Cardiotrophin-1 Activates a Distinct Form of Cardiac Muscle Cell Hypertrophy VIA gp130/LEUKEMIA INHIBITORY FACTOR RECEPTOR-DEPENDENT PATHWAYS*

Cardiotrophin-1 (CT-1) was recently isolated by ex- pression cloning based on its ability to induce an increase in cell size in neonatal rat ventricular cardio- myocytes. Sequence similarity data suggested that CT-1 is a novel member of a family of structurally related cytokines sharing the receptor component gp130. The present study documents that gp130 is required for CT-1 signaling in cardiomyocytes, by demonstrating that a monoclonal anti-gp130 antibody completely inhibits c- fos induction by CT-1. Similarly, a leukemia inhibitory factor receptor subunit (cid:98) (LIFR (cid:98) ) antagonist effectively blocks the CT-1 induction of c- fos , indicating a require- ment for LIFR (cid:98) in the hypertrophic response, as well. Upon stimulation with CT-1, both gp130 and the LIFR (cid:98) Involvement of gp130 and LIFR (cid:98) in CT-1 Induction of c-fos —


Cardiotrophin-1 (CT-1) was recently isolated by expression cloning based on its ability to induce an increase in cell size in neonatal rat ventricular cardiomyocytes. Sequence similarity data suggested that CT-1 is a novel member of a family of structurally related cytokines sharing the receptor component gp130. The present study documents that gp130 is required for CT-1 signaling in cardiomyocytes, by demonstrating that a monoclonal anti-gp130 antibody completely inhibits cfos induction by CT-1. Similarly, a leukemia inhibitory factor receptor subunit ␤ (LIFR␤) antagonist effectively blocks the CT-1 induction of c-fos, indicating a requirement for LIFR␤ in the hypertrophic response, as well.
Upon stimulation with CT-1, both gp130 and the LIFR␤ are tyrosine-phosphorylated, providing further evidence that CT-1 signals through the gp130/LIFR␤ heterodimer in cardiomyocytes. CT-1 induces a hypertrophic response in cardiomyocytes that is distinct from the phenotype seen after ␣-adrenergic stimulation, both with regard to cell morphology and gene expression pattern. Stimulation with CT-1 results in an increase in cardiac cell size that is characterized by an increase in cell length but no significant change in cell width. Confocal laser microscopy of CT-1 stimulated cells reveals the assembly of sarcomeric units in series rather than in parallel, as seen after ␣-adrenergic stimulation. CT-1 induces a distinct pattern of immediate early genes, and up-regulates the atrial natriuretic factor (ANF) gene, but does not affect skeletal ␣-actin or myosin light chain-2v expression. As evidenced by nuclear run-on transcription assays, both CT-1 and ␣-adrenergic stimulation lead to an increase in ANF gene transcription. Transient transfection analyses document that, in contrast to ␣-adrenergic stimulation, the CT-1 responsive cis-regulatory elements are located outside of the proximal 3 kilobase pairs of the ANF 5-flanking region. These studies indicate that CT-1 can activate a distinct form of myocardial cell hypertrophy, characterized by the promotion of sarcomere assembly in series, via gp130/LIFR␤-dependent signaling pathways.
Hypertrophy is one of the most important compensatory responses of the myocardium, as it is critical for the maintenance of normal cardiac function in response to pressure overload in the setting of hypertension or volume overload during valvular insufficiency. However, the mechanical stimuli of pressure versus volume overload produce distinct cardiac muscle phenotypes, with the former resulting in a concentric pattern of hypertrophy, and the latter promoting an eccentric form of hypertrophy with cardiac chamber dilation. On a single cell level, cardiac myocytes isolated from concentric hypertrophic hearts display an increase in cell diameter with the addition of new myofibrils in parallel (1,2), while myocytes derived from dilated hearts exhibit an increase in cell length, reflecting the addition of new sarcomeric units in series (3,4). These distinct morphological phenotypes are associated with different effects on global cardiac function. Chronic volume overload hypertrophy can result in an irreversible loss of cardiac function, while pressure overload hypertrophy is usually associated with the preservation of contractile function, and can often be reversible. While both forms of hypertrophy can lead to the induction of the atrial natriuretic factor gene, distinct molecular phenotypes have also recently been observed (5). Although it has long been accepted that there are distinct morphologic forms of hypertrophy in the setting of pressure or volume overload, the signaling pathways which mediate these distinct cardiac phenotypes remain unclear. One of the critical questions is whether divergent signaling pathways are responsible for the activation of these distinct forms of hypertrophy that display clear differences with respect to cell morphology, physiology, and the induction of molecular markers.
Our current insight into the mechanisms controlling cardiomyocyte hypertrophy has primarily been obtained from an in vitro model employing neonatal rat ventricular myocytes. A number of growth factors signaling through G-protein coupled receptors, including ␣-adrenergic agonists (6 -11), endothelin 1 (12,13), and angiotensin II (14), promote a hypertrophic response in this in vitro system. Following activation of the G-protein-dependent pathways, cardiomyocytes display a uniform enlargement of cell size, with increases in cell width and length, resulting from the assembly of new myofibrils in parallel (12,15,16). ras-dependent pathways appear to be both necessary and sufficient to activate the hypertrophic response following ␣-adrenergic stimulation in the in vitro model system (17), and a recent study in transgenic mice documented that the overexpression of a constitutively active ras protein in the ventricular chambers results in a concentric form of cardiac hypertrophy (18). Taken together, these studies indicate that G-protein coupled receptors and ras-dependent pathways activate a form of cardiac hypertrophy that resembles the hypertrophy seen in the setting of pressure overload. The question arises as to whether divergent signaling pathways would mediate a volume overload like hypertrophic phenotype, characterized by sarcomere assembly in series and myocyte enlargement primarily due to an increase in cell length versus width.
