TNFR1 and TNFR2 Signaling Interplay in Cardiac Myocytes*

Tumor necrosis factor α (TNFα) plays a major role in chronic heart failure, signaling through two different receptor subtypes, TNFR1 and TNFR2. Our aim was to further delineate the functional role and signaling pathways related to TNFR1 and TNFR2 in cardiac myocytes. In cardiac myocytes isolated from control rats, TNFα induced ROS production, exerted a dual positive and negative action on [Ca2+] transient and cell fractional shortening, and altered cell survival. Neutralizing anti-TNFR2 antibodies exacerbated TNFα responses on ROS production and cell death, arguing for a major protective role of the TNFR2 pathway. Treatment with either neutralizing anti-TNFR1 antibodies or the glutathione precursor, N-acetylcysteine (NAC), favored the emergence of TNFR2 signaling that mediated a positive effect of TNFα on [Ca2+] transient and cell fractional shortening. The positive effect of TNFα relied on TNFR2-dependent activation of the cPLA2 activity, independently of serine 505 phosphorylation of the enzyme. Together with cPLA2 redistribution and AA release, TNFα induced a time-dependent phosphorylation of ERK, MSK1, PKCζ, CaMKII, and phospholamban on the threonine 17 residue. Taken together, our results characterized a TNFR2-dependent signaling and illustrated the close interplay between TNFR1 and TNFR2 pathways in cardiac myocytes. Although apparently predominant, TNFR1-dependent responses were under the yoke of TNFR2, acting as a critical limiting factor. In vivo NAC treatment proved to be a unique tool to selectively neutralize TNFR1-mediated effects of TNFα while releasing TNFR2 pathways.

Tumor necrosis factor ␣ (TNF␣) plays a major role in chronic heart failure, signaling through two different receptor subtypes, TNFR1 and TNFR2. Our aim was to further delineate the functional role and signaling pathways related to TNFR1 and TNFR2 in cardiac myocytes. In cardiac myocytes isolated from control rats, TNF␣ induced ROS production, exerted a dual positive and negative action on [Ca 2؉ ] transient and cell fractional shortening, and altered cell survival. Neutralizing anti-TNFR2 antibodies exacerbated TNF␣ responses on ROS production and cell death, arguing for a major protective role of the TNFR2 pathway. Treatment with either neutralizing anti-TNFR1 antibodies or the glutathione precursor, N-acetylcysteine (NAC), favored the emergence of TNFR2 signaling that mediated a positive effect of TNF␣ on [Ca 2؉ ] transient and cell fractional shortening. The positive effect of TNF␣ relied on TNFR2-dependent activation of the cPLA 2 activity, independently of serine 505 phosphorylation of the enzyme. Together with cPLA 2 redistribution and AA release, TNF␣ induced a time-dependent phosphorylation of ERK, MSK1, PKC, CaMKII, and phospholamban on the threonine 17 residue. Taken together, our results characterized a TNFR2-dependent signaling and illustrated the close interplay between TNFR1 and TNFR2 pathways in cardiac myocytes. Although apparently predominant, TNFR1-dependent responses were under the yoke of TNFR2, acting as a critical limiting factor. In vivo NAC treatment proved to be a unique tool to selectively neutralize TNFR1-mediated effects of TNF␣ while releasing TNFR2 pathways.
Tumor necrosis factor ␣ (TNF␣) 2 is a potent proinflammatory cytokine produced by many cell types including cardiac myocytes (1). Silenced under normal conditions, myocardial TNF␣ expression is enhanced upon sustained hemodynamic overloading of the heart or ischemic injury. Levine et al. (2) were the first to correlate circulating levels of TNF␣ with the severity of chronic heart failure in patients and postulated that TNF␣ might contribute to the pathogenesis of heart failure. Thereafter, increased circulating TNF␣ has been shown to be associated with many forms of cardiac injury, including acute viral myocarditis, myocardial infarction, atherosclerosis, chronic heart failure, cardiac allograft rejection, and sepsis-associated cardiac dysfunction (1). These studies clearly highlight that a control of the TNF␣ destructive role in cardiovascular disease represents a realistic goal for clinical medicine.
Nevertheless, a large body of evidence indicates that TNF␣ also exerts beneficial effects on the heart (3). In fact, high levels of TNF␣ are detected in patients with well compensated heart failure, suggesting that TNF␣ may serve as a short term adaptive response and initiate cardiac remodeling (4 -7). This protective action of TNF␣ might explain why large scale, randomized, placebo-controlled trials with TNF␣ antagonists have failed to show any improvement in the clinical status of heart failure and even highlighted worsening of the clinical condition of patients with moderate to severe heart failure (8,9).
