Cytokine-inducible Nitric Oxide Synthase (iNOS) Expression in Cardiac Myocytes CHARACTERIZATION AND REGULATION OF iNOS EXPRESSION AND DETECTION OF iNOS ACTIVITY IN SINGLE CARDIAC MYOCYTES IN VITRO”

,

Cellular constituents of heart muscle contain both constitutive and inducible nitric oxide (NO) signaling pathways that modulate the contractile properties of cardiac myocytes. The identities of the inducible NO synthase (iNOS) isoform(s) expressed in cardiac muscle, and of the specific cell types expressing iNOS activity, remain poorly characterized. We amplified a 217-base pair cDNA by reverse transcriptase-polymerase chain reaction from primary cultures of inflammatory cytokine-pretreated adult rat ventricular myocytes ( A R V M ) that was nearly identical to other iNOS cDNAsequences. Using this 217-base pair cDNA as a probe in Northern blots, we found no evidence of iNOS mRNA in control myocytes, but both interleukin-lp and interferon-y individually increased iNOS mRNA abundance in primary cultures of A R V M , with maximal expression at 12 h. The half-life of iNOS mRNA in actinomycin C1-treated cells was 4 h. Both dexamethasone and transforming growth factor+ attenuated the induction of iNOS mRNA abundance and enzyme activity by IL-1P and INFy. Pretreatment with dexamethasone also abolished the induction of iNOS mRNA, but not the increase in GTP cyclohydrolase mRNA in purified cardiac myocytes from lipopolysaccharide-injected rats. In order to further characterize the specific cell type producing NO, we used a NO-specific porphyrinic/Nafion-coated microsensor to record NO release from a single, isolated ARVM pretreated with IL-1P and IFNy in L-arginine-depleted medium. NO release could be detected following microinjection of L-arginine in the vicinity of the cell juxtaposed to the NO microsensor, but not following microiqjection of D-arginine, and not from ARVM pretreated with L-Nmonomethylarginine. Cytokine-pretreated ARVM that had been maintained in L-arginine-depleted medium also exhibited a depressed contractile response to iso- diovascular Division, Brigham and Women's Hospital, 75 Francis St., proterenol after addition of L-arginine, but not ~-a r g inine. These results indicate that altered contractile function of cardiac myocytes following exposure to specific inflammatory cytokines is due to induction of myocyte iNOS.

~~~~ ~ ~ ~
Recent evidence indicates that nitric oxide (N0)l signaling pathways play a direct role in regulating the contractile properties of cardiac muscle in vitro and in vivo. We have previously established that both neonatal and adult cardiac myocytes contain a constitutive NO synthase (cNOS) activity that mediates the responsiveness of these two myocyte phenotypes to muscarinic and P-adrenergic agonists, respectively (1). Han et al. (2) also have shown that NO plays an essential role in mediating the effects of muscarinic agonists on the rate of depolarization of sinoatrial nodal cells. In addition to the modulation by NO of cardiac myocyte function under normal physiologic conditions, there is also accumulating evidence that specific soluble inflammatory mediators may cause a delayed but sustained increase in inducible NO synthase (iNOS) activity within cellular constituents of cardiac muscle (3-5). I n vitro, addition of IL-lP for at least 24 h to neonatal rat myocytes has been shown to decrease the spontaneous beating rate of these cells, a change that can be reversed by addition of the L-arginine analogue L-NMMA, a NO synthase antagonist. This effect of IL-1P was associated with the expression of iNOS protein in these primary isolates, and was antagonized by co-treatment with TGFp (6,7). Primary isolates of adult rat ventricular myocytes ( A R V M ) exposed to medium conditioned by LPS-activated rat macrophages become less responsive to the positive inotropic effect of P-adrenergic agonists, coincident with a sustained rise in nitrite release into the myocyte-conditioned medium, that can be completely reversed by L-NMMA (3).
While these and other reports in the literature provide compelling evidence for the existence of cNOS and iNOS activities within cellular components of cardiac muscle, the relative contributions of NO synthesized within cardiac myocytes themselves and NO released from neighboring cells to produce the functional effect mentioned above remain unclear. Also, the molecular identity of the isoform of NOS responsible for the sustained rise in NO production following exposure to cytokines in adult ventricular myocytes is poorly characterized: the previous report of a NO-mediated negative inotropic effect of cytokines on adult guinea pig hearts within minutes (8) was in apparent contradiction with the classical time dependence (in hours) for iNOS expression in neonatal preparations (61, but instead pointed to the possible activation of a constitutively expressed cNOS by inflammatory mediators. Another important point of regulation of the activity of both cNOS and iNOS from diverse cultured cell types is influenced by the availability of tetrahydrobiopterin, a co-factor for the NOS enzyme (9). GTP cyclohydrolase I, the rate-limiting enzyme for de novo synthesis of tetrahydrobiopterin was also shown to be inducible by cytokines in vascular smooth muscle and aortic endothelial cells (10,11). However, the relevance of the co-induction of GTP cyclohydrolase I and NOS seen in vitro to pathophysiological situations in vivo has been less clearly established (12).
