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.

INTRODUCTION

Myocardial infarction (MI) leads to the loss of cardiomyocytes due to necrosis. The subsequent healing process is at first dominated by an acute inflammatory response, followed by a repair phase during which inflammatory cell and myofibroblast proliferation and migration occur (1). In the first week after MI, cardiomyocytes in the infarcted region become almost completely replaced by fibroblasts (2). During the initial inflammatory phase, there is release of inflammatory cytokines and other mediators and degradation of extracellular matrix (ECM). Subsequent scar formation results from fibroblast proliferation, release of cytokines which promote fibrosis, and increased matrix synthesis (3). Three-dimensional scar remodeling (i.e., collagen cross-linking) then leads to the formation of a mature and stable scar. This process is essential for normal healing. However, during adverse remodeling, chronic inflammation can lead to continued collagen turnover and destabilization of the scar, leading to dilatation and heart failure (4).

Many of the ECM changes which occur during the remodeling process occur as a result of differentiation of fibroblasts into myofibroblasts, fibroblast hyperplasia, and collagen synthesis and deposition leading to fibrosis. Myofibroblast differentiation is typified by α-smooth muscle actin (αSMA) expression and increased migratory, proliferative, and secretory properties and occurs in response to inflammatory cytokines which are increased in the remodeling heart. These include tumor necrosis factor alpha (TNF-α), interleukin-1α (IL-1α), interleukin-6 (IL-6), transforming growth factor β (TGF-β), vasoactive peptide hormones such as angiotensin II (AngII) and endothelin-1 (ET-1), and neurohormones such as noradrenaline (5).

Cardiac injury also results in activation of the innate immune system, which plays an important role in wound healing and remodeling. Myocardial injury leads to “sterile” inflammation via activation of pattern recognition receptors (PRRs) by host-derived tissue components such as intracellular or ECM-derived proteins or fragments, referred to as damage-associated molecular patterns (DAMPs). Proposed DAMPs include Hsp60, Hsp70, GP96, fibrinogen, fibronectin (FN) fragments (including the extra domain A [FN-EDA] and certain other type III repeats), surfactant protein A, Tenascin C (6), HMGB1, heparan sulfate, and hyaluronan (7). Like bacterial lipopolysaccharide (LPS), DAMPs activate PRRs of the Toll-like receptor (TLR) family, particularly the LPS receptors TLR2 and -4. TLR4 is activated by FN-EDA (8) and a 70-amino-acid (aa) C-terminal fragment of the first type III repeat of fibronectin (FN-III_1-c) (9), which are extracellularly (i.e., ECM-) derived proteins rather than intracellular DAMPs. In fact TLR4 activation is strongly inhibited by intact ECM, and this inhibition is relieved when the matrix is degraded (10). Thus, the breakdown of ECM by proteases during the course of injury or infection is a critical step in the initiation of the immune response by TLRs. TLR2 and TLR4 have been demonstrated to be present in cardiac myocytes and fibroblasts (11). The activation of TLRs typically leads to the transcriptional activation of genes encoding proinflammatory cytokines and other mediators by Rel family transcription factors (TFs) such as nuclear factor κB (NF-κB) and nuclear factor of activated T cells (NFAT).

The role of TLR4 in cardiac remodeling following injury is highly complex. For instance, fibroblasts are activated by FN fragments such as EDA produced in the heart after MI (12). However, EDA^−/−, TLR4^−/−, and MyD88^−/− mice show reduced inflammatory cell infiltrate, fibrosis, and injury (12–14). Furthermore, the TLR4-selective antagonist Eritoran (E5564) reduces hypertrophy and inflammation in a transaortic constriction (TAC) model (15). These results suggest that the TLR4 signaling pathway may promote heart failure and postinjury remodeling. Contrary to this, TLR4 may also have positive effects on remodeling after MI. Myocardial function is preserved in TLR4^−/− mice (16), and TLR4-dependent induction of NOS2 (iNOS) has been proposed to be important for mitochondrial biogenesis (17) and to modulate survival via positive effects on remodeling after myocardial infarction (18). Loss of endothelium-dependent relaxation and increased left ventricular dilatation also occurs in TLR4^−/− hearts (19). These conflicting data suggest that the negative role of TLR4-dependent pathways in ischemia/reperfusion (I/R) injury appears to be due to monocyte/neutrophil-dependent effects on inflammation, cytokine production, and remodeling rather than cardiomyocytes or fibroblasts per se.

The effects of TLR4 activation are mediated by activation of downstream signaling pathways including protein kinase C epsilon (PKCε). The PKC family of phosphotransferases comprise some 12 isotypes which fall into classical (Ca²⁺- and diacylgylerol [DAG]-dependent), novel (Ca²⁺-independent and DAG-dependent), and atypical (Ca²⁺- and DAG-independent) subgroups (20). Of the novel isotypes (PKCδ, PKCε, and PKCθ), PKCε plays an important role in cardioprotection, particularly in the mechanism of delayed ischemic preconditioning (dIPC), which is in itself cyclooxygenase 2 (COX-2) dependent (21, 22). Importantly, PKCε couples TLR4 to NF-κB activation in macrophages via MyD88 (23) and therefore mediates distinct effects on gene regulation and hence cell fate. More recently, PKCε has also been shown to play an anti-inflammatory and antiapoptotic role in vascular endothelial cells which is Erk1/2 and NF-κB dependent (24).

COX-2 is essential for cardiovascular homeostasis and cardioprotection; it is a mediator of delayed ischemic preconditioning (dIPC) (22, 25) and inhibits fibrosis in kidney and lung following injury (25). Previous work has suggested that COX-2 expression is regulated by a cooperative interaction between the Ca²⁺/calmodulin (CaM)-dependent protein phosphatase calcineurin (CN) (ppp3c) and protein kinase C (PKC) in T cells (26) (27), and it thus serves as a useful marker of gene activation by CN/PKC cooperation.

In this study, we have identified an important bimodal cooperative signaling pathway in cardiac fibroblasts in which calcineurin (CN) and PKCε combine to suppress fibrotic gene expression and synergistically induce cardioprotective and antifibrotic gene expression for which COX-2 expression is a signature marker and is associated with synergistic prostaglandin E2 (PGE₂)/PGI₂ synthesis. This pathway is activated by ECM-derived DAMPs, including FN fragments downstream of TLR4, and is potentiated by the proinflammatory and profibrotic peptide hormone AngII via the AT1 receptor. Furthermore, dissection of the proximal COX-2 gene promoter identifies critical NFAT binding sites which are modulated by CN/PKCε cosignaling. Indeed PKCε appears to convert a profibrotic AngII/CN-alone signal into an antifibrotic signal characterized by downregulation of connective tissue growth factor (CTGF) and several collagen isoforms and which may be critical for positive remodeling/wound healing.

MATERIALS AND METHODS

Materials.

Salmonella enterica serovar Typhimurium LPS (L2262), fibronectin from human plasma (F0895), and human angiotensin II (A9525) were from Sigma, United Kingdom. Goat polyclonal antibodies to COX-2 (C-20: sc-1745), CTGF (L-20: sc-14939), and actin (I-19: sc-1616), rabbit polyclonal GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (FL-335: sc-25778), and NFATc1 (H-110: sc-13033) were from Santa Cruz Biotech; mouse monoclonal antibodies to PKCα, PKCδ, and PKCε were from BD-Transduction Labs, United Kingdom. Mouse monoclonal antibody to sarcomeric α-actinin was from Sigma, United Kingdom. Goat polyclonal antibody to vimentin was from Chemicon International, United Kingdom. Rat polyclonal antibody to CD44 was from BD Biosciences, United Kingdom. Alexa Fluor 568–goat anti-mouse antibody, To-Pro, and DAPI (4′,6′-diamidino-2-phenylindole) were from Molecular Probes, United Kingdom. General-purpose reagents were from Sigma-Aldrich, United Kingdom, and Merck-BDH, United Kingdom, unless otherwise stated. Plasmid and RNA preparation reagents were from Qiagen, United Kingdom. Collagenase was from Worthington Biochemicals. Pancreatin was from Gibco-BRL. Gelatin solution was from Sigma, United Kingdom. Lipofectin reagent was from Invitrogen, United Kingdom. Bisindolyl maleimide I (GF109203X), cyclosporine (CsA), PD123319, and ZD7155 were from Calbiochem, United Kingdom. General cell culture reagents, Dulbecco modified Eagle medium (DMEM), and fetal bovine serum (FBS) were from PAA Laboratories, United Kingdom, unless otherwise stated. Integrin-targeting peptide (peptide 6: [K]₁₆-GACRRETAWACG) was a gift from Stephen Hart, Institute of Child Health, United Kingdom, and has been described previously (28). Soluble fibronectin fragments (sFNf) were prepared by treating human plasma fibronectin (Sigma, United Kingdom) with urokinase-type plasminogen activator (uPA) (Sigma, United Kingdom) as previously described (29). FN-EDA and FN-III₁₂ were kind gifts from Hiroko Ushio, Juntendo University School of Medicine, Tokyo, Japan, and have been described previously (8). ΔCam-AI plasmid was a kind gift from Stephen O'Keefe, Merck, NJ, and has been described previously (30). Calcineurin Aα transgenic mice and ΔCam-AI adenovirus were kind gifts from Jeffery Molkentin, Cincinnati, OH, and have been described previously (31). PKCε null mice were a kind gift from Peter Parker, CRUK, London, United Kingdom, and have been described previously (32). Adenoviruses expressing wild-type (WT) PKCα and PKCδ were a kind gift from Chris Proud, Dundee, United Kingdom. Recombinant adenovirus expressing the full-length rabbit PKCε cDNA was kindly provided by Jody Martin (Loyola, Chicago, IL) and generated as previously described (33). The human COX-2 promoter-luciferase reporter plasmid was a kind gift from Miguel Iñiguez, Madrid, Spain, and has been described previously (26).

