an autocrine function for transforming growth …developmental biology 194, 99–113 (1998) article...

DEVELOPMENTAL BIOLOGY 194, 99–113 (1998)ARTICLE NO. DB978807

An Autocrine Function for Transforming GrowthFactor (TGF)-b3 in the Transformation ofAtrioventricular Canal Endocardium intoMesenchyme during Chick Heart Development

Yuji Nakajima,* Toshiyuki Yamagishi,* Hiroaki Nakamura,*Roger R. Markwald,† and Edward L. Krug†*Department of Anatomy, Saitama Medical School, Saitama, 350-0495 Japan;and †Department of Cell Biology and Anatomy, Medical Universityof South Carolina, Charleston, South Carolina 29425

Transformation of atrioventricular canal endocardium into invasive mesenchyme is a critical antecedent of cardiac septationand valvulogenesis. Previous studies by Potts et al. (Proc. Natl. Acad. Sci. USA 88, 1510–1520, 1991) showed that treatmentof atrioventricular canal endocardial and myocardial cocultures with TGFb3 antisense oligodeoxynucleotides blockedmesenchyme formation. Based on this observation, we sought to: (i) identify the target tissue of TGFb3 antisense oligosin this transformation bioassay, and (ii) more clearly define the mechanism of TGFb3 function in atrioventricular canalmesenchyme formation. In situ hybridization and immunohistochemistry showed little or no TGFb3 mRNA or proteinin the atrioventricular canal myocardium or endocardium prior to mesenchyme formation (stage 14; paraformaldehydefixation). However, by stage 18 transforming atrioventricular canal endocardial cells and mesenchyme as well as myocar-dium were positive for both TGFb3 mRNA and protein. In culture bioassays, atrioventricular canal endocardial monolayerspretreated with antisense phosphorothioate oligodeoxynucleotides to TGFb3 did not transform into invasive mesenchymein response to cardiocyte conditioned medium: the subsequent addition of exogenous TGFb3 protein relieved this inhibi-tion. Control cultures without pretreatment or those receiving missense oligos generated similar numbers of invasivemesenchyme in response to cardiocyte conditioned medium. Direct addition of TGFb3 protein to atrioventricular canalendocardial monolayers in the absence of cardiocyte conditioned medium resulted in loss of cell:cell associations andstimulated cellular hypertrophy, but did not engender invasive mesenchyme formation or alter endocardial proliferationafter 24 h of culture. Similar results were obtained with TGFb2 protein, either alone or in combination with TGFb3. Theresults of this study indicate that: (i) atrioventricular canal endocardium expresses TGFb3 in response to a myocardiallyderived signal other than TGFb3, (ii) atrioventricular canal endocardial TGFb3 functions in an autocrine fashion to elicitselected characteristics necessary for cushion tissue formation, and (iii) TGFb3 alone or in combination with TGFb2 isinsufficient to transform atrioventricular canal endocardium into invasive mesenchyme in culture. q 1998 Academic Press

Key Words: cardiogenesis; cushion tissue; TGFb3; chick embryo; antisense oligodeoxynucleotides; epithelial–mesenchy-mal transformation.

INTRODUCTION myocardium) separated by an expanded acellular extracellu-lar matrix (ECM),1 also referred to as cardiac jelly (Davis,1924; Markwald et al., 1977, 1978). By stage 14 the chickThe ability to generate migrating mesenchymal cell popu-

lations from epithelial sheets in a reproducible temporal andspatial pattern is a common mechanism used throughout 1 Abbreviations used: AV, atrioventricular; CCM, cardiocyteembryogenesis. One of the earliest events in organogenesis conditioned medium; ECM, extracellular matrix; OT, outflowis the formation of the primitive heart tube, which consists tract; PTO, phosphorothioate oligodeoxynucleotides; TGFb,

transforming growth factor b; PBS, phosphate-buffered saline.of two concentric epithelial layers (i.e., endocardium and

99

0012-1606/98 $25.00Copyright q 1998 by Academic PressAll rights of reproduction in any form reserved.

AID DB 8807 / 6x36$$$281 02-02-98 10:37:54 dbas

100 Nakajima et al.

embryonic heart is a right-side looped structure that is di- tion of AV endocardium, several questions remain regardingits involvement on a mechanistic level. None of the datavided into presumptive sequential segments, i.e., primitive

atrium, atrioventricular canal (AV) region, primitive ventri- to date differentiate between three possible mechanisms ofaction: (i) myocardially derived TGFb3 could function in ancle, and outflow tract (OT) (De la Cruz et al., 1989). At stage

16 a subpopulation of endocardial cells in the inferior AV autocrine fashion to elicit secretion of a paracrine transfor-mation activity; (ii) myocardially derived TGFb3 could tra-region transforms from an epithelial to a mesenchymal mor-

phology and invades the subjacent extracellular matrix, fol- verse the ECM to signal certain characteristics of the trans-formation process, with ultimate invasion regulated by a co-lowed by a similar transformation of the apposing superior

AV endocardium 10 h later (Moreno-Rodriguez et al., 1997). operative activity as yet unidentified; or (iii) endocardialexpression of TGFb3 is an essential response of target cellsThis invasive mesenchyme becomes the AV endocardial

cushion tissue that constitutes the primordial valves of the for epithelial-to-mesenchymal transformation.The focus of the present study was to assess the roleembryonic heart and is a critical component of valvular and

septal structures in the mature heart (Moller et al., 1994; of TGFb3 expression by AV endocardium relative to itstransformation into cushion mesenchyme. The resultantNakajima et al., 1996a,b; Wessels et al., 1996).

The regulation of epithelial-to-mesenchymal transfor- data are consistent with endocardially derived TGFb3 func-tioning in an autocrine manner to manifest endocardial hy-mations is likely to involve both paracrine and autocrine

factors. By using a three-dimensional collagen gel culture pertrophy and a loss of cell:cell adhesion, which are charac-teristic and necessary events in cushion mesenchyme for-model it has been shown that the epithelial–mesenchymal

transformation of AV endocardium is dependent upon co- mation. The role of myocardially expressed TGFb3 is yetto be elucidated.culture with AV myocardium; ventricular myocardium

does not elicit mesenchyme formation (Runyan and Mark-wald, 1983), suggesting a regionally restricted, paracrine-acting factor produced by myocardium. Subsequent studies MATERIALS AND METHODSshowed that transformation-promoting activity was pres-ent in conditioned medium from cardiocyte cultures. The In Situ Hybridizationgeneration of mesenchyme from AV endocardium by treat-