In this report, we describe the isolation and sequencing of a partial iNOS cDNA from cytokine-treated primary ARVM isolates that is identical to the other iNOS isoforms cloned from other cell types, as well as the regulation of iNOS mRNA and protein content and enzyme activity by specific cytokines and glucocorticoids. To determine whether cardiac myocytes were a source of NO in these primary cultures, we employed an NOselective microsensor (13,141 and experimental conditions that permitted quantitative detection of NO release from individual cytokine-treated adult ventricular myocytes in vitro. Finally, using a model of LPS injection in rats, we verified that the transcripts for iNOS and GTP cyclohydrolase are both induced in adult cardiac myocytes in uivo, but are differentially regulated by glucocorticoids.

EXPERIMENTAL PROCEDURES
Isolation and Preparation of ARVM-Calcium-tolerant ventricular myocytes were isolated from adult male Sprague-Dawley rats (225-275 g) as described previously ( E ) , with modifications to limit the number of non-myocyte cells contaminating primary isolates. These included two density gradient sedimentation steps through a 6% bovine serum albumin (Sigma) cushion followed by centrifugation of the resulting myocyte fraction through a Percoll suspension (Pharmacia Biotech Inc.) and differential attachment to laminin-coated tissue culture dishes, as described previously (16). This method typically limits the amount of nonmyocyte cell contamination to 2 4 % . All myocytes were cultured in a defined medium modified from that originally described by Volz et al. (17) consisting of Dulbecco's modified Eagle's medium with phosphatebuffered saline, including 25 m M HEPES and NaHCO, with L-glutamine (Life Technologies, Inc.), supplemented with 2 mdml bovine serum albumin, 2 m~ L-carnitine, 5 m M creatine, and 5 m M taurine, with 100 IU/ml penicillin, and 100 pdml streptomycin (Life Technologies, Inc.). This medium, which we have termed ACCT (X), was used for all primary cultures except for the NO microsensor and contractility experiments, for which ACCT was supplemented with 0.1 p~ insulin and 10"O M T, (referred to as ACCITT).
Since previous reports have shown that LPS contamination of culture media and reagents can profoundly influence the expression of iNOS in neonatal cardiac myocyte primary isolates (6), we measured the LPS levels in all the solutions used for the isolation and culture of our adult cardiac myocytes preparations using a Chromogenic Limulus Amebocyte Lysate assay kit (Whitaker Biochemicals, Walkersville, MD): all the physiologic buffers (70 pdml), Percoll gradients (stock solution, up to 1 ng/ml), and enzyme mixtures (stock solution, up to 2 ng/ml) contained significant amounts of LPS, from which, after repetitive washes of the myocytes in endotoxin-free culture media, only a fraction was still detectable in the supernatant of 24-h cultured myo-cyte~ in ACCITT (80 pg to 0.4 ng/ml); we then measured the cyclic GMP content (using a RIA kit, Biomedical Technologies Inc, Stoughton, MA), normalized per mg of protein, ofARVM cultured in the absence of added cytokines for 24 h in ACCITT containing either 0.6 m M L-arginine or 1 m M L-N-monomethyl arginine (L-NMMA), an inhibitor of NO synthase, and found no significant difference (10.6 f 0.57 and 10.3 0.41 pmol/mg protein, respectively, 12 determinations in each group from two differ-ent cultures) that would suggest a NOS-dependent increase in cyclic GMP by contaminating LPS in culture.
For experiments in which cardiac myocytes were isolated from animals pretreated with an intraperitoneal injection (4 mgkg) of a lipopolysaccharide (LPS) component of Salmonella typhimurium (87F402; Sigma), or with saline, 300-g male Sprague-Dawley animals were sacrificed 6-7 h after injection. A subgroup of rats were pretreated with intraperitoneal injection of dexamethasone (1.2 mgkg) 45 min before the injection of LPS. ARVM were isolated as described above, including purification through a Percoll sedimentation gradient, and total RNA was extracted from fresh isolates.
Measurement ofN0 Synthase (NOS) Activity-Myocyte NOS activity was quantified by measuring the conversion of ~-[~Hlarginine to ~-[~H]citrulline in the presence of saturating concentrations of the enzyme's cofactors. Total cellular homogenates were prepared from approximately 7 x lo6 ARVM that had been washed three times in warm stopped by the addition of 2 ml of ice-cold 20 m M HEPES (pH 5.5) and 5 m M EDTA, and the total volume was applied to a Dowex 50W-X8 column that had been pre-equilibrated with 20 m~ HEPES (pH 5.5). ~-[~HlCitrulline was eluted with 2 ml of deionized water, and radioactivity was quantified by scintillation counting. The results are expressed as counts (cpm)/mg protein.
Myocyte Contractility Measurements-Myocyte contractility studies were performed exactly as described previously (3). Briefly, ARVM plated on laminin-coated plastic coverslips following pretreatment with control or experimental (i.e. cytokine-containing) medium were placed in a superfusion chamber on the heated stage of an inverted phasecontrast microscope. Cell motion was detected by an optical-video system and digitized for later computer analysis. Cells were stimulated at 2 Hz at 37 "C and superfused with a HEPES-buffered physiologic salt solution with or without isoproterenol (final concentration, 2 nM). One cell per coverslip was examined.
