (19.40% ± 2.88%, n = 5) at 7 dps, and was minimal at 21 dps (6.59% ± 0.33%, n = 5) (Figure 1B). Functional assessment of the LV with echocardiography showed a strong reduction in the LV ejection fraction (EF) and fractional shortening (FS) at 1 dps compared with shams, which was associated with an increase in LV end-diastolic area (LVEDA) (Figure 1C and Supplemental Figure 1, C and D). Importantly, LV function almost completely recovered at 7 dps in P1 MI mice, and functional measurements were indistinguishable from shams at 14 and 21 dps (Figure 1C and Supplemental Figure 1, C and D).

Cauterization of the root of the LCA results in opposite outcomes in P1 and P3 MI mice. (A) P1 MI hearts and heart sections at 2, 7, and 21 days post surgery (dps). Visualization of scar areas (marked with dashed lines) with TNNI3 and DAPI or Sirius red and Fast Green costaining. Scale bars: 1 mm. (B) Quantitation of infarct size (percentage of the LV). (C) Ejection fraction (EF) measured with echocardiography in P1. (D) P3 MI hearts and heart sections, visualization of scar area (marked with dashed lines) with TNNI3 and DAPI or Sirius red and Fast Green costaining. Scale bars: 1 mm. (E) Quantitation of infarct size. (F) EF measured with echocardiography. (G) Kaplan-Meier survival curve of mice after P1 MI and P3 MI. (H) Echocardiographic short-axis view of MI and sham hearts; dashed lines mark endocardial and epicardial regions. Scale bar: 1 mm. (I) Echocardiographic measurements of LV wall thickness in MI and sham hearts. (J) Echocardiographic long-axis view of MI and sham hearts; red lines mark the endocardium of the LV at the end of diastole. Scale bar: 1 mm. (K) Echocardiographic quantitation of the LV end diastolic area (LVEDA) in MI and sham hearts. (L) Quantitation of LV cardiomyocyte (CM) cross-sectional area in MI and sham hearts. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by unpaired, 2-tailed Student’s t test (E) or 1-way ANOVA with Holm-Šidák post hoc test (B, C, F, I, K, and L). Data in G were evaluated with the log-rank test: P = 0.0006. NS, no significance.

By contrast, cauterization of P3 mice, in which a more adult-like compacted myocardium has developed (17), resulted in an enlarging lesion and loss of function (Figure 1, D–F). Lesion size in P3 MI 2-dps hearts was similar to P1 MI hearts (42.74% ± 1.64%, n = 6; Figure 1, B and E, and Supplemental Figure 1, B and E), but increased in size at 7 dps and became transmural (70.41% ± 2.99%, n = 6; Figure 1, D and E). Moreover, LV function remained compromised, often leading to spontaneous death or requiring euthanasia (Figure 1G and Supplemental Figure 1F). Echocardiography indicated a similar reduction in LV wall thickness, and an increase in LVEDA at 1 dps in P1 MI and P3 MI mice (Figure 1, H and I, and Supplemental Figure 1G), whereas at 7 dps both parameters had recovered in P1 MI, but not in P3 MI hearts (Figure 1, J and K, and Supplemental Figure 1H), and CM cross-sectional area was 40% higher in P3 mice at 7 dps, indicating hypertrophy and adverse remodeling (Figure 1L and Supplemental Figure 1I).

Surprisingly, we observed a prominent difference in the immediate functional response of the RV to MI in P1 MI and P3 MI mice when examining RV function through echocardiographic measurements of fractional area change and RV end-diastolic area. These measurements maintained function in the RV of P1 MI mice (Figure 2, A–E). This correlated with major morphological changes in the RV in P1 MI mice, as it became more elongated at 7 dps and increased further with age (Figure 2F). Moreover, echocardiography revealed that the RV even partially wrapped around the LV and formed the tip of the heart at 14 and 21 dps (Figure 2G and Supplemental Videos 1–6). An elongation of the RV was also found in histological sections of P1 MI hearts and a steady increase up to 120 dps occurred (Figure 2, F and H), but the RV CM length/width ratio remained very similar in MI and sham mice (Figure 2, I and J). In contrast to P1 MI mice, echocardiography revealed deterioration of RV function and global heart failure in P3 MI mice at 1 and 7dps (Figure 2, A–E). These data identified prominent adaptive changes in the RV that underlie the functional rescue from injury in P1 MI, but not P3 MI, mice.

