THE JOURNALBIOLOGICAL OF CHEMISTRY Q 1992 by The American Society for Biochemistry and Molecular Biology. Inc.

Vol. 267, No. 15, Issue of May 25, pp. 10551-10560,1992

Printed in U.S.A.

Molecular Characterizationof the Stretch-induced Adaptation of Cultured Cardiac Cells

AN I VITRO MODEL OF LOAD-INDUCED CARDIAC HYPERTROPHY* N
(Received for publication, December 31, 1991)

Jun-ichi Sadoshima$ll**, Lothar Jahn$1$$, Toshiyuki TakahashiST, Thomas J. KulikQII and , Seigo Izumo$lQ#
From the $Molecular Medicine Unit and CardiovascularDivision, Beth Israel Hospital, the Department of Cardiology, Childrens Hospital, and the Departments of lMedicine and IIPediatrics, Harvard Medical School, Boston, Massachusetts 02215

Although it is a well-known fact that hemodynamic plays a critical role in determining muscle mass andits load is a major determinant of cardiac muscle mass and phenotype in both cardiac and skeletal muscles in uiuo. Howits phenotype, little is known as to how mechanical ever, in intact animals it is extremely difficult to isolate and signals of gene evaluate the role of external load in the regulation of muscle load is converted into intracellular regulation. To address thisquestion, we characterized gene expression. The lack of a well-characterized in vitro the stretch-induced adaptation of cultured neonatal model of load-induced cardiac hypertrophy has hampered a further mechanistic investigation in this area, and little is cardiocytesgrown on a stretchablesubstratein serum-free medium. Static stretch (20%) of the cells known as to how mechanical load is initially converted into was applied without cell injury. Stretch caused hypertrophy in myocytes and hyperplasia in non-myocytes. intracellular signals of gene regulation (reviewed in Refs. 2, Stretch caused an induction of immediate-early genes 3). The firstdirect evidence that muscle cells are able to such as c-fos, c-jun, c-myc, JE, and Egr-1,but not sense the external load, in the absence of neuronal or Hsp70. Immunostaining showed that the stretch-innucleus of both myo- hormonal factors, came from a study by Vandenburgh and duced Fos protein localized in the of protein cytes and non-myocytes. Nuclear extracts from Kaufman (4) who demonstrated thattherate stretched myocytes contained DNA binding activity to synthesis of cultured chick skeletal muscle cells grown on 1 the AP- 1 and Egr- consensus sequences. In myocytes, elastic substrate increased significantly in response to static the induction of immediate-early genes was followed stretch of the substrate. A similar phenomenon was observed by expression of fetal genes such as skeletal a-actin, in adult cardiocytes plated on silicone sheet (5). More reatrial natriuretic factor, and &myosin heavy chain. cently, it has been shown that stretchof neonatal cardiocytes DNA transfection experiments showed that the in culture caused induction of c-fos protooncogene and skelestretch-response element of the c-fos gene promoter tal a-actin (6, 7), a phenotype also observed in the myocaris present within 366 base pairs of the 6flanking dium in response to pressure overload in uiuo (8).These region, whereas that of the atrial natriuretic factor studies suggest that stretching cardiocytes in vitro may be a and the &myosin heavy chain genes is probably located outside of 3412 and 628 base pairs of the 5-flanking suitable experimentalsystem with which to address the quesregion, respectively. These results demonstrate that tion of how mechanical load is transduced into intracellular the phenotype of stretched cardiocytes in this in vitro signals regulating gene expression. However, since previous analysis of the rateof protein model closely mimics that of hemodynamic load-in- studies have been limited to the duced hypertrophy in vivo. This model seems to be a and RNA synthesis and expression of c-fos and skeletal asuitable system with which to dissect the molecular actin genes (5-7), the appropriateness of this system as a mechanisms of load-induced hypertrophy of cardiac model with which to dissect molecular mechanisms of loadinduced hypertrophy can only be determined after further muscle. characterization of the system. For example, it has not been demonstrated whether or not the stretch-induced c-fosgene expression and increase in Hypertrophy, an increase in cell size without cell division, protein synthesis in cultured cardiocytes is a result of cell is a fundamental adaptive process employed by post-mitotic injury, rather than that of physiological stimulus. This is an muscle cells (1). It is a well-known fact that external load important question because the c-fosgene is known to be induced by a variety of nonspecific stimuli, such as cell injury * This work was supported in part by grants from the Whitaker or cell dispersion by trypsintreatment (9). Furthermore, Foundation and the National Institute of Health. The costs of pub- repetitive stretch of skeletal myotubes in culture is known to lication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertise- be associated with an initial cell injury (10).Thus, itis critical ment in accordance with 1 U.S.C. Section 1734 solely to indicate to determine whether or not the cardiocyte stretch model is 8 an injury repair model. this fact. ** Fellow of Human Frontier Science Program. Second, pressure overload in vivo causes hypertrophy of $$ Supported by a Fellowship in Basic Science Research at the myocytes and proliferation of non-myocytes (11).Smce neoHarvard-Thorndike Laboratory. 3s Established Investigator of the American Heart Association. To natal myocytes still retain some capacity to divide (12), it is whom correspondence should be addressed Molecular Medicine Unit, important to determine whether this in uitro stretch model Beth Israel Hospital, 330 Brookline Ave., Boston, MA 02215. Te1.l represents hypertrophy or hyperplasia of neonatal myocytes. 617-735-2656; 617-735-2913. Fax: It is also important to determinewhether stretch affects the

10551

10552

Stretch-induced Hypertrophy Cardiac

in Vitro
These are some examples of questions that have not been resolved by previous studies. In order to address these questions, we have performed a systematical analysis of the phenotype of rat neonatal cardiocytes undergoing stretch in vitro. The results demonstrate that the phenotype of this in vitro model closely mimics that of load-induced cardiac hypertrophy in intact animals, but there are a few important differences. Using this model, we have determined whether the transfected promoters of the c-fos, ANF, and /3-MHC genes are responsive to stretch inventricular cardiac myocytes.
EXPERIMENTAL PROCEDURES

growth of non-muscle cells in vitro. Third, theimmediate-early (IE)' genes induced by pressure overload in vivo are not limited to c-fos, but include a large number of genes such as c-jun, jun B, Egr-1, c-myc, C-Haras, nur77, and a major heat shock protein, Hsp7O (8 and reviewed in Ref. 13). It is not known whether these genes are also induced by in vitro stretch. It would be interesting to examine this because there have been several examples suggesting that the profile of IE gene expression is stimulusspecific, and distinct combinatorial expression of IE genes might confer phenotypic specificity in the cellular response to different stimuli (14, 15). Fourth, most previous studies concerning IE gene expression in cardiocytes have been done only at the mRNA level and not at the protein level. Recently, Roux et al. (16) reported that in the absence of serum, Fos protein could not be translocated into nucleus and stayed in cytoplasm in rat embryonic fibroblast and mouse fibroblast cell lines. If the same is true for cardiocytes, which are usually cultured in serum-free media, it will have important implications for our understanding of the roles of c-fos in cardiac hypertrophy because it would suggest that the c-fos gene product may not be necessary for cells to develop the hypertrophic phenotype. Thus, it is important to determine whether Fos protein is localized in the nucleus after the stretch stimulus, and whether there is an inducible DNA binding activity in thenuclear extracts to the target sequence of Fos/Jun proteins. Fifth, following induction of IE genes, expression of several genes are known to change in response to pressure overload in the rat ventricle. These include induction of the P-MHC, ANF, skeletal a-actin, smooth a-actin, &tropomyosin and Btype creatine kinase, as well as down-regulation of the SR Ca2+-ATPase(reviewed in Refs. 1, 13). A previous study has shown that theexpression of skeletal a-actin is up-regulated by in vitro stretch (7).However, skeletal a-actin gene expression may not be the best marker for hypertrophy because induction of this gene in vivo is usually transient (lasting 1 week) and is often no longer up-regulated in well-established hypertrophy (8, 17, 18). In contrast, up-regulation of the 8MHC and ANF genes persists in the chronic phase of hypertrophy (8, 13, 19). Therefore, it would be important to determine whether these "stable late markers" of hypertrophy are also expressed in the in vitro stretch model. Finally, the skeletal a-actin gene contains multiple CC(A/ T),GG motif(CArG box) in the proximal promoter (20), which is essential for the basal as well as growth factorinducible expression of this gene in cardiac myocytes (21). This element is very similar to the serum response element (SRE) of the c-fos geneand cross-competes for the binding of the serum response factor to SRE (22, 23). The factor that binds to the CArG box of the skeletal a-actin gene is apparently identical to serum response factor (24). It has been suggested that the activation of the c-fos and skeletal aactingenes by stretch may occur by a common mechanism (7). At present, however, a link between IE gene expression and the late phenotypic changes remains highly conjectural. Therefore, it would be interesting to examine whether other promoters that do not contain the CArG box or SRE, such as the /3-MHC and the ANF genes (25,26), arealso up-regulated by stretch.
The abbreviations used are: IE, immediate-early gene; ANF, atrial natriuretic factor; BrdU, bromodeoxyuridine; CAT, chloramphenicol acetyltransferase; FCS, fetal calf serum; MHC, myosin heavy chain; PBS, phosphate-buffered saline; kb, kilobase(s); bp, base pair(s); SR, serum response; SRE, serum response element; RSV, Rous sarcoma virus; PMSF, phenylmethylsulfonyl fluoride; HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid.

