Protein kinase Cε-calcineurin cosignaling downstream of toll-like receptor 4 downregulates fibrosis and induces wound healing gene expression in cardiac myofibroblasts

Abstract

The pathways which regulate resolution of inflammation and contribute to positive remodeling of the myocardium following injury are poorly understood. Here we show that protein kinase C epsilon (PKCε) cooperates with the phosphatase calcineurin (CN) to potentiate induction of cardioprotective gene expression while suppressing expression of fibrosis markers. This was achieved by detailed analysis of the regulation of cyclooxygenase 2 (COX-2) expression as a marker gene and by using gene expression profiling to identify genes regulated by coexpression of CN-Aα/PKCε in adult rat cardiac myofibroblasts (ARVFs) on a larger scale. GeneChip analysis of CN-Aα/PKCε-coexpressing ARVFs showed that COX-2 provides a signature for wound healing and is associated with downregulation of fibrosis markers, including connective tissue growth factor (CTGF), fibronectin, and collagens Col1a1, Col3a1, Col6a3, Col11a1, Col12a1, and Col14a1, with concomitant upregulation of cardioprotection markers, including COX-2 itself, lipocalin 2 (LCN2), tissue inhibitor of metalloproteinase 1 (TIMP-1), interleukin-6 (IL-6), and inducible nitric oxide synthase (iNOS). In primary rat cardiomyocyte cultures Toll-like receptor 4 (TLR4) agonist- or PKCε/CN-dependent COX-2 induction occurred in coresident fibroblasts and was blocked by selective inhibition of CN or PKC α/ε or elimination of fibroblasts. Furthermore, ectopic expression of PKCε and CN in ARVFs showed that the effects on COX-2 expression are mediated by specific NFAT sites within the COX-2 promoter as confirmed by site-directed mutagenesis and chromatin immunoprecipitation (ChIP). Therefore, PKCε may negatively regulate adverse myocardial remodeling by cooperating with CN to downregulate fibrosis and induce transcription of cardioprotective wound healing genes, including COX-2.

FIG 1

(A) Wild-type (WT) or PKCε knockout (KO) mice underwent sham operation or coronary artery ligation and reperfusion 3 times for 5 min each (3×IR). Hearts were isolated 24 h later and processed for Western blotting for COX-2 or actin as a loading control. These were compared to neonatal rat ventricular cardiomyocytes (NRVM) subjected to simulated ischemia/reperfusion (sI/R). (B) COX-2 expression was normalized to actin following densitometry and represented quantitatively as mean ± standard error of the mean (SEM). *, P ≤ 0.05 versus sham by 1-way ANOVA with Newman-Keuls multiple-comparison test. (C) Hearts from 4-week-old nontransgenic (NTG) or CN-transgenic (TG) mice were isolated and processed for Western blotting for COX-2, CTGF, TLR4, or actin as a loading control. (D) Protein-of-interest (POI) expression was normalized to actin following densitometry and represented quantitatively as mean ± SEM. **, P ≤ 0.01 versus NTG by 1-way ANOVA with Newman-Keuls multiple-comparison test.

FIG 2

Calcineurin- and PKC-dependent COX-2 induction. Induction of COX-2 expression in response to upstream stimuli was studied in ARVFs. Cells were harvested 24 h after treatment, and COX-2 expression was determined by Western blotting unless otherwise stated. Blots were probed for actin as a loading control. (A) ARVFs were treated with angiotensin II (AII) (10 μM) and harvested at different time points from 30 min to 24 h. Expression of COX-2 was determined by Western blotting. COX-2 expression was normalized to actin following densitometry and represented quantitatively as mean ± SEM (n = 3). ***, P ≤ 0.0001 versus control/vehicle (by 1-way ANOVA with Newman-Keuls multiple-comparison test). (B) ARVFs were treated with LPS (2 μg ml⁻¹) and harvested at different time points from 5 min to 48 h. Expression of COX-2 was determined by Western blotting. COX-2 expression was normalized to actin following densitometry and represented quantitatively as mean ± SEM (n = 3). **, P ≤ 0.001; ***, P < 0.0001 versus control/vehicle (by 1-way ANOVA with Newman-Keuls multiple-comparison test). (C) ARVFs were treated with LPS (2 μg ml⁻¹), fibronectin type III repeats (III₁₂), or EDA (5 μg/ml). Cells were harvested 24 h, later and blots were probed for COX-2 or connective tissue growth factor (CTGF). (D) ARVFs were pretreated with the selective calcineurin antagonist cyclosporine (CsA) (5 μM) or the selective PKC antagonist bisindolylmaliemide I (GF109203X) (1 μM) 30 min prior to LPS treatment (10 μg/ml). The positive control (+ve) was LPS-treated fibroblasts. (E) ARVFs were pretreated with CsA (5 μM) or GF109203X (1 μM) 30 min prior to treatment with soluble (plasma) fibronectin fragments (sFNf) (5 μg/ml).