PCR Cloning and Sequencing ofiNOS mRNA from ARVM-In order to identify the iNOS expressed in isolated A R V M , the following PCR and sequencing strategies were used. Total RNA was isolated from both cytokine-pretreated and control ARVM using the method of Chomczynski and Sacchi (18), and stored in diethylpyrocarbonate (Sigma)-treated water at -70 "C. %verse transcription ofARVM RNAwas accomplished using standard protocols. Briefly, 10 pg of total RNA were treated with 100 units of RNase inhibitor (Promega), and denatured at 65 "C for 10 min. A reverse transcription mixture containing desoxynucleotides trisphosphate (dNTPs; 1 mM), random hexamers (20 PM), and 1,000 units of reverse transcriptase (Superscript; Life Technologies, Inc.) was then added in buffer containing 50 m~ Tris-HC1 (pH 8.3), 40 m M KC1, and 2.5 m M MgC1,. For each experiment, parallel control samples were prepared in the same way, except that the reverse transcriptase enzyme was omitted. All samples were incubated at 26 "C for 10 min and at 42 "C for 35 min and the reaction was stopped by heating at 100 "C for 5 min.
Amplification of reverse transcriptase (RT) products was accomplished by subjecting 5 4 aliquots to 35 cycles of PCR (94 "C for 1 min; 55 "C for 1 min; 72 "C for 2 min) in the presence of dNTPs (0.125 mM) and 2.5 units of Taq polymerase (Promega) in a standard buffer containing 1.5 m M MgC1,. The sense oligonucleotide 5"GAGATCAATG-CAGCTGTG-3' corresponds to bp 1342-1359 and the antisense oligonucleotide 5"AGAATGGAGATAGGACGT-3' is complementary to bp 1541-1558 and lies in the 5'-half of the cDNA sequence of the rat vascular smooth muscle cell iNOS isoform (19). For each experiment, control samples were always run in the absence of cDNA template, as well as with RNA samples processed in parallel without reverse transcriptase.
The single PCR amplified product of the expected size (217 bp) was at Harvard Libraries on March 15, 2007 www.jbc.org Downloaded from purified by agarose gel electrophoresis, cloned into pBluescript (Stratagene, La Jolla, CAI, and used to transfect DH5a Escherichia coli. Two positive clones were selected for further characterization. Following plasmid purification, two inserts were sequenced in both directions by the dideoxy chain termination technique using a Sequenase kit (U. S. Biochemical Corp.). Northern Blots-After electrophoresis of 15 pg of total RNA through a 1.5% formaldehyde-agarose gel, and transfer onto a nylon membrane using a vacuum blotter (Bio-Rad model 789, Northern blots were hybridized with the 3ZP-radiolabeled 217-bp cDNA insert obtained by RT-PCR or the full-length cDNA of the rat GTP cyclohydrolase I (kindly provided by Dr. Hatakeyama) under standard conditions (20), and then washed with 2 x SSC, 0.1% SDS for 30 min at room temperature, followed by 1 x SSC, 0.1% SDS at 37 "C, and 0.2 x SSC, 0.1% SDS a t 65 "C. The blots were prepared for autoradiography at -70 "C with intensifier screens for at least 6 h, or as noted.
Western Blot of Inducible NOS Synthase (iNOSI Protein-ARVM were lysed directly in each well by application of a buffer containing 0.062 M Tris-HC1 (pH 6.8), 10% glycerol, 2% SDS (w/v), 5% (v/v) p-mercaptoethanol, and the mixture boiled for 5 min. Equal amounts (90 pg) of the denatured proteins per lane were loaded and separated on a 12% SDS-polyacrylamide gel (Mini Protean 11, Bio-Rad) and transferred to a nitrocellulose membrane that was reversibly stained with Ponceau red to verify equal loading andor transfer between lanes (Millipore, HATF 20200 membrane). The membrane was blocked with 1% bovine serum albumin in Tris-buffered saline with 0.05% (v/v) Tween 20 (TBST; Sigma). Membranes were incubated with rabbit polyclonal anti-mouse iNOS primary antibody that had undergone affinity purification on a synthetic peptide composed of a unique sequence (residues 1-20 of the N terminus of the murine macrophage iNOS (21)) for 3 h in TBST. After three washes (10 min each), the membranes were incubated for 1 h at room temperature with a 1251-coupled goat anti-rabbit IgG secondary antibody at a 1:2000 dilution in TBST with 1% bovine serum albumin.
After three additional washes in TBST, the membranes were rinsed twice in TBS and autoradiographed for 48-72 h.
Detection of NO by Porphyrinic Microsensor-Percoll-purified ARVM treated in vitro with cytokines were cultured on 13-mm diameter plastic coverslips (Thermanox; Nunc, Naperville, IL), and then transferred to a temperature-controlled stage of a Zeiss inverted microscope coupled to a video camera. Two micromanipulators were employed to position a micropipette (Femptopette, Eppendorf) coupled to a microinjector (Eppendorf 5242) and a porphyrinic microsensor in the vicinity of a single rod-shaped cardiac myocyte. The microsensor was based on the design originally described by Malinski and Taha (131, consisting of a porphyrinic Nafion-coated carbon fiber with a sensing tip of approximately 20 pm in length and 0.5-0.8 pm in diameter. A PAR model 264A voltammetric analyzer with a PAR model 181 current-sensitive preamplifier were used in a two-electrode system (NO sensor working electrode and platinum wire (0.25 mm) counterelectrode) for continuous recordings of generated currents under a constant, optimal voltage of 0.72 V in the amperometric mode. All experiments were carried out in Hanks' balanced salt solution with 10 mM Tris-HC1 (pH 7.4 a t 37 "C) in a final volume of 2 ml.