Isolated NRVM culture.

All animal experiments were performed in accordance with European Commission and United Kingdom Home Office guidelines. Neonatal rat ventricular cardiomyocytes (NRVMs) were isolated from 1- to 2-day-old Sprague-Dawley rats as previously described (34, 35) by collagenase-pancreatin digestion. Cells were plated at a density of approximately 2 × 10⁶ cells per well on gelatin-coated 6-well plates (Nunc) in medium containing DMEM-M199 (4:1) supplemented with 5% fetal calf serum, 10% horse serum, and 1% penicillin-streptomycin (Gibco/BRL) for 24 h and then transferred to maintenance medium (serum-free DMEM-M199 plus antibiotics). Cells were maintained for 2 days in maintenance medium (3 days after plating) prior to treatment. Cells were treated with LPS, sFNf, or angiotensin II for the indicated times (generally 4 or 24 h). For inhibitor experiments, cells were pretreated with 5 μM cyclosporine, 2 μM Bis I (GF109203X), 2 μM Bis V (inactive analogue), or vehicle (0.01% dimethyl sulfoxide [DMSO]) for 30 min prior to LPS or sFNf application.

Isolated ARVF culture.

Adult rat ventricular fibroblasts (ARVFs) were isolated during the isolation of adult rat ventricular myocytes by collagenase digestion of LV tissue. After allowing the filtered cell suspension to settle for 10 min, the supernatant was collected in a clean tube. The supernatant rich in fibroblasts was centrifuged for 6 min at 150 × g. The pellet was then resuspended in 25 ml full growth medium (FGM) containing 10% fetal bovine serum (FBS), and the process was repeated two further times. The pellet was finally resuspended in 75 ml of FGM and plated on six 6-well polystyrene plates. ARVFs were cultured for approximately 1 week in FGM. The medium was replaced 2 days after plating the cells to ensure that necrotic cells and cells which had not adhered to the plastic culture surface were removed. After 1 week, the medium was replaced with serum-free medium (SFM) prior to treatments. All treatments were done in SFM, and the cells remained in SFM until processed for further analysis.

Cloning of the 2-kb rat COX-2 promoter.

Whole adult rat heart was washed in saline and snap-frozen in liquid N₂. The heart was homogenized in 1.5 ml TRIzol. The homogenate was precipitated with chloroform and centrifuged at 13,000 rpm for 15 min at 4°C. The upper aqueous phase was removed and stored at −80°C, and 300 μl ethanol was added to the lower phase, mixed, and left at room temperature (RT) for 3 min. The sample was then centrifuged at 2,000 rpm for 5 min at 4°C. The supernatant was discarded and the pellet resuspended in 0.1 M Na citrate–10% (vol/vol) ethanol and left at RT for 30 min. The sample was centrifuged at 1,500 rpm for 5 min at 4°C. The last two steps were repeated. The pellet was resuspended in 1.5 ml 75% ethanol, left at RT for 20 min, and then centrifuged at 2,000 rpm for 5 min at 4°C. The pellet was air dried and then resuspended in 500 μl double-distilled water (ddH₂O) and quantified using a NanoDrop ND1000 spectrophotometer and stored at −20°C. A region 2 kb upstream of the transcription start site (TSS) of the rat COX-2 gene was determined using the Ensembl Genome Browser and primers designed manually. Three forward primers and one reverse primer were used: wild type (forward), 5′-976-gaggtctaGAGAATCTGCCTAAGCTCCCTCTTCTccctcagt-1051-3′; NheI (forward), 5′-976-gaggtctaGAGATCTGCCTAgctagcTCTTCTCCCTCAGT-1051-3′; MluI (forward), 5′-301-cacAGCAGGCACTTacgctgtGCCCAAA-376-3′; and XhoI (reverse), 5′-2935-GACACAGAGAGctcgagTTCCTTC-285-3′ (underlined lowercase text indicates the introduced restriction sites). The 2-kb region was cloned by PCR in a reaction mixture containing 4 μl deoxynucleoside triphosphates (dNTPs) (400 μM), 10 μl 5× buffer, 5 μl 25 mM MgCl₂, 1 μl forward primer (100 ng/μl), 1 μl reverse primer (100 ng/μl), and 0.5 μl Pfu Taq polymerase. PCR conditions were as follows: initial denaturation at 94°C for 5 min followed by 40 cycles of denaturation at 94°C for 45 s, annealing at 55 to 65°C for 45 s, and extension at 72°C for 45 s; final extension at 72°C; and holding at 4°C. The fragment was purified from a low-melting-point (LMP) agarose gel using the QIAquick gel extraction kit (Qiagen, United Kingdom) and subcloned into the pJET1.2/blunt vector by blunt-end ligation. pJET.COX2 was then linearized with EcoRV, and the insert was confirmed by diagnostic restriction digests using BglII and XhoI and sequencing. The 2-kb insert was excised using BglII and gel purified as described above. pGL3.luc was also linearized with BglII and gel purified and then the insert ligated into the BglII site, generating the pGL3.COX2.luc construct.

Transfections and reporter assays.

pSRα, ΔCam-AI, pCAGGS, PKCα, PKCε, pGL3.COX2.luc, and Renilla TK plasmids were transfected into NRVMs or ARVFs on 6-well plates using a Lipofectin-integrin targeting peptide-DNA (LID) protocol as previously described (28, 36). LID complexes were added dropwise to the cells in a total volume of 800 μl and left overnight, followed by 48 h in serum-free medium as previously described (28, 36). Luciferase reporter assays were performed after harvesting the cells in 200 μl passive lysis buffer (Promega, United Kingdom) and storing at −70°C. Assays were performed using the dual-luciferase reporter assay system (Promega, United Kingdom), and measurements were with a TD20/20 tube luminometer (Promega, United Kingdom). Transfections were performed in triplicate and data pooled from three independent experiments. A cotransfected Renilla luciferase-TK plasmid was used as a control for transfection efficiency. pGL3.COX2.luc activity was normalized to Renilla luciferase activity and expressed as a ratio of relative light units (RLU).

Adenovirus preparation and infection.

E1a-deleted, replication-deficient adenoviruses were amplified by large-scale multiplication in HEK293 cells cultured in low-serum medium (DMEM supplemented with 2% FBS). After 2 to 3 days, when a cytopathic effect was detected, cells were recovered by centrifugation at 400 × g for 5 min and resuspended in phosphate-buffered saline (PBS). After three cycles of freeze-thaw lysis, the lysate was centrifuged at 800 × g for 20 min to remove cellular debris. The supernatant was transferred onto a 1.25-g/ml (4 ml) to 1.4-g/ml (4 ml) cesium chloride step gradient and centrifuged at 90,000 × g using an SW41Ti swing-out rotor for 2 h at 20°C. The viral band was removed using a 21-gauge needle and the adenovirus dialyzed using dialysis cassettes against 1,000 ml dialysis buffer (400 ml glycerol, 4 ml of 1 M MgCl₂, and 40 ml of 1 M Tris [pH 7.4] in 4 liters of ddH₂O) for 24 h with 3 to 4 dialysis medium changes. The purified viruses were stored in aliquots at 70°C. Titers of the purified adenoviruses were determined by serial dilution on 293s cells. After 8 days of incubation, the number of wells containing plaques was counted and the results analyzed as previously described (37). The adenovirus stocks were also tested for replication-competent adenoviruses by infection of nonpermissive cells (3T3 cells) in an assay similar to titration (37). Adenovirus infection of the cells was performed at different multiplicities of infection (MOI) in 2 ml (6-well plates) of serum-free maintenance medium with SR (null virus) (MOI of 312 /μl), AdCnA (56/μl), AdPKCα (56/μl), or AdPKCε (56/μl).

sI/R.