FIG 3

Calcineurin- and PKC-dependent COX-2 induction. (A) Mouse embryo fibroblasts (MEFs) from wild-type (+/+) or PKCε knockout (−/−) mice were either untreated (C) or treated with LPS (2 μg ml⁻¹). (B) AVRFs were treated with sI/R or urokinase-type plasminogen activator (uPA). AVRFs were grown directly on 6-well plates or on plates precoated with matrix (collagen, fibronectin, laminin, and FBS). (C) ARVFs were pretreated with CsA (5 μM) or GF109203X (1 μM) 30 min prior to treatment with angiotensin II (AII) (10 μM). Cells were harvested 4 h after treatment, and COX-2 expression was determined by Western blotting. The positive control (+ve) was LPS-treated fibroblasts. (D) ARVFs (as described above) were pretreated with the selective angiotensin type 2 receptor antagonist PD123319 (1 μM) or the selective angiotensin type I receptor antagonist ZD7155 (1 μM) 30 min prior to treatment with AII (10 μM). Cells were harvested 2 h after treatment, and COX-2 mRNA expression was determined by qRT-PCR. Results show COX-2 mRNA quantitation (ng RNA) normalized to GAPDH mRNA. Results are expressed as mean ± SEM (n = 3). ***, P ≤ 0.0001 versus control by 1-way ANOVA with Newman-Keuls multiple-comparison test. NS, not significant.

FIG 4

Calcineurin- and PKC-dependent COX-2 induction. (A) ARVFs were treated with LPS, AII, or AII plus LPS. The positive control (+ve) was LPS-treated fibroblasts. (B) ARVFs were treated with sFNf, AII, or Ang II plus sFNf. (C to E) ARVFs were treated for 30 min with inhibitors for PKC (GF109203X [GF]) (1 μM), calcineurin (FK506) (1.3 μM), and NF-κB (PS-1145 dihydrochloride [PS]) (10 μM) or vehicle (DMSO) followed by FN-EDA (5 μg/ml) or AII (10 mM), and cells were harvested 24 h later (EDA) or 4 h later (AII). Graphs, COX-2 expression was normalized to actin following densitometry and represented quantitatively as mean ± SEM. **, P ≤ 0.01; *, P < 0.05 versus control/vehicle (1-way ANOVA with Newman-Keuls multiple-comparison test).

FIG 5

Differential gene expression patterns in ARVFs expressing CN and PKCε. AVRFs were transfected with null vector (pSRα) or cotransfected with ΔCam-AI (CN) plus WT PKCε. Total RNA was isolated, and the Affymetrix rat GeneChip gene 1.0 ST array was used to identify genes differentially regulated in ARVFs expressing CN plus PKCε. Using a false-detection rate (FDR) of <0.3 and a fold change (FC) of ±1.5 as stringency parameters, 979 gene changes were observed. (A) The results are shown as a dendrogram (heat map) where rows correspond to genes with their expression represented as a red-blue color scale (high-low expression). As expected, ARVFs transfected with CN plus PKCε showed high expression of both calcineurin (ppp3ca) (FC, 20; P ≤ 0.00007) and PKCε (Prkce) (FC, 56.7; P ≤ 2.09 × 10⁻⁷) themselves. (B) Significant gene changes in addition to ppp3ca and Prkce. (C) Changes in collagen gene expression. (D and E) Target validation by qRT-PCR for the targets identified in panel B as being significantly up- or downregulated by coexpression of CN plus PKCε. Data are presented as mean ± SEM (n = 3). *, P ≤0.01; **, P ≤ 0.001; ***, P ≤ 0.0001 by 1-way ANOVA with Newman-Keuls multiple-comparison test.