Digoxygenin-labeled single-strand RNA probes were prepared us-ment with cardiocyte conditioned medium could be ing a DIG RNA labeling kit (Boehringer Mannheim, Indianapolis,blocked by antibodies to EDTA-soluble cardiac ECM, dem- IN). A 462-bp fragment of chicken TGFb3 cDNA (base pairs 31–onstrating the presence of a soluble inductive factor (Krug 492 in Accession No. M31154) was subcloned into pBluescriptet al., 1987; Markwald et al., 1990a,b; Mjaatvedt et al., (KS0). An antisense probe was produced using EcoRI and T3 RNA

polymerase; a sense probe was generated using HindIII and T7 RNA1991; Rezaee et al., 1993).polymerase. Staged embryos (Hamburger and Hamilton, 1951) wereTransforming growth factor-b (TGFb) is a family of pro-fixed with 4% paraformaldehyde in PBS and embedded in paraffin.teins thought to regulate cellular proliferation and differenti-Sections of 6–8 mm were cut and mounted onto slides coated withation at different stages of embryogenesis in vertebrates. The3-triethoxysilylpropylamine (Merck, Darmstadt, Germany), depar-distribution patterns of several TGFbs and their receptors inaffinized, hydrated, refixed with 4% paraformaldehyde in 0.1 M PBdeveloping organs suggest that they may have signaling roles(phosphate buffer) for 15 min, rinsed with 0.1 M PB, and digested

in epithelial–mesenchymal interactions (Sporn et al., 1986; with proteinase K (3 mg/mL) for 10 min at 377C, then acetylatedIgnotz and Massague, 1986; Yang and Moses, 1990; Mac- with 0.25% acetic anhydride in 0.1 M triethanolamine (pH 8.0) forLellan et al., 1993; Akhurst, 1994). During chicken endocar- 10 min. Sections were dehydrated next through a graded seriesdial cushion tissue formation, TGFb is expressed by both of ethanol, air-dried, and hybridized (0.5 mg/mL of probe in 50%

deionized formamide containing 10 mM Tris–HCl, pH 7.6, 200 mg/endocardial cells and endocardially derived mesenchymalmL yeast tRNA, 11 Denhardt’s solution, 10% dextran sulfate, 600cells as well as myocardium (Choy et al., 1991; Runyan etmM NaCl, and 0.25% SDS) at 507C for 14–16 h in a moist chamber.al., 1992; Nakajima et al., 1994; Ghosh and Brauer, 1996).These sections were rinsed with 51 SSC followed by 50% for-Interestingly, the expression of TGFb1/b3 immunoreactivitymamide, 21 SSC at 607C for 30 min each (twice), then 0.21 SSC forby AV and OT endothelial cells is upregulated in vivo just20 min at 607C (twice). Hybridization was detected using alkaline-prior to mesenchyme formation and can be affected in cul-phosphatase-conjugated anti-digoxygenin antibody (Boehringer

ture by cardiocyte conditioned medium (Nakajima et al., Mannheim; diluted 1:2000) and BCIP/NBT according to the manu-1994). Potts et al. (1991) showed that TGFb3 expression was facturer’s instructions.critical to AV mesenchyme formation by treating coculturesof AV endocardium and myocardium with antisense oligo-

Indirect Immunofluorescence Microscopydeoxynucleotides to TGFb3. Although no mesenchyme wasformed by these treated cultures, the experimental design The antibody used to localize TGFb3 was produced againstprecluded a discrimination of the oligodeoxynucleotide tar- chicken recombinant protein (Cat. No. AB-244-NA, lot No. CK02;get tissue, i.e., AV endocardium vs myocardium or both. R & D Systems, Minneapolis, MN). According to the manufacturerAlthough these observations strongly suggest a critical func- this antibody neutralizes the bioactivity of TGFb3, but not that of

TGFb1, TGFb2, or TGFb5. Moreover, while it will react with alltion for TGFb3 during the epithelial–mesenchymal transi-

Copyright q 1998 by Academic Press. All rights of reproduction in any form reserved.

AID DB 8807 / 6x36$$$281 02-02-98 10:37:54 dbas

101TGFb3 in Early Heart Development

FIG. 1. Expression of mRNA for TGFb3 during AV endocardial cushion tissue formation. (A) At stage 14, the heart consists of twoepithelial layers, the endocardium (e) and myocardium (m), separated by cardiac jelly. The mRNA for TGFb3 was detected in the outflowtract (OT) myocardium; however, little or no signal was detected in OT endocardium and atrioventricular (AV) endocardium or myocardiumat this stage. (B) Stage 14 heart section processed with sense probe showed no detectable staining. (C) Stage 18 AV region showed positivestaining for TGFb3 mRNA in all three cell populations, i.e., the endocardial monolayer (e), cushion mesenchyme (me), and myocardium(m). (D) High magnification view of boxed area in C shows transforming endocardial cells (arrowheads) and invasive mesenchyme (halfarrows) stained strongly for TGFb3 mRNA. (E) Stage 18 AV section processed with sense probe control. All panels represent sagittalsections. AS, antisense probe; S, sense probe; ST, stage. Bar for (A–C and E), 100 mm; D, 20 mm.

three TGFb isoforms in ELISA and immunoblot analyses, it does studies of Ghosh and Brauer (1996) suggest that this antibody onlyrecognizes mature TGFb3.so with at least a 10-fold lower sensitivity than for TGFb3 (manu-

facturer’s description; Ghosh and Brauer, 1996). Immunoblot analy- Cultures were drained of medium, rinsed with phosphate-buf-fered saline (PBS), fixed with 4% paraformaldehyde in PBS (pHsis showed that this antibody detected at least 1 ng of recombinant

chicken TGFb3 with or without reduction (data not shown). The 7.4) for 60 min at room temperature, and then rinsed with PBS.


AID DB 8807 / 6x36$$$282 02-02-98 10:37:54 dbas

102 Nakajima et al.


AID DB 8807 / 6x36$$8807 02-02-98 10:37:54 dbas


Specimens were blocked for 2 h with 5% nonfat dry milk in PBS endocardial monolayer, which will remain epithelial without theaddition of transformation-promoting preparations (Krug et al.,containing 0.1% Triton X-100, incubated with anti-TGFb3 anti-

body (10 mg of IgG/mL in blocking solution) at 47C overnight, rinsed 1987). At this time, monolayers were treated with either antisenseor missense PTO (0.5–1 mM in 0.3 mL CM199) or with CM199extensively with PBS, incubated with tetramethylrhodamine 5-(and

6)-isothiocyanate (TRITC)-conjugated rabbit anti-goat IgG (Cappel, alone. After an additional 12 h, all cultures were treated with CCM,which can substitute for AV myocardium to elicit AV endocardialDurham, NC; diluted 1:100) for 2 h at room temperature, rinsed

with PBS again, transferred onto slides, coverslipped with mount- transformation into mesenchyme (Krug et al., 1987). Parallel cul-tures of AV endocardium treated with the antisense PTO to TGFb3ing medium (0.2 M n-propylgallate in 90% glycerol/10% PBS), and

observed using an Olympus BX60 fluorescent microscope. received both CCM and chicken TGFb3 protein (R & D Systems;5 ng/mL). After 48 h total incubation time, individual cultures wereNonfixed or paraformaldehyde-fixed stage 14–18 chicken em-

bryos were embedded in OCT (Miles, Elkhart, IN), and frozen in assessed for the presence of mesenchymal cells within the collagengel using Hoffman modulation optics. In this bioassay, ‘‘inhibition2-methylbutane cooled by liquid nitrogen. Frozen sections were

mounted onto 3-triethoxysilylpropylamine-coated slides and air- of mesenchymal formation’’ was defined as the situation when thenumber of invasive mesenchymal cells was less than 2 standarddried. After rinsing with PBS for 15 min, sections were blocked

with 5% nonfat dry milk in PBS for 60 min, incubated with primary deviations below that exhibited by control explants cultured withCCM alone.antibody in a moist chamber for 2 h at room temperature, rinsed

with PBS, incubated in TRITC-conjugated rabbit anti-goat IgG In the second type of AV endocardial transformation bioassay,AV endocardial monolayers were established on collagen gels as(Cappel; diluted 1:100) for 1 h, and then rinsed with PBS and

mounted in mounting medium. above and then treated with CM199 containing: (i) chicken recom-binant TGFb3 protein (R & D Systems; 0.5–50 ng/mL in 0.3 mLCM199), (ii) human purified TGFb2 protein (Nakarai, Tokyo, Ja-pan), or (iii) both growth factors together. After 24 or 48 h totalAV Epithelial–Mesenchymal Transformationincubation time, cultures were assessed microscopically for endo-Bioassay by Collagen Gel Culturethelial proliferation and for characteristics of epithelial–mesenchy-