The biological responses to TNF␣ are mediated through two structurally distinct receptors: type 1 (TNFR1) and type 2 (TNFR2), both expressed in cardiac myocytes (10). Although the exact functional significance of TNFR1 and TNFR2 in the heart is not known at present, the majority of the deleterious effects of TNF␣ are related to the activation of TNFR1, and include short term negative inotropic effects (10), and long term TNF␣-induced cell death (11). In contrast, activation of TNFR2 appears to exert protective effects (12,13). Although the cardiac TNFR1 downstream signaling system has been studied extensively (1), the transduction of signals from TNFR2 and its role in TNF␣ signaling remains far less well characterized.
TNF␣ has been shown to induce oxidant stress and to cause a drop in glutathione levels, which precedes and regulates its cytotoxic effects (14). Alternatively, in in vivo studies, we have previously shown that a glutathione precursor, antioxidant molecule, NAC prevents the deleterious effect of TNF␣ on cardiac myocyte contraction from control rats, and hinders the progression of cardiac injury in hypertensive L-NAME-treated rats and in post-myocardial infarction rats (15)(16)(17).
The present study was undertaken to further delineate the role and the signaling pathways of TNFR1 and TNFR2 in cardiac myocytes. Our working hypothesis was that NAC treatment might be a unique tool to characterize TNFR2-dependent signaling pathways insofar as protective action of NAC against TNF␣ might rely on TNFR1 signaling inhibition. In this study, we compared TNF␣ signaling pathways, in cardiac myocytes isolated from control or NAC-treated rats and investigated the impact of neutralizing TNFR1 or TNFR2 antibodies. Our results clearly demonstrate that TNFR1 mediates production of ROS, dual positive and negative effects on [Ca 2ϩ ] i handling and cell fractional shortening and cell death. In contrast, TNFR2 plays a major protective role through inhibition of ROS production and cell death. Treatment with either NAC or neutralizing TNFR1 antibodies reveal a new TNFR2-dependent positive effect on [Ca 2ϩ ] handling and cell fractional shortening mediated by activation of cPLA 2 , PKC, and CaMKII pathways, leading to threonine 17 phosphorylation of phospholamban.
Cardiac Myocyte Isolation-The care and the use of animals were in accordance with institutional guidelines. Adult, male Wistar rats (180 -250 g, Janvier, LeGenest St Isle, France) were used. Rats received, or not, NAC (Sigma) added to the drinking water (50 mg/d per animal), for 2 weeks. Calcium-tolerant myocytes were isolated by cardiac retrograde aortic perfusion as previously described (18). Freshly isolated cardiac myocytes were plated on laminin (10 g/ml, Sigma) and cultured up to 2 days, as described previously (18). Note that after isolation of cardiac myocytes from NAC-treated rats, experiments were performed in the absence of NAC or glutathione supplementation.

Measurement of [Ca 2ϩ ] i Transients and Cell Fractional
Shortening-Measurements of [Ca 2ϩ ] i transients and cell fractional shortening were performed in plated cardiac myocytes, loaded with Fura2-AM (Molecular Probes) and submitted to electrical stimulation (square waves, 0.5 Hz) as previously described (16,18,19). Results are shown as mean Ϯ S.E. on at least 10 cells obtained from three different isolations, or a typical representative.
Coordinated ROS and [Ca 2ϩ ] i Imaging-Imaging experiments were performed at room temperature in BSS buffer (in mM: 130 NaCl, 5 KCl, 5 MgCl 2 , 2 CaCl 2 , 200 glucose, and 50 HEPES, pH 7.4). Cell-permeant 2Ј,7Ј-dichlorodihydrofluorescein diacetate (H2DCF-DA) (Molecular Probes) was used to visualize intracellular ROS. Cells were exposed to 5 M H2DCF-DA and 1.5 M Fura2-AM, for 20 min at room temperature, to combine ROS and [Ca 2ϩ ] i imaging. After washing, cells were checked for appropriate [Ca 2ϩ ] i transient responses to electrical stimulation, as well as to TNF␣, before and after recording of the DCF fluorescence images, respectively. DCF fluorescence was recorded as previously reported (16), using 480 nm and 540 nm as excitation and emission wavelength, respectively. At the end of each experiment, maximal DCF fluorescence was determined in response to 2.5 mM H2O2. Results are shown as typical representation from experiments performed on at least 10 cells obtained from three different cell isolations.
Measurement of ROS Production in Isolated Cardiac Myocytes-Cardiac myocytes (7,000 cells/well in 96-well plates precoated with laminin) were allowed to attach overnight. After one wash with BSS buffer, cells were loaded for 30 min at 37°C with 5 M H2DCF-DA together with increasing concentrations of anti-TNFR1-or anti-TNFR2-Mabs or vehicle, in BSS buffer. After one wash with BSS buffer, either 25 ng/ml TNF␣, 2.5 mM H 2 O 2 , or vehicle were added, and fluorescence at excitation and emission wavelengths of 485 and 530 nm, respectively, was monitored every 5 min for 30 min (FL-600 multiplate fluorimeter, Biotek Instruments). Values were corrected for cell autofluorescence. Results were the mean of three different experiments performed on two different cell isolations.