Simulated ischemia/reperfusion (sI/R) was carried out using a modified Krebs buffer/anaerobic environment protocol as described previously (35, 38), which is a modification of the method described by Esumi et al. (39). Briefly, cells in 6-well tissue culture plates were exposed to 1.0 ml per well of sI buffer, consisting of modified Krebs-Henseleit buffer containing 137 mM NaCl, 12 mM KCl, 0.49 mM MgCl₂, 1.8 mM CaCl₂, 4 mM HEPES, 10 mM 2-deoxyglucose, and 20 mM Na lactate (pH 6.8). Cells were then exposed to a hypoxic environment by placing the plates into BBL GasPak anaerobic pouches (Becton, Dickinson, United Kingdom) for 1 h. Simulated reperfusion was achieved by removing the plates from sI and replacing the buffer with 2.0 ml of maintenance medium. Cells were thus recovered for 24 h prior to harvesting for Western blot analysis.

Sample preparation, SDS-PAGE, and Western blotting.

Samples were prepared from heart tissue by homogenization into 10-ml/g homogenization buffer (20 mM Tris HCl [pH 6.8], 5 mM NaF, 1 mM Na₃VO₄, and 1 tablet protease inhibitor cocktail [Roche] in 50 ml). One hundred microliters of homogenate was then added to an equal volume of 2× sample buffer (20% [vol/vol] glycerol, 6% [wt/vol] sodium dodecyl sulfate [SDS], 0.12 M Tris HCl [pH 6.8], 0.2% [wt/vol] bromophenol blue, 10% [vol/vol] β-mercaptoethanol) and heated to 100°C for 5 min. Cells were harvested directly from 6-well plates by lysis in 200 μl 2× sample buffer (as described above). Samples (15 μl) were run on 10% SDS-polyacrylamide gels using a Mini-Protean II apparatus (Bio-Rad, United Kingdom) and electroblotted overnight at 30 mA onto polyvinylidene difluoride (PVDF) membranes (Hybond P; Amersham, United Kingdom). Membranes were exposed to antibodies as follows. Membranes were blocked in phosphate-buffered saline (PBS) containing 0.05% (vol/vol) Tween20 and 2% (wt/vol) powdered milk (Marvel, United Kingdom) for 1 h at RT and then rinsed briefly in PBS containing 0.05% (vol/vol) Tween 20 and 0.1% (wt/vol) powdered milk (Marvel, United Kingdom) (PBSTM). Membranes were then incubated in primary antibody at a 1:1,000 dilution in PBSTM for 3 h at RT and then washed three times for 5 min each in PBSTM. Membranes were then incubated with the appropriate horseradish peroxidase (HRP)-conjugated anti-species secondary antibodies (Dako, United Kingdom) for 1 h at RT and then washed in PBSTM as described above. Membranes were then exposed to enhanced chemiluminescence (ECL) detection reagent (Amersham, United Kingdom) and exposed to ECL Hyperfilm (Amersham, United Kingdom). Films were scanned using a Perfection 1240U desktop scanner (Epson) and bands quantified (integrated optical density) using GelPro Analyzer6.3 software. Ratios for the protein of interest (POI) were expressed relative to actin in the same sample as a loading control.

Preparation of fibronectin fragments.

Soluble fibronectin (FN) fragments (sFNf) were prepared by digestion of human plasma fibronectin (Sigma) with urokinase-type plasminogen activator (uPA) essentially as described previously (29). FN-EDA and FN-III₁₂ with C-terminal 6× His tags were expressed in Escherichia coli BL21(DE3) as described previously (8) and captured using Ni-nitrilotriacetic acid (NTA) His-Bind resin (Merck, Germany). Proteins were eluted in 0.1 to 0.2 M imidazole and further purified using ion-exchange chromatography. EDA-His was applied to a HiTrap Q FF column (GE Healthcare, United Kingdom), and pure protein was eluted using 0.3 to 0.5 M NaCl. FN III₁₂ was loaded on to a HiTrap SP FF column (GE Healthcare, United Kingdom) and eluted using 0.05 to 0.1 M NaCl. Proteins were dialyzed against 20 mM sodium phosphate (pH 7.5) and stored at −20°C.

Immunocytochemistry and confocal fluorescence microscopy.

NRVMs or ARVFs were plated on plastic coverslips placed into 6-well plates. Following treatment, samples were fixed in 4% paraformaldehyde for 10 min at room temperature (RT). Coverslips were aspirated and left overnight in PBS at 4°C, washed three times for 5 min each in PBS at RT, and then blocked for 1 h in PBS–2% BSA–2% horse serum–0.1% Triton X-100 at RT. Primary antibody was added at a 1:800 dilution in blocking buffer and incubated overnight at 4°C. Coverslips were washed three times for 5 min each in PBS–0.1% Triton X-100 (PBST), and fluorophore-conjugated secondary antibodies were added at a 1:200 dilution in blocking buffer and left for 2 h at RT. Coverslips were washed three times for 5 min each in PBST, aspirated, and removed from the wells, mounting medium added, and the coverslips were inverted onto a glass slide. Coverslips were sealed with clear nail varnish and kept in the dark until analysis. Immunofluorescence confocal microscopy was performed using a Zeiss LSM 510 microscope. Where appropriate, the DNA stain ToPro was added to mounting medium for visualization of nuclei.

Prostanoid assays.

Cell culture supernatant was collected, transferred directly to a 1.5-ml tube, and centrifuged for 1 min at 13,000 rpm. The supernatants were then transferred to clean 1.5-ml tubes and stored at −20°C until further use. The PGE₂ concentration was measured using an Assay Designs Correlate-EIA PGE₂ enzyme immunoassay kit according to the manufacturers protocol. Samples were read on a microplate reader at 405 nm in comparison to PGE₂ standards. The calculation of the results was performed as follows: (i) the blank optical density (OD) was subtracted from each value, (ii) the average net OD was calculated by subtracting the average OD in non-substrate-bound wells, (iii) the binding of each pair of standard wells was calculated as a percentage of the wells with maximum binding, (iv) the different values of the concentration of PGE₂ standards were plotted against the percent bound and an exponential regression curve was calculated by Microsoft Excel, and (v) the concentration of PGE₂ in the samples was calculated from the respective bound percent based on the regression equation. The 6-keto-PGF_1α concentration was measured in a similar manner using an Assay Designs Correlate-EIA 6-keto-PGF_1α enzyme immunoassay kit as described above.

qRT-PCR analysis.

Following treatment, RNA was extracted from cells using RNeasy (Qiagen) according to the manufacturers protocol. RNA was reverse transcribed in 50-μl reaction mixtures using 10 μl 5× first-strand buffer (Promega), 5 μl 0.1 M dithiothreitol (DTT), 1 μl oligo(dT) (10 μg/ml), 1 μl random hexamers, 1 μl dNTPs (10 μg/ml), 28 μl ddH₂O, and 3 μl total RNA. Samples were mixed, heated to 72°C for 15 min (35), and cooled to RT, and 1 μl RNase inhibitor and 1 μl reverse transcriptase were added. Reverse transcription was performed at 42°C for 2 to 3 h. cDNA was stored at −20°C and 1 μl used for PCRs. Quantitative reverse transcription-PCR (qRT-PCR) was performed using an MJ Opticon DNA Engine in 20-μl reaction mixtures containing 10 μl 2× SYBR green master mix (Invitrogen), 7 μl ddH₂O, 2 μl primers (10 μM), and 1 μl template DNA. Cycle threshold conditions were as follows: initial denaturation at 95°C for 15 min followed by 44 cycles of denaturation at 95°C for 30 s, annealing at 59°C for 30 s (GAPDH), extension at 72°C for 30 s, and plate reading; and performing melting curve analysis from 65°C to 95°C with reading every 0.3°C and a hold for 1 s between reads. Primers used were as follows: GAPDH forward, 5′-TTCACCACCATGGAGAAGGC-3′; GAPDG reverse, 5′-GGCATGGACTGTGGTCATGA-3′; Ptgs2 (COX-2) forward, 5′-TCCTCCTTGAACACGGACTT-3′; Ptgs2 reverse, 5′-CTGCTTGTACAGCGATTGGA-3′; Il6 forward, 5′-CCGGAGAGGAGACTTCACAG-3′; Il6 reverse, 5′-ACAGTGCATCATCGCTGTTC-3′; Fn1 forward, 5′-TACCAAGGCTGGATGATGGT-3′; Fn1 reverse, 5′-TGTCGCTCACACTTCCACTC-3′; Pla2g2a forward, 5′-GCTATGGCTTCTACGGTTGC-3′; Pla2g2a reverse, 5′-CCCCTCGGTAGGAGAACTTG-3′; Lcn2 forward, 5′-GCTTTACCATGTACAGCACCA-3′; Lcn2 reverse, 5′-GAATATTCCCCCAGGGTGAACTG-3′; Col1a1 forward, 5′-ATCAGCCCAAACCCCAAGGAGA-3′; Col1a1 reverse, 5′-CGCAGGAAGGTCAGCTGGATAG-3′; Ctgf forward, 5′-CAGGCTGGGGAGAAGCAGAGTCGT-3′; Ctgf reverse, 5′-CTGGTGCAGCCAGAAAGCTCAA-3′; Timp1 forward, 5′-ATAGTGCTGGCTGTGGGGTGTG-3′; and Timp1 reverse, 5′-TGATCGCTCTGGTAGCCCTTCTC-3′. The absolute amount of RNA (ng) was quantified using the comparative cycle threshold (C_T) (2^−ΔΔCT) method.