Two types of endothelial bioassay were carried out to evaluate mal transformation (i.e., cell hypertrophy, loss of cell-to-cell con-the functional significance of TGFb3 in the transformation of AV tacts, formation of migratory processes, and gel invasion).endocardium into mesenchyme. These involved treatment of AVendocardial monolayer cultures with either (i) cardiocyte condi-tioned medium (CCM; Krug et al., 1987) following a pretreatment RESULTSwith antisense oligodeoxynucleotides to TGFb3 or (ii) basal culturemedium supplemented with chicken recombinant TGFb proteins.

TGFb3 Expression during AV Endocardial CushionPhosphorothioate oligodeoxynucleotides (PTO; Funakoshi, Tokyo,Tissue FormationJapan) used in these experiments were identical to those used pre-

viously by Potts et al. (1991). The antisense PTO (5*-CGTACA-In situ hybridization analysis of the stage 14 heart showedTCTTCATGTGC-3*) to chicken TGFb3 mRNA (Accession No.

little or no reactivity for TGFb3 transcripts in the myocar-M31154) spanned the initiation codon as determined by Jakowlewdium and endocardium of the primitive atrium and AV re-et al. (1988). Control cultures included treatment with a missensegion (Fig. 1A) relative to the sense probe control (Fig. 1B).PTO (5*-ATGGTGCACCTGACT-3*), which did not encode any

specific sequence in the GenBank database (Potts et al., 1991), or Interestingly, a strong signal was observed in the myocar-CM199 alone. dium at the distal extreme of the OT region (i.e., trucus

AV endocardial monolayers were established as described pre- arteriosis/aortic sac). TGFb3 mRNA was also detected inviously (Nakajima et al., 1994). Briefly, hearts from stage 140 the ventricle and proximal OT region, but the signal waschicken embryos were collected in Earle’s balanced salt solution less intense than that in the more distal OT (data not(EBSS; Gibco BRL). AV regions were resected and cut longitudinally shown). Only a very low level of immunoreactivity forto expose the lumen, then placed on top of drained collagen gels

TGFb3 protein was observed in the stage 14 heart (para-(0.3 mL/15 mm well) equilibrated in complete medium (CM199;formaldehyde-fixed specimens) (Fig. 2A).Medium 199 containing 1% chick serum, 5 mg/mL insulin, 5 mg/

By stage 18, a subpopulation of endocardial cells in themL transferrin, 5 ng/mL selenium (ITS, Becton Dickinson), andAV and OT regions has changed from an epithelial morphol-streptomycin/penicillin (Gibco BRL, Tokyo, Japan)). AV explants

were removed after 4 h of coculture leaving behind only an AV ogy to become mesenchymal cells that invade into the sub-

FIG. 2. Immunohistochemical staining for TGFb3 protein in stage 14 and 18 hearts with paraformaldehyde fixation. The antibody usedfor these studies has been reported to react with only the mature form of TGFb3 and does not recognize latent TGFb3 (Ghosh and Brauer,1996). (A) At stage 14 no TGFb3 immunoreactivity was detected in the heart. By stage 18, endocardial, myocardial, and mesenchymalcells all showed anti-TGFb3 antibody immunoreactivity in both the outflow tract (B) and atrioventricular canal (C). No detectable stainingin the cardiac jelly of either stage 14 or stage 18 hearts. The cardiac jelly in fresh-frozen tissue sections pretreated with 4 mM HCl orplasmin was negative as well (data not shown). (D) Control section, secondary antibody alone. Left panels show light microscopic viewof the immunofluorescent images in right panels. AV, atrioventricular canal; OT, outflow tract; e, endocardium; m, myocardium; me,cushion mesenchyme; cj, cardiac jelly. Bars in A, B, and D, 100 mm; C, 50 mm.


AID DB 8807 / 6x36$$$282 02-02-98 10:37:54 dbas

104 Nakajima et al.

jacent cardiac jelly, resulting in the formation of AV endo-cardial cushion tissue. In addition, the ventricular musclelayer has begun to form ventricular trabeculae, which re-quires loss of its epithelial morphology as well (Manasek,1970; Markwald et al., 1977, 1978). At this stage, TGFb3mRNA transcripts were readily detectable throughout theheart. In the AV region TGFb3 mRNA was present in myo-cardium, endocardium, and its associated cushion mesen-chyme (Fig. 1C), the latter of which stained somewhatstronger than its overlying endocardium (Fig. 1D). Thestongest staining occurred in the ventricular myocardium.The myocardium associated with the inferior and superiorAV cushions was moderately positive. By comparison, im-munohistochemical staining for TGFb3 protein in the stage18 chick heart was much more intense than that observedat stage 14 and revealed a cytoplasmic deposition of TGFb3in myocardium, endocardium, and mesenchymal cells ofboth the AV and OT regions (Figs. 2B and 2C). No extracel-lular staining was observed in the cardiac jelly of eitherstage embryos subjected to paraformaldehyde fixation: simi-lar results were obtained with nonfixed/frozen and Bouin-fixed embryos as well (data not shown). Conditions for con-verting latent TGFb3 to the mature form (e.g., heat or acidpretreatment) were not tested on these specimens nor wascryopreserved tissue, which might account for a lack of FIG. 3. Collagen gel culture model for evaluating PTO treatment

of AV endocardial monolayers. This diagram shows the temporalextracellular staining in the current study: Ghosh andrelationship of establishing AV endocardial monolayers from stageBrauer (1996) has used this same antibody from R & D140 chick embryos and their subsequent treatment with either anti-Systems to demonstrate the presence of latent TGFb3 (andsense or missense PTOs to TGFb3. Additional details are describedlack of the mature form) in the cardiac jelly of stage 14–16under Materials and Methods. Abbreviations: m, myocardium; e,chick embryos.endocardium; cj, cardiac jelly.

TGFb3 Antisense PTO Pretreatment of AVEndocardial Monolayers in Culture Inhibits 4B). Pretreatment of AV endocardial monolayers with anMesenchyme Formation in Response to Cardiocyte antisense PTO to TGFb3 blocked mesenchyme formationConditioned Medium in response to CCM (Table 1, Fig. 4C). This inhibitory effect

of the TGFb3 antisense PTO on endocardial transformationPrior studies showed that immunoreactivity for TGFb1/b3 in stage 14 chick OT/AV endocardial cells can be upregu- was reversed when the CCM was supplemented with re-

combinant chicken TGFb3 protein (Table 1 and Fig. 4D).lated by CCM (Nakajima et al., 1994). Whether this repre-sented new synthesis by the target endocardial cells or the AV endocardial monolayers pretreated with missense PTO

(identical length and G/C content) formed mesenchyme inacquisition of TGFb protein from the CCM was not deter-mined conclusively, although the data strongly suggested response to the addition of CCM similar to cultures without

PTO treatment (Table 1, Fig. 4E). Parallel experiments withthat a non-TGFb, EDTA-soluble protein(s) was responsiblefor affecting endocardial expression of TGFb. In order to AV endocardium/myocardium cocultures receiving TGFb3

antisense vs missense PTO treatment duplicated the resultsresolve this question, AV endocardial monolayers fromstage 140 embryos (a ‘‘pretransformed’’ state, i.e., will not of Potts et al. (1991) (data not shown).