Measurement of Cell Survival-Cardiac myocytes isolated from control rats were preincubated for 30 min with or without 3 g/ml anti-TNFR1-or TNFR2-Mabs before addition or not of 25 ng/ml TNF␣, and cultured for 18 h. In parallel, cardiac myocytes isolated from NAC-treated rats were cultured for 18 h, in the presence or in the absence of 25 ng/ml TNF␣. Cardiac myocytes were visualized using brightfield at ϫ100 magnification, and survival was estimated by counting viable rod-shaped cells versus contracted, nonrod-shaped, dead cells, in 20 random microscopic fields. At least 300 cells were counted in each dish, and results were the mean of two different experiments performed on two different cell isolations.

Measurement of [ 3 H]AA
Release-Freshly isolated cardiac myocytes were plated in 12-well plates previously coated with 10 g of laminin (30,000 cells/well). Following adhesion (4 h), the cell medium was replaced by fresh medium supplemented with [ 3 H]AA (0.3 Ci/well) and cells kept in culture for 24 h in humidified 6% CO 2 , 95% air at 37°C. Where indicated, neutralizing TNFR1 or TNFR2 antibodies were added during the last 30 min of incubation. Cells were kept at 37°C, washed twice with 1 ml of BSS buffer containing 0.2% fatty acid-free bovine serum albumin, and resuspended in 1 ml of the same buffer. At time 0, cells were exposed to TNF␣ or vehicle, and medium samples of 200 l were taken at time 5, 10, and 20 min, transferred to microcentrifuge tubes and diluted with 200 l of icecold EGTA (4 mM final). After centrifugation at 17,600 ϫ g for 10 min at 4°C, the amount of radioactivity in the supernatants was quantitated by liquid scintillation counting, as previously described (16,20). Results were obtained from quadruplicate determinations. Kinetics analyses of the data showed a linear [ 3 H]AA release during the 20-min period examined, in all conditions tested. This allowed determination of the rate of Immunoblot Analysis of the Phosphorylation Status of ERK, MSK1, cPLA 2 , CaMKII, PKC, and PLB-Freshly isolated cardiac myocytes were submitted to pretreatment with or without neutralizing antibodies or MAFP followed by incubation with or without 25 ng/ml TNF␣ in BSS buffer at 37°C for the indicated period of time. Following centrifugation, cellular pellets were dissolved in Laemmli-loading buffer and samples (30 g) subjected to SDS-PAGE (8% (P-cPLA 2 ), 10% (P-MSK1, P-PKC, and P-ERK) and 18% (P-PLB) acrylamide gels). Proteins were electrotransferred to PVDF membranes (0.45 m (P-ERK, P-MSK1, P-cPLA 2 , P-CaMKII, P-PKC) or 0.22 m (P-PLB) (Millipore)). Membranes were first incubated with antibodies against P-ERK, P-MSK1, P-cPLA 2 , P-PKC, or P-PLB, 1:1,000 dilution or antibodies against P-CaMKII, 1:2000 dilution. Blots were then treated with secondary peroxidaseconjugated goat anti-rabbit (1:500 dilution) or goat anti-mouse (1:500 dilution) antibodies. The peroxidase activity was visualized with an enhanced chemiluminescent detection kit (Supersignal West Dura). Signals were normalized to the actin signal (1:2000 dilution of primary antibodies).
Immunohistochemistry and Confocal Laser-scanning Microscopy-Indirect immunofluorescence was performed on freshly isolated myocytes fixed with 4% formaldehyde at room temperature as described previously (18). Briefly, myocytes were incubated in phosphate-buffered saline containing 5% bovine serum albumin for 30 min to block nonspecific binding sites, followed by overnight incubation with a solution of mouse monoclonal antibodies against cPLA2 (1: 250 dilution). After washing, cells were incubated for 1 h with an excess of the secondary FITC-conjugated donkey anti-mouse antibodies (1: 200 dilution). After washing, coverslips were mounted in Vectashield mounting medium. Labeling of cells with secondary antibodies alone was carried out as negative control.
Images were collected as previously described (18), with a Zeiss LSM-510 multitracking laser scanning confocal microscope (Carl Zeiss SAS, Frankfurt, Germany), laser line 488 nm and an oil objective ϫ63 (NA 1.4). We studied nine individual myocytes in each condition examined (control versus TNF␣, in the presence or absence of anti-TNFR1 or TNFR2 Mab antibodies). Note that we were only able to analyze TNF␣ -dependent redistribution of global cPLA 2 labeling due to the lack of availability of antibodies directed against P-cPLA 2 isoforms others than P-Ser505-cPLA 2 . Thus, only TNF␣-dependent intensification of the cPLA 2 signal was detectable without precise enzyme redistribution localization likely due to the limited portion of enzyme redistributed in response to TNF␣ compared with total cPLA 2 labeling.