Materials-Collagenase Type IV was purchased from Sigma. Silicone sheet (0.005- or 0.01-inch thick) was Silastic sheet from Dow Corning. Rat tail collagen was from Biomedical Technologies. The air brush used for collagen coating was from Badger Air-Brush. All other culture reagents were purchased from GIBCO. All radiochemicals were obtained from Du Pont-New England Nuclear. Anti c-Fos antibody was from Medac Molecular Biology (Hamburg, Germany). Anti c-Jun antibody was from Oncogene Science. All other chemicals were from Sigma. In Vitro Stretch Deuice-The stretch device we used was a modification of the system originally developed by Vandenburgh and Kaufman (4) for the embryonic skeletal muscle cells. Silicone sheet (50 X 73 mm') in a stretching frame was finely coated with rat tail collagen dissolved in 0.02 N acetic acid (0.25 mg/ml) using an air brush. The stretching frame was put in a notched Plexiglas support fixed ontoa 150-mm culture dish (Fig. lA).Uniaxial strain was applied by stretching the silicone sheet in the frame by 10 or 20% in the length along the Plexiglas support. Ten and 20% stretch of the frame gave mean stretch of10.6 and 20.0% of the silicone sheet, respectively, which was obtained from average of the substrate length at various points measured in control and stretched states. In the experiments of [3H]phenylalanine incorporation and I3H[thymidine uptake, we used smaller version of the silicone sheet (15 X 20 mm'). In the present experiments, we used 20% stretch unless otherwise indicated. To eliminate differences due to substrate materials, static control cells or serum-stimulatedcells were also grown on the silicone sheet. Preparation of Myocyte-rich Culture (Hereafter CalledMyocyte Culture)-A diagram of experimental protocol is shown in Fig. 1B. Primary culturesof cardiac myocytes were prepared using a variation of the method described by Simpson et al. (27) and Orlowski and Lingrel (28). Hearts were removed from 1-day-old Wistar rats anesthetized by ether underaseptic conditions. The ventricles were minced into 1-3-mm3 fragments. Digestion was performed by five to six 15-min periods of incubation at 37 "C with HEPES-buffered saline solution (mM): 20 HEPES-NaOH, pH 7.6, 130 NaCl, 3 KCl, 1 NaH2P04,4 glucose, 3.3 p~ phenol red containing 0.1% collagenase IV, 0.1% trypsin, 15 pg/ml DNase I, and 1.0% chicken serum at 37 "C. At the end of each cycle, the supernatant was stored on ice after addition of calf serum (10% v/v) to neutralize trypsin. The dissociated cells were collected by centrifugation and resuspended in Dulbecco's modified Eagle'smedium/F-12 (GIBCO) (l:l, v/v) suppleM mented with 5% horse serum, 3 m pyruvic acid, 100 p M ascorbic acid, 1 pg/ml insulin, 1 pg/ml transferrin, 10 ng/ml selenium, and 100 pg/ml ampicillin. To selectively enrich for myocytes, dissociated cells were preplated for 1 h, during which period the non-myocytes attached readily to the bottom of the culture dish. The resultant suspension of cardiocytes was plated ontothe collagen-coated silicone sheet a t a density of 1 X lo5 cell/cm'. Bromodeoxyuridine (BrdU) (0.1 mM) was added during the first 24-36 h to prevent proliferation of non-myocytes except for the cultures used for thymidine uptake measurement. The culture medium waschanged 24-36 h afterseeding to a defined serum-free Dulbecco's modified Eagle's mediump-12 medium which had the same composition as described above, except that 5% horse serum and BrdU were not added. To minimize the effect of remnants of serum, cultured cells were rinsed twice with the serum-free medium. In the serum-free culture medium, we did not observe proliferation of non-myocytes even in the absence of BrdU. Using this method, we routinely obtained contractile myocardial cell cultures with -90-95% myocytes, as assessed by microscopic observation of cell beating and by immunofluorescence staining with a monoclonal antibody againstsarcomeric myosin heavychain (MF 20) (29). All experiments were done 24-36 h afterchanging to theserum-

Stretch-induced Cardiac Hypertrophy in Vitro

10553

A
20% Stretch

Dispersion d cardiac ventricles (myocvto and ron-myocyte)

t
preplaling on plastic dlsh in5% HS for 1h

r
Adhecent calls Grown to 2ps~agesinlO%HS" plsted silicone wbaralo R i m twice with seturn free medium Medium change to serum free mediumt

\
"
"adherent
plated on siliccne substrate in 5% HS + BrdU. for 24 h Medium chmge to remove d " p

t t Connuence on

inlO%HS fCf481072h

t t

cslis

in5%HS+BrdU furl2h

R twice with serum free medium m i M i u r n change to m u m tree medium+

for24h froctlon

fCf24h fr8ction

Non-myocyto-rich

Yyocyto-rlch

t
Strotch oxp8rimonts

t
Strotch oxporimonto

FIG. 1. Scheme of stretch culture dish (A) and protocol of cell preparation procedures (myocytes and non-myocytes) ( B ) .A, Scheme of stretch culture dish details of the device were described in the text. Stretch was applied by gently pulling the frame t o the direction indicated by the arrow. It took 2-3 s for the stretch t o be accomplished. Morphology and beating of cardiocytes could be observed through the silicone membrane by microscopy. a, 53 mm; b, 75 mm. B , diagram of the cell preparation procedures (myocytes and non-myocytes). *BrdU was omitted in [3H]thymidine uptake experiment. **BrdUwas not used. ?The serum-free medium did not contain BrdU. HS, horse serum. free medium. Lactate dehydrogenase and creatine phosphokinase were measured by an IL Monac Auto Analyzer (Instrumental Laboratory). Preparation of Non-myocyte-rich Culture (Hereafter Called Nonmyocyte Culture)-Highly enriched cultures of non-myocytes were prepared by two passages of cells adhered to the culture dish during the preplating procedure (Fig. 1B). Until the second passage, cells were maintained in the same culture medium as above except that 10% horse serum was used and BrdU was not used. After the second passage, the same serum-free medium as above was used. Isolation and Analysis of RNA-Total cellular RNA was isolated from cardiocytes dissolved in 4 M guanidium thiocyanate, followed by centrifugation through 5.7 M cesium chloride solution (30). Aliquots of total cellular RNA were size-fractionated by 1%agarose gel electrophoresis, transferred to nitrocellulose membranes, and hybridized with cDNA probes labeled with [32P]dCTP(3000 Ci/mmol) by random priming (30). The following probes were used for Northern blot analysis in this study: 1) c-fos: 2.1-kb EcoRI fragment of the mouse c-fos cDNA clone pGEMfos3 (a gift from J. G . Belasco and M. E. Greenberg, Harvard Medical School); 2) glyceraldehyde-3-phosphate dehydrogenase: a 1.3-kb PstI fragment of the rat glyceraldehyde-3phosphate dehydrogenase cDNA (31); 3) c-jun: a 2.1-kb fragment of the mouse c-jun cDNA (32); 4) Egr-1 (zif/268): a 2.8-kb fragment of the mouse zif/268 (= Egr-1) cDNA (33); 5) c-myc: the mouse cDNA clone pG2 myc5 (gift from C. Stiles, Dana-Farber Cancer Institute); 6) JE: the mouse cDNA clone pJE (34); 7) Hsp70 the rat cDNA clone pDPF (35); 8) skeletal a-actin:asynthetic oligonucleotide complementary to the first 57 nucleotides of the 3"untranslated