By coupling expression cloning to a miniaturized version of the in vitro hypertrophy assay, we have recently isolated a novel 21.5-kDa protein that induces an increase in cell size and atrial natriuretic factor (ANF) 1 release in cardiomyocytes (19). The protein was designated cardiotrophin-1 (CT-1). Sequence similarity data and structural considerations suggest that CT-1 is a new member of a family of structurally related cytokines including interleukin (IL)-6 and IL-11, leukemia-inhibitory factor (LIF), ciliary neurotrophic factor (CNTF), and oncostatin M (OSM). The receptors of this cytokine family are multimeric and share the class-specific transmembrane signal transducing component gp130 (20 -23). Signaling is triggered through the homodimerization of gp130 (24), or the heterodimerization of gp130 with a related transmembrane signal transducer, the LIF receptor subunit ␤ (LIFR␤ (25,26). The gp130/LIFR␤ heterodimer constitutes the functional bipartite LIF receptor complex (22,27). Other members of this cytokine family first bind to a private receptor component and subsequently induce gp130 dimerization. For example, IL-6 binds to the receptor component IL-6R, and the IL-6⅐IL-6R complex then induces the homodimerization of gp130 (28,29); CNTF binds to the receptor component CNTFR␣ and induces gp130/ LIFR␤ heterodimerization (26,30,31). Consequently, the expression pattern of the various receptor components defines the spectrum of action of different members of the gp130 cytokine family.
The cloning of CT-1 based on its ability to induce an increase in cell size in cardiomyocyte culture (19) suggested that gp130dependent signaling pathways may be coupled to cardiomyocyte hypertrophy. In the present study we define the receptor system used by CT-1 in cardiomyocytes, and we examine the effects of CT-1 and two other members of the IL-6 cytokine family, LIF and CNTF, on morphological and molecular features defining a hypertrophic response in cultured cardiomyocytes. By using a monoclonal anti-gp130 antibody and a LIFR␤ antagonist, we establish that both gp130 and LIFR␤ are required for CT-1 signaling in cardiomyocytes. Immunoprecipitation and immunoblotting data show that CT-1 induces the tyrosine phosphorylation of gp130 and a higher molecular weight protein, corresponding in size to LIFR␤. Using confocal laser microscopy of cells stained for both thick and thin filament markers, our studies further indicate that gp130/LIFR␤dependent signaling pathways induce a hypertrophic phenotype in cardiomyocytes that is distinct from the response seen after ␣-adrenergic stimulation, and that shows resemblance to volume overload cardiac hypertrophy both with regard to cell morphology and gene expression pattern.

EXPERIMENTAL PROCEDURES
Murine LIF was produced at Genentech, Inc. Murine LIF was used throughout the study, except for the experiment employing the LIFR␤ antagonist (hLIF-04), in which both murine LIF and human LIF were used. Rat CNTF and human IL-6 were purchased from Boehringer Mannheim and from Genzyme, respectively. Human sIL-6R was prepared as outlined previously (21). The ␣-adrenergic agonist phenylephrine (PE) was obtained from Sigma.
Expression and Purification of the CT-1 Fusion Protein-The reading frame of murine CT-1 was cloned C-terminal to the sequence encoding the herpes simplex virus glycoprotein D followed by a factor Xa cleavage site. Following expression in 293 cells and cleavage of the herpes simplex virus secretion signal sequence, the CT-1 fusion protein contained a 34-amino acid N-terminal extension followed by the CT-1 sequence (19). The CT-1 fusion protein was purified from conditioned medium with a herpes simplex virus glycoprotein D-specific monoclonal antibody and quantified by a colorimetric assay (Bio-Rad).
LIFR␤ Antagonist hLIF-04-The LIFR␤ antagonist hLIF-04 was used to assess the requirement of the LIFR␤ for CT-1 signaling. Human LIF (hLIF) and the hLIF mutant, hLIF-04, were expressed as glutathione S-transferase fusion proteins, as described previously (32). hLIF-04 contains three amino acid substitutions (Q25A, S28A, and Q32A), within the A-helix of the hLIF molecule, resulting in a selective disruption of the gp130 binding site, but no impairment in LIFR␤ binding. A detailed characterization of the LIFR␤ antagonist hLIF-04 will be presented elsewhere. 2 Cell Culture-Hearts from 1-3-day-old Sprague-Dawley rats were removed, the ventricles were pooled, and the ventricular cells were dispersed by digestion with collagenase II (Worthington) and pancreatin (Life Technologies, Inc.). The cell suspension was purified by centrifugation through a discontinuous Percoll gradient to obtain myocardial cell cultures with Ͼ95% myocytes, as assessed by immunofluorescence with an anti-TrpE/MLC-2v antiserum (12,33). The cardiomyocytes were plated at a density of 5-6 ϫ 10 4 cells/cm 2 in Dulbecco's modified Eagle's medium (DMEM)/medium 199 (4:1) supplemented with 10% horse serum, 5% fetal bovine serum, glutamine and antibiotics (plating medium), and allowed to attach overnight. The cells were then washed twice with DMEM/medium 199 and further incubated in DMEM/medium 199 supplemented with glutamine and antibiotics (maintenance medium) in the presence or absence of various agents. Neonatal mouse ventricular myocytes were isolated from 1-4-day-old NIH Swiss mice, following the protocol outlined above. One litter (8 -12 pups) yielded ϳ5 ϫ 10 5 cells. Approximately 95% of the cells displayed spontaneous contractile activity in culture.