Drug Treatments-Note that a 25 ng/ml TNF␣ concentration was chosen to favor analysis of the biphasic effect of the cytokine on [Ca 2ϩ ] i transients and cell fractional shortening (16,19).
Myocardial TNF␣, TNFR1, and TNFR2 Protein Expression Levels-Rat hearts were rapidly frozen in isopentane cooled with liquid nitrogen, and stored at Ϫ80°C. Left ventricles (LV) were cut into 20-m sections. Homogenates were prepared from five frozen sections of each LV by homogenization at 4°C, in 200 l of 50 mM Hepes, pH 7.4, containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, 2 g/ml aprotinin), using disposable pestle/microtube devices (Fisher Scientific). After centrifugation at 4°C at 20,000 ϫ g for 20 min, sTNF-␣, sTNFR1, and sTNFR2 were quantified in the supernatant. The pellet, containing the membrane fraction, was resuspended by homogenization in the Hepes buffer containing protease inhibitors, and 1% Triton X-100. After 30 min of incubation on ice, the suspension was centrifuged at 20,000 ϫ g for 20 min, and solubilized membrane-bound TNF-␣, TNFR1, and TNFR2 in the supernatant were determined with double sandwich ELISA kits from R&D Systems (rat TNF␣ Quantikine, mouse sTNFR1, and mouse sTNFR2, respectively).
Statistical Analysis-Results were analyzed by the unpaired two-tailed Mann-Whitney test using GraphPad Prism 4 software. Differences were considered statistically significant at a value of p Ͻ 0.05.

Respective Role of TNFR1 and TNFR2 on TNF␣ Effects on ROS Production, [Ca 2ϩ ] i Transients and Cell Fractional Shortening and Cell
Death-The amplitude of [Ca 2ϩ ] i transients was measured in electrically stimulated adult rat cardiac myocytes, loaded with Fura 2-AM, alone or in combination with the ROSsensitive fluorescent indicator, H2DCF-DA, in response to TNF␣. TNF␣ (25 ng/ml) exerted a dual, early transient positive and late persistent negative effect on [Ca 2ϩ ] i transients, with a 85 Ϯ 42% increase over basal after 10 min followed by a 46 Ϯ 13 decrease below basal after 30 min of TNF␣ perfusion (Fig. 1, C  and A). Concurrently, as shown in a typical experiment performed in cardiac myocytes coloaded with Fura 2-AM and H2DCF-DA, TNF␣ elicited an early and sustained ROS production (Fig. 1A). Both early positive and late negative actions of TNF␣ on [Ca 2ϩ ] i transient amplitude were associated with parallel early increase (not shown) and late decrease (Fig. 1A) in cell fractional shortening. To determine the respective role of TNFR1 and TNFR2 pathways in TNF␣ effects, neutralizing Mabs, specific for TNFR1 or TNFR2, were added to cardiac myocytes, 30 min before subsequent addition of TNF␣. As shown in Fig. 1B, TNFR1-and TNFR2-Mabs exerted dose-dependent opposite effects on TNF␣-induced ROS production, with a full inhibitory effect triggered in the presence of 3 g/ml anti-TNFR1-Mab, contrasting with a maximal 286 Ϯ 73% amplification elicited by 3 g/ml anti-TNFR2-Mab. Neutralizing anti-TNFR1 Mab, which inhibited ROS production also abrogated the dual early positive and late negative effect of TNF␣ on the amplitude of [Ca 2ϩ ] i transients (Fig. 1C), uncovering a new persistent TNF␣-induced positive increase in the amplitude of [Ca 2ϩ ] i transients, with a mean 55 Ϯ 7% stimulation over basal after 30 min, associated with a parallel increase in cell fractional shortening (not shown). Likewise, in cardiac myocytes isolated from rats treated for 2 weeks with the antioxidant molecule, NAC, TNF␣ failed to induce ROS generation, producing 7 Ϯ 2% of the maximal H2O2 response as compared with 35 Ϯ 6% in cardiac myocytes isolated from control rats. In the absence of ROS production, TNF␣ elicited a persistent increase in the amplitude of [Ca 2ϩ ] i transients, with a mean 58 Ϯ 8% stimulation over basal after 30 min (Fig. 1C). In contrast, cardiac myocytes in vitro treated with neutralizing TNFR2-Mabs and exposed to TNF␣ did not only display increased ROS production (Fig. 1B) but also became rapidly unresponsive to electrical stimulation (Fig. 1C). The role of TNFR1 and TNFR2 was also evaluated on a long term effect of TNF␣, i.e. cardiac myocyte survival. After 18 h in culture, counting of contracted, nonrod-shaped, dead cardiac myocytes indicated 13.5 Ϯ 2% versus 54 Ϯ 3% in the absence and in the presence of TNF␣, respectively (Fig. 1D). Treatment with TNFR1-Mab suppressed the deleterious effect of TNF␣ and 11 Ϯ 4 and 7 Ϯ 2% nonrod-shaped cells were counted in the absence and in the presence of TNF␣, respectively (Fig. 1D). In  DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 vivo treatment of rats with NAC before cell isolation also protected cardiac myocytes from TNF␣-induced cell death since 18 Ϯ 8 and 24 Ϯ 4% nonrod-shaped cells were counted in the absence and in the presence of TNF␣, respectively (Fig. 1D). In contrast, treatment with TNFR2-Mab exacerbated the deleterious effect of TNF␣ on cell survival with 17 Ϯ 4 and 68 Ϯ 6% nonrod-shaped cells counted in the absence and in the presence of TNF␣, respectively. Taken together, these results supported the important role of the TNFR1/ROS pathway in cardiac myocytes but also highlighted a major protective effect of TNFR2 signaling. Interestingly, in regard to TNF␣ responses, the in vivo NAC treatment reproduced TNFR1-Mabs action, neutralizing TNFR1 and unmasking TNFR2. Note that, compared with control rats, NAC-treated rats displayed similar undetectable cardiac TNF␣ levels (Ͻ5 pg/mg prot) and comparable cardiac TNFR1 (6.3 Ϯ 0.47 and 5.1 Ϯ 0.4 pg/mg prot, respectively) and TNFR2 protein expression (12.4 Ϯ 1.22 and 8.7 Ϯ 0.5 pg/mg prot, respectively), giving a similar TNFR1/TNFR2 ratio (0.5 Ϯ 0.01 and 0.58 Ϯ 0.03, respectively). To characterize TNFR2-dependent mechanisms underlying the persistent positive effect of TNF␣ on [Ca 2ϩ ] handling and cell fractional shortening, we used cardiac myocytes isolated from NAC-treated rats as a trick to silence TNFR1 pathways.

TNFR1 and TNFR2 Signaling in Cardiac Myocytes
In  from NAC-treated rats, amplification of [Ca 2ϩ ] i transients and cell fractional shortening, in response to a 30-min perfusion with TNF␣, was unaffected by a pretreatment with anti-TNFR1-Mabs but blunted in the presence of neutralizing antibodies selectively directed against TNFR2 (Fig. 2, A and B). Among candidates likely to transduce positive TNF␣ responses, we focused on nitric-oxide synthase (NO synthase) and cytosolic phospholipase A 2 (cPLA 2 ) activities. In fact, both NO and AA have been shown to elicit positive contractile responses, at low doses, in cardiac myocytes (19,21,22). TNF␣ effects were unaffected by the presence of 1 mM L-NAME, the NO synthase inhibitor (not shown). In contrast, preincubation with the cPLA 2 inhibitor, MAFP, suppressed TNF␣-induced responses (Fig. 2, A and B). In the same line, inhibition of PKC, a recently identified target of cPLA 2 in cardiac cells (23), also blunted TNF␣-induced positive effects (Fig.  2, A and B).
Western blot analyses were performed to identify upstream and downstream signaling events in the cPLA 2 activation in response to TNF␣. Particular attention was payed to ERK and MSK1, previously identified as key elements of cPLA 2 activation in response to ATP and ␤ 2 -adrenergic stimulation, in cardiac cells (18,24). TNF␣ induced a time-dependent phosphorylation of both ERK (Fig. 3A) and MSK1 (Fig. 3B). cPLA 2 activation in response to TNF␣ was illustrated by direct assessment of the AA release, measured in cardiac myocytes labeled for 24 h with [ 3 H]AA. As shown in the right panel in Fig. 3C, basal release of AA was linear during the 20-min period examined. TNF␣ elicited a mean 33 Ϯ 6% increase in the rate of AA release (Fig. 3C, right panel). Pretreatment with MAFP suppressed TNF␣-induced AA release (not shown). In addition, confocal microscopy in cardiac myocytes immunostained with a cPLA 2 antibody clearly highlighted a redistri-  middle panels a, b, and c) and corresponding differential interference contrast (Nomarski) images (C, middle panel aЈ, bЈ, and cЈ) in cardiac myocytes isolated from NAC-treated rats, following a 10-min incubation without (C, middle panel, b and bЈ) or with TNF␣ (C, middle panel, c and cЈ). Negative control, in the absence of primary antibody and corresponding Nomarski image were presented (C, middle panel, a and aЈ, respectively) DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 bution of the cPLA 2 in response to a 10-min treatment with TNF␣ that was visualized as an intensification of fluorescent labeling (Fig. 3C, middle panel, c compared with b). In contrast, the phosphorylation of the Ser 505 residue of the cPLA 2 , which is currently referred to as an index of cPLA 2 activation, was not observed in response to TNF␣, under conditions in which the tumor promoter, TPA, was effective (Fig. 3C, left panel). The role of PKC as a target of TNF␣ action was confirmed by a time-dependent induction of its phosphorylation level in response to the cytokine (Fig. 3D).