RV function and morphology differ between P1 MI and P3 MI hearts; P1 MI hearts display a small persistent scar and adaptive hypertrophy at 120 dps. (A) Echocardiographic measurement of RV fractional area change (RV FAC). (B) Echocardiographic short-axis view of the ventricles; dashed blue lines indicate the RV end diastolic area (RVEDA). Scale bar: 1 mm. (C–E) Echocardiographic measurements of RVEDA (C and E) and RV FAC (D). (F) Echocardiographic measurements of the length of the RV free wall after P1 surgery. (G) Echocardiographic longitudinal axis view of P1 ventricles; blue lines indicate endo- and epicardial layers of the RV and the LV. Scale bar: 1 mm. (H) Heart sections of a P1 MI and sham heart; black dashed lines mark the RV free wall. The RV is elongated in the P1 MI heart. Black arrowhead indicates residual scar. Scale bar: 2 mm. (I and J) Heart sections (I) and quantitation (J) of length/width ratio of CMs costained with WGA and DAPI in the RV. Scale bar: 10 μm. (K and L) Images (K) and quantitation (L) of heart weight (HW)/tibia length (TL) ratio. Scale bar: 2 mm. (M and N) Heart sections stained for α-actin (ACTN2) and with WGA and DAPI (M), and quantitation of CM cross-sectional area (N) in both ventricles. Scale bar: 20 μm. (O) LV EF. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by 1-way ANOVA with Holm-Šidák post hoc test (A–F) or unpaired, 2-tailed Student’s t test (J, L, and O). NS, no significance.

LV and vascular adaption to left heart ischemia. To determine the extent of regeneration, we examined the hearts of P1 MI mice at 120 dps and detected only a small remaining scar in the LV (2.92% ± 0.10% of total LV area, n = 4) (Figure 2H). At this stage, the hearts displayed a roundish shape (Figure 2K). Heart weight/tibia length, heart weight/body weight ratios, and the average cross-sectional area of CMs in both ventricles were all significantly increased, consistent with an increase in heart cell mass (Figure 2, L–N, and Supplemental Figure 2A). However, no signs of pathological remodeling such as interstitial fibrosis were found, and EF, cardiac output (CO), and LVEDA were similar to those of sham hearts (Figure 2O and Supplemental Figure 2, B–D).

Regenerative adaptation to ischemic infarction in P1 mice was also associated with substantial vascular remodeling. Vessel casting at 21 dps demonstrated an atypical septal coronary artery (SCA), which was elongated and extended toward the ischemic area in the LV in 86% of hearts (n = 29) (Figure 3, A and B, and Supplemental Figure 2E). Other vascular adaptations included elongation and expansion of the right coronary artery (RCA) to the left side of the heart (48.3%; Supplemental Figure 2F). In some hearts, vascular connections to the lung and the chest wall (27.5%) or the coronary vein (13.79%) were observed (Supplemental Figure 2, G and H).

Atypical revascularization in P1 MI hearts; transcriptome analysis of the LV at 1 dps reveals a large overlap, but also striking differences between P1 MI and P3 MI compared with sham hearts. (A and B) Coronary artery vessel casting: red arrowheads mark LAD coronary artery or the residual of it after MI, green arrowheads left circumflex coronary artery or the residual of it after MI, red arrows indicate the direction of the blood flow from the septal coronary artery to the LV after MI, and green arrows indicate the direction of the blood flow from the right coronary artery to the LV after MI. Scale bar: 1 mm. (C and D) Heatmaps of differentially regulated genes and gene ontology (GO) of biological processes of upregulated and downregulated genes in the LV. (E) Venn diagram showing a high overlap in genes overexpressed in the LV; tables list selected uniquely and overlapping GOs of upregulated genes in LV. (F) Heatmap of genes related to biological processes of “negative regulation of cell proliferation” in LV.