RNA microarray studies.

Total RNA was isolated from ARVFs transfected with null vector (pSRα) or cotransfected with ΔCam-AI plus PKCε. RNA was quantified using a NanoDrop ND1000 spectrophotometer and quality assessed by RNA nano-LabChip analysis on an Agilent Bioanalyzer 2100 (Agilent Biotechnologies, CA). Processing and GeneChip analysis for microarray were performed on three samples for each of the two treatment groups. One hundred nanograms of total RNA was processed for hybridization to Affymetrix rat gene ST 1.0 microarrays using the Affymetrix GeneChip whole-transcript sense target labeling assay according to the manufacturer's protocol (Affymetrix, CA). The Affymetrix rat gene 1.0 ST array interrogates 27,342 well annotated genes with 722,254 distinct probes. Analysis of microarray data was performed using Partek Genomics Suite (GS) version 6.4 (Partek, MO) and normalized using robust multichip average (RMA). Expression levels for each gene were normalized to the median array intensity for all genes. Data were analyzed by two-way analysis of variance (ANOVA) to look for differences between the two treatment groups. False-detection rate (FDR) corrections for multiple comparisons were applied to reduce the total number of false positives. A P value of <0.05 was considered significant. The FDR threshold was determined from the observed P value distribution and was thus representative of the amount of signal in the data. Expertise, facilities, and instrumentation for the Affymetrix GeneChip experiments and analysis were provided by UCL Genomics (University College London).

ChIP.

Cross-linked chromatin immunoprecipitation (X-ChIP) was performed essentially as described by Pascussi et al. (40). NRVMs or ARVFs were cultured and transfected with pSRα, PKCε, ΔCNAα, or PKCε plus ΔCNAα as described above. Two 6-well plates were used per treatment. At 36 h after transfection, cells were cross-linked with 37% formaldehyde (54 μl/2 ml/well) for 10 min at 37°C. Cross-linking was stopped by a further 10-min incubation with 250 μl of 125 mM glycine at RT. Cells were washed with PBS at 4°C, scraped into 1 ml PBS, and centrifuged at 1,000 rpm for 10 min. The cell pellet was lysed with 1% SDS–10 mM EDTA–50 mM Tris HCl (pH 8.0) with protease inhibitors and incubated on ice for 10 min. To generate chromatin fragments of 500 bp to 1 kb, lysed cell pellets were sonicated on ice 4 times for 10 s each at 2 micrometers and kept on ice between cycles. Samples were then centrifuged at 13,000 rpm for 10 min at 4°C. Supernatants were removed into fresh tubes and 40 μl removed as input material. Input samples were diluted with 160 ml IP buffer (0.01% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris HCl [pH 8.0], protease inhibitors). The remaining sample was divided into 2 180-μl aliquots and diluted with 720 μl IP buffer. The samples were incubated with 5 μg control antibody (anti-mouse IgG) or test antibody (anti-NFATc1) overnight at 4°C with rotation. Twenty microliters of protein G-Sepharose beads was added to each sample and incubated for 1 h at 4°C. Beads were pelleted at 1,000 rpm for 1 min. The supernatant was removed and pellets washed sequentially for 5 min in wash buffer 1 (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 10 mM Tris HCl, 150 mM NaCl, pH 8.0), wash buffer 2 (same as wash buffer 1 but with 0.5 M NaCl), wash buffer 3 (250 mM LiCl, 1% NP-40, 10 mg/ml deoxycholic acid, 1 mM EDTA, 10 mM Tris HCl, pH 8.0), and finally Tris-EDTA (TE) buffer (10 mM Tris HCl, 1 mM EDTA, pH 8.0). Pellets were then resuspended in elution buffer (1% SDS, 0.1 M NaCO₃), vortexed briefly, and incubated at RT with rotation for 15 min. The suspension was centrifuged briefly at 13,0000 rpm and this step repeated. The eluates were combined and 20 μl 5 M NaCl was added. Input samples and IP mixtures were incubated overnight at 65°C to reverse cross-links, and then 10 μl 0.5 M EDTA–20 μl 1 M Tris HCl (pH 6.8)–20 μl 1-mg/ml proteinase K was added to each sample and 20 μl 1-mg/ml proteinase K to input samples and incubated for 1 h at 45°C. Input samples were adjusted to 500 μl with 300 μl ddH₂O and extracted with 500 μl 1:1 phenol-chloroform. The aqueous phase was recovered and extracted again with 500 μl chloroform. DNA was precipitated with from the aqueous phase with isopropanol. All samples were recovered at 13,000 rpm for 10 min at 4°C, air dried, and resuspended in 75 μl ddH₂O (input) or 30 μl (test samples). PCR primers were designed to 5 regions within the 2-kb COX-2 promoter region, P1 to P5 (see Fig. 9). ChIP DNA was then used as a template in PCRs using either P1, P2, P4, and P5 for NRVMs or P1, P2, and P5 for AVRFs. Primers were as follows: P1 (COX152) forward, 5′-CAGTTTGAGCCCATGCAATTC-3′; P1, reverse 5′-GACAGGGAGTCCTATGCCATC-3′; P2 (COX851) forward, 5′-GCAATGTTTCAGAAAGAAGTG-3′; P2 reverse, 5′-CAGGCTCCGCAGTCCTAGCA-3′; P4 (COX316) forward, 5′-GATTCCCTTAGTTAGGATCTCG-3′; P4 reverse, 5′-GAGCTGCACCGCCCTCTGCAT-3′; P5 (COX117) forward, 5′-AGGGAAGCTTCCTGGCTTCTC-3′; and P5 reverse, 5′-GACTCCACGTGACTTCGTGA-3′. PCR conditions were as follows: initial denaturation at 94.5°C for 5 min followed by 40 cycles of denaturation at 94.5°C for 45 s, annealing at 65°C for 45 s, and extension at 72°C for 45 s; final extension at 72°C; and holding at 4°C.

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.

Statistical analysis.

Statistical analysis was carried out by 1-way analysis of variance (ANOVA) followed by the Newman-Keuls multiple-comparison post hoc test using GraphPad Prism software. Unless otherwise stated, data were pooled from three independent experiments, each performed in triplicate and taken as n = 3. Results were considered significant when the P value was ≤0.05.

RESULTS

Myocardial injury following infarction is associated with an acute inflammatory response during which upregulation of COX-2 and iNOS and activation of innate immune receptors of the Toll-like family are integral components. Since COX-2 has been reported to be induced by brief, repetitive cycles of transient cardiac ischemia/reperfusion and plays a cardioprotective role in delayed ischemic preconditioning (dIPC), we investigated COX-2 expression in response to 3 cycles of brief coronary artery occlusion/reperfusion (3×I/R) in wild-type and PKCε knockout mouse hearts, since dIPC is a PKCε-dependent phenomenon (21, 41, 42). COX-2 was strongly induced at 24 h following 3×I/R in wild-type hearts (Fig. 1A and B). In contrast, COX-2 induction by 3×I/R was absent in hearts from PKCε knockout mice, and also basal levels of COX-2 were lower than in sham-treated hearts. However, COX-2 was not induced by simulated ischemia/reperfusion (sI/R) in isolated neonatal rat ventricular cardiomyocytes (NRVMs).