region of the mouse skeletal a-actin cDNA (36); 9) ANFa synthetic oligonucleotides (84-nucleotides long) complementary to the entire coding sequence of the rat ANF (37). The filters were washed under appropriate conditions of stringency. The relative amounts of specific mRNA were quantitated by laser densitometry of the corresponding autoradiograms in the linear response range of the x-ray films. The hybridization signals of specific mRNA were normalized to those of glyceraldehyde-3-phosphate dehydrogenase mRNA to correct for differences in loading and/or transfer. The levels of glyceraldehyde-3phosphate dehydrogenase mRNA were not significantly affected by stretch (not shown). To quantitate the relative levels of the a- and 8-MHC mRNAs, S1 nuclease mapping was performed using the 3' end PstI fragment of the rat &MHC cDNA clone pCMHC5 as described (38). Incorporation of [3H]Phenylalanine-As an index of protein synthesis, [3H]phenylalanine incorporation was measured as described (39). Cells were grown on 15 X 20-mm2 collagen-coated silicone membranes. After incubation in serum-free medium for24 h, the cells were stretched or stimulated with 20% fetal calf serum (FCS) for various periods of time with [3H]phenylalanine (10 pCi/ml) and unlabeled phenylalanine (0.36 mM) in the medium. For controls, cells were harvested at comparable times without stretch. The cells were washed with phosphate-buffered saline (PBS), and 10% trichloroacetic acid was added a t 4 "C for 60 min to precipitate protein. The precipitate was washed three times with 95% ethanol, then resuspended in 0.15 N NaOH. Aliquots were counted by a scintillation counter. The results were expressed as counts/minute/microgam of total protein, or counts/minute/dish. Protein concentration was determined by the method of Lowry (40), using bovine serum albumin as standard. Incorporation of r3H]Thymidine and Cell Count-Cells were grown on a 15 X 20-mm2collagen-coated silicone sheet without BrdU (Fig. 1B). Even in this condition, we generally observed less than 10% non-myocytes in myocyte culture preparations, as measured by cell beating and immunostaining with MF 20. This is probably due to high plating density (1 X 10' cells/cm2) and the use of horse serum (5%)which contains very low amounts of growth factors. After a 24h culture in the serum-free medium, cells were stimulated by stretch or by addition of 20% FCS. After 18 h, [3H]thymidine(5 pCi/ml) was added for 6 h. Cells were then washed with PBS and harvested with 10% trichloroacetic acid. Trichloroacetic acid-precipitable counts were measured as above. For cell counting, each dish was rinsed three times with PBS. Cells were detached with 1ml of trypsin-EDTA and dissociated by trituration. Cell counts were performed using a hemacytometer. Each sample was counted three times. Protein Content-Protein contents were measured using the methods of McDermott et al. (41). Each dish was rinsed three times with PBS. The cell layer was scraped with 1 ml of 1 X standard sodium citrate containing 0.25% (w/v) sodium dodecyl sulfate and frozen at -20 "C. Prior to use the extracts were thawed and vortexed extensively. Total cell protein was assayed by the method of Lowry et al. (40). The results were normalized by DNA content, which was measured fluorometrically in aliquots of each extract by the method of Labarca and Paigen (42) using calf thymus DNA as a standard. P h m i d Constructs and DNA Transfection-The mouse c-fos CAT (chloramphenicol acetyltransferase) construct was a generous gift of Dr. M. Gilman (Cold Spring Harbor Laboratory) (43, 44). The rat ANF CAT constructs were from Dr. C. E. Seidman (Brigham and Women's Hospital, Boston) (45). The rabbit OMHC CAT constructs were gifts from Dr. N. Shimizu (Beth Israel Hospital, Boston) (46) andtherat BMHC constructs from Dr. V. Mahdavi (Children's Hospital, Boston) (25). Transfection was performed by the calcium phosphate precipitationmethod (30). 15 pg of one of these DNAs and 5 pg of the RSV 8-galactosidase expression plasmid were transfected/ dish. Glycerol shock was performed 8 h after transfection. 24 h after transfection, the cells were stretched for the time indicated and harvested. Cell extracts were made by repeat freeze/thaw cycles, and the protein concentration of the extracts was measured. Equal amounts of protein extracts (100 pg) were incubated with [3H]chloramphenicol and butyryl-CoA. The butylated chloramphenicol was separated by a repeated phase extraction using xylene (47). Aliquots were counted by scintillation counter. To correct for transfection efficiency, @-galactosidaseactivity of cell extracts was measured as described (30). The activity of the RSV promoter, normalized by the luciferase activity of the cotransfected cytomegalovirus-luciferase construct, was not influenced significantly by stretch (1.16 & 0.1-fold, stretch uersus control). Extraction of Nuclear Protein-Nuclear proteins were extracted

10554

Stretch-induced Hypertrophy Cardiac

in Vitro
rable (20 f 1%) increase in cell length (or width) in the direction of stretch, regardless of the orientation of the cells (Fig. 2b). The cells returned to their prestretch length upon the release of stretch (Fig. 2c). A similar result was observed when confluent cells were stretched by 20% (data not shown). In order to examine whether stretch causes cell injury, we measured release of lactate dehydrogenase or creatine kinase in the culture mediumfollowing a 20% staticstretch of confluent cells. As shown in Fig. 3, there was no increase in release of lactate dehydrogenase or creatine kinase by stretch, either in the myocyte or in the non-myocyte culture. Thus, static stretch of the substrate membrane resulted in a comparable strain of adherent cells without causing significant cell injury. Stretch Causes True Hypertrophy Cardiac Myocytes-We of next examined whether stretch causes hypertrophy or hyperplasia of cardiac myocytes. Stretch of the cardiac myocytes caused an increase in protein synthesis in a time-dependent fashion, as measured by [3H]phenylalanine incorporation (Fig. a), in agreement with previous studies (5,7).In skeletal muscle cells,stretch in serum-free medium causes an increase in protein synthesis, but the protein content of the cells did not increase due to a concomitant increase in protein degradation (51). Serum factors were necessary for skeletal muscle cells to develop true hypertrophy in response to stretch (51). Whether stretch in serum-free medium causes true hypertrophy of cardiac myocytes has not been addressed in previous studies (5-7). Therefore, we measured the protein content and rate of DNA synthesis of cardiac myocyte culture in response to stretch. As shown in Fig. 4B, stretch caused a significant increase in protein content in serum-free medium. In contrast, stretch did not significantly increase the rate of DNA synthesis, as measured by [3H]thymidineuptake during 24 h (1.10-fold, n = 4, Fig. 4C). As a positive control, cells were stimulated by 20% FCS,whichcaused a significant increase in [3H]thymidineuptake (3.78-fold, n = 4, p < 0.001 uersus control), indicating that these neonatal cells are capable of DNA synthesis upon mitogenic stimulation. The effects of stretch are very similar to those of pressure overload on heart muscle cells in vivo, that is, increased load causes an increase in protein synthesis and protein content without an increase in DNA synthesis (11).

using a variation of the method of Dignam et al. (48). Cells were lysed directly on the culture dish in ml of cold lysis buffer (0.6% Nonidet 1.6 M M P-40,0.15 M NaCl, 10 m Tris, pH7.9,and 1m EDTA), transferred into a 2-ml Eppendorf tube, and incubated for 5 min on ice. The nuclei were pelleted (1250 X g, 4 "C, 5 min), and nuclear proteins were extracted from the pelleted nuclei in an equal volume of cold M M M extraction buffer (500 m HEPES, pH 7.9, 0.5 m EDTA, 0.75 m M M M MgC12,500 m KCl, 1 m dithiothreitol, 0.1 m phenylmethylsulfonyl fluoride (PMSF), 2 pg/ml leupeptin, 5 pg/ml aprotinin, and 12.5% glycerol) on ice for 20 min, and cellular debris was removed by centrifugation for 5 min a t 4 "C and 1250 X g. The supernatant containing nuclear proteins was dialyzed against buffer containing 10 M M mM Tris-HC1, pH 7.9, 1 m EDTA, 5 mMMgC12, 10 m KCl, 10% M M glycerol, 1 m dithiothreitol, 0.1 m PMSF. The entire procedure was carried out a t 4 "C. The samples were stored a t -85 "C. DNA Mobility Shift Assay-A double-stranded synthetic oligonucleotide containing the consensus sequence of AP-l(49) or Egr-1 (50) was labeled by [y3*P]ATP and polynucleotide kinase and was purified by a 20% polyacrylamide gel. Samples of 15 pl containing 5 pg of nuclear extract were incubated with 3 pg of poly(d1-dC) and 1 ng of probe (20,000 cpm) in the presence or absence of competitor oligonucleotides or antibodies for 20 min a t room temperature. Before adding probe antibodies were preincubated with nuclear extract for 15 min at room temperature. Binding reactions were electrophorated on a 4% non-denaturing polyacrylamide gel and visualized by autoradiography. Immunohistochemistry-Cells grown on silicone membranes were permeablized with methanol (-20 "C) for 5 min followed by a short dip in acetone (-20 "C). After air drying the cells were incubated with primary antibodies for 30 min in a humidified chamber, washed three times in PBS, pH 7.4, and exposed to thesecondary antibodies for 30 min. After another three washes in PBS, the cells were rinsed in deionized water, dehydrated in absolute ethanol, and mounted in mowiol (Polysciences, Warrington,PA).For double label experiments, both primary and secondary antibodies were applied (doubleconcentrated) simultaneously. Rabbit serum 456 against c-Fos was from Medac. Monoclonal antibody MF 20 against sarcomeric myosin was a generous gift of Dr. David Bader (Cornel1 University) (29). Secondary antibodies were fluorescein isothiocyanate-conjugatedor Texas Red-coupled goat antibodies to immunoglobulins of rabbit or mouse (Jackson, West Grove, PA). Statistics-Data are given as mean f S.E. Statistical analysis was performed using analysis of variance and unpaired Student's t test, as appropriate. Significance was accepted a t p < 0.05 level.
RESULTS

