0892-6638/97/0011 -0683/$01 .50 © FASEB 683

Three-dimensional reconstitution of embryoniccardiomyocytes in a collagen matrix: a new heartmuscle model system

ThOMAS ESCHENHAGEN,’ CHRISTINE FINK, UTE REMMERS, HASSO SCHOLZ,JENS WATFCHOW, JOACHIM WElL, WOLFRAM ZIMMERMANN, HANS H. DOHMEN,*HANSJORG SCHAFER,* NANETFE BISHOPRIC,f TETSURO WAKATSUKI,t AND

ELLIOT L ELSONt

Abteilung Allgemeine Pharmakologie and *Institut f#{252}rPathologie, Universit#{227}ts-KrankenhausEppendorf, Universit#{227}tHamburg, Germany; tSRI International, Menlo Park, California 94025-3434,USA; and Department of Biophysics and Molecular Biochemistry, Washington University School ofMedicine, St. Louis, Missouri 63110-1093, USA

ABSTRACT A method has been developed forculturing cardiac myocytes in a collagen matrix toproduce a coherently contracting 3-dimensionalmodel heart tissue that allows direct measurementof isometric contractile force. Embryonic chick car-diomyocytes were mixed with collagen solution andallowed to gel between two Velcro-coated glasstubes. During culture, the cardiomyocytes formedspontaneously beating cardiac myocyte-populatedmatrices (CMPMs) anchored at opposite ends tothe Velcro-covered tubes through which they couldbe attached to a force measuring system. Immu-nohistochemistry and electron microscopy re-vealed a highly organized tissue-like structure ofct-actin and cc-tropomyosin-positive cardiac myo-cytes exhibiting typical cross-striation, sarcomericmyofilainents, intercalated discs, desmosomes, andtight junctions. Force measurements of paced orunpaced CMPMs were performed in organ bathsafter 6-11 days of cultivation and were stable forup to 24 h. Force increased with frequency between0.8 and 2.0 Hz (positive “staircase”), increasingrest length (Starling mechanism), and increasingextracellular calcium. The utility of this system asa test bed for genetic manipulation was demon-strated by infecting the CMPMs with a recombinant3-galactosidase-carrying adenovirus. Transductionefficiency increased from about 5% (MOl 0.1) toabout 50% (MO! 100). CMPMs display more phys-iological characteristics of intact heart tissue thanmonolayer cultures. This approach, simpler and

faster than generation of transgemc animals,should allow functional consequences of genetic orpharmacological manipulation of cardiomyocytesin vitro to be studied under highly controlled con-ditions.-Eschenhagen, T., Fink, C., Remmers, U.,Scholz, H., Wattchow, J., Well, J., Zimmermann, W.,Dohmen, H. H., Schafer, H., Bishopric, N., Wak-atsuki, T., Elson, E. L. Three-dimensional reconsti-

tution of embryonic cardiomyocytes in a collagenmatrix: a new heart muscle model system. FASEBJ.

11, 683-694 (1997)

Key Words: heart . cell culture reconstituted tissue ade-noviru.s transgenic model

PRIMARY CULTURES OF EMBRYONIC OR NEONATAL car-diac myocytes have been used in cardiovascular re-search for about 40 years (1) and have become a

standard model system to investigate ion channelfunction, regulation of cardiac gene expression, andmolecular mechanisms of cardiac hypertrophy. Wide-spread use of primary cardiomyocyte cultures is dueboth to the lack of stable cardiomyocyte cell lines andto the fact that they exhibit stable spontaneous beat-ing for several days and allow pharmacological andgenetic manipulations without interference fromcompensatory mechanisms that operate in animal ex-

periments. Nevertheless, monolayer cultures in gen-eral have important methodological and conceptuallimitations. For example, skin fibroblasts markedlyproliferate and synthesize collagen (up to 15% of to-tal protein) in monolayer cultures but exhibit littleproliferative and synthetic activity in normal skin orin a 3-dimensional collagen matrix in vitro (2). Sim-ilarly, cardiomyocytes change their pattern of geneexpression toward an undifferentiated phenotype inmonolayer culture (3-6). Most important, culturedembryonic or neonatal cardiomyocytes do not allowmeasurement of their major function in the heart,namely, loaded isometric contraction.

Now that an increasing number of gene productshas been identified whose functional role in a phys-

Correspondence: Abteilung Allgemeine Pharmakologie,Universit#{227}ts-Krankenhaus Eppendorf, Martinistrasse 52, D-20246 Hamburg, Germany.

b

684 Vol. 11 July 1997 The FASEBJournal ESCHENHAGEN ET AL.

iological context is largely unknown, and becausemolecular techniques allow overexpression or knock-out of single genes or gene products, it is importantto have methods for assaying the results of these ge-netic modifications. This is especially true for at-

tempts to define the role of molecular alterations inpathophysiological states such as heart failure thatrepresent subtle quantitative changes of constitutivecardiac genes (e.g., changes in -adrenoceptors, -

adrenergic receptor kinase, Ci proteins, SR calciumATPase, phospholamban by ±30-50%; 7). Tissuemodels for studying functional consequences of ge-netic manipulations should therefore fulfill the fol-lowing criteria: stable, reproducible, and simplemeasurement of the essential heart functions, effi-cient gene transfer, and a relatively natural and con-trolled physiological environment. The aim of thepresent study was to develop and characterize a cellculture system that meets at least some of these cri-teria.

MATERIALS AND METHODS

Isolation of cardiomyocytes

Ventricles from 9-li day incubated chicken embryos wereminced to i mm pieces in Dulbecco’s minimal essential me-

dium (DMEM;2 Gibco-BRL, Eggenstein, Germany), washed

once with 0.25% trypsin/0.1% EDTA (Boehringer Mann-heim, Germany) in phosphate-buffered saline (PBS), pH

7.45, and then digested in fresh trypsin/EDTA for 15 mm

at 37#{176}C.The supernatant was discarded and the pellet wassubjected to digestion with 0.1% collagenase (144 U/mg,

Bibby Dunn, Asbach, Germany) in PBS, pH 7.45, for 30 mm

at 37#{176}C(8). This supernatant was discarded and the pelletdigested further with several cycles of collagenase for 10-

20 mm each until the pellet was completely digested. DNaseI (40 p.1, 1 mg/mI in PBS; Sigma, St. Louis, Mo.) was addedbetween cycles depending on the presence of viscous DNA.

The isolated cells were kept in petri dishes (Falcon Series3000) in DMEM supplemented with i5% heat-inactivated

fetal calf serum (FCS; Gibco BRL) in the CO2 incubator.After completion of the digestion, cells were incubated for

another 30-60 mm in the CO2 incubator (preplating). Thecell suspension was centrifuged at 250 rpm (i2Xg). The

pellet was resuspended in 10 ml culture medium (DMEM,iO% inactivated horse serum, 2% chicken embryo extract

(Gibco BRL), 2 mmol/l glutamine, 100 p.g/ml streptomy-

cm, and 100 U/mI penicillin G (Gibco BRL), recentrifugedat 250 rpm, and finally resuspended in culture medium at

2-3 X 10 cells per ml.