Immunoprecipitation and Immunoblotting-Cardiomyocytes were stimulated with various agents and solubilized with ice cold lysis buffer (0.5% Nonidet P-40, 150 mM NaCl, 10 mM Tris-HCl, pH 8.0, 5 mM EDTA, 2 mM sodium vanadate, 1 mM p-amidinophenylmethanesulfonyl fluoride (Sigma), 5 g/ml aprotinin). Clear lysates were obtained by centrifugation and immunoprecipitated with a polyclonal rabbit anti-gp130 cytoplasmic peptide antibody (34). Immunoprecipitates (protein A-Sepharose, Pharmacia Biotech Inc.) were subjected to SDS-PAGE under reducing conditions and immunoblotted with an anti-phosphotyrosine antibody (4G10, Upstate Biotechnology), employing an enhanced chemiluminescence detection system (Amersham Corp.). For the detection of protein tyrosine phosphorylation, the cell lysates were analyzed directly by SDS-PAGE and immunoblotting with 4G10.
were dual-stained for 1 h in 3% bovine serum albumin with 1) a polyclonal anti-TrpE/MLC-2v antiserum from rabbit (12), combined with a monoclonal mouse anti-rat ANF antibody (35,36), or 2) a monoclonal mouse anti-human ␤ myosin heavy chain (␤MHC) antibody (37), combined with rhodamine phalloidine (Molecular Probes). Following four phosphate-buffered saline washes, the slides were blocked with 5% normal donkey serum for 10 min, incubated for 1 h in 5% normal donkey serum using a Texas Red-conjugated donkey anti-rabbit IgG and a fluorescein isothiocyanate-conjugated donkey anti-mouse IgG as secondary antibodies (Jackson Laboratories), and finally mounted on glass coverslips. Cardiomyocytes stained against MLC-2v and ANF were viewed by fluorescence microscopy. Cell size was estimated by measuring the area to which individual MLC-2v-positive cells attached (planimetry), cell length and width were determined as outlined in Fig.  1, and the percentage of MLC-2v positive cells that displayed perinuclear staining for immunoreactive ANF was calculated. Cardiomyocytes stained against ␤MHC and F-actin (rhodamine phalloidine) were analyzed by confocal laser microscopy using a 60 ϫ oil immersion objective. Images of cardiomyocytes stained against ␤MHC were used to determine the effect of various agonists on the average sarcomere length.
ANF Radioimmunoassay-The concentration of immunoreactive ANF in cell culture supernatants was determined by radioimmunoassay, using the monoclonal mouse anti-rat ANF antibody with synthetic ANF-(99-126) as the standard and 125 I-ANF-(99-126) (Amersham) as the trace (35).
Isolation and Hybridization of RNA-Total cellular RNA was isolated by a modification of the acid guanidium/thiocyanate phenol/chloroform extraction method (38) using the RNA STAT 60 RNA isolation reagent (Tel-Test "B"). RNA was size-fractionated by formaldehyde agarose gel electrophoresis, transferred to nylon filters (Magna NT) by overnight capillary blotting, and hybridized with cDNA probes labeled with [␣-32 P]dCTP by random priming (Life Technologies, Inc.). The filters were washed under stringent conditions and exposed to x-ray film (Kodak X-Omat AR). Signal intensities were determined by densitometry (UltroScan XL, Pharmacia). The following cDNA probes were used: mouse c-fos (1.  (45) and rat MLC-2v (0.6 kb of coding region) (46). Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1.2-kb fragment) (47) and mouse 28s (5-kb fragment) cDNA probes were used to control for loading and transfer efficiency.
Nuclear Run-on Transcription Assay-Run-on transcription assays were performed as described previously (48). Cardiomyocytes were lysed in ice-cold Nonidet P-40 buffer (10 mM Tris-HCl, pH 7.5, 10 mM NaCl, 3 mM MgCl 2 , 0.5% Nonidet P-40) and further disrupted by 10 gentle strokes with a loose pestle in a Dounce homogenizer. The nuclei were recovered by centrifugation. The supernatants (representing the cytosolic fractions) were saved, and vanadylribonucleoside complexes (Life Technologies, Inc.) were added to inhibit RNase activity. Total RNA was isolated from the cytosolic fractions and subjected to Northern blot analysis using prepro-ANF, GAPDH, and 28s cDNA probes. The nuclei were washed once in Nonidet P-40 buffer, resuspended in 100 l of storage solution (50% glycerol, 50 mM HEPES, pH 7.4, 5 mM MgCl 2 , 0.1 mM EDTA, 5 mM dithiothreitol), and stored at Ϫ80°C. On the day of the assay, the nuclei were thawed, 27.5 l of run-on buffer (25 mM Tris-HCl, pH 7.5, 12.5 mM MgCl 2 , 750 mM KCl, 25 mM dithiothreitol, 1.25 mM of ATP, GTP, and CTP), 6 l (62.5 Ci) of [␣-32 P]UTP, and 1 l of RNasin (Promega) were added, and the nuclei were incubated at 30°C for 30 min. Following DNA and protein digestion (RQ-1 DNase 1, Promega; proteinase K, Life Technologies, Inc.), the 32 P-labeled RNA was extracted once with phenol/chloroform, ethanol-precipitated, and purified over a RNase-free Sephadex G50 column (Boehringer Mannheim). The total amount of 32 P-labeled RNA recovered from each reaction was determined by liquid scintillation counting (Beckman). Five micrograms of linearized, alkali-denatured pBluescript (Stratagene), pBluescript harboring the rat prepro-ANF cDNA (44), and pBluescript harboring the human GAPDH cDNA (47) were spotted onto nylon filters in advance. The filters were prehybridized for 1 h in 56.25% formamide, 0.25 M sodium phosphate, pH 7.2, 0.25 M NaCl, 1 mM EDTA, 7% SDS. An equal number of counts from each run-on reaction were added to the prehybridization solution, and hybridizations were carried out for 48 h at 42°C. Finally, the filters were washed under stringent conditions and exposed to x-ray film. The amount of ANF and GAPDHspecific hybridization was determined by densitometry. No hybridization to the empty pBluescript control was detectable (data not shown).