In order to determine if the antisense PTO treatment in-transform into mesenchyme without either AV myocardialcoculture or CCM treatment) were established on collagen hibited TGFb3 expression in target AV endocardial mono-

layers, immunohistochemistry was performed on these cul-gels and pretreated with antisense PTO identical to thoseused by Potts et al. (1991). After a 12-h incubation with tures. As shown in Fig. 5, TGFb3 immunoreactivity was

observed in AV mesenchyme in cultures receiving CCMeither antisense or missense PTO, CCM was added as asource of transformation-promoting activity (Fig. 3). As has alone or with missense PTO pretreatment (Figs. 5B and 5D),

while only a uniformly low level of reactivity was apparentbeen demonstrated previously (Krug et al., 1987), AV endo-cardial monolayers seeded invasive mesenchyme into the in the monolayer of those with CM199 or TGFb3 antisense

PTO pretreatment prior to the addition of CCM (Figs. 5Aunderlying collagen gel in the presence of CCM, while con-trols cultures remained epithelial (Table 1, Figs. 4A and and 5C). Given the paucity of material available in the trans-


AID DB 8807 / 6x36$$$283 02-02-98 10:37:54 dbas


TABLE 1Addition of TGFb3 Protein Relieves TGFb3 Antisense PTO Inhibition of AV Endocardial Transformation into Mesenchyme in Culture

Mesenchyme formed by 48 hCulture conditions

Mesenchymal cellsPretreatment Bioassay Frequencya per monolayerb

CM199 CM199 0/8 (0%) 0CM199 CCM 8/10 (80%) 17 { 1CM199 / TGFbPTO CCM 0/11 (0%) õ2CM199 / TGFb3 PTO CCM / TGFb3 4/5 (80%) 21 { 3CM199 / missense PTO CCM 7/9 (78%) 20 { 2

Note. State 140 AV endocardial monolayers were prepared on a collagen gel lattice and pretreated for 12 h with CM199 alone orcontaining 1 mM antisense or missense PTO as described under Materials and Methods. The resulting monolayers were then subjectedto various culture conditions including CM199, CCM, or CCM containing 5 ng/mL of chicken recombinant TGFb3.

a Number of monolayers forming mesenchyme after 48 h.b Mean number of mesenchymal cells formed per monolayer {SE.

formation bioassay, it was not technically feasible to per- 6A): these invasive cells were usually found at the peripheryof the monolayer instead of underneath the monolayer asform Northern blot or RNase protection assays.

The observation that CCM required exogenous TGFb3 is more common with CCM treated cultures (cf. Figs. 7Cand 5B). By comparison, AV endocardial monolayers treatedprotein supplementation to override the inhibitory effect of

TGFb3 antisense PTO treatment of AV endocardium with CCM generated 4–7 times the number of invasivemesenchymal cells (12 { 2 and 22 { 5 mesenchymal cellsstrongly suggests that active TGFb3 is absent in CCM.

However, it is possible that this CCM preparation contains per explant after 24 and 48 h in culture, respectively; Fig.6A). Control monolayers treated with CM199 alone did notlatent TGFb as found the conditioned medium of OT endo-

cardial/myocardial cocultures and in vivo in the cardiac show any phenotypic characteristics of epithelial–mesen-chymal transformation by 24 h in culture (Fig. 7A): after ajelly (Ghosh and Brauer, 1996). Immunoblot analysis of

CCM in the present study, either with or without acidifica- total of 48 h, some of these nontreated endocardial cells hadtion, failed to show any detectable reactivity at the level of apparently died or floated off the gel surface (not shown).sensitivity possible by this technique (lower limit of 1 ng/ Significant cellular proliferation of the monolayer was anmL). This observation is consistent with our earlier work additional effect of CCM treatment (about twofold after 24which also was unable to demonstrate precipitable TGFb h and four-fold after 48 h; Fig. 6B); however, endocardialfrom radiolabeled CCM with an antibody that recognizes cells treated with CM199 or TGFb3 alone did not show suchboth TGFb1 and TGFb3 (Nakajima et al., 1994). a dramatic proliferative response (cell number increased by

about 50% after 24 h, with no further increase by the 48-htime point). There was no significant difference in the num-

Exogenous TGFb3 Initiates Cell:Cell Separation in ber of cells in individual monolayers at the time of eitherPretransformed AV Endocardial Monolayers in CCM or TGFb3 addition.Culture It has been reported that TGFb2 is transcribed in the

developing chicken heart (Potts et al., 1992). Thus, oneIn order to further assess the role of endocardially derivedmight propose a hypothesis that TGFb2 and TGFb3 actTGFb3 in the epithelial–mesenchymal transformation pro-synergistically to affect complementary aspects of cushioncess, stage 140 AV endothelial monolayers cultures weretissue formation. For this reason, we examined the biologi-evaluated for their morphological response to purifiedcal effect of TGFb2 alone and in combination with TGFb3TGFb3. As shown in Figs. 6A, 7A, and 7B, the addition ofon pretransformed AV endocardial cells in culture. In bothrecombinant chicken TGFb3 over a wide range of concen-situations the results were similar to treatment withtrations (0.5–50 ng/mL) did not elicit invasive mesenchymeTGFb3 alone (Figs. 6C and 6D). These experiments usedformation after 24 h in culture. However, TGFb3 additionhuman TGFb2 because chicken TGFb2 protein was un-did result in significant cell:cell separation and cellular hy-available; however, the amino acid sequences of the twopertrophy of cells in the monolayer, as well as migratoryproteins are virtually identical. There is only one aminocells on the surface of the gel at the periphery of the mono-acid difference between the mature form of the human andlayer (Fig. 7B). After an additional 24 h (48 h total treatmentchicken TGFb2 homologs (H40 in chicken corresponds totime), 3–5 endocardial cells per monolayer had invaded into

the collagen gel lattice in the TGFb3 treated cultures (Fig. N40 in human).