Casting cardiomyocyte-populated collagen gels

The principal technique is shown in Fig. la (9). Strips of

Velcro were glued with silicone rubber (No. 734, Dow Corn-

2Abbreviations: DMEM, Dulbecco’s minimal essential me-

dium; PBS, phosphate-buffered saline; MOl, multiplicity of in-fection; CMPM, cardiac myocyte-populated matrices; CB,

cacodylate buffer; RT, room temperature; Ln,ax, maximal forcelength; bpm, beats/nsin.

ing, Wiesbaden, ERG) to glass tubes (13 mm length, 3 mm

outer, 2 mm inner diameter). Pairs of Velcro-coated tubes,kept at a fixed distance by a stainless steel wire spacer, wereplaced in rectangular wells (l5X17X4 mm) cut intoalayer

of silicone rubber in a 100 mm polymethylpentene petri

dish (Nalgene, Nalge Co., Rochester, N.Y.). This assembly

was autoclaved before use. For each gel, 1 ml of an ice-coldcollagen/cell mixture was poured into each well between

the Velcro-coated glass tubes. This mixture had the same

composition as the culture medium and contained in ad-dition 1 mg neutralized collagen I from rat tail (Upstate

Biotechnology Inc.), 1 X 106 cardiomyocytes, the acetic acid

in the collagen solution, and the NaOH used to neutralizeit. The mixture was allowed to gel at 37#{176}Cfor 60 mm before

culture medium was added to the dish. Medium changeswere performed after overnight incubation and then every

other day.

Force transducer

- - - - Cell matrix

S.

- Pacing electrode

S.Fixed pole

Figure 1. Illustration of the experimental arrangement. a)Photograph of a culture dish with six cardiomyocyte-popu-

lated collagen gels (day 5). The bottom of the dish contains

a layer of silicone rubber with six rectangular wells that have

been used to cast the CMPMs. The CMPMs show the typical

concave shape of the free edges. b) Schematic representationof the force measurement technique. After 10 days in culture,

one end of a CMPM was attached to a lower fixed pole, whichalso serves as an electrode; the other, via a noncompliant fil-

ament, was attached to an isometric force transducer. Thetransducer measurements were recorded on a thermal arrayrecorder. Preload of the CMPMs was adjusted by raising theforce transducer. CMPMs were paced with a Grass stimulator.The CMPM isimmersed in continuously gassed (95% 02/5%C02) Tyrode’s solution in a thermostatted organ bath.

CARDIAC MYOCYTE-POPULATED COLLAGEN MATRIX 685

Isometric force measurements

After 6-11 days in culture, the gels were removed from theculture dish, the spacers were withdrawn, and one of the glass

tubes was mounted on a fixed electrode (Fig. 1b); the othertube was connected by an inelasticsilkstring to an isometric

force transducer (Scientific Instruments, Heidelberg, Ger-many) attached to a Wekagraph thermal array recorder (FOhr

Instruments, Heidelberg, Germany). The preparation was ad-

justed to its original (spacer) length before it was immersed in

a conventional organ bath filled with modified Tyrode’s solu-

tion (NaCI 119.8, KCI 5.4, CaCI2 1.8, MgCI2 1.05, NaH2PO4 0.42,NaHCO3 22.6, Na2EDTA 0.05, ascorbic acid 0.28, glucose 5.0mmol/l) maintained at 35#{176}Cand continuously gassed with 95%

02 + 5% CO2, as described for papillary muscle previously (10).After a 30-60 mm equilibration period without pacing, force

and frequency reached a stable value that we define as baseline

force/frequency. Gels were then electrically stimulated with rec-

tangular pulses (10 ms, 20-40 V) at a standard frequency of 1.5

Hz. Preload was stepwise adjusted to L, the length at which

the preparation developed maximal force. After equilibration,the pacing frequency was varied from 0.8 to 2.5 Hz for 3 mm

each and then returned to 1.5 Hz, which was held constantthroughout the experiment. Inotropic interventions were per-

formed as indicated in the Results section. The following stocksolutions were used: 2.25 mol/l CaCl2, 10 mmol/l isoprenalinein 4 mmol/l HC1, 1 mmol/l propranolol, 30 mmol/l forskolin

(Calbiochem) in 100% ethanol (giving a final maximal concen-tration in the bath of 0.01%), 10 mmol/l carbachol inTyrode’s

solution, 2 mmol/l cytochalasin D in dimethyl sulfoxide

(DMSO, final concentration in bath 0.1%). Stock solutionswere prepared freshly (carbachol, ouabain, norepinephrine,

propranolol) and stored for up to 2 wk at 4#{176}C(isoprenaline,forskolin) or at -20#{176}Cfor up to 6 months (cytochalasin D).

The KC1-rich Tyrode’s solution was prepared by replacing 119.8mmol/I NaC1 with 125.2 mmol/l KCI. All gels were exposed toa concentration-response curve (CRC) for calcium (1.8-12.6

mmol/l) and to one or two additional inotropic stimuli. Thelength of contraction experiments was 3 to 5 h. In that timebasal force of contraction decreased by less than 30%.

Histology, immunohistochemistry, and electron microscopy

Gels were either removed directly from the culture dish orcarefully removed from the holding electrode after contrac-

tion experiments. For light microscopy, specimens were fixedovernight at 4#{176}Cin 4% formaldehyde in PBS, pH 7.4, with themetal wire spacer attached. After dehydration in graded con-

centrations of ethanol and paraffin infiltration according to

standard procedures, the specimens were cut from the Velcro-coated glass tubes, cut in half, and embedded in paraffinblocks in either of the three orientations (perpendicular and

parallel to the glass tubes and flat). Four-micrometer sectionswere cut and stained with hematoxylmn-eosin or trichrome

stain according to Masson and Goldner (11).For immunohistochemistry, 4 p.m sections were mounted

on Superfrost Plus slides (Menzel-Gl#{227}ser, Hamburg, Ger-many) and dried overnight at 55#{176}C.After deparaffinization

and rehydration through xylene and graded concentrationsof ethanol, antigen retrieval was performed by boiling the

slides in DAKO ChemMate buffer for antigen retrieval (DAKOA/S, Glostrup, Denmark) in a domestic pressure cooker for 4

mm at 120#{176}C.Endogenous peroxidase activitywas blocked byincubation in 3% hydrogen peroxide for 5 mm at room tem-

perature (RT). Nonspecific binding was masked by incubation

in normal goat serum (DAKO) diluted 1:5 in TBS, pH 7.6, for

20 mm at RT. Primary antibody incubation was done in a moist

chamber for 1 h at 37#{176}C.The following antibodies were used:anti-muscle-actin (monoclonal, mouse, clone HHF 35, Enzo

Diagnostics, Farmingdale, N.Y., purchased in ready-to-use di-

lution), anti-sarcomeric-tropomyosin (monoclonal, mouse,

clone CH1, SIGMA, dilution 1:50), and anti-c-sarcomeric-ac-

tin (monoclonal, mouse, clone 5C5, SIGMA; in TBS, pH 7.6,

with 3% BSA). The detection system consisted of the DAKODuett streptavidin/horseradishperoxidase kit used

according to the manufacturers instructions. Secondary

biotmnylated goat-anti-mouse/rabbit antibody (1:100) was in-

cubated for 30 mm at RT, the streptavidin-biotin complex for30 mm. The sections were developed in a solution of SIGMA

FAST diaminobenzidine and urea-H202 tablets in water for 5mm at RT. Sections were counterstained with hematoxylin,

dehydrated, and mounted in EUKITT. Negative and positive

controls were obtained by omission of primary antibody, sec-tions from chicken bowel, and whole heart.