TNFR1 and TNFR2 Signaling in Cardiac Myocytes
Finally, we examined PLB phosphorylation as a possible cascade signaling terminal component directly linked to the positive effects of TNF␣ on [Ca 2ϩ ] i transients and cell contraction, in cardiac myocytes isolated from NAC-treated rats. As shown in Fig. 3F, TNF␣ induced a time-dependent phosphorylation of PLB on the Thr 17 residue (PT17). Surprisingly, TNF␣ did not elicit phosphorylation of PLB on Ser 16 residue (PS16) (not shown). Note that control experiments performed in parallel with a ␤-AR agonist revealed efficient and predominant phosphorylation of PS16-PLB in cardiac myocytes (not shown). TNF␣ also produced phosphorylation of CaMKII, which ensures PT17-PLB phosphorylation in cardiac myocytes (Fig.  3E) (25).
As shown in Fig. 4A, the cascade of TNF␣-induced ERK, MSK1, PKC, CaMKII, and PLB phosphorylations was blunted upon TNFR2 neutralization but insensitive to the presence of anti-TNFR1 Mab antibodies. Similarly, the TNF␣-induced redistribution of the cPLA 2 , and the increase in [ 3 H]AA release appeared to be TNFR2-dependent but TNFR1-independent events (Fig. 4B).
PKC, CaMKII, and PLB phosphorylations, triggered by TNF␣, were sensitive to MAFP treatment, indicating that they occurred downstream of cPLA 2 activation (Figs. 5 and 6). In contrast, MAFP treatment did not suppress the effect of TNF␣ on either ERK or MSK1 phosphorylation and a 166 Ϯ 5% and a 168 Ϯ 38% increase in the level of ERK and MSK1 phosphorylation were induced in response to the cytokine, respectively. Noteworthy, the effect of TNF␣ on the phosphorylation of CaMKII was not affected by the inhibitor of PKC, with a 147 Ϯ 9% increase in the level of CaMKII phosphorylation in response to TNF␣ to be compared with 151 Ϯ 9% in the absence of inhibitor, clearly arguing for CaMKII and PKC activations as two independent signaling events.

DISCUSSION
This study defines the role of TNFR1 and TNFR2 with respect to TNF␣ effects in cardiac myocytes on ROS production, Ca 2ϩ signaling, cell fractional shortening, and cell survival. We also demonstrate the critical impact on TNF␣ signaling in cardiac myocytes of an in vivo treatment with the glutathione precursor NAC. NAC treatment is proving a valuable tool to promote and provide new insights into the mechanisms that contribute to TNFR2 signaling.
Our results indicate that NAC treatment blunts TNFR1dependent production of ROS, dual positive and negative action on both [Ca 2ϩ ] i handling and cell fractional shortening, and alteration of cell survival. A consequence of TNFR1 neutralization by in vivo NAC treatment is the emergence of TNFR2 signaling and unmasking of a strong stimulatory effect of TNF␣ on both [Ca 2ϩ ] i handling and cardiac myocyte fractional shortening. Analysis of the TNFR2 signaling in cardiac myocytes isolated from NAC-treated rats identifies ERK, MSK1, cPLA 2 , PKC, CaMKII, and PLB as the cascade signaling of TNF␣ (Fig. 6). In addition, we highlight a FIGURE 4. In cardiac myocytes isolated from NAC-treated rats, TNF␣ induced phosphorylation of ERK, MSK1, PKC, CaMKII, and Thr 17 -PLB, cPLA 2 redistribution and AA release, via activation of TNFR2. A, cardiac myocytes isolated from NAC-treated rats were pretreated for 30 min without or with anti-TNFR1 or anti-TNFR2-Mab antibodies and then exposed for 2 min (P-ERK), 5 min (P-MSK1), or 10 min (P-PKC, P-CaMKII, and P-T17-PLB) to TNF␣. Cellular pellets were dissolved in Laemmli-loading buffer, subjected to SDS-PAGE, transferred to PVDF membranes, before immunostaining with antibodies, as described under "Experimental Procedures." Representative immunoblots and densitometric evaluations are presented. Values are mean Ϯ S.E. of densitometric analysis of immunoblots obtained from at least three different isolations. *, p Ͻ 0.05 TNF␣ versus basal; #, p Ͻ 0.05 treatment versus no treatment. B, confocal microscopic analysis of cPLA 2 immunolabeling was performed in cardiac myocytes isolated from NAC-treated rats, pretreated for 30 min without or with anti-TNFR1 or anti-TNFR2-Mab antibodies and incubated for 10 min without or with TNF␣, as described under "Experi- selective Thr 17 phosphorylation of PLB following TNFR2 stimulation.