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.

A constitutively active CnAα catalytic subunit is generated during heart failure due to a proteolytic cleavage event which removes the N-terminal regulatory domain (43). Furthermore, transgenic mice overexpressing this truncated form of CnAα (CN-TG) develop cardiac hypertrophy, adverse remodeling, heart failure, and sudden death (31, 44). Therefore, we examined the expression of COX-2 and connective tissue growth factor (CTGF) as an early marker of fibrosis remodeling in CN-TG hearts compared to nontransgenic (NTG) controls. Figure 1C and D show that COX-2 was expressed at low levels basally and was not induced in CN-TG hearts at 4 weeks of age prior to the development of overt heart failure. In contrast, CTGF was highly upregulated in CN-TG hearts. These results are consistent with CN driving profibrotic remodeling of the heart.

We next investigated in more detail the induction of COX-2 in different cell types of the myocardium by different agonists using either isolated neonatal rat ventricular cardiomyocyte (NRVM) or primary adult rat ventricular fibroblast (ARVF) cultures. Screening of a battery of proinflammatory and growth stimuli, including cytokines and growth factors, revealed very restricted effects on COX-2 expression, with COX-2 induction observed only in response to LPS, angiotensin II (AngII), and soluble fragments of the extracellular matrix protein fibronectin (sFNf). These agonists showed distinct but overlapping time courses of COX-2 induction, whereby the G protein-coupled receptor (GPCR) agonist AngII induced early, transient expression between 2 and 6 h, peaking at 4 h, which declined rapidly thereafter (Fig. 2A). In contrast, the TLR4 agonists LPS, FN-EDA, and sFNf induced sustained expression of COX-2, peaking at 6 to 8 h but lasting at least 24 h (Fig. 2B).

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).

Fibronectin is a known substrate of urokinase-type plasminogen activator (uPA), which leads to the generation of sFNf and plays an important role in fibroblast activation and matrix degradation during metastasis (29, 45). Although COX-2 was not induced by uPA alone, its expression was strongly induced by LPS and sFNf at 24 h (Fig. 2C and E). Furthermore, COX-2 was also induced (4 to 24 h) by the alternatively spliced, wound healing-associated FN domain EDA (FN-EDA), which is a known selective TLR4 agonist, but not by the flanking type III repeat III₁₂ (Fig. 2C). Expression of CTGF was at high levels basally in ARVFs cultured in the presence of serum. This is because when cultured in vitro in the presence of serum, the cells spontaneously adopt a myofibroblast phenotype, expressing αSMA and vimentin (46). However, CTGF expression was inhibited following LPS or EDA treatment and therefore shows reciprocal expression compared to COX-2 (Fig. 2D). These results suggest that COX-2 is expressed primarily by cardiac fibroblasts following ischemia/reperfusion or in response to TLR4 agonists such as matrix-derived fibronectin fragments and is associated with downregulation of CTGF expression.

We next examined the downstream mechanisms of regulation of COX-2 expression in ARVFs. Based on previous findings, we tested the effects of various pharmacological kinase and phosphatase inhibitors. Figure 2D and E show that the sustained induction of COX-2 by LPS or sFNf (at 24 h) was inhibited by the selective calcineurin (CN) antagonist cyclosporine (CsA) and the selective PKC antagonist bisindolylmaleimide I (GF109203X), confirming a role for CN and PKC in COX-2 induction by TLR4 agonists. To further ascertain whether the PKCε isotype regulates inducible COX-2 expression in fibroblasts in a manner similar to that observed in the intact heart, mouse embryo fibroblasts (MEFs) from either wild-type (WT) or PKCε^−/− mice were treated with LPS and expression of COX-2 was determined by Western blotting. Figure 3A shows that although baseline levels were similar (in the absence of serum), COX-2 was not induced by LPS in PKCε^−/− compared to WT MEFs.

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.

COX-2 is induced in the myocardium by I/R in vivo but not in isolated purified NRVMs or adult cardiomyocytes by simulated ischemia/reperfusion (sI/R) (Fig. 1A and B). Therefore, since COX-2 induction may occur in other myocardial cell types, we tested whether sI/R induced COX-2 in ARVFs. Figure 3B shows that COX-2 was strongly induced by sI/R in ARVFs and that furthermore, COX-2 expression in ARVFs was potentiated when the cells were plated onto a complex matrix substrate containing fibronectin and collagen compared to on plastic alone.

AngII is a proinflammatory and pressor peptide hormone, and chronic elevation of AngII levels induces profound myocardial remodeling, including cardiomyocyte hypertrophy and interstitial fibrosis (47, 48). We therefore tested the effects of AngII on COX-2 expression in ARVFs. Figures 2A and 3C show that AngII strongly induced COX-2 in ARVFs, with maximal expression observed at 4 h. These results suggest that different mechanisms may be involved in the induction or maintenance of COX-2 expression downstream of TLR2/4 or AT1/2R. Furthermore, this acute, transient (4-h) AngII-induced expression was blocked by CsA but not GF109203X (Fig. 3C).

We next examined which AT receptor subtype may be involved in the AngII-mediated response in ARVFs by pretreatment with the selective AT2R or AT1R antagonist PD123319 or ZD7155, respectively, prior to AngII treatment. The results shown in Fig. 3D demonstrate that the induction of COX-2 mRNA was blocked by ZD7155 but not PD123319, confirming the role of AT1R in the early, transient induction of COX-2. Interestingly, cotreatment of ARVFs with AngII plus LPS resulted in synergistic induction of COX-2 at 24 h, despite the fact that AngII alone did not induce COX-2 at 24 h (Fig. 4A). Similarly, AngII also potentiated the response to sFNf at 24 h (Fig. 4B). These results suggest that mechanisms downstream of AngII are able to potentiate TLR2/4-dependent COX-2 upregulation.

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).

Since FN-EDA is a known TLR4 agonist, we compared the induction of COX-2 by EDA to that by LPS and sFNf. Figure 4C and D show that COX-2 induction by EDA was blocked by GF109203X and the highly selective calcineurin antagonist FK506. Furthermore, COX-2 induction by AngII at 4 h was also blocked by FK506 (Fig. 4D). To confirm that the NFAT rather than NF-κB pathway was involved, EDA treatment was performed in the presence of the NF-κB inhibitor PS1145. COX-2 induction was potentiated by PS1145 (Fig. 4E).

Having established this paradigm for the regulation of COX-2 expression in cardiac myofibroblasts and given that COX-2 is cardioprotective and antifibrotic, we posed the question of whether or not COX-2 represents a marker for a broader gene expression phenotype and whether other genes may be regulated in a similar manner by the CN-plus-PKCε costimulus. In order to test this hypothesis, we performed gene expression profiling of ARVFs transfected with CN plus PKCε compared to empty-vector controls (pSRα plus pCAGGS) using Affymetrix GeneChip microarrays. Stringency conditions were selected to include the Ptgs2 (COX-2) and Pla2g2a (PLA₂) genes, which we know are upregulated under these conditions. Thus, using an FDR of ≤0.3 and fold change (FC) of ±1.5, there were 979 gene changes (651 upregulated and 328 downregulated). These were used as computational input for hierarchical clustering, the results of which are shown in a dendrogram (heat map) (Fig. 5A) which identified a high degree of change in some genes. Further analysis showed that, as expected, both calcineurin (Ppp3ca) and PKCε (Prkce) genes themselves were highly expressed, with FCs of 20 (P ≤ 0.0007) and 56.7 (P ≤ 2.09 ×10⁻⁷), respectively. Ptgs2 (COX-2) was increased 9.9-fold (P ≤ 0.0007). Phospholipase A2 group IIA (pla2g2a) which is also involved in arachidonic acid metabolism was also upregulated (FC, 41.1; P ≤ 4.68 ×10⁻⁵).

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.

Figure 5B shows the changes associated with a selection of target genes associated with cardiac remodeling. As expected, COX-2 was upregulated (approximately 10-fold), as was interleukin-6 (IL-6) (24-fold). Lipocalin 2 (LCN2) was highly induced (47-fold) and is involved in innate immunity and induced by TLR stimulation. Other genes upregulated include those for fibronectin (Fn1) (2-fold), tissue inhibitor of metalloproteinase 1 (TIMP-1) (11-fold), interleukin-1α (IL-1α) (20-fold), and hypoxia-inducible factor 1α (HIF1α) (2-fold). Conversely, genes prominent in fibrosis were downregulated, including those for connective tissue growth factor (CTGF) (5-fold) and several collagen isoforms, including Col3a1 (19.5-fold), Col12a1 (16.7-fold), and Col11a1 (35.7-fold), as shown in Fig. 5C.