LV neonatal cardiac regeneration. RNA-seq analysis of the LV identified activation of distinct and highly overlapping (1585 genes, 70%) gene expression patterns in P1 and P3 MI hearts, including genes previously associated with pathologic hypertrophy (Figure 3, C–E) (18). These comprised gene categories identified by positive regulation of inflammatory responses, cell migration, angiogenesis, cell proliferation, and the ERK1 and ERK2 signaling cascade. However, important differences in P1 MI and P3 MI cardiac gene expression were also noted, as genes associated with positive regulation of apoptosis, cell cycle arrest, and Hippo signaling, and negative regulation of cell proliferation and canonical Wnt signaling were unique for P3 MI LVs (Figure 3, E and F). Interestingly, RNA-seq data of the septum showed that the gene expression pattern in P1 MI and P3 MI hearts was very similar to that in the LV (Supplemental Figure 3, A–C).

Because of the prominent upregulation of genes involved in cell proliferation and apoptosis in the LV of P1 MI hearts, we further investigated postinfarction CM proliferation. Increased cell cycle activity was indicated by significantly enhanced MKI67 staining in the LV of P1 MI hearts (Figure 4, A and B, and Supplemental Figure 4A), and cell division was confirmed using transgenic CAG-eGFP-anillin mice in which cells display cell cycle–specific cytosolic sublocalization of eGFP (19, 20). Infarction of P1 CAG-eGFP-anillin mice resulted in a 2-fold increase in eGFP-ANILLIN+/TNNI3+ interphase CMs in LV heart sections, whereas no increase was found in the LV of P3 MI hearts (Figure 4, C and D, and Supplemental Figure 4B). Finally, using midbody position analysis (19), we found that the number of CMs undergoing cytokinesis and binucleation in the LV of P1 MI hearts was more than 2-fold higher than in the LV of P1 shams, whereas neither cytokinesis nor binucleation was significantly increased in the LV of P3 MI hearts (Figure 4, E–G). Increased CM binucleation rates relative to P3 MI hearts were confirmed in CMs isolated from P1 MI 4-dps and P3 MI 2-dps αMHC-H2BmCherry mice (21), in which all CM nuclei are marked by red fluorescence (Figure 4, H and I). Thus, infarction at P1 causes CMs to proliferate and binucleate at an increased rate in the LV of P1 MI, but not P3 MI, hearts.

CM cytokinesis and binucleation strongly increase in the LV of P1, but not of P3, MI hearts. (A and B) Images (A) and quantitation (B) of MKI67+ CMs in LV marked by costaining for MKI67, TNNI3, and with DAPI. Scale bar: 20 μm. (C) Mosaic images of LV heart sections of CAG-eGFP-anillin mice costained for eGFP and TNNI3; arrowheads mark eGFP+ CMs, which are shown at higher magnification (×5.5) as confocal images in insets. Scale bar: 20 μm. (D) Quantitation of eGFP+ CMs in the LV. (E–G) Cytokinetic and binucleating CMs in heart sections costained for eGFP, AURKB, TNNI3, and with DAPI; midbodies were identified based on costaining for eGFP and AURKB (arrows). Two-sided arrows mark the distance between the 2 nuclei in the same CM; typical (yellow arrows) or atypical midbody location (green arrows) mark cytokinetic (F) and binucleating (G) CMs. Scale bar: 10 μm. (H) Images of mono- and binucleated CMs isolated from αMHC-H2BmCherry hearts at P5. Scale bar: 20 μm. (I) Quantitation of the percentage of binucleated/total CMs isolated from whole hearts. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by 1-way ANOVA with Holm-Šidák post hoc test. NS, no significance.

Cyclin-dependent kinase inhibitor 1A (Cdkn1a) was significantly upregulated in LV P3 hearts in RNA-seq analysis and CDKN1A staining (Figure 5, A and B), and expression of the upstream Cdkn1a signaling elements Foxo1 and Akt1 were consistent with cyclase-dependent kinase inhibition limiting cell cycle progression in P3 MI hearts (Supplemental Figure 4C). In addition to inhibition of cell cycle progression, FOXO1 has been implicated in caspase-mediated apoptosis (22), and P3 MI hearts evidenced upregulation of genes related to programmed cell death (Figure 5C). Accordingly, apoptosis rates in the lesion site of P3 MI increased at least 2-fold compared with P1 MI hearts, as assessed by cleaved caspase 3 (cCASP3) staining and terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL), indicating that apoptosis was a prominent feature of the MI response in P3 hearts (Figure 5, D–G). Conversely, immune response genes were upregulated in the injured area of P1 MI hearts, which persisted over time, and PTPRC+ cell invasion was increased 4-fold in P1 MI compared with P3 MI hearts (Figure 5, H and I, and Supplemental Figure 4D). P1 MI and P3 MI hearts were also distinguished by the expression of an angiogenesis gene program, and capillary density was reduced by almost 25% in the LV of P3 MI hearts at 7 dps, whereas P1 MI hearts showed no reduction in capillary density (Figure 5, J and K). These data suggest that in the LV the absence of CDKN1A-mediated inhibition of cell cycle progression and programmed cell death, as well as increased cell invasion and angiogenesis, underlie the maintenance of LV post-MI function at P1.