For electron microscopy, cardiac myocyte-populated matri-

ces (CMPMs), with the spacer attached, were fixed overnightin a 2.5% solution of glutaraldehyde in 0.1 mol/l sodium ca-

codylate buffer, pH 7.3 (CB), rinsed in GB containing 7.5%

sucrose, postfixed for 2 h in 1% Os04 in GB, rinsed with GB

with sucrose, dehydrated in graded series of ethanol, and in-

cubated for 1 h each in propylene oxyde and 1:1 mixture ofpropylene oxyde/epon. After infiltration and embedding in

epon 812, ultrathin sections were obtained in an Ultracut S(Reichert, Austria), double-stained by uranyl acetate and lead

citrate according to standard procedures, and visualized in aPhilips EM 400 electron microscope.

[3H]Thymidine incorporation, cell count, and RNA

preparation

One microliter of [3H]thymidine (NEN DuPont, Bad Hom-

burg, Germany; 20 Gi/mmol, 1 p.Ci/pi) was added to the

GMPMs in 1 ml culture medium in the absence or presenceof 5-bromo-2’-deoxyuridine (BrdU; 0.1 mM) at cultivation

days 0, 1, 2, 3, 4, 5, and 6. After 6 h of incubation at 37#{176}C,CMPMs were washed 3 X 10 mm with 2 ml PBS, dissected from

the Velcro, and dissolved in 1 ml 0.1% acetic acid. After over-night incubation at 4#{176}C,samples were rapidly filtered overglassfiber filters (GE 50, Schleicher & Schuell, Keene, N.H.).

Filters were washed twice with 6% trichloric acid and oncewith 95% ethanol. Radioactivity was measured after overnight

incubation in scintillation liquid. [3H]thymidine incorpora-

tion in the presence of BrdU amounted to less than 5% of

that in the absence of BrdU and was defined as nonspecific

binding. The number of intact nuclei was determined in in-dividual samples by adding 0.001% Hoechst 33258 (Sigma) tothe 0.1% acetic acid-solubilized CMPM and counting the num-

ber of intact nuclei under fluorescent light. Total RNA yieldwas determined in parallel samples by RNA extraction accord-

ing to Chomczynski and Sacchi (12). RNA was quantified pho-

tometrically and checked by gel electrophoresis.

Adenovirus-mediated gene transfer

A -galactosidase-encoding recombinant replication-deficientadenovirus type 5 (Ad5LacZ) was a kind gift from Dr. HaroldPrentice, Glasgow, U.K. The virus was propagated in human

embryonic kidney 293 cells (HEK 293), and the medium of

infected cultures was used for infection of CMPMs. The virus

titerwas determined by plaque assays on HEK 293 cellsand,

to biologically compare different preparations of viruses, byinfection of cardiac myocytes in monolayer cultures. Normal

titers were 1-5 X iO pfu/ml. To define the optimal trans-

duction efficiency, CMPMs were infected on day 4, 6, 8, 10, or12 with 1, 10, 100, and 1000 p.1 virus-containing medium for

18 h at 37#{176}Cand then incubated for an additional 48-72 hwith complete medium. Control gels were subjected to the

Figure 2. Light microscopy of GMPMs (days 9-10). a) Paraffin section of a CMPM parallel to the plane of the tissue, showing the

characteristic concentration of longitudinally oriented cells at one of the two free edges of the gel. Note the continuous cell layerat the surface and several cell layers that form cell-cell contacts (Masson-Goldner, ref ii). b) Same as panel a at higher magnifi-

cation. Brown immunohistochemical staining of muscle actin (c, d), a-sarcomeric actin (e), and sarcomeric tropomyosin (/) inparaffin sections similar to those in panels a and b counterstain with hematoxylmn. More than 80% of cells stain positive with each

antibody. The photograph in panel d shows a prominent concentration of highly cross-striated cells at a free edge of the CMPM

that is often found close to the Velcro-coated glass tubes. Magnifications: 128x (a), 256x (b c, e), 470x (d), 354X (f).

same conditions without the virus. Gels were thoroughlyrinsed with PBS pH 7.4 (2X20 mm; essential to get rid of I-galactosidase in the virus-containing medium), fixed for 30

mm in PBS-buffered 2% formaldehyde, 0.2% glutaraldehyde

at RT, washed 2 X 5 mm with PBS, and then incubated for 1-18 h in 2 mmol/l MgCl2, 20 mmol/l KFe(CN)6 II, 20 mmol/

I KFe(CN)6 III, 0.02% Ipegal (Sigma), 0.01% deoxycholate

(Sigma), and 1 mg/ml X-gal at 37#{176}C.The reaction was ter-minated by washing with PBS and overnight incubation in10% formaldehyde at 4#{176}C.Four-micrometer sections were cut

from paraffin-embedded CMPMs in three different orienta-

tions as described and stained with hematoxylin-eosmn, nuclear

fast red, or according to Masson-Goldner (11). In parallel, -

galactosidase activity was determined by a colorimetric assay.In short, CMPMs were rinsed 3 X 10 mm with PBS, dissected

from the Velcro, mechanically homogenized in 1 ml PBS at4#{176}C,and subjected to one freeze/thaw cycle. Fifty microliters

were incubated in 250 p.160 mmol/l sodium phosphate buffer,

pH 7.5, 1 mmol/l MgCl2, 45 mmol/l mercaptoethanol, and0.8 mg/ml 2-nitrophenyl-fl-D-galactopyranosid (ONPG; Boeh-ringer Mannheim) for 30 mm at 37#{176}C.The reaction was

stopped by addition of 350 p.1 1 mol/l Na2CO2. OD at 420 nm

686 Vol. 11 July 1997 The FASEB Journal [SCHENHAGEN ET AL.

aaa1x

E0C)

8

6

4.