AID DB 8807 / 6x36$$$283 02-02-98 10:37:54 dbas

106 Nakajima et al.


AID DB 8807 / 6x36$$8807 02-02-98 10:37:54 dbas


influences on AV endocardial transformation is regulated byDISCUSSIONa redundant mechanism (possibly TGFb3), (ii) myocardialTGFb2 plays an autocrine role in formation of the ventricu-

TGFb3 Is Implicated in the Formation of AV lar septum, or (iii) failure to form a ventricular septum andEndocardial Cushion Tissue the associate double outlet right ventricle is due indirectly

to a perturbation of outflow tract morphogenesis. In eitherPrevious experimental paradigms for examining the rolecase it appears that in the mouse TGFb2 alone is not essen-of TGFb3 in AV cushion mesenchyme formation have nottial to AV endocardial transformation. This may be due toallowed for a selective descrimination of endocardial vscompensation by TGFb3 which could be tested by crossingmyocardial functions because both tissues express this pro-TGFb2 and -b3 heterozygotes. It is important to note thattein. The present study has made use of the fact that AVthe distribution of TGFbs in chicken and mouse is differentendocardial monolayers treated with cardiocyte condi-before extrapolating these data between species.tioned medium faithfully recapitulate both morphological

and molecular characteristics or AV mesenchyme forma-tion (Krug et al., 1987; Mjaatvedt et al., 1991; Nakajima et AV Endocardial Expression of TGFb3 Regulatesal., 1994). Using this culture model the present study shows the Initial Aspects of Epithelial–Mesenchymalthat AV endocardial expression of TGFb3 is necessary for Transformationits subsequent transformation into mesenchyme, consis-tent with the observed upregulation of TGFb3 mRNA and The morphological characteristics of AV endocardial

transformation in vivo include cellular hypertrophy, lossprotein in this tissue. These data indicate that AV endocar-dial TGFb3 plays an autocrine role in the loss of cell:cell of cell–cell contact, migratory appendage formation, and

invasion into the subjacent myocardial basement mem-adhesions and in facilitating cellular hypertrophy, withoutwhich invasive mesenchyme will not form. In addition, it brane (Markwald et al., 1977, 1978). Exogenous addition of

either TGFb3 or -b2 protein to stage 140 AV endocardialwas shown that TGFb2 has similar biological effects asTGFb3 on cultured AV endocardial monolayers, suggesting monolayers induced only cellular hypertrophy and cell:cell

separation after 24 h. These responses in culture are veryfunctional redundancy of these TGFbs in the developingchick heart. No evidence was found to support a role for similar to those observed in vivo during the initial phase

of cushion tissue formation and temporally related to themyocardially derived TGFb as a paracrine inducer of AVendocardial transformation. upregulation of TGFb3 in AV endocardium both in vivo

and in culture (Nakajima et al., 1994). The identity of theIn the mouse AV region three TGFbs are expressed differ-entially, i.e., TGFb1 in endocardial/mesenchymal cells and factor that regulates this event is unknown, but a myocardi-

ally derived, non-TGFb protein has been suggested. ThisTGFb2 and TGFb3 in the myocardium (Heine et al., 1987;Lehnert and Akhurst, 1988; Akhurst et al., 1990; Millan activity is likely to be contained in 30-nm particulates

found only in the cardiac jelly of mesenchyme-forming re-et al., 1991; Dickson et al., 1993). Gene knock-out mouseexperiments show that there is no obvious primary cardio- gions. Culture experiments show that these 30-nm particu-

lates can elicit mesenchyme formation from AV endocar-vascular abnormality during early development in eitherTGFb1 or -b3 null embryos (Letterio et al., 1994; Dickson dial monolayers: antibodies to these particulates block this

event in cocultures with AV myocardium (Mjaatvedt andet al., 1995; Diebold et al., 1995; Kaartinen et al., 1995;Proetzel et al., 1995), consistent with a hypothesis of func- Markwald, 1989; Mjaatvedt et al., 1991).

The observation that TGFb3 facilitates only a subset oftional redundancy between these TGFb isoforms. However,severe ventricular septal defects (15 of 16 embryos), double events in AV cushion mesenchyme formation is consistent

with earlier data by Potts and Runyan (1989), who reportedoutlet right ventricle (3 of 16), and double inlet left ventricle(4 or 16) were reported in the TGFb2 null mouse (Sanford et that ventricular myocardium supplied an accessory activity

to TGFb1- or -b2-treated AV endocardial monolayers to com-al., 1997). Both mitral and tricuspid valves were apparentlynormal, but the outflow septum failed to muscularize. plete its transformation into invasive mesenchyme. These

experiments suggest that this complementary activity wouldThese observations suggest that: (i) any putative myocardial

FIG. 4. AV endocardial monolayers pretreated with TGFb3 antisense PTO do not transform into invasive mesenchyme in the presence ofcardiocyte conditioned medium. AV endocardial monolayers from stage 140 embryos were cultured for 12 h in the presence or absence ofTGFb3 antisense PTO and then evaluated 36 h later by optical sectioning for responsiveness to transformation promoting activity presentin CCM. (A) Monolayers cultured in CM199 did not show endothelial–mesenchymal transformation. (B) Monolayers pretreated with CM199transformed into invasive mesenchyme in response to CCM. (C) Monolayers pretreated with TGFb3 antisense PTO (AS) did not transforminto mesenchyme in response to CCM. (D) When AV endocardial monolayers pretreated with TGFb3 antisense PTO received CCM supple-mented with exogenous TGFb3 (5 mg/mL) mesenchymal invasion of the collagen gel lattice was observed. (E) Pretreatment of AV endocardialmonolayers with a missense PTO did not prevent mesenchyme formation in response to CCM. Bar, 100 mm.


AID DB 8807 / 6x36$$$284 02-02-98 10:37:54 dbas

108 Nakajima et al.

FIG. 5. Treatment of AV endocardial monolayers with antisense PTO to TGFb3 inhibits expression of TGFb3 immunoreactivity. AVendocardial monolayers from stage 140 embryos were treated under the various culture conditions shown in Table 1 and Fig. 3 andevaluated for the expression of TGFb3 protein by immunohistochemistry. Neither mesenchyme formation nor TGFb3 immunoreactivityoccurred in monolayers receiving CM199 alone (A) or pretreated with antisense PTO to TGFb3 prior to CCM addition (C). In contrast,both mesenchyme formation and TGFb3 immunoreactivity was observed in cultures receiving CCM alone (B) or in conjunction withmissense PTO pretreatment (D). In the latter two situations, TGFb3 immunoreactivity was strongest in invasive mesenchymal cells,with the next most intense signal coming from surface cells in the initial phases of transformation. Bar, 50 mm.

be expressed by all regions of the early myocardium, i.e., Several other differentiative events have been observedduring endocardial cushion mesenchyme formation, suchthose both with and without mesenchyme formation. This

as yet unidentified accessory factor would complement the as upregulation of urokinase (McGuire and Orkin, 1992),collagenase (Nakagawa et al., 1992), fibronectin (Icardo andaction of another protein(s) responsible for upregulation of

endocardial TGFb3 expression. Both transferrin (Isokawa et Manasek, 1983; Kitten et al., 1987), and type I procollagen(Sinning et al., 1988), and downregulation of NCAM expres-al., 1994) and basic fibroblast growth factor (Parlow et al.,

1991) have been identified in the early myocardium; how- sion (Mjaatvedt and Markwald, 1989). At present it is un-clear which, if any, of these are direct targets of AV endocar-ever, preliminary evaluations of a synergistic function with

exogenous TGFb3 have proven inconclusive. dial TGFb3 expression. The loss of cell:cell adhesion ob-