Plasmid Constructs-pANF(Ϫ3003)Luc5Ј, an ANF-luciferase reporter construct composed of the most proximal 3003 bp of the rat ANF gene 5Ј-flanking region inserted into the promoterless firefly luciferase reporter plasmid pSVOALuc5Ј was used to assess the CT-1-, LIF-, CNTF-, and PE-induced activity of the rat ANF promoter (11). pRSV-Luc5Ј, a RSV-luciferase reporter construct, and pSVOALuc5Ј served as positive and negative controls, respectively (49). To correct for transfection efficiency, cardiomyocytes were cotransfected with pON249, a ␤-galactosidase expression vector under the control of the human cytomegalovirus promoter (50).
Transfection and Luciferase/␤-Galactosidase Assays-Cells were plated in 3.5-cm dishes in duplicate, allowed to attach overnight, and switched to maintenance medium supplemented with 4% horse serum. Two hours later, the cells were cotransfected with 2.75 g of either pANF(Ϫ3003)Luc5Ј, pRSVLuc5Ј, or pSVOALuc5Ј, and 0.75 g of pON249 using the calcium phosphate method as described by Chen and Okayama (51). After 18 -20 h, the cells were washed twice to remove the fine layer of calcium precipitate and incubated in maintenance medium in the presence or absence of various agents. At different time points, the cells were washed twice with ice-cold phosphate-buffered saline and were lysed for 30 min on ice in 100 l of lysis buffer (0.1 M KH 2 PO 4 , pH 7.9, 0.5% Triton X-100, 1 mM dithiothreitol). Luciferase activities were determined in duplicate samples from each plate; 20 l of cell lysate were combined with 100 l of luciferase assay buffer (100 mM Tricine, pH 7.8, 10 mM MgSO 4 , 2 mM EDTA, 2 mM ATP, 1 mM dithiothreitol, 73 M ␤-luciferin), and luciferase activities were measured using a Monolight 401 luminometer (Analytical Luminescence Laboratory). ␤-Galactosidase activities were determined in duplicate samples from each plate; 20 l of cell lysate were added to 250 l of ␤-galactosidase assay buffer (0.1 M sodium phosphate, pH 7.3, 1.2 mM MgCl 2 , 60 mM ␤-mercaptoethanol, 1.8 mM chlorophenol red-␤-D-galactopyranoside (Boehringer Mannheim)), and incubated at 37°C for 15 min. ␤-Galactosidase activities were obtained by measuring OD at 574 nm, using lysis buffer as a blank.
Statistical Analysis-Data are presented as means Ϯ S.E. p values were determined using one-way analysis of variance.

Induction of a Distinct Pattern of Immediate Early Genes following CT-1 versus ␣-Adrenergic Stimulation of Cultured
Myocardial Cells-The induction of immediate early (I.E.) genes precedes the late phenotypic changes in cardiomyocytes following the stimulation of G-protein-coupled receptors in vitro (12)(13)(14)(15) and cardiac hypertrophy in vivo (52,53). We therefore determined whether stimulation of cardiomyocytes with CT-1, LIF, or CNTF would induce I.E. gene expression (Fig. 2). Stimulation with CT-1 resulted in the rapid and transient induction of a panel of I.E. genes, including c-fos, egr-1, c-myc, c-jun, jun-B, and tis11. The pattern of I.E. gene expression in response to stimulation with LIF was virtually indistinguishable from the pattern induced by CT-1. CNTF displayed a relatively weak induction of the same panel of I.E. genes as did CT-1 and LIF. As compared to stimulation with PE, CT-1, and LIF induced Ͼ10 fold higher levels of c-myc and tis11.

Involvement of gp130 and LIFR␤ in CT-1 Induction of c-fos-
Previous studies have employed blocking antibodies directed against gp130 to demonstrate that IL-6, LIF, OSM, CNTF, and IL-11 utilize the common receptor component gp130 (21)(22)(23). Sequence similarity data, as well as structural considerations, suggest that CT-1 might be a new member of this cytokine family, and might therefore share the receptor component gp130 (19). To directly test this hypothesis, we examined the effect of an anti-gp130 antibody on the CT-1 induction of c-fos in cardiomyocytes. A monoclonal anti-gp130 antibody (RX435) was generated by immunizing rats with recombinant murine gp130. 3 RX435 specifically blocks the action of LIF, IL-6, IL-11, and OSM on mouse myeloid leukemic M1 cells 3 and inhibits the binding of CT-1 to M1 cells (54). Since RX435 does not crossreact with rat gp130, we used neonatal mouse cardiomyocytes in this set of experiments. Similar to the rat system, murine cardiomyocytes responded to stimulation with CT-1, LIF, and PE with an induction of immediate early genes and typical morphology changes. However, CT-1, LIF, and PE did not induce ANF gene expression in murine cardiomyocytes (data not shown). This might relate to the fact that in mice, in sharp contrast to rats, the ventricular expression of ANF is not downregulated after birth, but instead remains at high levels from embryogenesis through adulthood (55). We first examined the effects of different concentrations of CT-1, LIF and PE on the induction of c-fos in murine cardiomyocytes. CT-1, LIF, and PE induced c-fos expression in a dose-dependent manner (Fig. 3, A  and B). In subsequent experiments, cardiomyocytes were incubated for 1 h with RX435 or, as a control, with immunopurified rat IgG prior to stimulation with CT-1, LIF, or PE. RX435 was used at a concentration that has previously been shown to inhibit ϳ80% of CT-1 (0.12 nM) binding to murine M1 cells (54). CT-1, LIF, and PE were used at concentrations that resulted in an approximately 4 -5-fold induction of c-fos expression. As shown in Fig. 3, C and D, RX435 completely abolished the induction of c-fos by CT-1 and LIF. By contrast, the c-fos in-duction by PE was not abolished by RX435 (although somewhat suppressed), confirming the specificity of the antibody (Fig. 3D).