As the purpose of these experiments was to determine whether NO release from a single cell could be detected following addition of Larginine, it was necessary to develop procedures for delivering the amino acid rapidly and locally adjacent to the cell. The microinjection duration and injection pressure were determined in pilot experiments in which fluorescein dye was injected under conditions similar to that for the microsensor experiments, adjacent to a n ARVM in primary culture, and the "cloud" of the fluorescence emission signal tracked to determine the time needed to envelope a single cardiac myocyte (i.e. 100 pm in length). These experiments were performed on the temperaturecontrolled stage of a digital imaging spectrophotometer (Inovision IC-300), attached to a Zeiss Axiovert epifluorescence microscope. A pressure of 6850 hecto-Pascal for at least 3 s was necessary to obtain a cloud of fluorescein 150 p m in diameter. These parameters were validated in was observed in ARVM loaded with the Ca2+-sensitive probe fura-2 separate experiments in which the change in intracellular Ca" activity (Molecular Probes), using previously reported techniques (22), following exposure by this technique to the p-adrenergic agonist isoproterenol. With the microinjection parameters shown above, only myocytes within 100-200 pm of the microinjector exhibited a significant change in the fura-2 emission spectrum indicative of a n increase in intracellular Ca2+ activity (data not shown), at the plating density used for the microinjection experiments (Le. 7,000 celldl3-mm coverslip). For NO measurements, the microsensor was allowed to stabilize for at least 15 min until a stable baseline recording was obtained.
All electrodes used were tested and calibrated with a NO standard in Tris-buffered Hanks' balanced salt solution at 37 "C.

RESULTS
iNOS Activity in ARVM Exposed to Recombinant Cytokines-In a previous report, we documented that primary isolates of ARVM incubated for 24 h in medium conditioned by LPS-activated rat alveolar macrophages demonstrated increased NO synthase activity as measured by detection of bioactive NO release by a reporter cell line (RFL-6 fibroblasts) and by nitrite accumulation in ARVM-conditioned medium (3). This was accompanied by a diminished inotropic responsiveness to the p-adrenergic agonist isoproterenol. Following a series of experiments designed to determine which cytokine(s) in the LPS-activated macrophage-conditioned medium was responsible for increased NOS activity,2 a combination of rhIL-lp and rmIFNy, with or without rhTNFa, was used for the experiments reported here. As shown in Fig. lA, this combination of cytokines increased NOS activity 5-fold in cellular homogenates of A R V M , as measured by the rate of conversion of ~-[~H]arginine to ~-[~Hlcitrulline. As expected, the L-arginine analogue L-NMMA (1 mM), a NO synthase inhibitor, or removal of NADPH from the enzyme assay buffer, decreased iNOS activity to control levels. As shown in Fig. lB, coincubation of rhTGFp, (100 ng/ml) with rhIL-lp (2 ng/ml) and rmIFNy (500 units/ml) reduced iNOS activation by about 50%, as does 3 p~ dexamethasone added to primary myocyte cultures 45 min before the cytokines.
Reverse nunscriptuse PCR Cloning a n d Sequence of a n iNOS cDNA from ARVM-To determine if an iNOS transcript could be amplified in cytokine-pretreated myocytes, PCR was performed on reverse-transcribed total RNA obtained from ARVM that had been isolated using sequential density gradient sedimentations and a Percoll centrifugation procedure, followed by differential attachment to laminin (16). Primary cultures of ARVM isolated in this fashion were used for all the experiments reported in this article.
Using amplimers derived from the nucleotide sequence of the iNOS cDNAfrom rat vascular smooth muscle (191, we amplified a 217-bp cDNA fragment from reverse transcribed RNA isolated from ARVM pretreated with cytokines for 24 h. No RT-PCR product could be identified in control myocytes not exposed to cytokines. Importantly, neither ARVM RNA that had not been reverse transcribed, nor samples from which the cDNA template had been deleted, generated any positive PCR products when otherwise handled identically. The 217-bp PCR product was cloned in pBluescript (Stratagene) and the nucleotide sequence of two independent clones was determined in both directions and yielded identical 217-bp sequences. This sequence (GenBank accession number L36063) was nearly identical to the published murine macrophage (5 mismatches) and rat vascular smooth muscle iNOS (2 mismatches) sequences (19,24). The few nucleotide differences we observed were present in the sequence of both strands in two independent clones.
Detection of iNOS mRNA by Northern Analysis-In order to explore further the abundance and regulation of transcripts corresponding to this PCR product, total RNA from ARVM induced by inflammatory cytokines in vitro was hybridized in Northern blot experiments with the 32P-labeled insert described above. As shown in Fig. 2  signal was found in ARVM exposed in vitro to control medium alone. The transcript size is consistent with that reported for the murine macrophage iNOS mRNA (241, both human and rat hepatocyte iNOS (25) mRNAs, and for rat vascular smooth muscle iNOS mRNA (19).