AID DB 8807 / 6x36$$$284 02-02-98 10:37:54 dbas


FIG. 6. Quantitative evaluation of invasive mesenchyme formation and proliferation from AV endocardial monolayers cultured withCCM vs TGFb. AV endocardial monolayers were prepared on collagen gel lattice from stage 140 embryos and incubated in CM199 withor without TGFb supplementation (TGFb2 vs TGFb3, alone or in combination). (A, B) Monolayers treated with TGFb3 alone over a 100-fold range of concentration were similar to control cultures in CM199 alone: by 48 h no more than 3–5 mesenchymal cells had formedper monolayer with only a slight increase in cell number. In contrast, cultures endothelial cells treated with CCM alone generated anaverage of 23 mesenchymal cells per monolayer and had more than four times the number of endothelial cells. (C, D) Monolayers treatedwith TGFb2 protein alone or in combination with TGFb3 were similar to those receiving only TGFb3. There was no significant differencein the initial number of endothelial cells in any of the target AV monolayers (Mann–Whitney U test). Number of explants tested: (A, B)17 for CM199, 7 for each TGFb3 concentration, and CCM; (C, D) 8 and 9 for TGFb2 (5 and 50 ng/ml, respectively); 10 and 9 for TGFb2/ b3 (5 and 50 ng/ml, respectively). Values represent average numbers of cells {SEM.

served in the present study would be consistent with a population of endocardial cells prestimulated in ovo. Theformer postulate seems unlikely as there is no apparentreciprocal relationship between TGFb3 and N-CAM expres-

sion. The antibody perturbation experiments for the TGFb difference in the amount of mesenchyme between cocul-tures of AV myocardium and endocardium in the presencetype II receptor by Brown et al. (1996) showed that invasion

could be blocked in early AV endocardium/myocardium co- or absence of exogenous TGFb (Potts and Runyan, 1989).Relative to the later possibility, unpublished studies bycultures. However, a lack of procollagen expression by inva-

sive mesenchyme from older explants suggests that both Ramsdell and Markwald have shown that AV mesenchymeconditioned medium is one of the few sources that cantransformation and differentiation may be dependent upon

TGFb expression in AV endocardium and mesenchyme. substitute for AV myocardium or CCM (notochord and limbectoderm being the only other known sources, Krug et al.,When target AV endocardial cells were cultured with

TGFb3 alone for 48 h, a small number of mesenchymal 1995). Interestingly, the developmental stage of the embryoused to establish the AV endocardial monolayers and thecells invaded the collagen gel (though much fewer than with

CCM treatment). This could be the result of two different duration of coculture with myocardium are directly linkedto the percentage of endocardial cells competent to trans-scenarios; i.e., TGFb3 is inhibitory to invasion or a missing

hypothetical migratory factor is produced locally by a sub- form into mesenchyme in response to these signals. For


AID DB 8807 / 6x36$$$284 02-02-98 10:37:54 dbas

110 Nakajima et al.

FIG. 7. Morphological analysis of AV endocardial monolayers treated with TGFb3 protein. (A) AV endocardial monolayers from stage140 embryos cultured with CM199 alone remained epithelial and showed no mesenchymal phenotypic characteristics. (B) In contrast,although invasive mesenchyme was not formed, AV monolayers treated with TGFb3 alone (5 ng/mL) showed cellular hypertrophy, cell:cellseparation, and formation of migratory appendages by 24 h in culture. (C) After 48 h in culture a few cells (arrows) were observed withinthe collagen gel lattice. Mesenchymal cells formed by TGFb3 treatment generally occurred at the periphery of the monolayers in contrastto CCM treated and myocardial coculture experiments that generate mesechyme primarily underneath the monolayer.

example, stage 140 embryos (18–20 somites) do not produce cells secrete TGFb as a high-molecular-mass, latent com-plex which contains mature TGFb associated with its la-mesenchyme in the standard bioassay (removing the myo-

cardium after 4 h of coculture), whereas parallel cultures tency-associated protein and latent TGFb-binding protein(LTBP; Miyazono et al., 1988; Kanzaki et al., 1990; Tsuji etthat start with embryos only 4 h older generate invasive

mesenchyme without supplementation, defining a critical al., 1990). The large latent TGFb complex is associated withthe extracellular matrix via the latent TGFb-binding pro-point of in ovo maturation (Krug, unpublished observa-

tions). For this reason, embryos beyond stage 140 should tein (Flaumenhaft et al., 1993; Saharinen et al., 1996). Thisis a component of extracellular microfibrils and concen-not be used to evaluate candidate transforming agents of

myocardial origin. trates the TGFb around cells (Taipale et al., 1996; Nakajimaet al., 1997). During endocardial cushion tissue formationin the mouse embryo (pc day 9.5) the large latent TGFb

Potential for Posttranslational Regulation of complex is distributed surrounding a subpopulation of AVTGFb3 Activity endocardial and all cushion mesenchymal cells, but is ab-

sent in the acellular cardiac jelly (Nakajima et al., 1997).Active TGFbs are highly hydrophobic, basic proteins thatare rapidly lost from solution by binding to many proteins Antibodies to LTBP-1 blocked both invasive mesenchyme

formation and endocardial proliferation in cocultures ofwith a relatively high affinity (Brown et al., 1990). Most


AID DB 8807 / 6x36$$$284 02-02-98 10:37:54 dbas


Davis, C. L. (1924). The cardiac jelly of the chick embryo. Anat.mouse AV myocardium and endocardium: lower concentra-Rec. 27, 201.tions of anti-LTBP-1 selectively inhibited cell invasion,

De La Cruz, M. V., Sanchez-Gomez, C., and Palomino, M. A. (1989).which is relieved by the addition of TGFb1 protein. ThisThe primitive cardiac regions in the straight tube heart (stage 90)may be mirrored in the chicken given the presence of latentand their anatomical expression in the mature heart: An experi-TGFb in the premesenchyme cardiac jelly (Ghosh andmental study in the chick embryo. J. Anat. 165, 121–131.

Brauer, 1996). Activation of this pool of latent TGFb may Dickson, M. C., Slager, H. G., Duffie, E., Mummery, C. L., and Ak-require proteases secreted by the AV endocardium and cush- hurst, R. J. (1993). RNA and protein localization of TGF beta 2ion mesenchyme (McGuire and Orkin, 1992). in the early mouse embryo suggest an involvement in cardiac

In summary, we have shown that AV endocardial mono- development. Development 117, 625–639.layers pretreated with an antisense PTO specific for TGFb3 Dickson, M. C., Martin, J. S., Cousins, F. M., Kulkarni, A. B., Karls-

son, S., and Akhurst, R. M. (1995). Defective haematopoiesis andfailed to carry out the complete epithelial–mesenchymalvasculogenesis in transforming growth factor b1 knock out mice.transformation in response to CCM. The conclusions of thisDevelopment 121, 1845–1854.study are: (i) that endothelially synthesized TGFb3 plays

Diebold, R. J., Eis, M. J., Yin, M., Ormsby, I., Boivin, G. P., Darrow,an essential autocrine function for cushion mesenchymeB. J., Saffitz, J. E., and Doetschman, T. (1995). Early-onsetformation in culture; (ii) there may be functional redun-multifocal inflammation in the transforming growth factor b1-dancy between TGFb3 and -b2 in vivo; and (iii) it appearsnull mouse is lymphocytic mediated. Proc. Natl. Acad. Sci. USA

unlikely that the myocardially derived stimulus for AV en- 92, 12215–12219.docardial transformation is a TGFb-like protein. Resolution Flaumenhaft, R., Abe, M., Sato, Y., Miyazono, K., Harpel, J., Heldin,of this final point awaits using a similar antisense PTO CH., and Rifkin, D. B. (1993). Role of the latent TGF-b bindingstrategy with cardiocyte cultures to determine if this leads protein in the activation of latent TGF-b by co-cultures of endo-

thelial and smooth muscle cells. J. Cell Biol. 120, 885–1002.to a loss of transformation-promoting activity in the condi-Ghosh, S., and Brauer, P. R. (1996). Latent transforming growthtioned medium and, if so, does this result from a selective

factor-b is present in the extracellular matrix of embryonic heartsinhibit of TGFb3 secretion by myocardial cells. If not, myo-in situ. Dev. Dyn. 205, 126–134.cardially derived TGFb3 may have an autocrine function

Hamburger, V., and Hamilton, H. L. (1951). A series of normalto affect secretion of other proteins that elicit endocardialstages in the developing chick embryo. J. Morphol. 88, 49–92.transformation and/or proliferation.