NAC appears as a unique tool to elucidate cardiac TNFR2 signaling. In fact, TNF␣-induced phosphorylations of ERK, MSK1, PKC, CaMKII, and PLB were almost undetectable in cardiac myocytes isolated from control rats, probably due to the concomitant dominant opposite impact of the TNFR1 pathway on these targets. Direct indications of TNF␣-induced TNFR2 activation in cardiac myocytes isolated from control rats were restricted to TNFR2-dependent [ 3 H]AA release similar to that produced in cardiac myocytes isolated from NAC-treated rats (not shown). Other evidences remained indirect but argued for a critical regulation of TNFR1-activated pathways by TNFR2. Thus: (i) the positive effect of TNF␣ on [Ca 2ϩ ] handling and cell fractional shortening, unraveled after anti-TNFR1 Mab treatment, was neutralized by anti-TNFR2 Mabs, (ii) TNF␣-induced ROS production was amplified by anti-TNFR2 Mabs, and (iii) the deleterious effect of TNF␣ on cell survival was exacerbated by anti-TNFR2 Mabs.
Taken together, our results clearly point out the stimulation of the TNFR2 receptor subtype in response to TNF␣. However, all experiments were performed using human sTNF␣, which has been considered as efficiently active on TNFR1 only, in contrast to membrane-bound TNF␣ that is able to activate both TNFR1 and TNFR2. Long term NAC treatment could favor human sTNF␣ binding to TNFR2. However, TNFR1-independent and TNFR2-dependent activation of [ 3 H]AA release, which was measured in response to human sTNF␣, was similar in cardiac myocytes isolated from either control (data not shown) or NAC-treated rats. In fact, recent flow cytometry based assays gave evidence for the interaction of human sTNF␣ with mouse TNFR2 (26).
Essentially considered as a cardiodepressant mediator, TNF␣ in vivo elicits a delayed and marked negative inotropic effect on cardiac contraction, that is however preceded by an early and limited positive inotropic effect (27,28). TNF␣-negative cardiac effects would result from disturbance of Ca 2ϩ homeostasis, disruption of excitation-contraction coupling, desensitization of the ␤-receptor as well as feedback-induction of other myocardial depressants such as IL1-␤. Our results show a clear association of negative effects of TNF with ROS production (28).
According to the literature, the TNFR1 receptor subtype clearly mediates major signaling mechanisms by which TNF␣ influences cardiac function in normal heart, overwhelming functional expression of the TNFR2 receptor subtype. However, studies using mice lacking either TNFR1 or TNFR2 or both receptors, suggest that, not only TNFR1, but also TNFR2, participate in the pathophysiology of heart failure (12). Studies using double transgenic mice with cardiac-specific overexpression of TNF␣ and TNFR1 or TNFR2 deletion have demonstrated that alterations in the balance of TNFR1 and TNFR2 signal transduction pathways, defined the severity of TNF␣-   DECEMBER 7, 2007 • VOLUME 282 • NUMBER 49 induced heart failure and cardiac remodeling. TNFR1 activation would promote adverse remodeling whereas TNFR2 would mediate cardioprotective effects (12). Our data clearly argue for a predominant role of TNFR2 in mediating TNF␣ effects in cardiac myocytes obtained from NAC-treated rats, or from control rats after neutralization of TNFR1.

TNFR1 and TNFR2 Signaling in Cardiac Myocytes
In contrast with the signaling of TNFR1-mediated negative effects of TNF␣, knowledge on signaling mechanisms mediating the positive effects of the cytokine remains sparse. Our results argue for the role of a TNFR1-induced ROS production associated with the early transient positive response elicited by TNF␣ in cardiac myocytes isolated from control rats. Note that ROS production has already been associated with positive inotropic effect of endothelin in cardiac myocytes, via Na ϩ /H ϩ exchanger stimulation and Na ϩ /Ca 2ϩ exchanger reverse mode activation (29). In contrast, the delayed positive effect of TNF␣ released after TNFR1 neutralization or NAC treatment clearly relies on ROS-independent activation of TNFR2. We have previously reported that TNF␣-induced short-term activation of cPLA 2 supported a positive effect of the cytokine in cardiac myocytes isolated from control rats (19). In the present study we provide evidence that TNF␣ activates cPLA 2 via TNFR2. Note that phosphorylation of the cPLA 2 on Ser 505 has long been considered as a prerequisite for cPLA 2 activation. Accordingly, TNFR2 used to be claimed unrelated to activation of the enzyme. Only recently, a study performed by the group of MacEwan highlighted distinct regulations of cPLA 2 phosphorylation, translocation, proteolysis, and activation by TNFR subtypes. cPLA 2 stimulation by TNFR2 was shown to be unrelated to Ser 505 phosphorylation of the enzyme, in contrast to the TNFR1-dependent regulation (30). The absence of TNFR2-induced cPLA 2 Ser 505 phosphorylation and our finding that MSK1 mediates TNFR2-induced cPLA 2 stimulation, agree with previous observation that ␤ 2 -adrenergic-and ATP-induced cPLA 2 activation, both mediated by MSK1, occurred in the absence of Ser 505 phosphorylation, (18).