Further experiments were designed to determine the time course of iNOS induction in ARVM pretreated with recombinant cytokines in vitro, as shown in Fig. 2B. iNOS mRNA transcripts were detectable within 6 h and appeared to plateau after 12 h.
The half-life of iNOS mRNA was examined using actinomycin C1, an analog of actinomycin D, which effectively prevented Detection of iNOS mRNA in ARVM exposed to cytokines in vitro. A, Northern analysis of iNOS mRNA from isolated myocytes induced in vitro. Total RNA from ARVM in primary cultures that had been exposed for 24 h to a 50% dilution of LPS-activated rat alveolar macrophage-conditioned medium in ACCT, prepared as described previously (Ref. 3; lane 21, or to 2 ng/ml rhIL-lp, 100 ng/ml rhTNFa, and 500 units/ml rmIFNy (lane 3 ) or to defined medium (i.e. ACCT) alone (lane 1 ), was hybridized with the 217-bp iNOS :lYP-labeled cloned cDNA fragment described under "Results." After autoradiography for 14 h, a band of the expected size, at 4.6 kb, was detected in RNA from myocytes exposed to either the LPS-activated macrophage supernatant or the combination of recombinant cytokines, but not in myocytes exposed to ACCT medium alone. The blot was rehybridized to a "YP-labeled cDNA probe for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Clontech) RNA to reflect the amount of RNA loaded per lane (15 pg). This experiment was repeated three times with similar results.
B, time course of iNOS mRNA induction in vitro. Northern analysis of time course of iNOS induction in primary cultures ofARVM exposed to ACCT with 100 ng/ml rhTNFa, 2 ng/ml rh-IGlp, and 500 units/ml IFNy for the indicated times. The blots were hybridized to the 217-bp iNOS "'P-labeled cloned cDNA fragment described above, and then rehybridized to a "P-labeled cDNA probe for glyceraldehyde-3-phosphate dehydrogenase RNA to reflect the amount of RNA loaded per lane (15 pg). the appearance of iNOS transcripts when given concurrently with recombinant cytokines (data not shown). When actinomycin Cl(10 pg/ml) was added at 18 h following ARVM exposure to cytokines, iNOS transcript levels decreased to the detection limit by Northern analysis at 10 h, and completely disappeared at 24 h. As a control for the possible spontaneous decline of iNOS transcript, RNA was isolated at each time point from ARVM treated in parallel with cytokines under the same conditions, but in the absence of actinomycin (Fig. 3A). The representative experiment shown in Fig. 3A was repeated seven times, and after correction for 18 S rRNA hybridization, the signal intensity for iNOS, measured by densitometry, was stable over time in cytokine-treated ARVM in the absence of actinomycin; however, in the presence of actinomycin, the iNOS transcript declined over time with a calculated half-life of 4 h (Fig. 3B).
Regulation of iNOS mRNA Abundance and Protein Content-Both IL-10 and IFNy alone were sufficient a t 18 h to increase iNOS mRNA abundance in primary isolates of ARVM showed that the antagonistic effect of TGFB on iNOS mRNA three times with similar results. C. Western blot analvsis of iNOS in enzyme activity in myocyte cellular homogenates as measwith similar results. B, effect of IL-lP and IFNy: cardiac myocfles were abundance was bimodal with almost complete inhibition of the iNOS signal at 1 ng/ml TGFp and a gradual decrease of the inhibitory effect at 5 and 10 ng/ml (data not shown).
To determine whether the changes in iNOS mRNAabundance with dexamethasone were paralleled by similar relative changes in iNOS protein content by Western analysis, we used an iNOS-specific polyclonal antibody raised against epitopes on the murine macrophage iNOS isoform (21). Primary ARVM isolates were exposed to a combination of IL-10 and IFNy with and without 3 PM dexamethasone. As shown in Fig. 4C, while no protein can be detected in ARVM exposed only to control medium for 24 h, there is a detectable protein band in IL-1p and IFNy-treated cells at the size expected for the iNOS isoform. iNOS protein content in dexamethasone-and cytokine-treated cells was markedly decreased compared to myocytes treated with cytokines alone, consistent with the decline in iNOS mRNA abundance in Fig. 4A and enzyme activity in Fig. lB. Regulation of iNOS and GTP Cyclohydrolase mRNA Abundance by Dexamethasone in Vivo-iNOS transcripts were detectable in purified ARVM from LPS-injected rats a t 6-7 h, protein abundance: protein extracts were prepared as described under "Experimental Procedures." iNOS protein was detected using an aflinity-purified anti-murine macrophage iNOS antibody (21), from control myocytes exposed for 24 h to ACCT alone (lane I ) , or from myocytes exposed to 2 ng/ml rhIL-lp and 500 unitdm1 rmIFNy (lane 2 ) , or to these two cytokines with 3 PM dexamethasone added 45 min before IL-1p and IFNy (lane 3).
Separate experiments showed that iNOS mRNA abundance continued to increase at 12 h postinjection (not shown). In LPS-injected animals, transcripts for GTP cyclohydrolase I also increased a t 6-7 h in purified ARVM. Pretreatment with dexamethasone nearly abolished the increase in iNOS mRNA in A R V M , but had little effect on the transcript for GTP cyclohydrolase (Fig. 5).