Substrate Stretch Causes Cell Deformation withoutCell Injury-We first examined whether stretching thesilicone substrate actually results in cell stretch or rather causes a "slippage" of the cellsfrom the substrate. Fig. 2a shows rat neonatal cardiac myocytes(plated sparsely in this experiment for precisecelledge detection) observedusing differential interference microscopy prior to stretch. A longitudinal stretch of the silicone substrate by 20% resulted in a compa-

Time (h)

Time (h)

FIG. 3. Effect of stretch on release of lactate dehydrogenase (LDH) a n d creatine kinase ( C P K )into culture media in myocyte (left)and non-myocyte(right) culture. 8 X lo6 cells were plated on a 53 alent strain in cultured cardiac cells. Cellswere plated on a collagen-coated silicone sheet a i low density. The silicone substrate was stretched in the direction indicated by the long arrow in b by 20%. Pictures of the same field taken before (a), during ( b ) ,and after (c) release of the stretch are shown. In this example, the cells had been stretched for 60 min before letting them to relax. The cells are highlighted by short arrows. Note the increase in the cell length with stretch (b). The mean increase in cell length with stretch was 20 f 1%. Theelongation of the cells by stretch was reversible ( c ) and did not cause detachment of the cells.
X

75-mm2 silicone membrane. After 48 h in culture,

FIG. 2. Stretch of the silicone substrate resulted in an equiv- medium was changed to serum-free medium (5.0 ml), and culture
time pointsindicated, with medium samples (50 pl) were taken at the and without 20% stretch of the silicone membrane. Release of lactate dehydrogenase and creatine kinase was less than 0.02% of the total cell lactate dehydrogenase and creatine kinase obtained from cell lysate. There was no significant difference in release of lactate dehydrogenase or creatinekinase in stretched versus nonstretched cultures, except that at h, alarger amount of lactate dehydrogenase 12 release was observed in nonstretchednon-myocyte culture. The physiological significance of this observation is not clear. Each data point represents the mean from three samples.

Stretch-induced Cardiac;Hypertrophy in Vitro

10555

16 TIME (HRB)

32

48
I "

CONTROL

STRETCH

SERUM

STRETCH ROL CONT

SERUM

CONTROL

STRETCH

SERUM

C
*+

4r I

"

CONTllOL

STHkTCH

St""M

CONTROL

STRETCH

SERUM

FIG. 4. Effect of stretchandserumstimulationon ["HI phenylalanine incorporation ( A ) , proteincontent ( B ) , and ["Hlthymidine uptake ( C ) in cardiac myocytes. A , ["Hjphenylalanine incorporation. Myocyte cultures inserum-free mediumwere a preparedasin Fig. IH. A t eachtimepoint,thecpm of .'H was normalized to meanof nonstretched control. Data were also normalized to protein contentof the dish (to adjust the small variahility for of the cell counts hetween dishes), although we observed the same results without this normalizat.ion. Each data point represents mean + S.E.from four to six samples (*p < 0.05, **p < 0.01 uer.su.s control). H , protein content. Cardiac myocytes were stimulated for 48 h. Data were normalized by mean of nonstimulated control. Data were also normalized hy DNAcontent of the dish to normalizethesmall variability of the cell counts between dishes. Each data represents mean S.E. from 9 to 10 samples (*p < 0.05, *'p < 0.01 uersus control, +p < 0.05 ucrsus stretch). C ["Hlthymidine uptake. Cardiac , myocytes were stimulated for 24 h. For serum stimulation, medium with 20% fetal calf serum was used. Data are expressed as relative cpm/dish normalized to the mean cpm control cells in each experof iment. Data are mean S.E. from four to five samples (**p < 0.001 ucrsus control).

F ~ G5. Effect of stretch andserum stimulation on ["Hlphen. ylalanine incorporation ( A ) , ["Hlthymidine uptake ( B ) ,and cell number (C) in the non-myocyte. Non-myocyte cultures in a serum-free medium were prepared a s in Fig, IH. For serum stimulation, medium containing 20% fetal calf serum was used. For A and H , cells were stimulated for 24 h. For C, cells were st.imulated for 48 h. Data are expressed as cpm of "H/dish and normalized to the mean o f cprn in control in each experiment. Data are mean + S.E. from four to eight samples < 0.05 and **p < 0.001 uc'rsu.s control, +p < (*p 0.05 uersus &etch).

Stretch CausesHyperplasia of Non-myocytes-Effects of stretch on protein synthesis and the rate of DNA synthesis were also examined in non-myocyte cultures obtained by two passages of the preplating dish. Microscopical observation showed no beating cells, and immunostaining with a monoclonal antibody against sarcomeric myosin (MF 20) showed that thisfraction generally contained less than 15%of MF 20 positive cells. 24 h of stretch of the non-myocyte culture caused a small but statistically significant increase (1.14 f 0.01-fold, n = 6, p < 0.05 uersus control) in ["Hlphenylalanine incorporation (Fig. 5 A ) . The same stimulation alsocaused 1.45-fold increase ( n = 5, p < 0.05 uersus control) in ["HI

thymidine uptake (Fig. 5B). This increase in DNA synthesis seemed to be associatedwith cell division, rather than an increasein cell ploidy, because cell numbers were also increased by stretch (Fig. X). As positive control, stimulation of ["Hlphenylalanine incorporation by 20% FCS (1.71-fold), [:3H]thymidine uptake(3.77-fold), and cell number (3.4-fold) in parallelcultures arealso shown (Fig. A - C ) . Thus, stretch5, ing non-myocytes i n serum-free medium in. uitm is associated with a small but significant increase i n mitogenic activity. It is interesting to note that in the in uiuo heart non-myocytes are known to undergohyperplasia in response to pressure overload (11). Induction of Immediate EarlyGenes--It is well known that a large number of IE genes are induced by a variety of growth stimuli as one of the earliest nuclear responses in many cell types (14, 52). In cardiocytes in uitro, it has been reported that the c-fos protooncogene is induced by stretch (6), peptide growth factors (13),and by aI- and @receptor stimulation (15). Interestingly, another immediate early gene, Rgr-1 (Zif268) is induced by ru,-receptor stimulationbutnot by 8receptor stimulation ( X ) , suggesting that the profile of protooncogeneexpression is stimulus-specific (13).Thus,the effect of stretch on expression of different classes of IE genes