Increased rates of apoptosis and CDKN1A+ CMs in the LV of P3 MI compared with P1 MI hearts. (A) Mosaic whole-heart and magnified LV images of heart sections costained for CDKN1A, TNNI3, and with DAPI. (B) Quantitation of CDKN1A+ CMs in heart sections. White scale bars: 1 mm; yellow scale bars: 40 μm. (C) Heatmap showing expression of genes related to GO “positive regulation of apoptosis process” in LV. (D) Mosaic whole-heart and magnified infarcted-area images of heart sections costained for cleaved caspase 3 (cCASP3), TNNI3, and with DAPI; dashed lines mark infarct areas. White scale bars: 500 μm; yellow scale bars: 20 μm. (E) Quantitation of cCASP3+ and DAPI+ cells in the infarct area. (F) Mosaic whole-heart and magnified infarcted-area images of heart sections costained for TUNEL, TNNI3, and with DAPI; dashed lines mark infarct area. White scale bars: 200 μm; yellow scale bars: 20 μm. (G) Quantitation of TUNEL+ and DAPI+ cells in the infarct area. (H) Mosaic whole-heart and magnified infarcted-area images of heart sections costained for PTPRC, TNNI3, and with DAPI. White scale bars: 500 μm; yellow scale bars: 40 μm. (I) Quantitation of PTPRC+ and DAPI+ cells in the infarct area. (J) Capillaries in LV heart sections costained for PECAM1, TNNI3, and with WGA. Scale bar: 10 μm. (K) Quantitation of capillary density assessed by number of PECAM1+ cells per CM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by 1-way ANOVA with Holm-Šidák post hoc test (B and F) or unpaired, 2-tailed Student’s t test (E, G, and I).

RV CM expansion and adaption are triggered by LV infarction. Unlike the highly overlapping gene expression pattern (70%) in the LV and septum of P1 MI and P3 MI mice, we found in the RV an overlap of only 17% in upregulated genes between P1 MI and P3 MI hearts (Figure 6, A–C). The number of uniquely upregulated genes in the RV of P1 MI hearts (total 799 genes, 658 unique to P1) was more than 2-fold greater than in P3 MI hearts (total 325 genes, 184 unique to P3), indicating a global cardiac response underlying neonatal heart regeneration. For instance, cell cycle–related genes were uniquely upregulated in the RV of P1 MI mice, whereas genes related to negative regulation of cell proliferation characterized the expression pattern of the RV of P3 MI mice (Figure 6, D and E). MKI67+ CMs and, to a lesser degree, MKI67+ non-CMs, were highly increased in the RV relative to the LV of P1 MI mice (Figure 7, A and B, and Supplemental Figure 5A), whereas MKI67+ CMs were fewer in the RV of P3 MI hearts and more similar to the level found in the LV of P3 MI hearts (Figure 7C). Analysis of eGFP-ANILLIN+ CMs in RV of P1 MI hearts confirmed a strong increase in CM division and binucleation rates, unlike in P3 MI hearts (Figure 7, D–F, and Supplemental Figure 5, B–D). Consistent with the critical role of cyclin-dependent kinase inhibition in heart regeneration, CDKN1A+ CMs were strongly increased in the RV of P3 MI hearts (Figure 7G and Supplemental Figure 5E).