2

0

I

9

I

5

I6

15

10

5

0

*

*

3

*

3

*

30 1 2 3 4 5 6

Time (days)

CARDIAC MYOCYTE-POPU LATED COLLAGEN MATRIX 687

sion and reshaping of the gel from an original size of(thickness X width X length) 5 X 13 X 15 mm (1 ml

drop) to biconcaval CMPMs, with an average size of0.18 ± 0.06 mm (n=25) X 6-10 X 15 mm. Thus, the

initial drop formed a stable tissue spanning the gapbetween the two glass rods (see Fig. 1 a, Fig. 11 a). Asseen in fibroblast-populated collagen lattices, the freeedges are concave, perhaps due to the balance offorces in this region (9). Spontaneous contractionsof single cells were seen after 1-2 days, coordinatedcontractions of the entire CMPM after 3-5 days.CMPMs remained stable for at least 2 wk. After 6-8days they did not change their gross macroscopic ormicroscopic appearance, nor did they stop beating.

Light microscopy revealed a reproducable patternof cell growth (Fig. 2). Sections parallel to the plane

a 3H-thymidine incorporation

Figure 3. Electron microscopy of a CMPM (day 10). a) Typicalmyocardial cell presenting cross-striation with Z-clisks (Z),

transversal and longital tubules, and mitochondria (M);

16,500x. b) Adjacent myocardial cells presenting intercellularcontact organelles. Note desmosomes (arrow) and tight junc-

tions (arrowhead) adjacent to the peripheral part of a T tu-

bule (T); 30,100x. c) Formation of a typical intercalated disk(arrow); 23,000x.

was referred to a standard curve of purified 3-galactosidase

(Boehringer Mannheim, 300 U/mg).

Statistics

All values presented are arithmetic means ± SEM. Student’s I

test for paired observations was used in all cases to compareforce of contraction, resting tension, or beating frequency be-fore and after other interventions. Student’s I test for un-

paired observations was used for all other experiments. A Pvalue of less than 0.05 was considered significant.

RESULTS

Morphological aspects

Soon after gelation of the collagen-cell mixture, aremodeling process began that resulted in compres-

0123456

b RNA yield per CMPM

* p<0.05 vs. day 1

Figure 4. Time course of [3HJ thymidine incorporation and RNA

yield of CMPMs. a) Incorporation of [tHjthymidine into GMPMs

over a period of 6 h, determined directly after casting of the gel

(day 0) to day 6.The ordinate indicatescpm per CMPM. b) Total

RNA yield per CMPM, expressed as p.g/CMPM. Individual

GMPMS from the same cell preparation series were randomly

assigned to [2H] thymidine incorporation and RNA preparation.Number in bars = number of independent experiments.

‘I, ‘I,

*p<005 vs. Ctr

Ctr Iso CarblOpmot/l lOpmol/l

688 Vol. 11 July 1997 The FASEB Journal ESCHENHAGEN ET AL.

a Isoprenaline Carbachol10 pmol/l 10 pmol/I

1 mm

b

CEU,

a)

>‘0Ca)

a)I.-IL

muscle differentiation. Many cells contained two nu-clei, as typical for differentiated chick myocardium.The remaining cells appeared less mature (Fig. 2c-

f). Some nonmuscie cells lay intermingled withcardiac myocytes.

By electron microscopy (Fig. 3), most of the cellsshowed to a greater or lesser extent characteristicmyocardial elements such as well-organized myofila-ments with cross-striations, Z-bands, and longitudinaland transversal systems. Intercellular junctions werevisible presenting intercalated discs, desmosomes,and tight junctions. Thus, the ultrastructure of theCMPMs resembles the well-known muscular structureof myocardial tissue.

Proliferative activity

[3H]Thymidine incorporation was determined as amarker of DNA synthesis. As expected for embryonicheart cells, [3H]thymidine incorporation was detecta-ble at all stages of cultivation (Fig. 4a).[3H]Thymidineincorporation into CMPMs showed a biphasic time

course with an increase from day 0 to day 1, followedby a gradual decrease to day 5 and 6 (Fig. 4a). In thesame time period, total RNA yield increased more

Figure 5. Spontaneous beating rate. a) Original tracing of arepresentative experiment. Cardiomyocyte-populated colla-

gen gels were individually suspended in organ baths (35#{176}C),stretched to and after equilibration exposed to isopren-

aline (10 p.mol/l) and then carbachol (10 p.mol/l). b) Statis-tical evaluation of all experiments in which the effects of

isoprenaline and carbachol were tested. Predrug values offorce of contraction were 0.11 ± 0.02 mN (n=13). The mean

spontaneous beating rate of all experiments evaluated was72.6 ± 2.6 bpm (n=35). Number in bars = number of CMPMs

measured.

of the tissue showed a characteristic concentration oflongitudinally orientated cells at the free edges of thegel, forming a structure comprised of several cell lay-ers similar to embryonic heart tissue (Fig. 2a, b). Inthe central region, cells were less concentrated andless regularly orientated, forming a loose reticularnetwork (not shown). Cross sections parallel to theglass tubes confirmed the cell concentration at thefree edges as a thick, caplike multicellular layer withtransition to a thinner (1-3 cells thick) but still con-tinuous cellular cover at the surfaces of the centralparts of the gels (not shown). In immunohistochem-ical stains, more than 80% of the cells showed a typ-ical marker spectrum of cardiac myocytes: muscleactin (Fig. 2c, d), cx-sarcomeric (cardiac and skeletal),actin (Fig. 2e), and sarcomenc ct-tropomyosin (Fig.2]). Of these, about 30-50% were elongated cell fi-bers with clear cross-striation (Fig. 2d) and network-like end-to-end contacts, demonstrating terminal

1 iJIiL

30 mm

] O.2mN

Figure 6. Stability of force measurements. Original tracing of

a representative force measurement of a CMPM (day 10). TheCMPM was suspended in a organ bath, paced at 1.5 Hz, al-

lowed to equilibrate, and then subjected to different physio-logical and pharmacological manipulations as indicated by

arrows and numbers. 1: Stepwise increase in resting tension(Frank-Starling); 2: cumulative addition of carbachol, 0.01,0.1, 1, 10, 100 p.mol/I; 3: cumulative addition of pirenzepine

0.01, 0.1 p.mol/l; 4: nadolol 1 p.mol/l; 5: cumulative addition

of norepinephrine 0.01, 0.1, 1, 10 p.mol/l; 6: cumulative in-creases in extracellular CaGI2 to 3.6, 5.4, 7.2, 9.0, 10.8, 12.6,

14.4 mmol/l; 7: increase in extracellular CaG12 to 5.4 mmol/I and start of overnight measurement. Asterisks indicate

changes in bathing solution with control Tyrode’s solution.

Note that for overnight measurements, a continuous flow ofTyrode’s solution with increased CaCl2 concentration (5.4

mmol/I) was used, which caused a mechanical disturbance of

the mechanogram.

a

b

[0.2 mN

5 mih

0.25E

0.2004-C)I.-C0C)

9-

0.5

*

electrically paced (-.20 V, 10 ms, rectangular pulses).In contrast to intact muscle preparations such as iso-lated papillary muscle, force of contraction of paced

CMPMs remained remarkably stable for up to 20 h(Fig. 6). In addition, when CMPMs were detachedfrom the force transducer, recultured for 1-2 days,

and resuspended in organ baths, they still beat spon-taneously and developed similar force of contractionas long as semisterile conditions were maintained.