NO Release by Single Cytokine-exposed ARVM Detected by a NO-selective Porphyrinic Microsensor-In order to further explore the cell of origin producing NO in the ARVM preparations, we took advantage of the sensitivity and selectivity for NO of a porphyrinic/Nafion-coated microsensor. In preliminary experiments, NO-specific signals had been observed in the buffer covering myocytes that had been exposed to LPS-activated macrophage-conditioned medium for 24 h (data not shown). We could also obtain rapid changes of an analytical current in the amperometric mode if cytokine-treated myocytes were first incubated for 4 h in L-arginine-depleted ACCITT medium and the recording initiated in buffer before and after L-arginine, the substrate for NO synthase, was reintroduced.
Experimental conditions were then established that permitted recordings to be obtained of NO release from a single Larginine-depleted cardiac myocyte (Fig. 6). For these experiments, L-arginine was reintroduced through a micropipette with injection parameters designed to deliver the substrate to a single cell (see "Experimental Procedures"). Following exposure to recombinant cytokines and a 4-h incubation in L-arginine-depleted medium, the microinjection pipette and NO microsensor were aligned adjacent to an isolated rod-shaped ARVM. The preparation was then left until a stable baseline was observed for at least 10 min. As shown in Fig. 6B, c, the subsequent microinjection of L-arginine with the same parameters as described above resulted in an increase of the NO signal that gradually returned to baseline over 15-20 min. This experiment was repeated four times with similar results. The mean estimated concentration of NO at the plasma membrane was 104 2 15.8 nM. In comparison, control experiments performed using the same technique on single rat alveolar macrophages induced in vitro with LPS, as described previously (31, resulted in readings between 155 and 800 nM ( n = 3). Other controls were performed with cardiac myocytes, prepared as described above, after microinjection of D-arginine (Fig. 6B, b ) or L-arginine following a 4-h incubation with the L-arginine analog L-NMMA (Fig. 6B, a ) , both conditions in which no signal could be recorded.
In order to verify that the experimental conditions in which NO release was detected from single cardiac myocytes with the NO microsensor also would result in decreased ARVM contractile responsiveness to P-adrenergic agonists, as expected from our previous studies (1, 31, both control and cytokine-pretreated myocytes were incubated in L-arginine-depleted medium for 4 h, and then either D-arginine or L-arginine was added in the superfusion buffer and myocyte contractile responses at baseline and in response to isoproterenol were recorded. The contractile responses to isoproterenol were not different between the two control, non-cytokine-treated groups of cells exposed to either L-or D-arginine (2.8 2 0.45 and 2.1 2 0.26 x the baseline value, respectively, p > 0.05; Fig. 6C, bars 1 and  3). There was no decreased contractile response to isoproterenol in cytokine-pretreated ARVM superfused with D-arginine (Fig. 6C, bar 2 1, but cytokine-pretreated ARVM demonstrated a decline in their inotropic response to isoproterenol in L-arginine-containing buffer compared to uninduced cells, consistent with the data shown above indicating induction of iNOS and NO release from these cells.

DISCUSSION
In the experiments described here we have determined that specific recombinant cytokines added to primary cultures of adult ventricular myocytes leads to an increase in iNOS mRNA abundance and NO synthase activity that is decreased by dexamethasone and TGFP. The induction of iNOS in ARVM by recombinant cytokines had a magnitude and time course that was consistent with the iNOS induction we observed in ARVM incubated in conditioned medium from LPS-activated rat alveolar macrophages (3). Sequence analysis of a partial cDNA showed that the iNOS that is expressed in these cells is highly similar to the iNOS described in other tissues following activation by LPS or cytokines. Cytokine-pretreated cardiac myocytes were clearly identified as a source of NO release in ARVM primary cultures, confirming that these cells contain iNOS activity. We verified that purified adult myocytes expressed iNOS mRNA after injection of LPS in vivo and that this stimulus coincided with an increase in GTP cyclohydrolase mRNA abundance; pretreatment with dexamethasone in vivo decreased iNOS mRNA, while having little effect on GTP cyclohydrolase mRNA.