FIG 5

Differential gene expression patterns in ARVFs expressing CN and PKCε. AVRFs were transfected with null vector (pSRα) or cotransfected with ΔCam-AI (CN) plus WT PKCε. Total RNA was isolated, and the Affymetrix rat GeneChip gene 1.0 ST array was used to identify genes differentially regulated in ARVFs expressing CN plus PKCε. Using a false-detection rate (FDR) of <0.3 and a fold change (FC) of ±1.5 as stringency parameters, 979 gene changes were observed. (A) The results are shown as a dendrogram (heat map) where rows correspond to genes with their expression represented as a red-blue color scale (high-low expression). As expected, ARVFs transfected with CN plus PKCε showed high expression of both calcineurin (ppp3ca) (FC, 20; P ≤ 0.00007) and PKCε (Prkce) (FC, 56.7; P ≤ 2.09 × 10⁻⁷) themselves. (B) Significant gene changes in addition to ppp3ca and Prkce. (C) Changes in collagen gene expression. (D and E) Target validation by qRT-PCR for the targets identified in panel B as being significantly up- or downregulated by coexpression of CN plus PKCε. Data are presented as mean ± SEM (n = 3). *, P ≤0.01; **, P ≤ 0.001; ***, P ≤ 0.0001 by 1-way ANOVA with Newman-Keuls multiple-comparison test.

FIG 6

Effect of coexpression of calcineurin and PKC isoforms on COX-2 expression in NRVMs. (A) Cell extracts from NRVMs infected with adenoviruses expressing ΔCam-AI (CN) or CN in combination with wild-type PKCα, PKCδ, or PKCε were analyzed by Western blotting and probed with antibodies for COX-2 or the respective PKC isotypes. Cells were cultured postinfection in the presence (+) or absence (−) of FBS. In each case blots were probed for actin as a loading control. (B) Cell extracts from NRVMs infected with adenoviruses expressing null vector (SR), ΔCam-AI (CN), or wild-type PKCα were analyzed by Western blotting and probed with antibodies for COX-2 or PKCα. Cells were cultured postinfection in the presence (+) or absence (−) of FBS. (C) Western blots of NRVMs expressing WT PKCδ or WT PKCε alone in the presence (+) or absence (−) of FBS probed for COX-2 or the respective PKCs. Graphs show quantitation of COX-2 normalized to actin. Data are expressed as mean ± SEM (n = 3). *, P ≤ 0.01 versus C; **, P ≤ 0.01 versus C plus FBS; #, P ≤ 0.05 versus CN; §, P ≤ 0.01 versus CN; $, P ≤ 0.001 versus CN plus PKCα; ¥, P ≤ 0.01 versus CN plus FBS; ¥¥, P ≤ 0.05 versus CN plus FBS; §§, P ≤ 0.01 versus CN plus PKCε (1-way ANOVA with Newman-Keuls multiple comparison test). (D) confocal immunofluorescence images of NRVM cultures infected with control Ad-SRα probed for the myofibroblast marker vimentin (green), the cardiomyocyte marker α-actinin (red), or nuclei (ToPro) (blue). (E) Confocal immunofluorescence images of NRVM cultures coinfected with Ad-ΔCam-AI plus Ad-wtPKCα probed for COX-2 (green), the myofibroblast marker vimentin (red), or nuclei (ToPro) (blue). (F) Confocal immunofluorescence images of NRVM cultures coinfected with Ad-ΔCam-AI plus Ad-wtPKCε probed for COX-2 (green), the myofibroblast marker vimentin (red), or nuclei (ToPro) (blue). In each case the bottom right panel shows the overlaid (merged) image. (G) Confocal immunofluorescence images of NRVM cultures infected with Ad-ΔCam-AI plus Ad-wtPKCε under control conditions (left) or treated with 5′-cytosine arabinoside C (AraC) (right) to eliminate fibroblasts and probed for COX-2 (green), α-actinin (red), or nuclei (ToPro) (blue). (H and I) COX-2 levels in cardiomyocytes (H) and nonmyocytes (I) were expressed semiquantitatively as mean ± SEM of fluorescence intensity (bar graphs) (n = 3 fields/slide of 75 to 100 cells per field). *, P ≤ 0.01 versus SRa; ***, P ≤ 0.001 versus SRa; ς, P ≤ 0.01 versus CN (1-way ANOVA with Newman-Keuls multiple-comparison test).