Physiological characteristics

CMPMs were suspended in continuously gassed Ty-rode’s solution in organ baths and allowed to equil-

ibrate at the original (spacer) length, which wasdefined as 100% or L0. Rest length was stepwise in-creased from this position to the length of maximalforce development which amounted to about120% of the original length. Stretching increasedresting tension from 1.15 ± 0.03 to 2.81 ± 0.11 mN

(n=69), accompanied by an increase in twitch am-

a 2 pmol/I

L0 Lm

2.5 sec 2.5

b4mN

zE3C0U,Ca)4-0)C

CI)a) 1

*

0

Ctr Cyto D2 pmol/l*p<005 vs. Ctr

CARDIAC MYOCYTE-POPU LATED COLLAGEN MATRIX 689

p<0.05 vs. L0

Figure 7. Starling behavior of CMPMs. a) Original tracing of

a representative experiment. CMPMs were electrically pacedat 1.5 Hz, stretched to the original (spacer) length, and equil-

ibrated. The length (preload) was increased stepwise by about

3% of the original length per step until no further increaseor decrease in twitch amplitude (force) was seen. Force was

evaluated after equilibration (5-10 mm). Note the gradualincrease in twitch amplitude with increasing preload. b) Sta-

tistical evaluation of all experiments. L0 is the original spacerlength, Lm, the length at which twitch amplitude was maxi-

mal (ca. 120% of spacer length). Number in bars = numberof CMPMs measured.

than threefold, from 4 to 14 .i.g per CMPM (Fig. 4b).

In addition, the total number of nuclei (counted byfluorescent stain) remained stable between 240,000and 360,000 per CMPM (not shown).

Contractile behavior

Baseline parameters

After 7-11 days in culture, isometric force was mea-sured in continuously gassed (95% 02, 5% C02) Ty-rode’s solution at 35#{176}Cin standard organ baths.Resting tension was 1.15 ± 0.03 mN (n=69) at theinitial (spacer) rest length. Under these conditions,the cardiomyocyte-populated gels contracted spon-taneously with a frequency of 72.4 ± 2.6 beats perminute (bpm; n=35; Fig. 5a). Preparations could be

Figure 8. Effect of actin depolymerization. a) Original tracing

of a representative experiment. CMPMs were electrically

paced (1.5 Hz), stretched to L,,, equilibrated, and then ex-posed to cytochalasin D (2 p.mol/I). Despite the large decreasein resting tension, significant twitch amplitude remains after

exposure to cytochalasin D. b) Statistical evaluation of the rest-

ing tension and the tension after addition of cytochalasmn D.

Number in bars = number of CMPMs measured.

a 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.5 1.5 Hz

+ + + + + +++ +

ri.iiI1ih2.5 mm 2.5 sec

b 0.3

* * * *

0.1 ___________

n=67IL

0 .....i I I I I I

0 0.8 1 1.2 1.4 1.6 1.8 2 2.5

Frequency (Hz)*p<005 vs. 0.8 Hz

.4f-, Pharmacological responses

*

b 0.4

z

E0

0

C

o.

0 1.8 3.6 5.4 7.2 9 10.8 12.6 14.4

Concentration of Ca2’ (mmo/I)

p’c0.05 vs. 1.8 mmol/I

Figure 10. Inotropic response to calcium, a) Original tracingof a representative experiment. After equilibration, stretching

to L,,,,, and testing of the force-frequency relationship,CMPMs were electrically paced at 1.5 Hz and exposed to a

cumulative increase in the extracellular calcium concentra-

tion from 1.8 mmol/l (Tyrode’s solution) to 14.4 mmol/l, 5mm each. Force was evaluated after equilibration (5 mm). b)Statistical evaluation of all experiments.

n=33

690 Vol. 11 July 1997 The FASEB Journal [SCHENHAGEN ET AL.

mammals including humans and chickens show anincreased twitch amplitude with increasing fre-

quency. The force-frequency relationship forCMPMs containing chick cardiomyocytes resembledthat of the chicken. CMPMs were equilibrated with-out pacing to test spontaneous frequency, then pacedat 1.5 Hz, stretched to L,,,ax, and again equilibrated.Pacing frequency was then reduced to 0.8 Hz andincreased stepwise for 5 mm. Twitch amplitude in-creased with the pacing frequency between 0.8 and

2.0 Hz (0.15±0.01 to 0.21 ±0.014 mN; n=67; Fig. 9)and decreased at higher frequencies. The responseto frequency changes was not completely reproduc-ible. Of the 104 CMPMs tested, 67 responded withan increase in maximal twitch amplitude; the rest re-mained unchanged. Only 1/104 gels showed a neg-ative response. The reason for this variability isunknown. CMPMs casted with rat neonatal cardio-myocytes showed a negative staircase, which resem-

bles that of intact rat cardiac tissue (14; not shown).

Cumulative increases in the extracellular calcium

concentration from 1.8 mmol/l (Tyrode’s solution)

Figure 9. Force-frequency relationship. a) Original tracing of a 3.6 5.4 7.2 9.0 10.8 12.6 14.4 mmol/Ia representative experiment. After stretching the CMPMs to + + + + + + +L,, the electrical pacing rate was decreased to 0.8 Hz andthen stepwise increased to 2.5 Hz, 5 mm per step. Force was ‘i

evaluated after equilibration. Note the gradual increase intwitch amplitude with increasing frequency and the steep de-crease between 2.5 and 1.5 Hz. b) Statistical evaluation of allCMPMs that showed a change in force of contraction with 2.5 sec 2.5 mmincreasing pacing frequency (67/104).

plitude from 0.09 ± 0.006 mN to 0.2 ± 0.01 mN(n=69; Fig. 7). Further stretching decreased twitchamplitude. Thus, the CMPMs exhibit a positivelength-tension relationship (Frank-Starling mecha-nism), which is characteristic of heart tissue. Todifferentiate how much of the resting tension wasdue to passive elements (matrix) and how much dueto active forces, CMPMs were stretched to Lmax andthen exposed to 2 Rmol/l cytochalasin D, an agentknown to disrupt nonsarcomeric actin organization(Fig. 8). Cytochalasin D markedly reduced restingtension from 3.1 ± 0.1 to 1.31 ± 0.11 mN (n=26).When collagen gels were cast in the absence of cells,cultured as CMPMs, and stretched to about 120% ofLi,, passive tension amounted to about 1 mN (notshown). This indicates that the resting tension ofCMPMs is derived to about 2/3 from (cytochalasin-sensitive) active forces and to about 1/3 from passiveforces.