In order to validate these findings, qRT-PCR was performed to quantify the changes in the mRNAs of a selection of these genes. Figures 5D and E show that the fold changes in the mRNAs corresponded to the changes observed in the microarray analysis, with Ptgs2 (COX-2), il6 (IL-6), Pla2g2a (PLA2), lcn2 (LCN2), fn1 (FN), and timp1 (TIMP-1) showing upregulation in ARVFs cotransfected with CN plus PKCε. Conversely, CTGF was downregulated 4-fold. A more detailed analysis of changes in mRNA expression for these markers was carried out in ARVFs following transfection with pSRα, CN, PKCε, or CN plus PKCε. Figure 5D shows that Ptgs2, IL-6, and LCN2 were all induced by CN plus PKCε. Interestingly, Col1a1 and Fn1 were upregulated by CN alone but downregulated by CN plus PKCε (Fig. 5E). Col1a1 was not altered in the microarray analysis, in contrast to Col3a1, Col6a3, Col11a1, Col12a1, and Col14a1. However, these results are consistent with type I collagens being key players in myofibroblast proliferation and fibrosis. The high levels of CTGF expression under control conditions (pSRα) were not affected by CN alone, but CTGF expression was downregulated both by PKCε alone and in an additive manner by CN and PKCε together. In addition, Lcn2 was induced modestly by CN or PKCε alone but synergistically by CN plus PKCε (Fig. 5D). In contrast, TIMP-1 was induced significantly by CN alone (5-fold) but more strongly by PKCε (12-fold), but there was no additive effect of CN plus PKCε, suggesting that TIMP-1 may be highly regulated by PKCε independently of CN. These results further confirm that PKCε may play an antifibrotic role either independently or by cooperation with CN, depending on the specific gene promoter involved.

To further examine the cooperativity between CN and PKCε, NRVMs were transduced with adenovirus constructs expressing CN (ΔCamAI) either alone or in combination with wild-type PKC isotype α, δ, or ε, which are the three most abundant isotypes present in the heart. The results showed that transduction with CN alone in the absence of serum (FBS) only slightly induced COX-2 (Fig. 6A) but that addition of serum to the culture potentiated the induction of COX-2 expression when CN was expressed (Fig. 6A and B). However, coexpression of CN with PKCα or PKCε significantly induced COX-2 even in the absence of serum, and it was further potentiated by the addition of serum (Fig. 6A). These results suggest that PKCα or -ε can synergize with CN to induce COX-2 and can substitute for a signal elicited by serum that is required to drive CN-mediated COX-2 expression. In contrast, PKCδ inhibited COX-2 induction by CN plus serum (Fig. 6A), suggesting that the cooperative induction of COX-2 by different PKC isotypes can drive different effects when coexpressed with CnA and, furthermore, that the CnA cooperation with PKCα/ε is isotype specific. The results in Fig. 6C show that PKCε alone in the presence or absence of serum was insufficient to induce COX-2.

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).

Confocal immunofluorescence analysis of the cells transduced with CN plus PKCα (Fig. 6E) or CN plus PKCε (Fig. 6F) demonstrated that COX-2 expression was observed primarily in the nonmyocyte cells in the culture, since COX-2 colocalized with the myofibroblast marker vimentin but not with the cardiomyocyte-specific myofilament marker α-actinin. Furthermore, treatment of the cultures with the DNA replication inhibitor 5′-cytosine arabinoside C (AraC), which eliminated the fibroblasts from the cultures, abolished COX-2 expression (Fig. 6G). Quantitation of COX-2 expression by fluorescence intensity showed that CN alone induced COX-2 compared to the null vector control (SRα) but that CN plus PKCα and CN plus PKCε resulted in cooperative induction of COX-2. In contrast, PKCα alone had no effect on COX-2 expression, whereas PKCε alone resulted in modest induction of COX-2 (Fig. 6H and I).

COX-2 (Ptgs2) is a rate-limiting step in prostanoid synthesis, and in the heart PGE₂ and PGI₂ (prostacyclin) are the major COX-2-derived prostanoids. Therefore, we measured the production of PGE₂ and 6-keto-PGF_1α (a stable metabolite of PGI₂) to establish the relationship between COX-2 expression levels and activity under the different experimental conditions. The results shown in Fig. 7 demonstrate that CN had no significant effect on PGE₂ synthesis, whereas CN plus PKCα or CN plus PKCε resulted in synergistic PGE₂ production (Fig. 7A). Furthermore, analysis of 6-keto-PGF_1α showed that CN, PKCα, or PKCε alone caused a significant increase in 6-keto-PGF_1α production but that coexpression of CN plus PKCα or CN plus PKCε resulted in a synergistic increase in 6-keto-PGF_1α production (Fig. 7B). As expected, the highly selective COX-2 inhibitor NS398 abolished prostanoid synthesis in all cases, confirming that the effects were dependent on COX-2 expression. COX-2 mRNA levels quantified by real-time qRT-PCR show that CN and CN plus PKCα induced a modest increase in COX-2 mRNA levels, whereas CN plus PKCε induced COX-2 mRNA in a synergistic manner (Fig. 7C).

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.

In order to determine whether TLR4-dependent downstream effects on fibrosis and wound healing gene expression were dependent on COX-2-derived prostanoids, ARVFs were pretreated with NS398 prior to LPS or FN-EDA treatment and harvested 24 h later for analysis of COX-2 and CTGF expression. Figure 7D shows that COX-2 and CTGF demonstrated reciprocal expression in response to LPS, as shown in Fig. 2C. NS398 slightly increased COX-2 expression under all conditions, suggesting negative feedback regulation of COX-2 expression by prostanoids. However, CTGF expression was unaffected by NS398 pretreatment prior to LPS. In contrast, CTGF expression was potentiated in the presence of FN-III₁₂ or FN-EDA following NS398 pretreatment. These results suggest that effects on downstream fibrosis gene expression are at least partially dependent on COX-2-derived prostanoids, particularly downstream of FN-EDA, but that differences exist between TLR4-dependent signaling elicited by LPS and FN-EDA.

To confirm that CN and PKCε cooperate to induce COX-2 in fibroblasts, we performed coexpression experiments in ARVFs as for NRVMs. Figure 8A shows low COX-2 expression in cells transduced with CN or PKCε alone in the absence of serum (FBS). However, COX-2 expression was synergistically increased by coexpression of CN plus PKCε. Addition of serum to the culture further potentiated COX-2 induction under all conditions. Quantitation of COX-2 protein (Fig. 8B) and mRNA (Fig. 8C) expression confirmed that although CN and PKCε could increase COX-2 mRNA, coexpression of CN plus PKCε potentiated COX-2 expression, confirming that PKCε can synergize with CN to induce COX-2 at the mRNA and protein levels in ARVFs and may substitute for a similar signal elicited by serum.

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).

We next sought to determine whether cooperation between CN and PKCε operates at the level of transcriptional control by regulation of the COX-2 promoter. A bioinformatic approach was used to identify the location and structure of the rat Ptgs2 gene using ENSEMBLE and GenBank databases. Phylogenetic relationships between mammalian and nonmammalian orthologous genomic sequences were established using the multiple-sequence alignment and visualization tool MULAN using a 60-kb interval region of the human Ptgs2 at Chr1 (q31.1) as the base sequence. There was a high degree of conservation within the 5′ untranslated regions (UTRs) of the rat and mouse genes, but these were divergent from the human sequence. Therefore, we analyzed the putative transcription factor (TF) binding sites within this region of the rat, mouse, and human genes. Previous studies of the human COX-2 proximal promoter identified key binding sites for NF-κB, NFIL-6, CRE, and NFAT TFs, and MatInspector analysis of an 8-kb region of the rat Ptgs2 gene identified NFAT, NF-κB, AP-1, CAAT, CEBP, CTCF, and GATA sites, as shown in Fig. 9A. Using CLUSTALW alignment and spatial mapping to analyze the orthologous loci in the rat and human Ptgs2 promoters, we identified TF binding sites which shared a common orientation across the 5′ flanking interval, although there was variation in the spatial distribution (Fig. 9B). Of particular note are the close proximity of AP-1 and NFAT sites and the overall conservation of their arrangement. This is in contrast to the case for the rat and mouse promoters, which are approximately 83% aligned by identity and show perfect conservation of the identified TF binding sites. These results imply evolutionary significance and conserved regulatory function.