Heine, U. I., Munoz, E. F., Flanders, K. C., Ellingsworth, L. R., PeterLam, H. Y., Thompson, N. L., Roberts, A. B., and Sporn, M. B.(1987). Role of transforming growth factor-b in the development

ACKNOWLEDGMENTS of the mouse embryo. J. Cell Biol. 105, 2861–2876.Icardo, J. M., and Manasek, F. J. (1983). Fibronectin distribution dur-

ing early chick embryo heart development. Dev. Biol. 95, 19–30.The authors acknowledge Joshua Spruill for assistance in prepara-tion of figures. This work was supported in part by a grant from Ignotz, R., and Massague, J. (1986). Transforming growth factor-bthe NIH (HL52813; E.L.K. and R.R.M.). stimulates the expression of fibronectin and collagen and their

incorporation into the extracellular matrix. J. Biol. Chem. 261,4337–4345.

Isokawa, K., Rezaee, M., Wunsch, A., Markwald, R. R., and Krug,REFERENCESE. L. (1994). Identification of transferrin as one of multiple EDTA-extractable extracellular proteins involved in early chick heartAkhurst, R. J. (1994). The transforming growth factor b family inmorphogenesis. J. Cell. Biochem. 54, 207–218.vertebrate embryogenesis. In ‘‘Growth Factors and Signal Trans-

Jakowlew, S. B., Dillard, P. J., Kondaiah, P., Sporn, M. B., and Rob-duction in Development’’ (M. Nilsen-Hamilton, Ed.), pp. 97–122.erts, A. B. (1988). Complementary deoxyribonucleic acid cloningWiley–Liss, New York.of a novel transforming growth factor-b messenger ribonucleicAkhurst, R. J., Lehnert, S. A., Faissner, A., and Duffie, E. (1990).acid from chick embryo chondrocytes. Mol. Endocrinol. 2, 1186–TGF beta in murine morphogenetic processes: The early embryo1195.and cardiogenesis. Development 108, 645–656.

Jakowlew, S. B., Ciment, G., Tuan, R. S., Sporn, M. B., and Roberts,Bernanke, D. H., and Markwald, R. R. (1982). Migratory behaviorA. B. (1994). Expression of transforming growth factor-b2 and b3of cardiac cushion tissue cells in collagen lattice culture system.mRNAs and proteins in the developing chicken embryo. Differ-Dev. Biol. 91, 235–245.entiation 55, 105–118.Brown, P. D., Wakefield, L. M., Levinson, A. D., and Sporn, M. B.

Kaartinen, V., Voncken, J. W., Shuler, C., Warburton, D., Bu, D.,(1990). Physiochemical activation of recombinant latent trans-Heisterkamp, N., and Groffen, J. (1995). Abnormal lung develop-forming growth factor-betas 1, 2, and 3. Growth Factors 3, 35–ment and cleft palate in mice lacking TGFb-3 indicates defects43.of epithelial–mesenchymal interaction. Nature Genet. 11, 415–Brown, C. B., Boyer, A. S., Runyan, R. B., and Barnett, J. V. (1996).421.Antibodies to the type II TGFb receptor block cell activation and

Kanzaki, T., Olofsson, A., Moren, A., Wernstedt, C., Hellman, U.,migration during atrioventricular cushion transformation in theMiyazono, K., Claesson-Welsh, L., and Heldin, C.-H. (1990).heart. Dev. Biol. 174, 248–257.TGFb-1 binding protein: A component of the large latent com-Choy, M., Armstrong, M. T., and Armstrong, P. B. (1991). Trans-plex of TGF-b1 with multiple repeat sequences. Cell 61, 1051–forming growth factor beta-1 localized within the heart of the

chick embryo. Anat. Embryol. 183, 345–352. 1061.


AID DB 8807 / 6x36$$$284 02-02-98 10:37:54 dbas

112 Nakajima et al.

Kitten, G. T., Markwald, R. R., and Bolender, D. L. (1987). Distribu- blocks the transformation of cardiac endothelium into mesen-chyme in culture. Dev. Biol. 145, 219–230.tion of basement membrane antigens in cryopreserved early em-

bryonic hearts. Anat. Rec. 217, 379–390. Moller, J. H., Taubert, K. A., Allen, H. D., Clark, E. B., and Lauer,R. M. (1994). Cardiovascular health and disease in children: Cur-Krug, E. L., Mjaatvedt, C. H., and Markwald, R. R. (1987). Extracel-

lular matrix from embryonic myocardium elicits an early mor- rent status. Circulation 89, 923–930.phogenetic event in cardiac endothelial differentiation. Dev. Biol. Moreno-Rodriguez, R. A., de la Cruz, M. V., and Krug, E. L. (1997).120, 348–355. Temporal and spatial asymmetries in the initial distribution of

mesenchyme cells in the atrioventricular canal cushions of theKrug, E. L., Rezaee, M., Isokawa, K., Turner, D. K., Litke, L. L.,Wunsch, A. M., Bain, J. L., Riley, D. A., Capehart, A. A., and developing chick heart. Anat. Rec. 248, 84–92.Markwald, R. R. (1995). Transformation of cardiac endothelium Nakagawa, M., Terracio, L., Carver, W., Birkedal-Hansen, H., andinto cushion mesenchyme is dependent on ES/130: Temporal, Borg, T. K. (1992). Expression of collagenase and IL-1a in devel-spatial, and functional studies in the early chick embryo. Cell. oping rat hearts. Dev. Dyn. 195, 87–99.Mol. Biol. Res. 41, 263–277. Nakajima, Y., Krug, E. L., and Markwald, R. R. (1994). Myocardial

Lehnert, S. A., and Akhurst, R. J. (1988). Embryonic expression pat- regulation of transforming growth factor beta expression by out-tern of TGF beta type-1 RNA suggests both paracrine and auto- flow tract endothelium in the early embryonic chick heart. Dev.crine mechanisms of action. Development 104, 263–273. Biol. 165, 615–626.

Letterio, J. J., Gieser, A. G., Kulkarni, A. B., Roche, N. S., Sporn, Nakajima, Y., Morishima, M., Nakazawa, M., and Momma, K.M. B., and Roberts, A. B. (1994). Maternal rescue of the trans- (1996a). Inhibition of outflow cushion mesenchyme formation informing growth factor beta knockout. Science 264, 1936–1938. retinoic acid-induced complete transposition of the great arteries.