Another physiological parameter tested was theforce-frequency relation or “staircase” phenome-non (13). Cardiac muscle preparations from most

CARDIAC MYOCYTE-POPULATED COLLAGEN MATRIX 691

Figure 11. Adenovirus-mediated gene transfer into CMPMs. a) Photograph of a native CMPM in the culture dish. Note the

concentration of cells (white appearance) at the biconcaval edges. b) Photograph of a paraffin-embedded CMPM that had been

infected on day 6 of cultivation with 5 x i07 pfu Ad5LacZ, washed, briefly formaldehyde-fixed, and stained with 0.5 mg/mIX-gal for 18 h at 4#{176}C.After overnight postfixation with 10% formaldehyde, the CMPM was embedded in paraffin and sectioned

parallel to the plane of the tissue. Note the homogeneous distribution of blue cells throughout the CMPM with the typicalconcentration of (blue) cellsat the free edges. V = Velcro, T = silicone tube. c) 5 sm section from an uninfected CMPM treatedidentically and in parallel with that in panel d. Note the absence of nonspecific staining for 13-galactosidase.Counterstain with

nuclear fast red. d) 5 m.tm section from a paraffin block similar to that in panel b) showing one of the two free edges. Note thatabout 50% of cells are stained blue, indicating the presence of 3-galactosidase. Magnifications: 256X (c, d).

to 12.6 mmol/l resulted in a marked and highly re-producible increase in twitch amplitude (0.11±0.013to 0.34±0.05 mN; n=33; Fig. 10) without a changein resting tension. The increase reached its maxi-mum in about 3 mm and was completely reversibleafter washout in about 10 mm. CMPMs also showeda slowly developing positive inotropic response to cu-mulative addition of the digitalis glycoside ouabain(0.1 to 10 j.tmol/l) that was reversible after washoutin about 60 mm (not shown).

The response of CMPMs to the 3-adrenergic ago-nist isoprenaline was complex (Fig. 5). As expectedfor heart muscle, isoprenaline (10 imol/l) increasedbeating frequency by 24.7% (51±6 to 64±6 bpm,n = 13, Fig. 5). However, unexpectedly, it also loweredresting tension and only marginally increased maxi-mal twitch amplitude (0.11±0.02 to 0.13±0.02 mN,n=13, P<0.05, not shown). Addition of the musca-rinic agonist carbachol (10 jimol/l) decreased beat-ing frequency to below baseline values (29±6 bpm,Fig. 5), increased passive tension back to baseline val-

ues, and increased twitch amplitude to 0.14 ± 0.03mN (n=13, P<0.05). Effects of both agonists werereversible (not shown). Cumulative addition of iso-

prenaline (0.01 to 10 p.mol/l, not shown) to pacedpreparations resulted in a concentration-dependentand reproducible decrease in resting tension (from0.93 ± 0.09 mN to a minimum of 0.71 ± 0.08 mN at0.3 Rmol/l, n=9, P<0.05), a small increase in maxi-mal twitch amplitude (from 0.15±0.03 to maximally

0.2±0.03 mN at 0.3 l.tmol/l, n=9; P<0.05), and a pos-itive lusitropic effect, i.e., shortening of contractiontime from 241 ± 14 to 183 ± 14 ms (n=8; P<0.05).The receptor-independent adenylyl cyclase activatorforskolin (3 imol/l) and dibutyryl cAMP (1 mmol/1) exerted effects that were qualitatively and quanti-tatively similar. The effect of cAMP-increasing agentsor carbachol showed no apparent changes with timein culture between days 4 and 15 and was also seenin CMPMs cultured in the presence of BrdU or theabsence of glutammne, which resulted in growth in-hibition of noncardiomyocytes (not shown).

692 Vol. 11 July 1997 The FASEB Journal ESCHENHAGEN ET AL.

Genetic manipulation

To the extent described, we were able to show thatCMPMs display important histological and functionalcharacteristics of intact heart muscle. To test theirapplicability as an in vitro transgenic model, CMPMswere exposed for 48-72 h to a recombinant adeno-virus type 5 carrying the cDNA for the bacterial 1-galactosidase driven by a CMV promotor (Ad5LacZ)and then evaluated for the presence of 3-galactosi-

dase. Infection with Ad5LacZ resulted in a dose-de-pendent, highly efficient gene transfer (Fig. 11).Transduction efficiency depended on the multiplic-ity of infection (MO!; 0.1-100) and the day of infec-tion (Table 1). Optimal gene transfer was obtainedwhen CMPMs were infected at day 6 with a MOl of100 and incubated for 3 days. Then 30-50% of cellsstained positive for f3-galactosidase. Accordingly, thebiochemically determined 3-galactosidase activity in-

creased from a minimum of 49 ± 16 U/b6 cells atday 8 to 253 ± 59 U/b6 cells at day 6 (Table 1).Histological sections revealed rather homogeneous

cell staining throughout the gel without an apparentpreference for specific cell types (Fig. 1 1D). The con-

centration of longitudinally oriented cells at the freeedges (Fig. 1 1A) was reflected by the higher numberof 13-galactosidase-positive cells (Fig. 11B).

DISCUSSION

This paper is the first description of a new method

for reconstituting embryonic cardiomyocytes in a 3-dimensional collagen matrix in a way that allows mea-surement of the isometric force of contractiongenerated by the resulting CMPMs. Three conse-quences of these results are noteworthy. First, this isthe first demonstration that embryonic cardiomyo-cytes can be reconstituted in a 3-dimensional colla-gen matrix to form a netlike, coordinately beatingheartlike tissue. Second, the experimental approachprovides simple and stable measurement of isometricforce of contraction with standard force transducers.Third, the reconstituted matrix allows highly efficient

adenovirus-mediated gene transfer into the cardio-myocytes. These unique features should allow theconsequences of genetic manipulation of cardiomyo-cytes to be measured under controlled conditions interms of their effects on contractility and beating fre-quency of the model heart tissue. Furthermore, theCMPMs should provide a simplified model withwhich to investigate cell and matrix interactions, celland tissue mechanics, and processes involved in tis-sue development.

The idea of growing or maintaining cells in a 3-dimensional collagen matrix is not new and has beenstimulated by the limitations of monolayer culturesdiscussed earlier. Early experiments have shown thatendothelial cells form capillary-like structures in col-lagen gels (15). Since then, several cell types includ-ing dermal fibroblasts (2), chondrocytes (16),mammary cells (17), endothelial cells (15), and skel-etal myotubes (18) have been grown in collagen gels.

Earlier attempts to grow cardiomyocytes in a 3-cu-mensional collagen matrix did not yield a coherentcontractile tissue (19). These investigators reportedthat neonatal rat cardiomyocytes attach to preformedcollagen gels but do not spread and form electricalcontacts inside the matrix. By systematically varying

the conditions of cell isolation, casting of the gels,and culturing we have identified factors that criticallydetermine the performance of the reconstituted tis-sue and are suitable to prepare spontaneously andcoordinately beating CMPMs. These factors are 1) avery gentle cell isolation with a tested batch of colla-

genase, 2) the use of DNase I between the collage-nase cycles to allow cells to detach from tissueresidues, 3) washing of cells before the casting of gels,and 4) the use of chick embryo extract.