10556

Stretch-induced Cardiac Hypertrophy Vitro in

was examined byNorthern blot analysis. These included the in rat embryonic fibroblasts and mouse fibroblast cell lines. leucine-zipper class (c-fos, c-jun) (53), the helix-loop-helix Therefore, we examined whether or not stretch of cardiocytes family (c-myc)(54), the zinc fingerclass (Egr-l/Zif-268) (50), (myocytesand non-myocytes)grown in serum-freeconditions and a cytokine-like gene, J E (34). Stretch of myocytes pro- actually leads to translocation of Fos protein to the nucleus duced a rapid and transient induction of all IE genes exam- after its synthesis in the cytoplasm. Immunostaining of nonined c-fos, c-jun, and J E (Fig. 6A), as well as c-myc and Egr- stretched cardiocytes with an antibody against Fos showed 1/Zif-268 (Fig. 6B). Although the expression of c-fos, c-jun, no specific signals (Fig. 9a). In contrast, staining of cardiJE, and Egr-1 showed a peak at around 30 min, that of c-myc ocytes stretched for 60 min showed specific signals localized revealed a littlelater peak around 1 h.Fig.6C shows a in the nucleus (Fig. 9, b and c). Thus, in cardiocytes stretch quantitative analysis of multiple Northern blots by laser induces translocation of Fos to the nucleus in the absence of densitometry. Mean expression of c-fos,c-jun, and J E mRNAs serum. In order to confirm that both myocytes and nonwere increased by 7.7-, 3.2-, and 3.0-fold, respectively. The myocytes express Fos protein in response to stretch, double induction of IE genes by stretch was also observed in primary immunofluorescence staining was performed using anti-Fos non-myocyte fraction (Fig. 7 and data not shown), although and anti-sarcomeric myosin (MF 20) antibodies. Both myosin (arrowheads) for reasons not clear the levels of expression of IE genes in positive cells (arrows) and myosin negative cells non-myocyte culture weremore variable (sometimes even (Fig. 9d) expressed Fos protein in response to stretch (Fig. higher) than those of myocyte culture (data not shown). 9d'). Induction of Fos protein was a transient response beIt is known that heat shock proteins are induced by a cause little Fos signals were detectable 2 h after stretch (data variety of stimuli including high temperature, hypoxia, oxi- not shown). dative stress, and heavy metals (55). Heat shock proteins are Induction of Nuclear DNA Binding Activities-To confirm involved in protecting cells under adverse conditions by mech- that stretch induces DNA binding activities to the AP-1 anisms not yet fully understood.The stress protein Hsp7O is element, we performed a DNA gel mobility shift assay. Nuone of the principal constituents of the mammalian heat shock clear extracts were prepared from control and stretched myoconsensus AP-1 response and can be promptly up-regulated at thetranscrip- cytes and were incubated with a 32P-labeled tional level in the absence of new protein synthesis, as in IE oligonucleotide (49).The nuclear extract obtained from congenes (55). A marked expression of Hsp7O mRNA by acute trol myocytes showed no significant binding (Fig. lOA, lane pressure overload in vivo has been demonstrated previously 1). In contrast, that obtained from myocytes stretched for 1 (8).Interestingly, however, stretching myocytes in vitro failed h showed significant binding ( l a n e 4 ) which can be competed to elicit the heat shock response, although 1 h of incubation by an excess of the unlabeled AP-1 oligonucleotide ( l a n e 5 ) of myocytes at 42 "C strongly induced expression of Hsp70, but not by unrelated sequences ( l a n e 6). Furthermore, preinparticularly the larger transcript (inducible form) (Fig. 8). cubation of the nuclear extracts with antibodies against Fos Unlike other IE genes, Hsp70 was not significantly induced and Junmarkedly diminished the DNA binding activity ( l a n e 7), but antibodies against myosin had no effect ( l a n e 8). The by serum stimulation either. Localization of Fos Protein in the Nucleus-The c-fos and residual DNA binding activity in lane 7 probably represents c-jun genes encode nuclear proteins Fos and Jun, respectively, other members of AP-1 (such as fra, jun-B, jun-D, etc.). When which are known as transcriptional regulators (9). Fos and the Egr-1 consensus oligonucleotide was used a probe (Fig. as Jun form a heterodimer and bind to AP-1 sites, present in 1OB) (50), detectable binding activity was observed in the the promoter region of many genes (52).Recently Roux et al. extracts from control myocytes ( l a n e 2), consistent with a (16) reported that in the absence of serum, the Fos protein detectable level of expression of Egr-1 mRNA in control state could not be translocated into nucleus and stayed in cytoplasm (Fig. 6B). On the other hand, there was a marked increase in
A B C

p i o.w*

i Control
E#
Stretch 30'

-c-fos

-c-myc

-c-jun

-2if-268
(Egr-11

- JE
-28s
-18s

-28s

- 18s
c-fos
("=I 1)

c-jun
(".7)

JE
(k7)

FIG. 6. Stretch-induced expression of immediate early genes in cardiac myocytes as determined by Northern blot analyses.A, expression of c-fos, c-jun, and JE genes. B, expression of c-rnyc and Zif 268 (Egr-1) genes. Myocyte cultures in a serum-free medium were preparedas in Fig. 1B.Myocytes were stretched for the time indicated above. 15 pg of total RNA were loaded in each lane. Ethidium bromide staining of 18 S and 28 S RNA showed that equal amounts of RNA was loaded in each lane. Gene expression 60 min after addition of 20% fetal calf serum to the medium is also shown as a positive control. C, quantitative analysis of the stretch-induced expression of c-fos, c-jun, and JE genes. Each hybridization signal was quantitated by laser densitometry and is expressed as the ratio of the experimental value to that score obtained from nonstretched samples, which was set as 1.0. For each analysis, the hybridization signal was normalized to the signal obtained with a glyceraldehyde-3phosphate dehydrogenase probe as an internal control. The magnitude of induction of JE gene was more variable than that of other IE genes. Results represent the mean + S.E. of 7-11 independent experiments.

VitroHypertrophy Stretch-induced Cardiac in

C .-

10557

Stretch

:-fos

FIG. 7. Stretch-induced expression of the c-fos gene in the non-myocyte as determined by Northern blot analysis. Nonmyocyte cultures in a serum-free medium were prepared as in Fig. 1H. Non-myocytes were stretched for the times indicated. c-fos expression 30 min after addition of 20% fetal calf serum is also shown as apositivecontrol. Hybridization withaglyceraldehyde-3-phosphate dehydrogenase probe showed equivalent amounts of signal in each lane (not shown).

28s-

FIG. 9. Immunofluorescent staining of cells with anti-Fos and anti-myosinantibodies. Cardiac myocytes were platedon 18ssilicone membrane in serum-free conditions as in Fig. 1B. After a 60minstretch,the cells were fixed andtreatedas described under Experimental Procedures. a, staining of control (nonstretched)cells FIG.8. Lack of stretch-induced expression of Hsp7O gene with the anti-Fos antibody. Note that the bright spots are artifact as as determined by Northernblot analysis. Cardiac myocytes were they correspond to non-cellular materials in the phase contrast mistretched for the times indicated. As a positive control, Hsp7O expres- croscopy image (not shown). b, staining by the anti-Fos antibody of sion by 1 h of incubation a t 42 C is also shown. Although 1 h of cells stretched for 60 min (low power view). Granular nuclear staining incubation a t 42 C caused a marked induction of both inducible and was observed in most of the cells in the field. c, the high power view of b. c, phase contrast image of c. Arrows correspond to the same constitutive transcripts, stretch for 1 or 2 h, or 1 h of treatment with nuclei in c and c. d, immunostaining of stretched cells with the anti20% fetal calf serum failed to induce Hsp70 mRNA significantly. sarcomeric myosin antibody using Texas Red as a secondary antibody. Hybridization with a glyceraldehyde-3-phosphate dehydrogenase usinga probe showed equivalent amounts of signal in each lane (not shown). d, staining of thesame field by theanti-Fosantibody fluorescein isothiothiocyanate-labeledsecondary antibody. Both myosin positive cells (arrows) and myosin negative cells (arrowheads) the binding activity to the Egr-1 target sequence in extracts in d had specific nuclear staining by the anti-Fos antibody ( d ) . FOS from the stretchedcells (lane 4 ) , which was specifically com- staining isweaker in dthan in probably due totechnical difficulties b, peted by excess unlabeled EGR-1 oligonucleotide (lane 5 ) but in performing double immunofluorescence on the silicone membrane. not by unrelated sequences (lane 6). Similarresults were Bars indicate 20 pm.