Transcriptomic analysis of the RV in P1 MI, P3 MI, and sham hearts at 1 dps reveals a unique gene expression pattern. (A and B) Heatmaps show differentially regulated genes and GO of biological processes of upregulated and downregulated genes in the RV. (C) Venn diagram depicting uniquely overexpressed genes in the RV, and overlapping genes compared to sham. Tables list selected GO biological processes. (D and E) Heatmaps showing gene expression in the GO “cell cycle” (D) and “negative regulation of cell proliferation” (E) in the RV.

Increased CM proliferation in the RV free wall of P1, but not of P3, MI hearts. (A) Confocal images of RV heart sections costained for MKI67, TNNI3, and with DAPI. Scale bar: 20 μm. (B and C) Quantitation of MKI67+ CMs in the RV and LV. (D) Quantitation of cycling CMs (eGFP+) in the RV of CAG-eGFP-anillin hearts. (E and F) Quantitation of CMs undergoing cytokinesis (E) and binucleation (F) in the RV of CAG-eGFP-anillin hearts. (G) Quantitation of CDKN1A+ CMs in the RV. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by 1-way ANOVA with Holm-Šidák post hoc test. NS, no significance.

These changes were consistent with prominent histological changes in the RV in P1 MI and P3 MI hearts. The RV wall was over 40% thicker than sham controls in P1 MI hearts, whereas it was 20% thinner in P3 MI hearts at 4 dps (Figure 8, A and B), further suggesting a major role of the RV in the functional rescue. Myocyte hypertrophy was not a feature of RV rescue of infarction, as RV CM cross-sectional area was unaltered in P1 MI hearts, but more than doubled in P3 MI hearts (Figure 8, C and D). In addition, capillary density was increased in the RV of P1 MI hearts at 7 dps, consistent with an upregulation of angiogenic genes, but decreased in the RV of P3 MI hearts (Figure 6, A–C, and Figure 8, E and F). Thus, LV ischemia evokes global cellular responses and prominent myocyte expansion in the RV, which are sufficient to rescue cardiac function in P1, but not P3, mice. However, unlike the ischemic LV, the RV lacks the direct effects of cell death, blood extravasation, inflammatory cell influx, and cytokine release (see also Supplemental Figure 4D). We postulated that postcapillary pulmonary arterial hypertension secondary to LV failure immediately after MI could likely be responsible for such prominent RV adaptation in P1 hearts (Figure 1C). With high-resolution echocardiographic assessment, as recently described for neonatal mice and human infants (23, 24), pulmonary artery acceleration time (PAAT) and pulmonary artery ejection time (PAET) were measured at 1 dps to assess increases in pulmonary arterial pressure (Supplemental Figure 5, F and G). Both PAAT and the PAAT/PAET ratio were significantly decreased in P1 MI mice compared with sham controls, consistent with an increase in pulmonary arterial pressure (Figure 8G and Supplemental Figure 5H).

Different adaptive response of the RV of P1 MI and P3 MI hearts, and functional evidence of pulmonary arterial hypertension in P1 MI mice. (A and B) Images (A) and quantitation (B) of the wall thickness of heart sections stained for TNNI3 and costained with DAPI; dashed lines mark infarct area, double arrows indicate measuring sites. Scale bar: 1 mm. (C and D) RV heart sections stained for TNNI3 and costained with WGA and DAPI showing CM sizes (C) and quantification (D) of CM cross-sectional area in the RV. Scale bar: 20 μm. (E and F) RV heart sections costained for PECAM1, TNNI3, and with WGA and DAPI showing capillaries (E) and quantitation of capillary density (F). Scale bar: 20 μm. (G) Pulmonary artery acceleration time/pulmonary artery ejection time (PAAT/PAET) ratio obtained with Doppler measurements of flow velocity in the main pulmonary artery in P1 MI and sham mice. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by 1-way ANOVA with Holm-Šidák post hoc test (B, D, and F) or unpaired, 2-tailed Student’s t test (G). NS, no significance.

Transcriptome analysis further indicated that after infarction, the RV in P1 hearts shifts to a gene expression pattern similar to the noninfarcted LV of P1 sham hearts (Supplemental Figure 5, I and J), suggesting that upon LV injury, systemic signals of cardiac stress trigger a critical shift in the expression program associated with a higher end-diastolic pressure. This response was absent in P3 MI hearts, which displayed typical features of a failing RV, indicating the narrow developmental window in which regenerative capacity and functional adaptation are available.