Adherence to these conditions ensured formationof CMPMs with highly reproducible mechanical andphysiological characteristics. Compared to experi-ments on monolayers and preformed collagen gelspopulated with cardiomyocytes, this system has sev-eral important advantages.

1) The 3-dimensional structure of CMPMs leads tomore physiological intercellular contacts, not onlyend-to-end (intercalated discs) but also side-to-side(desmosomes and tightjunctions). These histological

TABLE 1. Time dependenrv of adenovirns-mediated gene transfer into CMPMs”

Infection day

Monolayer4 6 8 10

13-Galactosidase activity (U/b6 cells) 101* 253* 49 76 522*SEM 8 59 16 27 40Number of experiments 6 6 8 8 4

CMPMs were infected with 1 ml of Ad5LacZ-containing culture supernatant (ca. 5 X iO pfus) atcultivationday 4, 6,8, or 10 and culturedfor an additional72 h with complete medium. f3-Galactosidase activity was determined by a colorimetric assay using ONPG as the substrate.Monolayer culturesof embryonic chick cardiomyocytes were infectedin parallelas positive controls. Note that in these cultures transductionefficiency, quantified by histochemical means, approached 100%. Each experiment was performed in triplicate.*J)< 0.05 vs. day 8.

CARDIAC MYOCYTE-POPULATED COLLAGEN MATRIX 693

features better resemble the architecture of the myo-

cardium and appear to maintain a higher degree ofcellular differentiation than the monolayer. In linewith this reasoning are a high percentage of cardio-myocytes with well-organized sarcomeric structuresand the decrease in [5H]thymidine incorporationinto the CMPMs over time in spite of the permanentpresence of serum. Under identical conditions,monolayer cultures maintained a high proliferativecapacity. The decrease in [3H]thymidine incorpora-tion in CMPMs is not due to cell death or a decreasein biological activity, because during cultivation the

number of cells remained more or less unchangedand the RNA yield increased threefold. Yet, the de-gree of differentiation should be better defined by

measuring the isoform expression pattern of markerproteins.

2) In contrast to floating cell-populated collagengels that shorten over time, resisted only by the vis-

coelasticity of the matrix, CMPMs that are held atfixed length subject the cardiomyocytes to a stable

isometric load (stress), which can be either imposedexternally or defined by the cardiomyocytes them-selves. Again, this better simulates the situation insitu. The alignment of the cells at the free edges ofthe CMPMs in contrast to the isotropic distribution

of the cells in the interior suggests that the load isnot evenly distributed throughout the gel. Load is animportant growth regulator in the developing heart

(20, 21). Steady unidirectional stress aligns the cellsparallel, repetitive stretch-relaxation perpendicularto the direction of stress. This has been shown con-sistently for skeletal myoblasts (18), neonatal rat car-diomyocytes (22, 23), and rat aortic smooth muscle

cells (24, 25). The model of CMPMs will be ideallysuited to test directly the effect of different mechan-ical stresses on cardiomyocyte alignment and growth.In contrast to other models, such as cardiomyocytescultured on flexible silicone matrices (26), it will alsoallow us to measure the functional consequences ofthese interventions.

3) A unique feature of the CMPMs is that they allow

easy quantitative measurement of isometric force ofcontraction without further manipulation of the ma-trix. In contrast to monolayer preparations, preload

can be adjusted, enabling measurements of the de-pendence of contractile performance on load usingstandard force transducers in conventional organbath experiments. The system is robust and allowsforce measurement of paced preparations over manyhours (and even days, as long as CMPMs are inter-

mittently recultivated). Whereas normal papillary orwhole heart preparations exhibit a pronounced de-crease in force development over these time periods,decreases in basal force exerted by CMPMs are quitemodest. We interpret these results as indicating the

absence of ischemia or nutritional deficits inside thematrix.

Despite the simplicity of their composition and or-ganization, CMPMs are similar to heart muscle in

their responses to several standard inotropic inter-ventions. The reconstituted tissue showed an increasein maximal twitch amplitude with increasing restlength (Frank-Starling mechanism), with increasingfrequency (“positive staircase”), with increasing ex-tracellular calcium, and after application of digitalisglycosides. These responses resemble both qualita-tively and quantitatively the physiological behavior ofintact embryonic heart tissue under similar condi-

tions. In contrast, CMPMs showed only a small posi-tive inotropic response to isoprenaline, which wasaccompanied by a decrease in resting tension. Thiswas also seen in response to the direct adenylyl cy-

clase activator forskolin and the cAMP-derivative di-butyryl-cAMP, excluding specific defects of3-adrenoceptors, stimulatory G-proteins, or the ade-nylyl cyclase. The decrease in resting tension resem-

bles the well-known response of smooth muscle cells,which in view of their high proliferative capacity inthe presence of serum may have increased in numberrelative to the cardiac myocytes. However, no appar-ent difference in the isoprenaline or carbachol re-sponse was seen after cultivation of CMPMs in thepresence of bromodeoxyuridine, the absence of glu-

tamine, or at earlier (days 6-7) or later (days 13-15,not shown) stages of cultivation. In addition, stan-dard hematoxylin stain and immunohistochemistiyrevealed a clear predominance of striated muscle

cells. The increase in resting tension of CMPMs afteraddition of a high potassium Tyrode’s solution (125mmol/l KC1 instead of NaC1) was not sensitive to thecalcium channel blocker verapamil (10 l.tmol/l; notshown). Thus, it appears as if the model of CMPMsdemasks an unusual contractile behavior of primary

cardiomyocytes in culture that has been largely over-looked in monolayer experiments and, at this point,partly limits conclusions regarding the 13-adrenergicsystem. On the other hand, CMPMs should providea suitable model system to investigate underlying rea-sons and to define conditions under which the 3-ad-renergic effect better resembles that in intact heart

tissues.We have shown that gene transfer with a recom-

binant adenovirus is highly efficient. Gene transferreached 30-50% of all cells at a MO! of about 100and at early stages of the formation of a tissue-likestructure. At later stages transduction efficiency de-

creased, most likely due to the more compact struc-ture of the matrix that prevented the virus from

approaching the cells. At this stage we did not usepurified virus preparations, which are expected toyield even higher transduction rates. The simplicityof gene transfer is an important advantage over ex-periments in vivo or with isolated intact tissues inwhich transfection or infection efficiency is report-edly low (27).

694 Vol. 11 July 1997 The FASEB Journal ESCHENHAGEN E AL.

Thus, the newly developed heart muscle cell

culture system combines the feasibility of directmolecular genetic, pharmacological, or mechanicalmanipulation of cultured cells with the advantages of

a 3-dimensional simplified model tissue and the pos-sibility of measuring contractile forces. As a newly de-veloped experimental system, many aspects of the

CMPMs remain to be studied. These include opti-mization of their size and shape, exact definition of

the percentage of cardiac myocytes as compared tofibroblasts, smooth muscle cells, endothelial, andneural cells, the degree of differentiation, the geneexpression pattern, possibilities of transferring thetechnique to cardiac cells of different species and ma-turity (and also to other cell types), as well as a fineanalysis of functional consequences of infection withrecombinant adenoviruses.