FIG 6

Effect of coexpression of calcineurin and PKC isoforms on COX-2 expression in NRVMs. (A) Cell extracts from NRVMs infected with adenoviruses expressing ΔCam-AI (CN) or CN in combination with wild-type PKCα, PKCδ, or PKCε were analyzed by Western blotting and probed with antibodies for COX-2 or the respective PKC isotypes. Cells were cultured postinfection in the presence (+) or absence (−) of FBS. In each case blots were probed for actin as a loading control. (B) Cell extracts from NRVMs infected with adenoviruses expressing null vector (SR), ΔCam-AI (CN), or wild-type PKCα were analyzed by Western blotting and probed with antibodies for COX-2 or PKCα. Cells were cultured postinfection in the presence (+) or absence (−) of FBS. (C) Western blots of NRVMs expressing WT PKCδ or WT PKCε alone in the presence (+) or absence (−) of FBS probed for COX-2 or the respective PKCs. Graphs show quantitation of COX-2 normalized to actin. Data are expressed as mean ± SEM (n = 3). *, P ≤ 0.01 versus C; **, P ≤ 0.01 versus C plus FBS; #, P ≤ 0.05 versus CN; §, P ≤ 0.01 versus CN; $, P ≤ 0.001 versus CN plus PKCα; ¥, P ≤ 0.01 versus CN plus FBS; ¥¥, P ≤ 0.05 versus CN plus FBS; §§, P ≤ 0.01 versus CN plus PKCε (1-way ANOVA with Newman-Keuls multiple comparison test). (D) confocal immunofluorescence images of NRVM cultures infected with control Ad-SRα probed for the myofibroblast marker vimentin (green), the cardiomyocyte marker α-actinin (red), or nuclei (ToPro) (blue). (E) Confocal immunofluorescence images of NRVM cultures coinfected with Ad-ΔCam-AI plus Ad-wtPKCα probed for COX-2 (green), the myofibroblast marker vimentin (red), or nuclei (ToPro) (blue). (F) Confocal immunofluorescence images of NRVM cultures coinfected with Ad-ΔCam-AI plus Ad-wtPKCε probed for COX-2 (green), the myofibroblast marker vimentin (red), or nuclei (ToPro) (blue). In each case the bottom right panel shows the overlaid (merged) image. (G) Confocal immunofluorescence images of NRVM cultures infected with Ad-ΔCam-AI plus Ad-wtPKCε under control conditions (left) or treated with 5′-cytosine arabinoside C (AraC) (right) to eliminate fibroblasts and probed for COX-2 (green), α-actinin (red), or nuclei (ToPro) (blue). (H and I) COX-2 levels in cardiomyocytes (H) and nonmyocytes (I) were expressed semiquantitatively as mean ± SEM of fluorescence intensity (bar graphs) (n = 3 fields/slide of 75 to 100 cells per field). *, P ≤ 0.01 versus SRa; ***, P ≤ 0.001 versus SRa; ς, P ≤ 0.01 versus CN (1-way ANOVA with Newman-Keuls multiple-comparison test).