Cardiovasc. Res. 31, E77–E85.MacLellan, W. R., Brand, T., and Schneider, M. D. (1993). Trans-forming growth factor-b in cardiac ontogeny and adaptation. Nakajima, Y., Hiruma, T., Nakazawa, M., and Morishima, M.Circ. Res. 75, 783–791. (1996b). Hypoplasia of cushion ridges in the proximal outflow

tract elicits formation of a right ventricle-to-aortic route in reti-Manasek, F. J. (1970). Embryonic development of the heart. I. Alight and electron microscopic study of myocardial development noic acid-induced complete transposition of the great arteries in

the mouse: Scanning electron microscopic observations of corro-in the early chick embryo. J. Morphol. 125, 329–366.sion cast models. Anat. Rec. 245, 76–82.Markwald, R. R., Fitzharris, T. P., and Manasek, F. J. (1977). Struc-

tural development of endocardial cushions. Am. J. Anat. 148, 85– Nakajima, Y., Miyazono, K., Kato, M., Takase, M., Yamagishi, T.,and Nakamura, H. (1997). Extracellular fibrillar structure of la-120.tent TGFb binding protein-1: Role in TGFb-dependent endothe-Markwald, R. R., Fitzharris, T. P., Bank, H., and Bernanke, D. H.lial-mesenchymal transformation during endocardial cushion tis-(1978). Structural analyses on the matrical organization of glycos-sue formation in mouse embryonic heart. J. Cell Biol. 136, 193–aminoglycans in developing endocardial cushions. Dev. Biol. 62,204.292–316.

Markwald, R. R., Mjaatvedt, C. H., Krug, E. L., and Sinning, A. R. Parlow, M. H., Bolender, D. L., Kokan-Moore, N. P., and Lough, J.(1991). Localization of bFGF-like protein as punctate inclusions(1990a). Inductive interactions in heart development: Role of car-

diac adherons in cushion tissue formation. Ann. NY Acad. Sci. in the pre-septation myocardium. Dev. Biol. 146, 139–147.588, 13–25. Potts, J. D., Dagle, J. M., Walder, J. A., Weeks, D. L., and Runyan,

R. B. (1991). Epithelial–mesenchymal transformation of embry-Markwald, R. R., Mjaatvedt, C. H., and Krug, E. L. (1990b). Induc-tion of endocardial cushion tissue formation by adheron-like, onic cardiac endothelial cells is inhibited by a modified antisense

oligodeoxynucleotide to transforming growth factor b3. Proc.molecular complexes derived from the myocardial basementmembrane. In ‘‘Developmental Cardiology: Morphogenesis and Natl. Acad. Sci. USA 88, 1516–1520.Function’’ (E. B. Clark and A. Takao, Eds.), pp. 191–204. Futura, Potts, J. D., and Runyan, R. B. (1989). Epithelial–mesenchymal cellNew York. transformation in the embryonic heart can be mediated, in part,

by transforming growth factor b. Dev. Biol. 134, 392–401.McGuire, P. G., and Orkin, R. W. (1992). Urokinase activity in thedeveloping avian heart: A spatial and temporal analysis. Dev. Potts, J. D., Vincent, E. B., Runyan, R. B., and Week, D. L. (1992).Dyn. 193, 24–33. Sense and antisense TGFb3 mRNA levels correlate with cardiac

valve induction. Dev. Dyn. 193, 340–345.Millan, F. A., Danhez, F., Kondaiah, P., and Akhurst, R. J. (1991).Embryonic gene expression patterns of TGF b1, b2 and b3 sug- Proetzel, G., Pawlowski, S. A., Wiles, M. V., Yin, M., Boivin, G. P.,gest developmental functions in vivo. Development 111, 131– Howles, P. N., Ding, J., Ferguson, M. W. J., and Doetschman, T.144. (1995). Transforming growth factor-b3 is required for secondary

palate fusion. Nature Genet. 11, 409–414.Miyazono, K., Hellman, U., Wernstedt, C., and Heldin, C.-H. (1988).Latent high molecular complex of latent transforming growth Rezaee, M., Isokawa, K., Halligan, N., Markwald, R. R., and Krug,factor b1. J. Biol. Chem. 263, 6407–6415. E. L. (1993). Identification of an extracellular 130 kDa protein

involved in early cardiac morphogenesis. J. Biol. Chem. 268,Mjaatvedt, C. H., Lepera, R. C., and Markwald, R. R. (1987). Myo-cardial specificity for initiating endothelial–mesenchymal cell 14404–14411.transition in embryonic chick heart correlates with a particulate Runyan, R. B., and Markwald, R. R. (1983). Invasion of mesen-distribution of fibronectin. Dev. Biol. 119, 59–67. chyme into three-dimensional collagen gels: A regional and tem-

poral analysis of interaction in embryonic heart tissue. Dev. Biol.Mjaatvedt, C. H., and Markwald, R. R. (1989). Induction of an epi-thelial–mesenchymal transition by an in vivo adheron-like com- 95, 108–114.plex. Dev. Biol. 136, 118–128. Runyan, R. B., Potts, J. D., and Weeks, D. L. (1992). TGF-b3-medi-

ated tissue interaction during embryonic heart development.Mjaatvedt, C. H., Krug, E. L., and Markwald, R. R. (1991). An antise-rum (ES1) against a particulate form of extracellular matrix Mol. Reprod. Dev. 32, 152–159.


AID DB 8807 / 6x36$$$285 02-02-98 10:37:54 dbas


Saharinen, J., Taipale, J., and Keski-Oja, J. (1996). Association of the components of extracellular matrix microfibrils. J. Histochem.Cytochem. 44, 875–889.small latent transforming growth factor-b with an eight cysteine

repeat of its binding protein LTBP-1. EMBO J. 15, 245–253. Tsuji, T., Okada, F., and Nakamura, T. (1990). Molecular cloningof transforming growth factor-b masking protein and expressionSanford, L. P., Ormsby, I., Gittenberger-de Groot, A. C., Sariola, H.,

Friedman, R., Boivin, G. P., Cardell, E. L., and Doetschman, T. of the mRNA in various rat tissues. Proc. Natl. Acad. Sci. USA87, 8835–8839.(1997). TGFb2 knockout mice have multiple developmental de-

fects that are non-overlapping with other TGFb knockout pheno- Yang, E. Y., and Moses, H. L. (1990). Transforming growth factortypes. Development 124, 2659–2670. b1-induced changes in cell migration, proliferation, and angio-

Sinning, A. R., Lepera, R. C., and Markwald, R. R. (1988). Initial genesis in the chicken chorioallantoic membrane. J. Cell Biol.expression of type I procollagen in chick cardiac mesenchyme is 111, 731–741.dependent upon myocardial stimulation. Dev. Biol. 130, 167– Wessels, A., Markman, M. W. M., Vermeulen, J. L. M., Anderson,174. R. H., Moorman, A. F. M., and Lamers, W. H. (1996). The devel-

Sporn, M. B., Roberts, A. B., Wakefield, L. M., and Assoian, R. K. opment of the atrioventricular junction in the human heart. Circ.(1986). Transforming growth factor-b: Biological function and Res. 78, 110–117.chemical structure. Science 233, 532–534.

Received for publication September 11, 1997Taipale, J., Saharinen, J., Hedman, K., and Keski-Oja, J. (1996). La-tent transforming growth factor-b1 and its binding protein are Accepted November 14, 1997


AID DB 8807 / 6x36$$$285 02-02-98 10:37:54 dbas

an autocrine function for transforming growth …developmental biology 194, 99–113 (1998) article...

Documents