To determine whether LIFR␤ is required for CT-1 signaling, we studied the effect of a LIFR␤ antagonist, hLIF-04, on c-fos induction by CT-1. hLIF-04 was generated by introducing three amino acid substitutions in the human LIF molecule, thereby abrogating the gp130 binding site. hLIF-04 has been shown to antagonize the action of hLIF in a Ba/F3 cell line, stably transfected with human gp130 and LIFR␤ (see "Experimental Procedures"). As shown in Fig. 3E, preincubation with hLIF-04 blocked the c-fos induction by human LIF, murine LIF, and by CT-1. By contrast, c-fos induction by PE was not affected by hLIF-04. Taken together, the results obtained with the anti-gp130 blocking antibody and the LIFR␤ antagonist strongly suggest that CT-1, like LIF, requires both gp130 and LIFR␤ to activate downstream signaling pathways leading to the induction of c-fos in cardiomyocytes.
CT-1 and LIF Induce Tyrosine Phosphorylation of gp130 and a 200-kDa Protein in Cardiomyocytes-To examine whether CT-1 would induce tyrosine phosphorylation of gp130, as it has been shown for the other members of the IL-6 cytokine family, cardiomyocytes were stimulated with soluble IL-6R, IL-6, a combination of IL-6 and sIL-6R, LIF, CT-1, and PE. Cell lysates were immunoprecipitated with a polyclonal anti-gp130 antiserum, analyzed by SDS-PAGE, and immunoblotted with an anti-phosphotyrosine antibody. As shown in Fig. 4A, the IL-6/ sIL-6R complex, LIF, and CT-1 induced gp130 tyrosine phos-3 M. Saito and T. Taga, unpublished data. phorylation. By contrast, no gp130 phosphorylation was detectable after stimulation with PE, and with IL-6 or sIL-6R alone. To compare the pattern of protein tyrosine phosphorylation induced by the various agonists, cell lysates were subjected to SDS-PAGE and immunoblotting with the anti-phosphotyrosine antibody (Fig. 4B). Stimulation with the IL-6/sIL-6R complex, LIF, and CT-1 resulted in the tyrosine phosphorylation of a protein, corresponding in size to gp130 (compare to Fig. 4A). An additional, higher molecular mass protein (approximately 200 kDa) was tyrosine-phosphorylated in response to LIF and CT-1 only, but not in response to the IL-6⅐sIL-6R complex.
CT-1 Induces a Morphologically Distinct Hypertrophic Response Characterized by the Assembly of Sarcomeric Units in Series-Neonatal rat ventricular cardiomyocytes respond to various hypertrophic stimuli by an increase in individual cell size, assembly of myofibrils, and perinuclear accumulation of ANF. To assess the effects of gp130/LIFR␤-dependent cytokines on these morphological features of a hypertrophic response, cardiomyocytes were treated with CT-1, LIF, or CNTF and dual-stained with an anti-TrpE/MLC-2v antiserum and a monoclonal anti-ANF antibody. Unstimulated cells and PEtreated cells served as negative and positive controls, respectively. The effects of CT-1, LIF, CNTF, and PE on cardiomyocyte area, length, and width and the percentage of cells displaying perinuclear accumulation of ANF are summarized in Table 1A. CT-1 and LIF induced a hypertrophic response, as evidenced by increases in cardiomyocyte area and the percentage of cells with perinuclear accumulation of immunoreactive ANF. As compared to PE stimulation, CT-1 and LIF had a more pronounced effect on cardiomyocyte length but did not significantly affect cardiomyocyte width. Treatment of cardiomyocytes with CNTF did not induce significant morphology changes as compared to control. Cardiomyocytes were dualstained with a monoclonal anti-␤MHC antibody and rhodamine phalloidine, to allow a simultaneous assessment of thick filament (␤MHC) and thin filament (F-actin) assembly (56). Images were obtained by confocal laser microscopy. Representative high power fields are depicted in Fig. 5. Cardiomyocytes stimulated with CT-1 (Panels E and F) and LIF (Panels G and H) displayed a high degree of sarcomeric organization: myofibrils were oriented along the longitudinal cell axis, and extended into the tips of the cytoplasmic projections. As compared with PE stimulation (Panels C and D), sarcomeric units were assembled predominantly in series, rather than in parallel (most striking in cytoplasmic projections containing only a few myofibrils (see Panels E-H)). As shown in Table 1B, the average sarcomeric length did not differ significantly among the experimental groups.
CT-1 Induces ANF Gene Expression-The reactivation of an embryonic pattern of gene expression is a hallmark of cardiomyocyte hypertrophy. We therefore determined whether stimulation of cardiomyocytes with CT-1, LIF, or CNTF would induce embryonic gene expression (Fig. 6). As shown previously (11), stimulation of neonatal rat ventricular cardiomyocytes with PE resulted in a significant (5.1-fold) increase in prepro-ANF mRNA levels. Likewise, treatment with CT-1 or LIF stimulated prepro-ANF mRNA expression (5.1-and 5.0-fold, respectively). By contrast, CNTF did not significantly affect prepro-ANF mRNA levels. As previously reported (8), PE stimulation resulted in an increase (4.5-fold) in skeletal ␣-actin mRNA expression. CT-1, LIF, and CNTF, however, did not significantly affect skeletal ␣-actin steady state mRNA levels. Finally, we examined the effects of CT-1, LIF, CNTF, and PE on the expression of MLC-2v, a constitutively expressed contract-

Cardiotrophin-1 Activates Cardiac Muscle Cell Hypertrophy
ile protein gene. PE stimulation resulted in a marginal 1.5-fold induction of MLC-2v mRNA levels. By contrast, CT-1, LIF, and CNTF did not induce MLC-2v mRNA expression (data not shown). We next determined whether the increase in prepro-ANF mRNA expression in response to CT-1 and LIF is accompanied by an increase in secretion of immunoreactive ANF. As noted in the initial study reporting the cloning of CT-1 (19), CT-1 induced ANF release from cardiomyocytes in a dose-dependent manner (Fig. 7). Likewise, LIF increased the secretion of ANF; maximum ANF release was observed with CT-1 and LIF concentrations of 0.1-10 nM. Consistent with the ANF Northern blot results, stimulation with CNTF did not induce ANF release at any of the concentrations tested (Fig. 7).