FIG 7

Effect of coexpression of calcineurin and PKCα or PKCε on prostanoid production. (A) PGE₂ and (B) 6-keto-PGF_1α were measured in supernatants of NRVM cultures 24 h following infection with Ad-SRα, Ad-CN, Ad-wtPKCα, Ad-wtPKCε, Ad-CN plus Ad-wtPKCα, or Ad-CN plus Ad-wtPKCε either without (−) or with (+) treatment with the selective COX-2 inhibitor NS398. Results are expressed as mean ± SEM (pg/ml) (n = 3 independent experiments). *, P ≤ 0.001; ***, P ≤ 0.0001 (versus SRα) (1-way ANOVA with Newman-Keuls multiple-comparison test). (C) qRT-PCR results showing quantitation of COX-2 mRNA following the different treatments. Results are expressed as mean ± SEM (n = 3 independent experiments) of C_T (2^−ΔΔCT). *, P ≤ 0.05 versus Ad-SRα. (D) ARVFs were pretreated with NS398 for 30 min prior to treatment with LPS, FN-III₁₂, or FN-EDA and harvested 24 h later. Western blots were probed for COX-2 and CTGF. Actin was used as a loading control.

FIG 8

Cooperative induction of COX-2 by CN and PKCε in ARVFs. AVRFs were transfected with null vector (pSRα), pCAGGS, pΔCam-AI, pwtPKCε, or pΔCam-AI plus pwtPKCε using the P6/LID method. Following transfection, cells were cultured in the presence or absence of 1% FBS. Cells were harvested at 48 h after transfection. (A) Western blots were probed for COX-2 and actin as a loading control. (B) COX-2 expression was quantified by densitometry and expressed as the COX-2/actin ratio (arbitrary units). Results are expressed as mean + SEM (n = 4 independent experiments each in triplicate). *, P ≤ 0.05; ***, P ≤ 0.0001 (versus the respective null vector control) (1-way ANOVA with Newman-Keuls multiple-comparison test). (C) COX-2 mRNA levels were quantified in cell extracts from ARVFs transfected with pSRα, pΔCam-AI, pwtPKCε, or pΔCam-AI plus pwtPKCε. COX-2 mRNA levels (ng) were normalized to GAPDH. Results are expressed as mean SEM (n = 3 independent experiments each in triplicate). *, P ≤ 0.01; **, P ≤ 0.001; ***, P ≤ 0.0005 versus the null vector control (1-way ANOVA with Newman-Keuls multiple-comparison test).

FIG 9

Chromatin immunoprecipitation (ChIP) assay of ARVFs expressing calcineurin, PKCε, or calcineurin and PKCε. (A) Consensus transcription factor binding sequences in the rat COX-2 promoter as identified using MatInspector analysis (GenBank accession number L11611). A proximal composite NFAT/AP-1 site is shown between −79 and −57. (B) Schematic representation of the rat and human COX-2 proximal 2-kb promoter regions. (C) ARVFs were infected with Ad-SRα (null vector), Ad-ΔCam-AI, Ad-wtPKCε, or Ad-ΔCam-AI plus Ad-wtPKCε. Cells were harvested 48 h later for ChIP analysis. DNA was immunoprecipitated with antibodies to NFATc1 (test) or GAPDH (control). (D) Quantitation of NFAT binding relative to the control and normalized to input (n=3) for regions P1 and P5.