cHsp 70 (inducible) Hsp 70 (constitutive)

observed when myocytes were stimulated by 20% FCS (lanes 7 and 8). and &MHCgenes contain stretch-responsive elements. The Induction of Fetal Genes-To examine whether stretchin -356 c-fos CAT construct contains up to base pairs (bp) 356 this model is accompanied by the late phenotypic changes upstream of the mouse c-fos promoter including two major that are characteristic in in uiuo models of hypertrophy (8, inducible elements, the SRE and the calcium/cAMP-respon13, 19), we examined expression of several fetal genes by a sive element (Ca+/CRE) (43, 44, 52). When this construct long term stretch. For this experiment, a physiologic dose of was transfected into myocytes, there was 4-5-fold induction Ta(1nM) was added to the medium to accelerate maturation of CAT activity after 2 h of stretch (Fig. 12), in agreement of neonatal cardiac myocytes in uitro (56) because neonatal with the previous reports (6). TheANF-BglII-CAT construct ventricular cells are known to express/3-MHC, ANF, skeletal contains 3412 bp of ANF 5-flanking sequence including APa-actin. and other fetal genes (8, 19, 56). In nonstretched 1 as well as a putative Ca*+/CRE (45). It has been shown cells, ANF and skeletal cy-actin geneshowed low levels of previously that this construct is expressed in atrial cells but expression in this culture condition. However, 12-48 h of that its expression is very low in neonatal ventricular cells stretch caused a progressive accumulation of ANF andskele- (26). When this ANF construct was transfected to neonatal tal cy-actin mRNAs (Fig. 1lA). S1 mapping analysis revealed ventricular myocytes, there was no induction of CAT activity that 48 h of stretch also induced expression of the @MHC by stretch for 48 h (Fig. 12). Similarly, ANF A HindIII-CAT, mRNA (Fig. 11B).Thus,thepattern of thephenotypic producedfromANF-BglII-CAT by deletion of an internal changes induced by stretch was very similar to that in vivo 556-bp Hind111 fragment containing the atrial-specific eleof hypertrophy produced by hemodynamic stress (8, 19). Fur- ment and AP-1 (45), failed to show induction by stretch (data thermore, genes whose expression is not dependent of CArG not shown). The @-MHC-1, and -3 constructs containup -2, box or SRE elements (ANF and @-MHC)(25, 45) were also to 295, 348, and 628 bpupstream of the rabbit@-MHC inducible by in uitro stretch. promoter, respectively (46), and the rat -673 @-MHCconTransfection Experiments-The above results suggest that struct contains673-bp 5-flanking sequence the rat @-MHC of this in uitro model closely mimics the phenotype of in uiuo gene(25). Allof these constructs have been shown to be hypertrophy in expression of IE genes and late fetal genes. highly active in skeletal muscle cells, although in cultured Therefore, we examined whether the promotersc-fos, ANF, cardiac myocytes their levels of expression are low (25, 46, of

10558
A
control
Sfretch

Stretch-induced Hypertrophy Cardiac

B
Conlrol Stretch Serum

in Vitro
Stretch

?m

4-EGR.1

Sk. Aclln

4 AP-1

FIG. 10. DNA gel mobility shift analysis of cardiac nuclear which extracts. A , stretching myocytes inducesanuclearfactor specifically binds totheAP-1 consensus sequence. Myocyte-rich cultures in a serum-free medium were prepared as inFig. 1B. Nuclear extracts ( N E ) were obtained from nonstretched (control) and stretched cardiac myocytes (60 min), and bindingreactions were carried out asdescribed in the textwith an end-labeled AP-1 consensus oligonucleotide. The specific complex formed by the factor binding to the AP-1 probe is indicated the arrow marked AP-I. Competitors by used are indicated above each lane. Lanes 1 and 4 , NE alone; lanes 2 and 5 , NE with 100-fold molar excess of cold AP-1 oligonucleotide; lane 6, NE with 100-fold molar excess of the thyroid hormone receptor-binding site of the a-myosin heavy chain ( M H C )oligonucleotide (61) used as a nonspecific competitor; lanes 3 and 7, NE with 2 pl of anti-Fos and Jun antibodies; lane 8, NE with 2 pl of anti-MHC antibody. Antibodies were preincubated withNE at room temperature for 15 min. B, stretch and serum stimulation significantly increases a nuclear factor which specifically binds to the Egr-1-binding site A sequence. The same NE as in were used. In addition, NE were also prepared from cardiac myocytes stimulated with 20% fetal calf serum for 60 min. Binding reactions were carried out as described in the text with an end-labeled oligonucleotide containing thesequence 5CGCCCCCGC-3 (50). The specific complex formed by the factor binding to the Egr-1probe is indicated by the arrow marked EGR-I. Lane I, probe alone; lanes 2,4, and 7, NE alone; lanes 3 , 5 , and 8, NE with 100-fold excess of cold Egr-1 oligonucleotide; lane 6, NE with 100-fold molar excess of the a-MHColigonucleotide (61).

Po

PolyIAl

aa18S
8-301nt

COOu-lsont

FIG. 11. Stretch-induced expression of fetal cardiac genes. A, induction of ANF and skeletal (Sk) a-actin genes by stretch as determined by Northern blot analyses. Myocyte cultures in a serumfree medium were prepared as in Fig. 1B. Myocytes were stretched for the times indicated. Hybridization was performed with 5 endlabeled specific oligonucleotide probes (36, 37). Hybridization with glyceraldehyde-3-phosphate dehydrogenase probe showed equivalent amount of signals in each lane (not shown). B, effects of long term stretch on a- and @MHC gene expression as determined by S1 mapping analysis. S1 mapping was performed as previously as described (38). aMHC, a-myosin heavy chain; P,PstI; w, amino acid; nt, nucleotide.
DISCUSSION

57).Whenthesep-MHC constructs were transfected into neonatal cardiac myocytes, none of them showed significant induction of CAT activity after 48 h of stretch (Fig. 12 and data not shown). A trivial explanation for the apparent lack of stretch response of the ANF and the p-MHC promoters might be low transfection efficiency of cardiac myocytes, and a lack of sufficient sensitivity of the CAT assay (as opposed to luciferase assay). We believe this is unlikely for the following reasons. First, we were able to demonstrate that the c-fos promoter is stretch inducible, which serves as a positive control of the system. Second, changing the method of transfection to electroporation or to lipofectin, which have been reported to give a higher transfection efficiency in cardiocytes (26, 58), did not change our results (data not shown). Third, the method of CAT assay we used is suitable to analyze weak promoter activities because it is 25-fold more sensitive than the conventional method due to a much lower background (47). Fourth, the base-line levels of expression of the ANF and the p-MHC constructs we observed (see Fig. 12 legend) are quite comparable to those reportedin the literature few (a % of RSV promoter) (25, 26, 57, 59). Taken together, it is highly unlikely that we had failed to detecta significant induction of the ANF and p-MHC promoters by stretch.

Our results demonstrate that the phenotype of stretched cardiac myocytes in vitro highly resembles that of load-induced hypertrophy in vivo. More specifically, stretch causes an increaseinproteinsynthesis andcontent without an increase in DNA synthesis, a rapid and transient induction of a variety of IE genes, followed byactivation of fetal genes, such as skeletal a-actin, /3-MHC, and ANF. Thus, thismodel seems to be a suitablesystem with which to dissect the signal transduction pathway of load-induced hypertrophy. A very important question is whether the 20% stretch we used is a physiologically relevant stimulus. We confirmed that the cells did not slip from the substratewith this degree of stretch, and there no evidence of cell injury. In thewhole was heart, a 20% increase in cell length would result in maximum a of 1.728 (U3)-fold increase in theend-diastolic volume. This magnitude of change in end-diastolic volume is well within that range observed in the intact heart (60). The second question concerns whetherstatic stretchof the membrane imposed a pure diastolic stress or a combination of diastolic and systolic stress. In thecase of non-muscle cells there is no systolic component. On the other hand, myocytes in ourculture system actively contracted. Therefore, it is likely that static stretch of myocytes resulted in an increase in bothsystolic and diastolic stress. However, IE gene expression was also observed in myocytes arrested with high K or tetrodotoxin in the culture medium, indicating that active tension generation is notnecessary for myocytes to sense the stretch stimulus.

* J. Sadoshima, T. Takahashi, L. Jahn, and S. Izumo, manuscript submitted.