We previously located cPLA 2 in caveolae/sarcoplasmic reticulum functional platforms together with MSK1, PLB, and SERCA that are major effectors of Ca 2ϩ cycling (18). Our results clearly show that TNFR2 increases the amplitude of [Ca 2ϩ ] i transients and cell fractional shortening in cardiac myocytes. PLB phosphorylation on Ser 16 or Thr 17 residues leads to the release of SERCA inhibition exerted by unphosphorylated PLB and determines the contractile function in cardiac myocytes. Ser 16 -PLB phosphorylation by PKA, that triggers an increase in [Ca 2ϩ ] i and subsequent activation of CaMKII, potentiates Thr 17 -PLB phosphorylation by CaMKII, in particular in response to ␤-adrenergic stimulation. However, Ser 16and Thr 17 -PLB can be phosphorylated independently, and both phosphorylations contribute to the contractile function in cardiac myocytes (31). Thr 17 -PLB phosphorylation state is the result of phosphorylation by CaMKII and dephosphorylation by protein phosphatase 1 (PP1) (25). We identify CaMKII and Thr 17 -PLB phosphorylations downstream of TNFR2-induced cPLA 2 activation. Thr 17 -PLB phosphorylation occurs independently of Ser 16 -PLB phosphorylation. Of note, recent evidence indicates that Thr 17 -PLB phosphorylation on its own participates in a protective mechanism that favors Ca 2ϩ han-dling and limits intracellular Ca 2ϩ overload and is implicated in the mechanical recovery under some pathological conditions, like acidosis and stunning (25).
We show that PKC activation also participates in the selective phosphorylation of Thr 17 -PLB in response to TNF␣, independently of CaMKII action, because treatment with the PKC inhibitor suppresses TNF␣-induced Thr 17 -PLB phosphorylation (186 Ϯ 26% versus105 Ϯ 4% of the control level, in the absence and in the presence PS-PKC, respectively) without affecting TNF␣-induced CaMKII phosphorylation. PKC are Ser/Thr protein kinases, members of the atypical group of PKCs, characterized by insensitivity to both diacylglycerol and calcium, but activation by other phospholipid cofactors such as AA or ceramide (32). One hypothesis might be that PKC activation, downstream cPLA 2 activation, favored Thr 17 -PLB phosphorylation via protein phosphatase inhibition. In fact, in smooth muscle cells, AA-induced activation of PKC triggers PP1 inhibition (33).
Force et al. (23) reported a hypertrophy of cardiac and skeletal muscles in a mouse model genetically invalidated for cPLA 2 . The authors concluded to the negative regulation of IGF 1 signaling by cPLA 2 , and the implication of PKC as a crucial target for cPLA 2 (23). PKC has also been shown to play a pivotal role in the catabolic pathways initiated by proinflammatory cytokines, IL-1␤ and TNF␣ (34). Our study points out a beneficial role of PKC in cardiac myocytes, observed in the absence of ROS production and supporting positive impact on Ca 2ϩ handling and cell fractional shortening. Its additional potential protective role against alteration of cell survival warrants future examination since a recent study provided evidence that PKC abrogated proapoptotic action of Bax via phosphorylation (35). In contrast, a deleterious role of PKC on cardiac function was reported in ischemia-reperfusion injury which is, in particular, characterized by ROS-induced oxidative stress (36). Hence, defining PKC cardiovascular impact is of major importance, all the more as enzyme inhibition has been proposed as a therapeutic option for the chronic treatment of osteoarthritis (37).
In the present study, possible direct inhibition of TNF␣ binding to TNFR1 by NAC (38), or direct antioxidant effect of the NAC molecule, could be ruled out because NAC was given to the rats per os, for 2 weeks, but was absent from all experiments performed in isolated cardiac myocytes. More likely, in vivo NAC treatment resulted in an increased intracellular glutathione level, as previously described (16). Neutralization of the TNF␣-induced TNFR1-dependent depressant effect might derive from glutathione-induced sphingomyelinase inhibition and glutathione antioxidant action. Thus, our results argue in favor of glutathione as an anti-inflammatory compound, combining anti-TNFR1 and pro-TNFR2 properties. Because inflammation has been clearly linked to cardiovascular disorders (8), the use of NAC as an anti-inflammatory drug precursor, and not as a mere antioxidant, warrants evaluation. In addition, our study contributes to the elucidation of TNFR2mediated pathways in the cardiac myocyte, and may provide novel insights into the role of TNF␣ and/or TNF␣ receptors as targets for therapeutic interventions in patients with heart failure.