In contrast with the two constitutive isoforms of NOS (cNOSi which show considerable sequence differences and are encoded by two different genes (26,271, the cDNAs encoding the inducible isoforms of NOS (iNOS) show a high degree of identity across tissues (19,24,25). The partial cDNA reported here is almost identical to previously published sequences and the deduced amino acid sequence, using the same reading frame as for the rat vascular smooth muscle iNOS, is completely identical to the equivalent portion of the originally identified iNOS sequence from murine macrophages (24), as well as other rat iNOS sequences identified in other cells and tissues (19,281. It is also similar (i.e. 81% identity in amino acid sequence) to human iNOS sequences published to date (25,29,30). Using the same amplimers, we identified a similar 217-bp cDNAfragment following RT-PCR amplification of RNA from cytokinepretreated primary cultures of microvascular endothelial cells isolated from adult rat ventricular muscle (31). However, subsequent Northern analysis showed that iNOS is differentially regulated in ARVM and microvascular endothelial cells (e.g. interferon-y alone induces iNOS in ARVM (see Fig. 4B) but not in cardiac microvascular endothelial cells) (31). Although we  6. iNOS activity and contractile dysfunction detected in single, isolated ARVM. A, Percoll-purified ARVM were plated at low density on Inminin-coated coverslips (approximately 7,000 celldl3-mm diameter coverslip), thcn cultured in ACCITT for 24 h with 2 ng/ml IL-Ip, 100 ng/ml TNFtr, and 500 unitshl IFNy. Two separate micromanipulators. one of which was coupled to a microinjector, allowed the placement of a micropipette ( a ) and a porphyrinic microsensor (13,14) ( b ) in the vicinity of a single cardiac myocyte under microscopic control. The electrode (sensing tip: 0.5-0.8 pm) was juxtaposed to the sarcolemmal membrane. Preliminary experiments had indicated that NO release from cytokine-pretreated ARVM primary cultures could be detected if myocytes were first incubated in 1.-arginine-free ACCITT medium (Select-Amine, Life Technologies, Inc.) for 4 h, followed by rapid reintroduction of the amino acid. Microinjection parameters of the micropipette were developed to permit directional release of a cloud of [<-arginine that would rapidly envelop a single cell (see "Experimental Procedures"). R. tracings of the electrode output in the amperometric mode in the experimental setting shown in A. At the point indicated by did not clone the entire cDNA, our data, taken together with previous reports, support the likelihood that iNOS expressed in these different tissues represent the product of a single gene.
The RT-PCR approach we used clearly does not exclude the possibility that we amplified a cDNA product from non-myoc-yte cells in these primary cultures, despite careful attempts at obtaining homogenous preparations of cardiac myocytes. However, subsequent Northern analysis of total RNA from the same cultures using our PCR product as a probe revealed an abundant distinctive transcript, supporting the view that cardiac myocytes themselves express an isoform of iNOS close to, or identical to other iNOS.
The time course of the increase in iNOS mRNA abundance is consistent with that reported for iNOS induction in hepatoc-ytes (32) and murine macrophages (24,33) in response to LPS and cytokines. As shown in Fig;.  Although we (as others (Ref. 6 ) ) measured detectable levels of LPS in our culture reagents, we never found detectable levels of iNOS mRNA in ARVM cultured in the absence of added cytokines either by Northern analysis or RT-PCR, which argues against a significant iNOS induction by traces of LPS alone in our cultured adult myocytes; this is consistent with the absence of the characteristic iNOS-mediated attenuation of the contractile response to isoproterenol when ARVM are cultured in the presence of added LPS alone (3). Of note, in our serum-free defined medium, the ARVM are never exposed to serum that could provide exogenous lipopolysaccharide-binding protein and/or CD 14 molecules, known to enhance endotoxin bioactivity (35); this factor (or developmental differences between neonatal (6) and adult cardiocytes) may provide an explanation for the absence of any effect at relatively low LPS concentrations (35). We cannot exclude, however, that traces of LPS may have potentiated, or even attenuated the induction of iNOS with added cytokines, as shown with other cell types (21. 3 6 ) .
By contrast with its effects in uitro, LPS injected in viuo caused an increase in iNOS mRNA in freshly isolated ARVM the urrow, either L-arginine or D-arginine (1 mu) was microinjected. n-Arginine (1 mu) failed to produce any Rignal under idrntical conditions ( h ) . but L-arginine resulted in an immediate release of SO that continued for several minutes ( c ) . There was no effect of 1 mal !,-arginine in cytokine-pretreated myocyte8 incubated for 4 h in I.-argininc-depleted medium to which 1 mu I.-NMMA had been added la I. The tracings are representative of at least three exprrirnents with the same results. C , contractile function of single ARVhf exposed to 1)-or m r pnine. In order to verify that myoc-vtc contractile dycifunction occurr~d under the same experimental conditions that elicited a pmitivr signal for NO using the NO microsensor, Prrcoll-purified ARVM wrre cultured on coverslips and incubated with cytokines exactly as drscribrd ahovr. After 18-22 h, they were transferred in 1.-arginine-deplrted ACCITT medium for an additional 4 h. then transferred on the hrntcd stape of a video microscope and perfused with Hepes-huffrred salinr solution containing either 1 mM D-or L-arginlne. Control myocytes not exposed to cytokines were cultured and studied in parallel. The . v axis reprrsrnts the increase in amplitude ofcell shortening in response to isoprntrrenol. in the presence of I)-(first and second bars) or [.-arginine cthrrd and fourth bars). The response to isoproterenol was not d~ffrrent (see text) in control myocytes exposed to 1)-or I.-ar*nine (first and third h n r s~. and the response to isoproterenol was not different brtwrrn cytokinetreated ARVM and control non-treated ARVhl in rbarginine-rxpowd cells (first and second bars) . 5); this is consistent with the previous observation of increased NOS activity in various organs after LPS injection (37), including heart (4); importantly, this model of induction differs by the fact that, in addition to LPS, it involves the secondary release of endogenous cytokines, among which TNF-a and IL-lp seem to play an important role for iNOS expression (37). LPS in vivo induced the co-expression of GTP cyclohydrolase I mRNA in freshly isolated ARVM, with the appearance of two transcripts (1.1 and 3.5 kb) consistent with previous descriptions from other cell types in culture (38, 39) (Fig. 5); although we did not specifically measure the functional impact of GTP cyclohydrolase I expression on iNOS activity in vivo, this observation nevertheless lends credence to the view that the co-regulation of both enzymes may be of physiological importance upon immune stimulation and does not only represent the consequence of a culture-induced artifact (9, 12). Werner-Felmayer et al. (40) had also observed variable increases in GTP cyclohydrolase I activity in different organ tissues upon LPS challenge in vivo, although heart tissue was not examined at that time.