CT-1 Activates the Transcription of the ANF Gene-To determine whether the induction of ANF by CT-1 results from an increased ANF gene transcription, we performed nuclear run-on transcription assays. We compared the ANF gene transcription rate during stimulation with CT-1 and PE. A typical experiment is presented in Table II and Fig. 8. Cardiomyocyte nuclei were isolated at various time points during stimulation with CT-1 or PE. Unstimulated cells served as a control. CT-1 and PE increased the overall RNA transcription rate, as measured by the total amount of 32 P incorporation into purified run-on RNA (Table II). The overall RNA transcription rate was higher at the early time points (0, 6, and 12 h), i.e. shortly after serum withdrawal (Table II). An equal number of cpm from each run-on reaction was hybridized to denatured prepro-ANF and GAPDH cDNAs immobilized onto nylon filters. The ANF autoradiograms are shown in Fig. 8A. The CT-1 and PE induced ANF transcription rates were calculated as -fold induction over the corresponding control values from each time point (Fig. 8B). ANF transcription reached a maximum 2.0-fold (CT-1) and 2.1-fold (PE) increase versus control at 24 h and returned to control levels at 48 h. By contrast, GAPDH gene transcription was not affected by CT-1 or PE (not shown). The

Cardiotrophin-1 Activates Cardiac Muscle Cell Hypertrophy
CT-1-and PE-induced increase in ANF gene transcription was accompanied by a time-dependent accumulation of prepro-ANF mRNA in the cytosol, reaching maximum levels between 24 and 48 h (Fig. 8C). By contrast, CT-1 and PE did not affect GAPDH mRNA expression (not shown).
CT-1 Responsive Cis-regulatory Sequences Are Located Outside a 3003-bp ANF Promoter Fragment-The transcriptional up-regulation of the ANF gene in response to a number of G-protein coupled receptor agonists, including PE, is mediated by cis-regulatory elements, which are located within the proximal 3 kb of ANF 5Ј-flanking region (11,12,57). We performed transient transfection analyses to determine whether gp130/ LIFR␤-dependent cytokines and PE utilize common cis-acting elements within the ANF promoter region. Cardiomyocytes were transfected with an ANF-luciferase reporter construct containing 3003 bp of the rat ANF 5Ј-flanking region. Cardiomyocytes transfected with a RSV-luciferase construct or a promoterless luciferase construct were used as positive and negative controls, respectively. To control for transfection efficiency, cells were cotransfected with a ␤-galactosidase expression vector. The cells were then stimulated with PE, CT-1, LIF, or CNTF; unstimulated cells served as controls. At various time points, cardiomyocytes were harvested for luciferase and ␤-galactosidase assays. As previously reported (11), the 3003-bp ANF promoter fragment conferred PE-inducible expression to the luciferase reporter gene (maximum 23.7-fold induction at 48 h) (Fig. 9). By contrast, the 3003-bp ANF promoter fragment did not mediate inducible expression by CT-1, LIF, or CNTF ( Fig. 9). Luciferase activities in cells transfected with the RSVluciferase construct were on average 80-fold higher as compared to unstimulated cells transfected with the ANF-luciferase construct and were not induced by any of the treatments (not shown). Cells transfected with the promoterless reporter construct displayed luciferase activities that were not different from background values and were not affected by any of the treatments (not shown). Considering that CT-1 as well as PE increase the ANF gene transcription rate (as demonstrated by nuclear run-on assays), the transfection analyses suggest that, in contrast to PE, the CT-1 responsive cis-regulatory elements are located outside of 3 kb of ANF 5Ј-flanking region. Thus, divergent pathways appear to mediate the induction of the ANF gene in response to CT-1 as compared to ␣-adrenergic stimulation. DISCUSSION CT-1 Signaling through the gp130/LIFR␤ Heterodimer-Cardiotrophin-1 was recently isolated by expression cloning based on its ability to induce an increase in cell size in cardiomyocyte culture (19). The deduced amino acid sequence suggested that CT-1 is a novel member of the IL-6, IL-11, LIF, CNTF, and OSM family of structurally related cytokines, that trigger downstream signaling pathways in multiple cell types through the homodimerization of gp130 or the heterodimerization of gp130 and LIFR␤ (22,24,26,27). To determine whether CT-1 shares the receptor components gp130 and/or LIFR␤ with the other members of the IL-6 cytokine family, we employed a FIG. 8. CT-1-and phenylephrine-induced ANF gene transcription. Same experiment as described in Table II. 1.2 ϫ 10 6 cpm from each run-on reaction were hybridized to linearized and denatured pBluescript harboring the prepro-ANF cDNA, immobolized onto nylon filters. The filters were washed under stringent conditions, and exposed to x-ray film for 5 days (Panel A). The signal intensities were analyzed by densitometry to quantitate the amount of ANF gene transcription; data are presented as -fold induction over control (Panel B). Total RNA was isolated from the cytosolic fractions of the same cells, that were used for the run-on assays. RNA was subjected to Northern blot analysis (7.5 g/lane), using prepro-ANF and 28s cDNA probes (Panel C). Data from one experiment are presented. One additional experiment yielded comparable results.