FIG 10

Analysis of cooperative COX-2 promoter activation by CN and PKCε. (A) HEK293 cells were cotransfected with the rat COX-2 promoter (pGL3.rCOX2.luc) together with a reference plasmid (Renilla.TK.luc), null vectors (pSRα; pCAGGS), WT PKCε, ΔCam-AI (CN), or ΔCam-AI plus wtPKCε (C+P). At 30 h following transfection cells were harvested for dual-luciferase assay. pGL3.rCOX2.luc activity was normalized to constitutive Renilla luciferase activity as a control for transfection efficiency. The fold activation is expressed relative to control (empty) plasmid (set to 1). Data are expressed as mean ± SEM (n = 3 independent experiments, each in triplicate). **, P ≤ 0.001; ***, P ≤ 0.0001 (1-way ANOVA with Newman-Keuls multiple-comparison test). (B) Comparison of activation of the rat and human proximal COX-2 promoters following cotransfection with ΔCam-AI (CN), WT PKCε, or ΔCam-AI (CN) plus WT PKCε. The fold activation is expressed relative to control (empty) plasmid (set to 1). Data are expressed as mean ± SEM (n = 3 independent experiments, each in triplicate). *, P ≤ 0.01; ***, P ≤ 0.0001 (1-way ANOVA with Newman-Keuls multiple-comparison test). (C) Schematic representation of the SmaI deletion of the rat COX-2 promoter construct. (D) Activation of the COX-2 promoter SmaI deletion construct by ΔCam-AI (CN), WT PKCε, or ΔCam-AI (CN) plus WT PKCε (right) compared to activation of the intact full-length promoter (FLP) construct (left). Data are expressed as mean ± SEM (n = 3 independent experiments, each in triplicate). *, P ≤ 0.01; **, P ≤ 0.001; ***, P ≤ 0.0001 (1-way ANOVA with Newman-Keuls multiple-comparison test). (E and F) Proximal NFAT sites located at positions −79 and −92 (Fig. 9A and C and corresponding to 1711 and 1698 in panel C) were mutated in the intact (FLP) promoter (FLP-m) (E) or the SmaI deletion construct (SmaI-m) (F). These were transfected with the respective plasmids and activation assessed relative to that with empty (null) plasmid (pSRα) set at 1. Data are expressed as mean ± SEM (n = 3 independent experiments, each in triplicate). *, P ≤ 0.01; **, P ≤ 0.001; ***, P ≤ 0.0001 (1-way ANOVA with Newman-Keuls multiple-comparison test).

FIG 10

Analysis of cooperative COX-2 promoter activation by CN and PKCε. (A) HEK293 cells were cotransfected with the rat COX-2 promoter (pGL3.rCOX2.luc) together with a reference plasmid (Renilla.TK.luc), null vectors (pSRα; pCAGGS), WT PKCε, ΔCam-AI (CN), or ΔCam-AI plus wtPKCε (C+P). At 30 h following transfection cells were harvested for dual-luciferase assay. pGL3.rCOX2.luc activity was normalized to constitutive Renilla luciferase activity as a control for transfection efficiency. The fold activation is expressed relative to control (empty) plasmid (set to 1). Data are expressed as mean ± SEM (n = 3 independent experiments, each in triplicate). **, P ≤ 0.001; ***, P ≤ 0.0001 (1-way ANOVA with Newman-Keuls multiple-comparison test). (B) Comparison of activation of the rat and human proximal COX-2 promoters following cotransfection with ΔCam-AI (CN), WT PKCε, or ΔCam-AI (CN) plus WT PKCε. The fold activation is expressed relative to control (empty) plasmid (set to 1). Data are expressed as mean ± SEM (n = 3 independent experiments, each in triplicate). *, P ≤ 0.01; ***, P ≤ 0.0001 (1-way ANOVA with Newman-Keuls multiple-comparison test). (C) Schematic representation of the SmaI deletion of the rat COX-2 promoter construct. (D) Activation of the COX-2 promoter SmaI deletion construct by ΔCam-AI (CN), WT PKCε, or ΔCam-AI (CN) plus WT PKCε (right) compared to activation of the intact full-length promoter (FLP) construct (left). Data are expressed as mean ± SEM (n = 3 independent experiments, each in triplicate). *, P ≤ 0.01; **, P ≤ 0.001; ***, P ≤ 0.0001 (1-way ANOVA with Newman-Keuls multiple-comparison test). (E and F) Proximal NFAT sites located at positions −79 and −92 (Fig. 9A and C and corresponding to 1711 and 1698 in panel C) were mutated in the intact (FLP) promoter (FLP-m) (E) or the SmaI deletion construct (SmaI-m) (F). These were transfected with the respective plasmids and activation assessed relative to that with empty (null) plasmid (pSRα) set at 1. Data are expressed as mean ± SEM (n = 3 independent experiments, each in triplicate). *, P ≤ 0.01; **, P ≤ 0.001; ***, P ≤ 0.0001 (1-way ANOVA with Newman-Keuls multiple-comparison test).