We wish to thank Dr. Harold Prentice, Glasgow, for the kindgift of the Ad5LacZ adenovirus, and Dr. Michael Kolodney forhelpful discussions. We are grateful tojochen R#{252}diger,Ham-

burg, for his help in graphics. This work was supported by

grants from the Deutsche Forschungsgemeinschaft (Es 88/6-

1 and Es 88/8-1) and from the National Institutes of Health

(E.L.E.). Part of this work was published in abstract form atthe 68th Scientific Sessions of the American HeartAssociation;Circulation, Vol. 92 (Suppl. I), p. 1-590, 1995. The work is partof the doctoral thesis ofJ.W. at the University of Hamburg.

REFERENCES

I. Cavanaugh, M. W. (1955) Growth of chick heart cells in mono-layer culture.]. Exp. Zol. 128, 573-581

2. Mauch, C., Hatamochi, A., Scharffetter, K., and Krieg, T. (1988)Regulation of collagen synthesisin fibroblastswithin a three-dimensional collagen gel.Exp. Gel!. Res. 178, 493-503

3. Allen, I. S., Gaa, S. T., and Rogers, T. B. (1988) Changes inexpression of a functional Gi protein in cultured rat heart cells.Am.]. Physiol. 255, C51-C59

4. Eppenberger-Eberhardt,M., Flamme, I., Kurer, V., and Eppenber-ger. H. M. (1990) Reexpres.sion of a-smooth muscle actin isoformin cultured adult rat cardiomyocytes. Dev.BioL 139, 269-278

5. Fitzgerald, M., Neylon, C. B., Marks, A. R.,amd Woodcock, E. A.(1994) Reduced ryanodine receptor content in isolated neonatalcardiomyocytes compared with the intacttissue.jMol. Cell. Car-diol. 26, 1261-1265

6. Vincan, E., Neylon, C. B., Graham, R. M., and Woodcock, E. A.(1995) Isolation of neonatal cardiomyocytes reduces the expres-sion of the GTP-binding protein, Gh. j Mol. Cell. Cardiol. 27,2393-2396

7. Hasenfuss, G., and Just, H. (1994) Myocardial phenotypechanges in heart failure: cellular and subcellular adaptations andtheir functional significance. Br. Heart]. 72 (Suppl.), S10-S17

8. Eschenhagen, T.,Hollmann, A., Gsell, S., Fnedrichsen, M., Schmitz,W., Scholz, H., Weil, J., and Weinstein, L S. (1995) Regulation ofthe human Gia-2 gene promotor actMty in embryonic chicken car-diomyocytes.BasicRet.CardioL 91 (Suppl. 2), 41-46

9. Kolodney, M. S., and Elson, E. L. (1993) Correlation of myosinlight chain phosphorylation with isometric contraction of fibro-blasts.]. Biol. Che,n. 268, 23850-23855

10. Eschenhagen, T., Mende, U., Diederich, M., Geertz, B., Hertle,B., Memmesheimer, C., PohI, A., Schmitz, W., Scholz, H., Stein-fath, M., B#{244}hm,M., Michel, M. C., and Brodde, 0. E. (1996)

Chronic treatment with carbachol sensitizes the myocardium tocAMP-induced arrhythmias. Circulation 93, 763-771

11. Goldner, J. (1938) Modification of the Masson trichrome tech-nique for routine laboratory purpose. Am.J. Pat/jo!. 14, 237-243

12. Chomczynski, P., and Sacchi, N. (1987) Single-step method ofRNA isolation by acid guanidinium thiocyanate-phenol-chloro-form extraction. Anal. Biochem. 162, 156-159

13. Bowditch, H. P. (1871) Uber die Eigenthumlichkeiten der Reiz-barkeit, welche die Herzmuskelfaserns des Herzens zeigen. Ber.S#{224}chs.Get. Wiss. 23, 652-689

14. Borzak, S., Murphy, S.,and Marsh,J. D. (1991) Mechanisms of ratestaircase in rat ventn cular cells. Am.]. PhysioL 260, H884-H892

15. Montesano, R., Orci, L., and Vassalli, P. (1983) In vitro rapidorganization of endothelial cells into capillary-like networks ispromoted by collagen matrices.]. Cell Biol. 97, 1648-1652

16. Gibson, G. J., Schor, S. L., and Grant, M. E. (1982) Effects ofmatrix molecules on chondrocyte gene expression: synthesis ofa low molecular weight collagen species by cells cultured withincollagen gels.]. Cell Biol. 93, 7190-7194

17. Rocha, V., Ringo, D. L., and Read, D. B. (1985) Casein produc-tion during differentiation of mammary cells in collagen gel cul-ture.Exp. Cell Res. 159, 201-210

18. Vandenburgh, H. H., Karlisch, P., and Fan, L. (1988) Maintai-nance of highly contractile tissue-cultured avian skeletal myotu-bes in collagen gel. In Vitro Cell Dee. Biol. 24, 166-174

19. Souren,J. E. M., Schneijdenberg, C., Verkele1j, A.J., and Van Wijk, K(1992)Factors controllingthe rhythmic contraction of collagen gelsby neonatal heart cells. In Vilm Cell Dev. BioL 28A, 199-204

20. Cooper, G. (1987) Cardiocyte adaptation to chronically alteredload. Annu. Rev. PhysioL 49, 501-18

21. Morgan, H. E., and Baker, K M. (1991)Cardiac hypertrophy. Me-chanical, neural, and endocrine dependence. Circulation 83, 13-25

22. Terracio,L., Miller, B., and Borg, T. K. (1988) Effects of cyclicmechanical stimulation of the cellular components of the heartin vitro.In Vitro Cell. Den. Biol. 24, 53-58

23. Vandenburgh, H. H.,Solerssi,R., Shansky,J., Adams,J. W., Hen-derson, S. A., and Lemaire,J. (1994) Response of neonatal ratcardiomyocytes to repetitive mechanical stimulation in vitro.Ann. N.Y. Acad. Sci. 752, 19-29

24. Schnetzer, K., Delafontaine, P., and Nerem, R. (1995) Effectofuniaxial, cyclic stretch on the morphology of monocytes/mac-rophages in culture.]. Biornech. Eng. 118, 420-422

25. Ziegler, T., and Nerem, R. M. (1994) Tissue-engineeringa bloodvessel: regulation of vascular biology by mechanical stresses. J.Cell. Biochem. 56, 204-209

26. Komuro, I., Kudo, S., Yamazaki, T., Zou, V., Shiojima, I., andYazaki. Y. (1996) Mechanical stretch activates the stress-activatedprotein kinases in cardiac myocytes. FASEBJ. 10, 631-636

27. Acsadi, G., Massie, B., and Jani, A. (1995) Adenovirus-mediatedgene transfer into striated muscles.]. Mo!. Med. 73, 165-180

Recei ved for publication February 19, 1997.Accepted for publication May 2, 1997.