To determine whether CN and PKCε cooperation was mediated by NFAT binding to the COX-2 promoter, chromatin immunoprecipitation (ChIP) was performed on NRVMs or AVRFs transfected with CN, PKCε, CN plus PKCε, or the vector control SRα using an antibody to recognize NFAT2 (NFATc1) or control antibody. PCR was then performed using primers to amplify preselected promoter regions that contained key NFAT binding sites, as shown in Fig. 9C. In NRVMs, expression of CN or CN plus PKCε resulted in NFAT binding on the three central NFAT sites (region P2) and to the NFAT site at position −316 and the proximal NFAT sites at positions −117 to −79 (region P5) (results not shown). Furthermore, coexpression of CN plus PKCε resulted in enrichment of NFAT binding on the NFAT site in region P2. Highly similar results were obtained in ARVFs, as shown in Fig. 9C, whereby CN alone or CN plus PKCε induced strong binding in regions P1 and P5 and weaker binding in region P2, consistent with all three regions containing NFAT sites. PKCε alone induced binding only in region P1, consistent with this region containing two AP-1 sites. The strongest binding was seen in the ChIPs with CN and CN plus PKCε in region P5, which contains three closely adjacent NFAT sites within 117 bp upstream of the transcription start site (TSS). Furthermore, this is consistent with the most proximal site at −79 consisting of a composite NFAT/AP-1 element as identified by Iñiguez et al. (26). These results confirm that CN or CN plus PKCε enhances NFAT binding to different sites on the COX-2 promoter.

In order to analyze the regulation of the proximal COX-2 promoter, 2 kb upstream of the TSS of the rat COX-2 promoter was cloned into pGL3 in order to drive expression of firefly luciferase reporter gene (pGL3.rCOX2.luc). This reporter construct was then cotransfected into HEK293 cells with either CN or PKCε alone or CN plus PKCε together. HEK293 cells were used to reduce nonspecific background effects on the proximal promoter, since HEK cells do not express TLR4 or COX-2 in response to any of the agonists and because the 2-kb proximal promoter had a high basal activation in ARVFs. Figure 10A shows that promoter activity was increased by PKCε and CN alone, whereas CN plus PKCε additively enhanced its activity. Comparison of the orthologous rat and human proximal promoter regions showed an identical pattern of activation whereby coexpression of PKCε and CN together produced an additive induction of both promoters (Fig. 10A and B). In order to further investigate the role of the proximal NFAT sites, we next designed a deletion construct in which SmaI digestion removed a 500-bp region, leaving the 3′ end of the core promoter containing the three NFAT sites closest to the TATA box (Fig. 10C). Cotransfection studies with this construct showed that CN-mediated induction was reduced by approximately 4-fold compared to that with the full-length promoter (FLP), whereas PKCε alone had no effect on the activity of this truncated promoter. Furthermore, CN plus PKCε resulted in a level of activation similar to that for CN alone, suggesting that removal of upstream regions abolished the cooperative effect (Fig. 10D). These results indicate that upstream NFAT sites were necessary for both activation by CN and maximal activation by CN plus PKCε.

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).

In order to analyze these proximal NFAT sites in more detail, we performed site-directed mutagenesis of the two most proximal sites at positions −69 to −79 and −82 to −92 in the intact full-length promoter (FLP). Mutation of these two sites in the intact promoter abolished the response to CN and PKCε (Fig. 10E). However, CN plus PKCε activated the mutated FLP-m promoter, but the additive effect was lost. These results suggest that although CN can mediate its effects via distal NFAT sites within the promoter, the proximal NFAT sites are required for the cooperative effects seen upon coexpression of CN plus PKCε. Mutation of the same two sites in the SmaI construct abolished the response to CN and prevented activation by PKCε alone, whereas activation by CN plus PKCε was similar to that by CN alone in the intact full-length promoter and was also abolished in the SmaI deletion construct. These results indicate that while CN can activate upstream and downstream NFAT sites, the proximal NFAT sites are essential for activation by CN. Furthermore, the additive effects are mediated by sites outside the proximal 500-bp region. The above results demonstrate that CN and PKCε can cooperate on the proximal promoter of the COX-2 gene to mediate an additive effect on COX-2 transcription which results in a synergistic increase in COX-2 protein levels and prostanoid (PGE₂/PGI₂) synthesis.

DISCUSSION

The results of this study show that PKCε is able to cooperate with calcineurin (CN) to regulate the expression of genes associated with a wound healing phenotype in cardiac myofibroblasts. Furthermore, we have shown that COX-2 expression serves as signature marker for this phenotype and that mechanistically the cooperative signaling was mediated by proximal NFAT transcription factor (TF) binding sites in the COX-2 promoter. Furthermore, CN/PKCε cooperation results in the synergistic induction of the COX-2-derived prostanoids PGE₂ and PGI₂. Interestingly and importantly, CN/PKCε-dependent upregulation of a panel of genes, including those for COX-2, iNOS, Pla2g2a, Lcn2, IL-6, and TIMP-1, occurred in parallel to downregulation of profibrotic genes, including those for at least eight collagen isoforms (Col1a1, Col1a2, Col3a1, Col6a1, Col6a3, Col11a1, Col12a1, and Col14a1), fibronectin, and connective tissue growth factor (CTGF). Since CTGF is a powerful regulator of fibrosis and collagen synthesis (49), these result show that PKCε is able to negatively regulate the fibrotic gene expression program, in part by collaborating with CN. This is supported by the fact that PKCε, apart from its known cardioprotective role, is antifibrotic in vivo, since mice deficient in PKCε (PKCε^−/−) have increased myocardial fibrosis and remodeling in response to injury (50).

Many of these upregulated genes are involved in regulation of the innate immune system. LCN2, or neutrophil gelatinase-associated lipocalin (NGAL), is involved in cellular homeostasis in heart, lung, and kidney (51) and is key in protection against bacterial infection (52). LCN2 is highly induced in macrophages by LPS and antagonizes TNF-α-dependent inflammation (53). LCN2 forms complexes with hydrophobic molecules, including fatty acids, steroids, prostaglandins, and also MMP9. LCN2 enhances MMP9 proteolytic activity and collagen degradation; however, in our study there was no parallel change in MMP9 expression. LCN2 has been proposed to play a role in heart failure, is expressed in cardiomyocytes, and is induced by IL-1β and TLR2/4 agonists (54, 55). Interestingly, IL-1β and TLR2 were both upregulated in ARVFs coexpressing ΔCN and PKCε (36-fold and 7-fold, respectively). Although COX-2, PLA₂, and IL-6 are markers associated with coronary artery disease (56, 57), the innate immune system has recently been linked to adaptation to ischemia (58) and may play roles in delayed and/or remote ischemic preconditioning (59, 60). Other genes of potential interest upregulated >1.5-fold in our study include those for HIF1α, IL-1α, SOD2 (MnSOD), Ddx60, and NOS2 (iNOS).

Interestingly, in our study overexpression of a constitutively active N-terminal deletion mutant of CnAα (ΔCN) alone drove expression of profibrotic markers, including fibronectin and Col1a1, and these were inhibited by PKCε coexpression. In addition, PKCε alone had a strong effect on TIMP-1 induction. Expression of CTGF was high basally in fibroblasts cultured under serum-containing conditions and was associated with a myofibroblast phenotype as indicated by αSMA expression. CTGF expression was strongly inhibited by PKCε expression and further suppressed by coexpression of PKCε and CN. Constitutive activation of CnAα has also been show to be associated with heart failure in humans (43). Furthermore, transgenic mice harboring the same catalytic fragment in a myocardially restricted manner develop adverse cardiac remodeling, heart failure, and sudden death (31, 44). This phenotype is rescued by crossing into a transgenic mouse line expressing the endogenous calcineurin repressor myocte-enriched calcineurin inhibitor protein 1 (MCIP-1) (61).

Our results show that in fibroblasts induction of COX-2 by activation of CN/PKCε cosignaling occurs downstream of Toll-like receptor 4 (TLR4) activation by extracellular matrix (ECM)-derived ligands, including soluble fibronectin fragments (sFNf). Furthermore, the response to sFNf or the classical TLR4 agonist LPS was potentiated by angiotensin II (AngII). Interestingly, AngII alone induced transient expression of COX-2, peaking at 4 h, which contrasted with the more sustained COX-2 induction (up to 24 h) by sFNf or LPS. Furthermore, COX-2 induction by AngII was inhibited by the CN inhibitors CsA and FK506 but not by PKC inhibition, again in contrast to sFNf and LPS, which were inhibited by both CsA and GF109203X, suggesting cooperation between CN and PKC. These results are consistent with AngII being a proinflammatory, profibrotic agonist in myofibroblasts (62) on the one hand and also with the ability of AngII to mediate complex effects on COX-2 expression by other agonists via posttranslational increases in mRNA stability and decreased proteasome-dependent degradation of COX-2 (63). Furthermore, AngII and TLR4 have been suggested to cooperate or be involved in a canonical pathway downstream of osteocalcin in the transformation of vascular myofibroblasts, which also involves COX-2 and PKC (64). Therefore, whereas AngII/AT1R-dependent signaling alone is profibrotic, TLR4/AT1R cross talk may modify the response to alter the fibrosis/wound healing balance.