Stretch-induced Cardiac Hypertrophy in Vitro

10559

of the microcoronary circulation or possibly by an increase in myocardial temperature due to excess workload. On other the hand, in our in vitro system, myocytes were well oxygenated and at constant temperature condition even in the stretched state. These differences may contribute to the lack of induction of Hsp7O in i n vitro stretch. In agreement with a previous report (6), we were able to demonstrate the stretch-induced activation of the c-fos gene promoter. However, we did not observe significant induction of the promoter activities of the transfected ANF or the BMHC genes by long term stretch despite activation of the endogenous genes by this procedure. In the case of ANF, the c-10s ANF P-MHC FIG. 12. Effect of stretch on the transfected -366 c-fos, construct we used contained a 3.4-kb fragment from the 5-3412 ANF, and -628 @-MHCreporter constructs in primary flanking sequence of the rat ANF genes which contains sufcardiac myocytes. Primary ventricularmyocytes plated on silicone ficient sequence to direct high levels of atrial-specific tranmembranes were cotransfectedwith 15 pg of eachCATreporter scription in primary atrial cell cultures (26). However, the construct and5 pg of the RSV-&galactosidase expression plasmid by activity of this construct in neonatal ventricular cells is very calcium phosphateprecipitationmethod. 24 h after transfection, was static stretch applied to the dishes for 2 h (for the c-fos construct) low and not significantly inducible by glucocorticoids,potent or 48 h (for the ANF and B-MHC constructs). Cells were harvested inducers of the ANF promoter in atrial cells (26). These data and CATactivity was determined by a modifiedphase extraction suggest that DNA sequences necessary for inducible ventricmethod as described (47). To normalizefor transfection efficiency, 8- ular expression of the ANF gene may be located outside of galactosidase activitywas measured as described (30). The activity this 3.4-kb sequence. In support of this interpretation, Rockof the RSV promoter was not influenced significantly by stretch (1.16 man et al. (66) demonstrated that, in a transgenic mouse line & 0.1-fold,stretch versus control). To calculate the magnitude of induction by stretch, thenormalized CAT activity of stretched cells in which 500 base pairs of the human ANF promoter region was divided by that of nonstretched cells foreach construct. Results directed the atrial specific expression of a marker gene, thoare expressed as mean f S.E. of three to seven independent experi- racic aortic banding led to no detectable expression of the ments done in pairs (*p < 0.01 compared to nonstretchedcontrols). marker gene in the ventricle despite a 20-fold increase in The mean base-line (nonstretched) CAT activities each construct of relative to that of RSV-CAT (which has a very strong activity in endogenous ANF mRNA. This suggests that the DNA elements necessary for the atrial-specific and pressure-inducible, cardiocytes) pUC9CAT were: (a promoterless negative control), 0.27%;-356 c-fos, 9.0%;-3412 ANF, 2.0%; -628 P-MHC, 1.8%. The ventricular expression of the ANF gene are distinct. The 3.4activities of the ANF and 0-MHC promoters were very similar to kb fragment of the ANF promoter contains the AP-1 consenthose of comparable constructs reported in the literature (25, 26,57, sus sequence which has been shown to bind to in vitro trans59). lated c-foslc-jun heterodimers (45). However, the fact that the activity of the ANF construct could not be induced by IE genes have been proposed to function as mediators stretch suggests that induction of c-fos and c-junalone is not (third messengers) of long term cellular responses (9, 52). sufficient to confer the stretch responsiveness to this proIn some systems, downstream genes regulated by IE genes moter. In fact, an increase in ANF mRNA in response to have been identified. In the rat fibroblast, c-fos expression overload i n vitro and i n vivo takes place when Fos protein is was necessary for induction of stromelysin gene expression no longer detectable. These observations raise the question of by epidermal growth factor (62). However, in cardiac hyper- whether the AP-l-binding site of physiological significance is trophy, whether IE genes direct the subsequent phenotypic in regulating the ANF promoter by hemodynamic overload. changes remains to be elucidated. We demonstrated that the The P-MHC promoter constructs used in this study are stretch-induced c-fos expression led to expression of Fos pro- known to contain two muscle-specific positive elements and tein in the nucleus in this model.DNAgel mobility shift their expression is high in cultured skeletal myotubes (46). assay showed that stretch induced a DNA binding activity to When transfected to cultured rat neonatal ventricular cells, the AP-1 consensus sequence as well as to the Egr-1 target however, their levels of expression are very low even in the sequence. Since many genes seem to contain the binding sites absence of thyroid hormone, a condition that causes a marked for AP-1 and Egr-1, these IE gene products may work as the induction of the endogenous B-MHC gene (38). Therefore, it third messenger for some part of the hypertrophic response. On the other hand, it is very difficult to explain long term is possible that additional positive cis-elements may exist that changes in the expression of cardiac genes on thebasis of IE confer a high level of expression in thecardiac myocytes. The gene expression alone because it occurs only transiently in presence of cardiac-specific elements, separate from the skeletal muscle positive elements, has been demonstrated in response to overload in vitro and in vivo. mouse the case of the chicken cardiac troponin T gene and the Among the IE genes which are known to be induced in cardiac troponin C gene (67, 68). It is possible that the presponse to external load in the intact heartonly Hsp7O was not induced in thisin vitro stretch model. It has been reported MHC promoter may require such additional cardiac-specific that the human Hsp70 gene is inducible by serum stimulation elements in order to respond to stretch. Stretching primary non-myocyte caused a small but signifbut, unlike other IE genes, it takes 12 h before the induction by serum to occur (63). Furthermore, &-regulatory sequences icant increase in mitogenic activity. This result is consistent required for heat shock response are different from those with the observation in vivo that the cardiac interstitium, required for serum induction (64,65). Thus, is not it surprising which comprises cardiac fibroblasts, endothelial cells, vascuthat Hsp7O and other IEgenes are regulated differentially in lar smooth muscle cells, macrophages, mast cells, and others, cardiac myocytes. The reason for the lack of Hsp7O induction undergoes hyperplasia in response to pressure overload (11). by stretch i n vitro is not clear yet. It is possible that pressure These primary non-myocytes express IE genes in response to overload in vivo may be accompanied by relative myocardial stretch in vitro, thus mimicking the early mitogenic response hypoxia due to an increase in oxygen demand and squeezing to growth factors in a variety of cells (9). Our model may be

10560

Stretch-induced Cardiac Hypertrophy in Vitro

23. Taylor, M., Treisman, R., Garrett, N., and Mohun, T. (1989) Deuelopment 106,67-78 Roeder, R. G., and Kedes, L. (1989) Mol. Cell 24. Boxer, L. M., Prywes, R., Biol. 9,515-522 25. Thomnson. W. R.. Nadal-Ginard. B.. and Mahdavi. V. (1991) J. Biol. Chem. . . ~,~ 268,~25678-22688 26. Seidman, C. E., Wong, D. W., Jarcho, J. A., Bloch, K. D., and Seidman, J. G. (1988) Proc. Natl. Acad. Sci. U.S. A. 86,4104-4108 27. Simpson, P., McGrath, A., and Savion, S. (1982) Circ. Res. 6 1 , 787-801 28. Orlowski, J., and Lingrel, J. B. (1990) J. Biol. Chem. 2 6 6 , 3462-3470 29. Bader, D., Masaki, T., and Fischman, D. A. (1982) J. Cell. BioL 9 5 , 763~ ~~ ~ ~~~ ~~~~ ~ ~

useful to elucidate the mechanism of the interstitialresponse to external load. It should be noted, however, that stretch responsiveness may not be a general phenomenon because the stretch-induced IE gene expression was observed only in primary cultures but not in established cell lines such as NIH 3T3 and PC12 or in cultured pulmonary arterial smooth muscle or aortic endothelial cells that have been passaged multiple times (data notshown). These resultssuggest that thesignal transduction pathway for the stretch-induced response may utilize components which are susceptible to down-regulation during multiple passages or in cell lines. In summary, the phenotype of the in vitro model of the load-induced cardiac hypertrophy described here very closely mimics that of the load-induced response in vivo. in vitro This model, whichis amenable to molecular biological approaches, should allow us to elucidate further the precise mechanisms of gene regulation by mechanical stimuli in muscle cells and in other cell types.
Acknowledgments-We thank E. Wu for her technical assistance, R. D. Rosenberg and W. Grossman for their support and encouragement, C. E. Seidman, V. Mahdavi, M. Z. Gilman, M. Greenberg, B. Rollins, C. Stiles, M. Karin, N. Shimizu, and D. Nathans for plasmids, and D. Bader for MF 20 antibody.
REFERENCES Morgan, H. E., and Baker, K. M. (1991) Circulation 83,13-25 Vandenburgh, H. H. (1987) Med. Sei. Sports Exercise 1 9 , S142-Sl49 Watson, P. A. (1991) FASEB J. 6,2013-2019 Vandenburgh, H. H., and Kaufman, S. (1979) Science 203,265-268 Mann, D. L., Kent, R. L., and Cooper, G., IV (1989) Circ. Res. 6 4 , 10791090 6. Komuro, I., Kaida, T., Sbibazaki, Y., Kurabayashi, M., Katoh, Y., Hob, E., Takaku, F., and Yazaki, Y. (1990) J.Biol. Chem. 265,3595-3598 7. Komuro, I., Katoh, Y., Kaida, T., Shibazaki, Y., Kurabayashi, M., Hob, E., Takaku, F., and Yazaki, Y. (1991) J.Biol. Chem. 266,1265-1268 8. Izumo, S., Nadal-Ginard, B., and Mahdavi, V. (1988) Proc. Natl. Acad. Sci. U. A. 86,339-343 S. 9. Reedy, E. P.,Skalka, A. M., and Curran, T. (1988) The OncogeneHandbook, Elsevier, Amsterdam 10. Vandenburgh, H. H., Hatfaludy, S., Karlisch, P., and Shansky, J. (1989) Am. J.Physiol. 2 5 6 , C674-C682 11. Zak. R. !1984) in Growth of the Heart in Health and Disease (Zak, R., ed) Raven Press, New York . 12. Ueno, H., Penyman, B., Roberta, R., and Schneider, M. D. (1988) J. Cell Biol. 107,1911-1918 13. Parker, T. G., and Schneider, M. D. (1991) Annu. Rev. Physiol. 6 3 , 1791. 2. 3. 4. 5.