TABLE II Total RNA transcription in control cells and CT-1-or phenylephrine-stimulated cells
Neonatal rat ventricular cardiomyocytes were plated into 15-cm dishes (10 ϫ 10 6 cells/dish), allowed to attach overnight, and switched to maintenance medium. Thereafter, nuclei from one dish were harvested (0 h, control). The remaining cells were further incubated with no additions (controls), 1 nM CT-1, or 100 M PE. Cardiomyocyte nuclei were isolated at the indicated time points, and nuclear run-on transcription assays were performed as described under "Experimental Procedures." The total amount of 32 P incorporation into purified run-on RNA was determined by liquid scintillation counting as an index of overall transcriptional activity. Data are presented as cpm ϫ 10 6 . Cardiotrophin-1 Activates Cardiac Muscle Cell Hypertrophy monoclonal anti-gp130 antibody and a mutated human LIF protein, acting as a LIFR␤ antagonist. The c-fos induction by CT-1 and LIF in cardiomyocytes was antagonized by the monoclonal anti-gp130 antibody as well as the LIFR␤ antagonist, strongly suggesting that CT-1, like LIF, utilizes the receptor components gp130 and LIFR␤. As demonstrated by gp130 immunoprecipitation and subsequent anti-phosphotyrosine immunoblotting, stimulation of cardiomyocytes with CT-1, LIF, and a combination of IL-6 and sIL-6R resulted in the rapid tyrosine phosphorylation of gp130. These data indicate that tyrosine phosphorylation of gp130 is an early step in CT-1 signaling, as it has previously been shown for the other members of the IL-6 cytokine family (21)(22)(23). As determined by immunoblotting with an anti-phosphotyrosine antibody, LIF induced the tyrosine phosphorylation of an additional ϳ200 kDa protein, which was not phosphorylated upon stimulation with the IL-6/sIL-6R complex. Based on previous results, this protein most likely corresponds to LIFR␤ (22,26,58). Stimulation of cardiac cells with CT-1 resulted in the tyrosine phosphorylation of a protein, indistinguishable in size from LIFR␤, as well. Considering the ability of the LIFR␤ antagonist to block the action of CT-1 in cardiomyocytes, the immunoblotting results strongly suggest that CT-1, like LIF, induces the tyrosine phosphorylation of LIFR␤. In support of this conclusion, CT-1 has recently been shown to bind to purified LIFR␤ in solution with about the same affinity as LIF, and to interact with a cell surface protein in M1 cells with a mobility expected for LIFR␤ (54). In summary, the present study indicates that CT-1 signals through and induces tyrosine phosphorylation of the gp130/LIFR␤ heterodimer in cardiomyocytes.

CT-1 Induces a Hypertrophic Phenotype in Cultured Myocardial Cells Characterized by Sarcomere Assembly in Series and a Selective Up-regulation of ANF Gene Expression-
The present study provides clear evidence that the CT-1 induced hypertrophic phenotype is distinct from the hypertrophic phenotype observed following G-protein-dependent stimulation. On a single cell level, G-protein-dependent pathways induce a form of hypertrophy with a relatively uniform increase in myocyte size and the addition of new myofibrils in parallel (12,15,16). By contrast, as noted in the present study, the gp130/LIFR␤-de-pendent cytokines CT-1 and LIF induce an increase in myocyte size characterized by a marked increase in cell length, but little or no change in cell width. To characterize the effects of gp130/ LIFR␤-dependent stimulation on the myofibrillar cytoarchitecture, cardiomyocytes were dual-stained against thick (␤MHC) and thin (F-actin) myofilaments, and viewed by confocal laser microscopy (56). Cardiomyocytes stimulated with CT-1 and LIF displayed a high degree of myofibrillar organization; myofibrils were organized in a strictly sarcomeric pattern, were oriented along the longitudinal cell axis, and extended to the cell periphery. Importantly, the increase in cell size and length was associated with no change in the average sarcomere length, strongly suggesting that the cell elongation in response to gp130/LIFR␤-dependent stimulation results from an addition of new sarcomeric units in series. The morphologic changes induced by gp130/LIFR␤-dependent stimulation in vitro are reminiscent of the changes observed in cardiac myocytes isolated from hearts subjected to chronic volume overload (3,4). By contrast, the pattern of cardiomyocyte hypertrophy induced by ␣-adrenergic stimulation more closely resembles a pressure overload-like phenotype (1,2).
On a molecular level, gp130/LIFR␤-dependent stimulation and ␣-adrenergic stimulation resulted in distinct patterns of embryonic gene, MLC-2v, and immediate early gene expression. Stimulation of cardiomyocytes with CT-1 and LIF induced prepro-ANF mRNA expression, and perinuclear accumulation and secretion of immunoreactive ANF. However, in contrast to ␣-adrenergic stimulation, CT-1 and LIF did not induce skeletal ␣-actin expression. Growth factors, signaling through G-protein coupled receptors induce ANF and skeletal ␣-actin in a coordinate fashion (8,(11)(12)(13)(14). A recent study compared the expression pattern of distinct members of the embryonic gene program in pressure overload versus volume overload hypertrophy in vivo in the rat myocardium (5). As shown previously (52), pressure overload resulted in the coordinate induction of ANF and skeletal ␣-actin. However, volume overload hypertrophy was associated with a selective increase in ANF expression, and no induction of skeletal ␣-actin, suggesting that the regulation of distinct embryonic genes in vivo is related to the hypertrophic stimulus (5). The pattern of embryonic gene expression induced by CT-1 and LIF in cardiomyocyte culture therefore resembles the pattern observed in volume overload hypertrophy in vivo.
