Pericardial delta like non-canonical NOTCH ligand 1 (Dlk1) augments fibrosis in the heart through epithelial to mesenchymal transition

Abstract

Background: Heart failure due to myocardial infarction (MI) involves fibrosis driven by epicardium-derived cells (EPDCs) and cardiac fibroblasts, but strategies to inhibit and provide cardio-protection remains poor. The imprinted gene, non-canonical NOTCH ligand 1 (Dlk1), has previously been shown to mediate fibrosis in the skin, lung and liver, but very little is known on its effect in the heart.

Methods: Herein, human pericardial fluid/plasma and tissue biopsies were assessed for DLK1, whereas the spatiotemporal expression of Dlk1 was determined in mouse hearts. The Dlk1 heart phenotype in normal and MI hearts was assessed in transgenic mice either lacking or overexpressing Dlk1. Finally, in/ex vivo cell studies provided knowledge on the molecular mechanism.

Results: Dlk1 was demonstrated in non-myocytes of the developing human myocardium but exhibited a restricted pericardial expression in adulthood. Soluble DLK1 was twofold higher in pericardial fluid (median 45.7 [34.7 (IQR)) μg/L] from cardiovascular patients (n = 127) than in plasma (median 26.1 μg/L [11.1 (IQR)]. The spatial and temporal expression pattern of Dlk1 was recapitulated in mouse and rat hearts. Similar to humans lacking Dlk1, adult Dlk1^-/- mice exhibited a relatively mild developmental, although consistent cardiac phenotype with some abnormalities in heart size, shape, thorax orientation and non-myocyte number, but were functionally normal. However, after MI, scar size was substantially reduced in Dlk1^-/- hearts as compared with Dlk1^+/+ littermates. In line, high levels of Dlk1 in transgenic mice Dlk1^fl/fl xWT1^GFPCre and Dlk1^fl/fl xαMHC^Cre/+Tam increased scar size following MI. Further mechanistic and cellular insight demonstrated that pericardial Dlk1 mediates cardiac fibrosis through epithelial to mesenchymal transition (EMT) of the EPDC lineage by maintaining Integrin β8 (Itgb8), a major activator of transforming growth factor β and EMT.

Conclusions: Our results suggest that pericardial Dlk1 embraces a, so far, unnoticed role in the heart augmenting cardiac fibrosis through EMT. Monitoring DLK1 levels as well as targeting pericardial DLK1 may thus offer new venues for cardio-protection.

Keywords: Delta like non-canonical NOTCH ligand 1 (Dlk1); cardiac fibrosis; epicardium-derived cells (EPDCs); epithelial to mesenchymal transition; myocardial infarction; myocardial remodelling.

FIGURE 1

Delta like non‐canonical notch ligand 1 (Dlk1) in the human heart is restricted to cells in the pericardium and soluble sDLK1 is secreted into the pericardial fluid. (A) Relative quantitative real time PCR of DLK1 in human myocardial specimens during development (days post‐fertilisatioin, dpf) and adulthood (n = 1–3). B2M, ATP6A and COX4A were used for normalisation. (B) DLK1 immunohistochemistry of normal human myocardium (n = 3) and *pituitary gland as positive control. Arrows point at discrete non‐muscle cells expressing DLK1. (C) DLK1 immunohistochemistry of human pericardium (n = 12) and positive control tissue (pituitary gland). Arrows and arrowheads point at DLK1+ and DLK1− mesothelial cells, respectively, while * marks vasculature with a DLK1+ endothelium. (D and E) Soluble DLK1 was measured in corresponding plasma and pericardial fluid samples from patients (n = 127) undergoing heart valve or coronary artery surgery. Wilcoxon matched‐pairs signed rank test (D) and simple linear regression (E) were used to test statistical significance. See Figure S1 for further details.

FIGURE 2

Dlk1 localises to the secondary heart field during mouse cardiac development. (A and B) Relative mRNA expression levels of full‐length and Dlk1^{Protease site} mRNA (Dlk1 variants that comprise the protease site resulting in soluble Dlk1 isoforms) during mouse heart development. Number of animals for each sample is depicted on graphs. Raw data were normalised against Gapdh and beta‐actin (qbase+: M = 0.542; CV = 0.187). (C) DLK1 (red) and sarcomeric Myosin (green) immunofluorescence of hearts during development and adulthood (n = 2–4 animals deriving from different litters for each timepoint were analysed, representative pictures shown. See Figure S2 for further details. (D and E) DLK1 is co‐expressed with markers of epicardial‐derived progenitor cells (EPDCs) at E10.5 in EPDCs and descendants residing in (D) the outflow tract (OFT) and within (E) the peri‐/myocardium as visualised by immunofluorescence (n = 3 animals deriving from different litters were analysed, representative pictures shown). DAPI (blue) was used for staining nuclei.

FIGURE 3

Dlk1^−/− hearts display modest abnormal growth. Stereomicroscopic dark field pictures of (A) E10.5− and (B) adult mouse Dlk1^+/+ and Dlk1^−/− hearts. Independent of sex (Figure S5), heart height/width was increased in adult Dlk1^−/− hearts, whereas (C) heart/body weight was increased in young (P1–P3) and decreased in adult‐old (P79–P436) Dlk1^−/− hearts as compared with Dlk1^+/+ hearts. (D) Electrocardiography (representative) showed that (E) electrical heart activity was similar in adult Dlk1^+/+ and Dlk1^−/− mice independent of sex. (F) Microfil‐injected coronary beds of Dlk1^+/+ and Dlk1^−/− hearts were visualised by CT scanning and used for (G) quantifying coronary artery length, which was further (H) normalised to heart weight. For all sets of data, we confirmed sex independency by two‐way ANOVA, and then in the accumulated data sets statistical significance was tested using unpaired t‐test (B and G) or Mann–Whitney (C, E and H) depending on data normality (D'Agostino & Pearson). LV, left ventricle; LA, left atrium; OFT, outflow tract; RV, right ventricle; RA, right atrium.