.~

7717

30. Sah;bvrook,J., Fritsch, E. F., and Maniatis, T. (1989) in Molecular Cloning, Cold S ring Harbor Laboratory Press, New York 31. Fort, P., harty, L., Piechaczyk, M., Sabrouty, S. E., Dani, C., Jaenteur, P., and Blanchard, J. M. (1985) Nucleic Acids Res. 13,1431-1442 32. An el, P , Allegretto, E. A., Okino, S. T., Hattori, K., Boyle, W. J., Hunter, and Karin. M. (1988) Nature 332.166-171 33. Lau,'L.F.,and-Nathans, D. (1985) EMEO J. 4,3145-3151 34. Rollins, B. J., Morrison, E. D., and Stiles, C.D. (1988) Proc. Natl. Acad. Sci. U. A. 85,3738-3142 S. 35. Thomas, G. P., and Pascucci, D. D. (1985) in Heat Shock: From Bacteria to Man (Schlesineer. M. J.. Ashburner. M.. and Tissieres. A.. eds) Cold Sprin Harbor Labratory Press, Cold S ring Harbor, NY 36. Hu, M. &-T., Sharp, S. B., and Davidson, (1986) Mol. Cell. Bwl. 6, 15-

4..

~~~

~~~

~~~

9.6

b.

'

'

37. Seidman, C. E., Duby, A. D., Choi, E., Graham, R. M., Haber, E., Homcy, C. J., Smith, J. A., and Seidman, J. G. (1984) Science 225,324-326 38. Izumo, S., Nadal-Ginard, B., and Mahdavi, V. (1986) Sctence 231, 597cnn 39. Btxk, B. C., Vekshtein, V., Gordon, H. M., and Tsuda,T. (1989) Hypertension 13,305-314 H., 40. Lowry, 0. Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Bwl. Chem. 193,265-275 P. .I.. Rothblum, L;J,Smith, S. D., and Morgan, H. E.(1989) 41.

"-

.WZL I

42. 43. 44.

3 8 0 ) A d . Biochem. 102,344-352 14-402 ilman, M. Z.(1989) Mol. Cell. Biol.

2nn

14. Ba-Gl. D. P.. Shene. M.. Lan. L. F.., and Greenbere. M. E. (1989) Genes & . . , Dev.' 3,304-313 "' ' 15. Iwaki, K., Sukhatme, V. P., Shubeita, H. E., and Cbien, K. R. (1990) J. Biol. Chem. 266.1380%13187 16. Roux,P., Blanih&d,-J-M., Fernandez, A., Lamb, N., Jeanteur, P., and Piechaczyk, M. (1990) Cell 63,341-351 17. Schiaffino, S., Samuel, J. L., Sasson, D., Lompre, A. M., Garner, I., Marotte, F., Buckingham, M., Rappaport, L., and Schwartz, K. (1989) Circ. Res. 64,937-948 18. Black, F. M., Packer, S. E., Parker, T. G., Michael, L. H., Roberts, R., Schwartz, R. J., and Schneider, M. D. (1991) J. Clin. Invest. 88, 15811588 19. Izumo, S., Lompre, A. M., Matsuoka, R., Koren, G., Scbwartz, K., NadalGinard, B., and Mahdavi, V. (1987) J. Clin. Inuest. 79,970-977 . 20. Muscat, G. E O., and Kedes, L. (1987) Mol. CeU. Biol. 7,4089-4096 21. Parker, T. G., Chow, K.-L., Schwartz, R. J., and Schneider, M. D. (1990) Proc. Natl. Acad. Sci. U. A. 87,7066-7070 S. 22. Norman, C., Runswick, M., Pollock, R., and Treisman, R. (1988) Cell 66, 989-1003
I I

9,4272-4281 45. Rosennvei A,, Halamnetis, T. D., Seidman, J. G., and Seidman, C. E. (1991) crculation 8 4 , 1256-1265 46. Shimizu, N., Dizon, E., and Zak, R. (1992) Mol. Cell. Bid. 12,,619-630 47. Kingston, R. E., and Sheen J. (1990) in Current Protocol In Molecular E., D. Bzology (Ausabel, F. M 'Brent R., Kingston, R. Moore, D., Seidman J. G., and Struil, K. e&) pp. 9.6.6-9.6.9 Wile , New York Acds Res. 48. Dignam, J. D., Lebovitz, R. M., and Roeder, R. G. (1683) 1 1 . 1475-1-489 - -,- - .- - -49. An el, P , Imagawa, M., Chiu, R., Stein, B., Imbra, R. J., Rahmsdorf, H. J., fonat,'C., Herrlich, P., and Karin, M. (1987) Cell 49,729-739 . 50. Cao. X . Koski. R. A,. Gashler. A,. McKiernan. M.. Morns. C. F.. Gaffnev. R:, Hay, R. V., and Sukhatme,'V. P. (199Oj Mol. Cell. Biol. io, 193iL 1939 51. Vandenburgh, H. H. (1983) J. Cell. Physiol. 116,363-371 52. Sheng, M., and Greenberg, M. E. (1990) Neuron 4,477-485 53. Landschulz, W. H., Johnson, P. F., and McKnight, S. L. (1988) Science 240,1759-1764 54. Luschner, B., and Eisenman, R. N. (1990) Genes & Deu. 4,2025-2035 55. Lindquist, S. (1986) Annu. Rev. Biochem. 66,1151-1191 56. Parker. T. G.. Packer. S. E.. and Scbneider. M.D. (1990) J. Clin. Inuest. 85,607-514 57. Kariva. K.. Karns. L. R.. and Simuson. P. C. (1991) J. Biol. Chem. 266, . . . . 10'023-lbO26 ' 58. Lin. H.. Parmacek. M. S.. hlorle, G., Bolling, S., and Leiden, J. M (1990) Circulation 8 2 , 2217-2221 P. Harris, A. N., Knowlton, K. U., 59. Shubeita H. E., McDonough, M., Glembbtski, C. C., Brown, J. H., and Chien, K. R. (1990) J. Biol . Chem. 265,20555-20562 60. Braunwald, E., Sonnenblick, E. H., and Ross, J., Jr. (1988) in Heart Disease (Braunwald E., ed) pp. 383-425, Saunders, Phlladelphla 61. Izumo, S., andMahdavi, V. (1988) Nature 334.539-542 c 62. McDonnell. S. E.. Kerr. L. D., and Matrisian. L. M. (1990) Mol. C dl. Bwl. 10,428414293' 63. Wu. B. J.. and Morimoto, R. I. (1985) Proc. Natl. Acad. Sci. U.S. A. 8 2 , 6070-6074 64. Bienz M. and Pelham, H. R. (1987) Ada Genet. 24,31-72 65. Willidms,'G. T., and Morimoto, R. I. (1990) MOL Cell. Biol. 10,3125-3136 66. Rockman, H. A., Ross, R. S., Harris, A. N., Knowlton, K. U., Steinhelper, M,E., Field, L. J., Ross, J., Jr., and Chlen, K. R. (1991) Proc. Natl. Acad. SCLU.S. A. 88,8277-8261 67. Mar, J. H., Antin, P. B., Cooper, T. A,, and Ordahl, C. P. (1988) J. Cell. Biol. 107,573-585 68. Parmacek, M. S., and Leiden, J. (1991) Circulation 84,991-1003

dlu we:

~~~~