As mentioned previously, COX-2 plays a role in the negative regulation of fibrosis in the lung and kidney (25, 65). However, it is not clear from the microarray studies whether the CN/PKCε-dependent changes in gene expression occur in parallel or downstream of COX-2 expression via autocrine/paracrine mechanisms. Therefore, in order to ascertain whether or not the effects on gene expression were dependent on COX-2-derived prostanoids, we used the highly selective COX-2 inhibitor NS398. Interestingly, NS398 slightly increased COX-2 expression, suggesting that COX-2-derived prostanoids have a negative feedback on COX-2 expression. However, NS398 had no effect on the reciprocal downregulation of CTGF expression following LPS stimulation. In contrast, CTGF expression was enhanced in the presence of NS398 following stimulation with FN-III₁₂ or FN-EDA. This suggests that while COX-2 upregulation is unequivocally associated with CTGF downregulation, LPS and FN-EDA are able to invoke different mechanisms of CTGF suppression. It is likely that differences in the mode of LPS and FN-EDA binding to TLR4 and different coreceptor requirements lead to the recruitment of different signaling adaptor molecules which couple to different downstream signaling. For instance, our results suggest that while LPS activates NF-κB, there is no indication that FN-fragments activate NF-κB in these cells. However, a more detailed analysis of possible differences in downstream signaling by these agonists is warranted.

These data support the conclusion that NS398 delays skeletal muscle wound healing and promotes fibrosis (65) and are consistent with the observation that prostanoids, including PGE₂ and prostacyclin, antagonize CTGF expression (66). Furthermore, our data are consistent with those of Maqbool et al. (46), which show a similar downregulation of CTGF (CCN2) and upregulation of tenascin C (TNC) expression in human cardiac myofibroblasts in response to IL-1α. Interestingly, IL-1α expression was significantly upregulated by CN/PKCε coexpression as shown by cDNA microarray analysis (Fig. 5), and treatment of human cardiac myofibroblasts with the NF-κB inhibitor IMD-0354 dramatically upregulated basal CTGF expression (46). This is further consistent with a role for NF-κB in negative regulation of CTGF expression.

PKCε has previously been shown to play a crucial role in signaling downstream of TLR4 in macrophages. PKCε^−/− mice have a profound defect in the macrophage response to LPS, are prone to infection, and have a greatly reduced survival in the face of a bacterial challenge (32). Furthermore, the induction of iNOS (NOS2) by LPS in PKCε^−/− macrophages is greatly reduced. MyD88-dependent coupling of TLR4 to downstream signaling, including activation of the Rel family TF NF-κB, is deficient in PKCε^−/− mice (23). In macrophages PKCε coupling to NF-κB appears to be the major mechanism of inflammatory cytokine induction. However, COX-2 induction by CN/PKC cosignaling has also been described in T cells and shown to be due to potentiation of NFAT-dependent transcription (26), possibly by stabilization of NFAT binding and transcriptional activity by interaction with the b-zip family TF activator protein 1 (AP-1), as also occurs on the IL-2 promoter in T cells in response to PKC/Ca²⁺ (CN) cosignaling (27). In our study, the loss of COX-2 induction in PKCε^−/− hearts and MEFs suggests that PKCε is critical for COX-2 induction in the heart. Interestingly, loss of PKCε impairs dermal fibroblast function and wound healing (67), and PKCε^−/− mice also show increased myocardial fibrosis in response to injury (50). These results suggest that PKCε plays a crucial beneficial role in modulating the balance between wound healing and fibrotic responses.

Comparison of a 60-kb genomic interval encompassing the human Ptgs2 locus (1q31.1) was used to identify regions of high constraint and conservation. Phylogenetic analysis using MULAN identified a 20-kb region of high constraint across multiple phyla directly upstream of the Ptgs2 TSS. This was further narrowed down to a 2-kb region immediately upstream of the TSS which showed high conservation between human, rodent, and canine lineages. Position-weighted matrix analysis of TF binding sites of this 2-kb region of the rat and human Ptgs2 genes using MatInspector identified the locations of key TF binding sites, including NFAT, AP-1, NF-κB, and c/EBP. The rat and human promoters showed overall conservation in the number and arrangement of TF binding sites, but not with high positional conservation. However, ChIP analysis revealed a very similar pattern of NFATc1 binding to this region in extracts from both NRVM and ARVF cultures.

ΔCN induced binding to all regions containing NFAT sites. Interestingly, cells transfected with PKCε showed increased binding of NFATc1 where AP-1 sites were adjacent to NFAT sites, for instance, in the region P1 (position −1594 to −1475) and the proximal P5 region (−220 to −45), which contains two NFAT sites plus an NFAT/AP-1 composite element. These results are similar to those observed for the human COX-2 promoter in T cells (26). These sites were necessary for activation by PMA and Ca²⁺ and were CsA sensitive. Therefore, in agreement with Iñiguez et al. (26), we demonstrated that the proximal NFAT sites (−117, −92, and −79) were necessary for activation by CN, and this was also confirmed by site-directed mutagenesis. The positions of the first and third sites (−117 and −79 from the TSS) are conserved in both the rat and human promoters. The 2-kb region of the rat promoter was cloned and analyzed using cotransfection and reporter assays. The promoter was responsive to CN or PKCε, and activity was potentiated by coexpression of CN plus PKCε. The SmaI deletion confirmed a lack of responsiveness to PKCε in the proximal 500 bp as suggested by the ChIP assay. Mutation of the first and third proximal NFAT sites abolished activation by CN and PKCε individually. However, since coexpression of CN plus PKCε still activated the full-length promoter with these sites mutated, this suggests that NFATc1 can be recruited to other distal sites and stabilized by PKCε, which would agree with an NFAT/AP-1 cooperative mechanism where AP-1 sites are in close proximity to NFAT sites. PKCε was confirmed to activate an AP-1.luc reporter construct in these cells. Furthermore, potentiation of CN-dependent activation of an NFAT.luc reporter by PKCε suggests that mechanisms independent of AP-1 may also operate, such as direct posttranslational modification of NFAT itself (68, 69). Both LPS and AngII activated pGL3.rCOX2.luc and NFAT.luc reporters as expected. However, there was no potentiation with LPS plus AngII, agreeing with our conclusion that AngII probably potentiates COX-2 induction by LPS or sFNf via posttranscriptional and/or posttranslational mechanisms on mRNA/protein stability as previously reported (63).

In our study NF-κB did not appear to play a role in COX-2 induction by CN/PKCε in cardiac fibroblasts, and in fact, NF-κB was inhibited by CN. Furthermore, COX-2 induction by FN-EDA was potentiated by the NF-κB inhibitor PS1145, suggesting that the NF-κB and NFAT pathways are mutually antagonistic in this setting. Therefore, it is likely that in myofibroblasts the NFAT/AP-1 mechanism is dominant over NF-κB, which may contrast to the situation in macrophages. TLR4-dependent mechanisms appear to aggravate cardiac ischemia/reperfusion injury, since TLR4^−/− and MyD88^−/− mice have reduced injury and negative remodeling, which is associated with reduced neutrophil/monocyte infiltration as well as fibrosis (13, 14, 16). Furthermore, the TLR4 antagonist eritoran reduced cardiac hypertrophy and remodeling following TAC (15). In addition, the extradomain A (EDA) of fibronectin is a known TLR4 agonist. EDA is spliced into the cellular/matrix (secreted) form of FN following inflammation/injury. EDA^−/− mice are also protected against adverse remodeling after MI (12). However, it is important to point out that these all appear to be monocyte/macrophage-dependent effects on myocardial remodeling. As mentioned above, PKCε^−/− mice have increased fibrosis and negative remodeling postinjury; therefore, the situation is complex. Our results and those of others suggest that the TLR4-dependent activation of CN/PKC cosignaling in myofibroblasts may be beneficial and promote wound healing (18, 19). Therefore, further careful analysis of the cell type- and compartment-specific regulation of fibrosis and wound healing gene expression programs is warranted.