FIGURE 4

Dlk1 increases scar size following myocardial infarction (MI). (A–C) MI (LAD) or sham surgery was performed in 10‐weeks old female C57bl/6 mice (cardiac stress markers, see Figure S7A). (A) Immunofluorescence images of hearts from three sets of generated transgenic mice: Dlk1^−/− (dlk1 deletion), Dlk1^fl/flxWT1^GFPCre (EPDC lineage Dlk1 overexpression) and Dlk1^fl/fl xαMHC^Cre/+Tam (EPDC environment Dlk1 overexpression) and their corresponding controls. MI was verified at 6 weeks by PET scanning (Figure S8‐S10), and (B) after 8 weeks by scar size measurements using serial Massons’ Trichrome stained sections from apex to base, which was used to (C) quantitate scar size in a 3D‐like manner by calculating the area under the curve of all steps. For statistical testing, we used two‐way ANOVA. Each MI series of transgenic and littermate control was performed by different operators and different blinded analysers, limiting comparisons between sets.

FIGURE 5

Dlk1^−/− EPDCs exhibit a reduced ability for TGFβ‐mediated EMT. (A) Typical cobblestone morphology of undifferentiated EPDCs obtained from neonatal Dlk1^+/+ and Dlk1^−/− hearts (n = 4) that (B) except for DLK1, display very similar marker expression and culture purity (CD45, CD31 and TroponinT) as judged by flow cytometry and furthermore, (C) genome wide profiling only identified a limited number of differentially expressed genes. (D and E) Relative quantitative RT‐PCR validated Itgb8 gene array data in in vitro cultured EPDCs and in Dlk1^+/+ and Dlk1^−/− hearts during development. For D, Gapdh, beta‐actin and B2m were used for normalisation. (F and G) Dlk1^+/+ and Dlk1^−/− EPDCs were stimulated with TGFβ or vehicle (control) to undergo fibroblast differentiation, which was verified by (F) a morphology change and (G) qRT‐PCR. Dlk1 and Itgb8 was as expected reduced in Dlk1^−/− EPDCs as compared with Dlk1^+/+ EPDCs, which was accompanied by lower expression of the cardiac fibroblast EMT markers aSMA and Procollagen I. (H and I) Itgb8 knock down by 75−90% in Dlk1^+/+ EPDCs was accompanied by a decrease in cardiac fibroblast differentiation (preserved epithelial morphology, low Procollagen I), but with no effect on Dlk1 itself. For (G and I) Procollagen I levels were normalised within each experiment to compensate the observed inter‐experiment variation and tested by non‐parametric Mann–Whitney. For all qRT‐PCR data (as exemplified in (D)), normalisation was performed against several stably expressed endogenous controls according to the qBase platform (see Material and Methods section). For all other statistical testing, we used two‐way ANOVA/Fisher's LSD (B), paired t‐test (D), two‐way ANOVA/Holm‐Sidak's test (E, G and I).

FIGURE 6

Dlk1 affects heart fibrosis through increased collagen I expression in the pericardium and in interstitial cardiac fibroblasts. (A) The pericardial sac was dissected from adult Dlk1^+/+ and Dlk1^−/− animals (n = 4–6) at day 7 after ± pericardial lesion and analysed by (B) qRT‐PCR for Dlk1, Itgb8, Procollagen I and Tcf21 expression. Expression was normalised against B2M and Gapdh (M:0.34 and CV:0.12) using qBase+ (C–E) Interstitial cardiac ventricle fibroblasts were isolated from adult mouse hearts, cultured and transfected with the extracellular cleavable Dlk1 gene (DLK1E‐pLHCX‐HA) or an empty vector (pLHCX‐HA). Dlk1 expressing cells were stimulated with ±TGFβ and treated with vehicle, isotype‐antibody (Ab) or anti‐DLK1‐Ab before qRT‐PCR for Procollagen I expression. For all qRT‐PCR data (as exemplified in (B)), normalisation was performed against several stably expressed endogenous controls according to the qBase platform (see Materials and Methods section). For statistical testing, we used two‐way ANOVA/Holm‐Sidak's test (B) and non‐parametric Kruskal–Wallis/two‐stage linear step‐up procedure of Benjamini, Krieger and Yekutieli (D and E).

FIGURE 7

Proposed mechanism of Dlk1 action in the heart during development and disease (MI). In heart development DLK1 (Soluble (DLK1^S) and membrane (DLK1^M)) is expressed in all mesothelial cells lining the pericardial space as well as in delaminating EPDCs in the subepicardial space, in developing vasculature, in OFT cardiomyocytes, and in neonatal (pre)fibroblasts. In adulthood, DLK1 is restricted to mesothelial cells (some are negative‐not shown) and the majority of cells in the fibrous pericardium, but also exist in its soluble form shedded into the pericardial fluid. During both development and disease (MI), DLK1 inhibits EPDC proliferation while mediating EMT of EPDCs including their maturation into fibroblasts and myofibroblasts, and in the end, this increase collagen I expression and scar formation after MI (encircled by a dotted line—‐). The underlying mechanism of DLK1 in these EMT scenarios (marked by a star) may be complex and vary but seems to include Itgb8/ TGFβ regulation, Tcf21 enhancement and eventually also regulation of the previously shown Dlk1 target Sox9 as well as other factors.