Dexamethasone largely prevented the increase in iNOS mRNA abundance and protein content in cytokine-pretreated ARVM (Fig. 4), as well as maximal iNOS activity in these cells in culture. This is consistent with reports of dexamethasone reducing iNOS mRNA abundance in cytokine-pretreated hepatocytes (32). This report and our data on dexamethasone's effects in ARVM contrast with our observations in cytokine-pretreated microvascular endothelial cells isolated from adult rat ventricular muscle, in which dexamethasone caused only a small (15%) decline in iNOS mRNA abundance, protein content, or maximal iNOS activity in cell lysates (31). Importantly, we verified that dexamethasone produced the same effect on iNOS mRNA from ARVM in vivo; however, dexamethasone had little (if any) effect on GTP cyclohydrolase I mRNA (Fig. 5).
Similarly, the effects of dexamethasone were shown to differ on the production of NO and tetrahydrobiopterin in macrophages (41). The mechanism of iNOS mRNA suppression by dexamethasone is not completely understood (9); we have observed a parallel suppression of IRF-1 mRNA, a critical transcription factor for iNOS expression (42), by glucocorticoids in cytokinetreated A R W L 3 The elucidation of GTP cyclohydrolase I gene regulation will await full characterization of its promoter.
Roberts et al. (6) reported that TGFp suppressed the induction of iNOS protein by LPS and IL-lP in primary cultures of neonatal rat cardiac myocytes. The increased and continuous rate of NO release following iNOS induction by IL-1p caused a marked sustained decrease in beating rate a t 20 h that could largely be prevented by TGFp, L-NMMA, or methylene blue. TGFp also decreased iNOS activity (Fig. 1) and iNOS mRNA abundance (Fig. 5 ) in IL-1p and IFNy-pretreated ARVM. The magnitude of the decrease is similar to that which we observed in IL-1p and IFNy-pretreated cardiac microvascular endothelial cells (31). Concentration-response experiments showed a maximal inhibitory effect at 1 ng/ml TGFP consistent with a bimodal effect; interestingly, this concentration is also in the range of activity that we found to be endogenously released in heterotypic co-cultures of ARVM and cardiac microvascular endothelial cells (43). We cannot comment on whether TGFp is acting to reduce iNOS transcription or increase iNOS mRNA breakdown, or acting by additional post-translational mechanisms. Vodovotz et al. (34) noted that TGFp suppressed murine macrophage iNOS mRNA abundance by reducing iNOS mRNA stability and not by inhibiting transcription. They also noted that both iNOS translation and protein half-life appeared to be reduced by TGFP.
J.-L. Balligand, unpublished observations. Finally, a NO-selective microsensor was used in order to verify that iNOS activity, detected at the cell membrane, was present in ventricular myocytes in vitro. The NO microsensor technology has been described and validated in both tissue and cell culture systems, with a detection limit of approximately 10 nM (i.e. approximately lo-'' mol of NO in the volume of the average cell). The amperometric mode used in our experiments has been successfully used to measure the kinetics of NO release from endothelial cells in response to cNOS activation by calcium-raising agonists. In contrast with cNOS, the activity of the largely calcium-insensitive iNOS is not rapidly regulated by changes in intracellular calcium, but is usually sustained for several hours after it has been induced. In order to take advantage of the sensitivity and time resolution of the electrode, we deprived the cells of L-arginine, the substrate for NOS, in order to reduce their production of NO and then abruptly reintroduced the amino acid to observe an outburst of NO release with an acceptable signal-to-noise ratio. Microinjection parameters and micropipette placement were designed to minimize the likelihood of significant NO release from neighboring cells. The detection of NO release following the injection of L-arginine into the medium immediately adjacent to a cytokine-pretreated myocyte, as shown in Fig. 6, demonstrates that these cells contain iNOS activity. That the amperometric signal recorded by the microsensor was specific for NO released by myocytes following L-arginine is supported by the failure of the sensor to detect NO release following either D-arginine or L-NMMA. Under identical conditions, myocytes pretreated with cytokines for 24 h in defined medium, and then deprived of L-arginine for 4 h, exhibited the typical decline in inotropic responsiveness to isoproterenol following readdition of L-arginine, but not D-arginine, that we had originally observed in ARVM exposed t o LPS-activated macrophage-conditioned medium (3).
While the role of the constitutive NO signaling pathway in cardiac myocytes is becoming clarified, it remains unclear what role the much higher levels of NO released from cytokine-exposed cardiac myocytes and other cellular constituents of cardiac muscle play in the normal physiologic response to injury or stress or in the pathogenesis of disease. In addition to its cytotoxic and antiviral effects (441, NO has been shown to either complex directly or to facilitate post-translational modification of a number of cellular and extracellular proteins that could affect myocyte phenotype or function (23,4547). These effects may be relevant to the pathophysiology of certain cardiac diseases.