Nuclear run-on transcription assays revealed that the induction of prepro-ANF mRNA by CT-1 and ␣-adrenergic stimulation is mediated, at least in part, by an increased transcription of the ANF gene. The fact that the ANF mRNA levels increased 5-fold, whereas the ANF gene transcription rate increased only 2-fold, suggests that post-transcriptional mechanisms(s) participate in the up-regulation of ANF gene expression in response to CT-1 and PE (59). In agreement with previous results (11), a 3003-bp ANF promoter fragment conferred PE-inducible expression to a luciferase reporter gene. Likewise, the cisregulatory elements mediating ANF inducibility in response to endothelin 1 and ␣-thrombin reside within the proximal 3 kb of the ANF 5Ј-flanking region, suggesting that G-protein-coupled receptor agonists may utilize common signaling pathways for the induction of ANF (12,57). By contrast, none of the gp130/ LIFR␤-dependent cytokines (CT-1, LIF, and CNTF) activated the 3003-bp ANF promoter fragment. Considering the increased ANF gene transcription rate in response to CT-1 and PE, we conclude that in contrast to ␣-adrenergic stimulation, the CT-1 responsive cis-regulatory elements are located outside of the proximal 3 kb of the ANF 5Ј-flanking region. In recent studies, transgenic mice harboring 3003 bp of ANF 5Ј-flanking region upstream of a luciferase reporter gene were subjected to transverse aortic banding. Despite a marked increase in left ventricular prepro-ANF mRNA expression, no significant increase in reporter activity was observed (60). These findings demonstrate that the induction of ANF during cardiac hypertrophy in vivo occurs through a distinct mechanism as compared to G-protein-mediated cardiomyocyte hypertrophy in vitro. Unraveling the mechanisms of ANF induction in response to gp130/LIFR␤-dependent stimulation may therefore provide insight into the mechanisms governing the induction of ANF during in vivo cardiac hypertrophy.
CT-1 and LIF did not induce MLC-2v mRNA expression, whereas phenylephrine stimulation resulted in a marginal induction of MLC-2v. The induction of MLC-2v on the mRNA level may be a feature unique to G-protein-coupled receptor stimulation of cardiomyocytes in vitro (10,12): MLC-2v mRNA expression is not up-regulated in the hypertrophic right ventricles of pulmonary artery-banded mice (61) and in the hypertrophic ventricles of transgenic mice overexpressing a constitutively active ␣-adrenergic receptor in the myocardium (62).
A distinct response to gp130/LIFR␤-dependent versus ␣-adrenergic stimulation was also observed at the level of immediate early gene induction. As compared to ␣-adrenergic stimulation, CT-1 and LIF induced Ͼ10-fold higher levels of c-myc and tis11 expression. Immediate early genes encode known or putative transcription factors, and potential binding sites have been identified in the promoter regions of a number of cardiac genes (for review, see Ref. 63). The distinct combinatorial expression of immediate early genes may therefore relate to the differences in cell morphology and gene expression pattern in response to gp130/LIFR␤-dependent versus ␣-adrenergic stimulation.
Taken together, these studies indicate that gp130/LIFR␤-dependent pathways can activate a hypertrophic phenotype in cardiac muscle cells which is distinct both at a morphological and a molecular level from the phenotype seen following Gprotein-dependent stimulation, and that is more consistent with a volume overload, as opposed to pressure overload, hypertrophic phenotype. Future studies in transgenic mice will be required to rigorously test whether gp130/LIFR␤-dependent signaling pathways are implicated in volume overload and/or pressure overload ventricular hypertrophy in vivo, in a manner analogous to recent studies exploring the in vivo role of ras-dependent pathways (18). In this regard, double transgenic mice overexpressing IL-6 and IL-6R have recently been generated (64). These mice display a constitutive tyrosine phosphorylation (i.e. activation) of gp130 in the myocardium and left ventricular hypertrophy with increases in cardiac weights and cardiomyocyte volumes. These results suggest that the induction of cardiomyocyte hypertrophy through gp130-dependent signaling pathways is not confined to the in vitro hypertrophy assay, but may also be observed in the in vivo context.
The question arises as to the relative role of individual members of the IL-6 cytokine family in the activation of a hypertrophic response in vivo. The expression of IL-6R in the heart of wild type mice is extremely low, making IL-6 an unlikely mediator of cardiac hypertrophy in vivo. 4 Accordingly, a combination of IL-6 with sIL-6R was required to induce tyrosine phosphorylation of gp130 in cardiomyocytes in the present study. Likewise, the myocardium does not express substantial levels of CNTFR␣ (30), and cardiomyocytes did not respond to CNTF stimulation in the present study. It should be noted, however, that CNTF exerted marginal effects on immediate early gene expression in cardiomyocytes. This might relate to a weak interaction of CNTF with the gp130/LIFR␤ heterodimer in the absence of CNTFR␣ (65,66). Alternatively, low level expression of CNTFR␣ in cardiomyocytes may account for a weak responsiveness to CNTF. The CT-1 mRNA is expressed in the heart from adult mice (19). RNase protection analysis of cardiac muscle cells and cardiac-derived fibroblasts from neonatal rats reveals a Ͼ10-fold higher level of CT-1 mRNA expression in cardiac muscle cells versus cardiac fibroblasts, and CT-1 is expressed primarily in cardiac muscle cells during murine cardiogenesis (67). By contrast, the expression of LIF mRNA in the adult murine heart is extremely low as documented by RNase protection analysis (68,69). These data would suggest that CT-1 is a candidate cytokine activating gp130/LIFR␤-dependent signaling pathways within the myocardium, potentially through an autocrine pathway. The generation of mice containing a null-mutation of the CT-1 gene should provide a more definitive understanding of the in vivo role of CT-1 in development and in cardiac hypertrophy in the adult animal.