DEVELOPMENTAL

BIOLOGY

143, 135-148 (1991)

Expression of Transforming Growth Factor-/% 1-4 in Chicken Embryo Chondrocytes and Myocytes SONIAB.JAKOWLEW,PAMELA J. DILLARD,THOMASS.WINOKUR,KATHLEENC.FLANDERS, MICHAELB.SPORN,ANDANITA B. ROBERTS Laboratwy

of Chemoprevention,

Natimal Cancer Institute. Bethesda, Maryland YO8.92 Accepted September 1.9,1990

cDNA probes and antibodies for TGF-0s 1, 2, 3, and 4 were used to study the expression of these different TGF-/3 isoforms in cultured chicken embryo chondrocytes and cardiac myocytes, as well as in developing cartilage and heart tissues, TGF-0s 2, 3, and 4 mRNAs, but not TGF-pl mRNA, were detected in cultured chondrocytes and myocytes. Expression of TGF-0s 2 and 4 mRNAs increased with age, while expression of TGF-03 mRNA was independent of age in chondrocytes cultured from 12. to 17-day-old embryos. In contrast, expression of TGF-0s 2, 3, and 4 mRNAs was constitutive in myocytes cultured from 7- to g-day-old embryonic hearts; expression of TGF-ps 3 and 4 mRNAs increased, while expression of TGF-P2 mRNA remained unchanged in myocytes from lo-day-old embryos. Immunoprecipitation studies demonstrated expression of TGF-8 in both the conditioned media and the cell lysates of metabolically labeled chondrocyte and myocyte cell cultures. Immunohistochemical staining of cultured chondrocytes and myocytes and of cartilage and heart tissues of developing chicken embryos with antibodies specific for each TGF-P isoform showed immunoreactive TGF-ps 1, 2, 3, and 4. Our results demonstrate coordinate expression of these four TGF-P isoforms in chicken embryo chondrocytes and myocytes, both in vitro and in viva, with expression of TGF-Ps 2,3, and 4 ‘6’ 1991 Academic Press, Inc mRNA and protein more prominent than that of TGF-61. INTRODUCTION

The original isolation of transforming growth factor/3 (TGF-/3) from human platelets (Frolik et ah, 1983), human placenta (Assoian et al., 1983), and bovine kidney (Roberts et al., 1983a) resulted in the identification of a single form of the peptide, a 25,000 molecular weight (MW) homodimer, now called TGF-fll. The cloning of human TGF-/31 has led to the identification of four other forms of TGF-fi (TGF-fis 2, 3, 4, and 5) and the definition of a larger gene family comprising several other structurally related, but functionally distinct, proteins. The original narrow definition of TGF-0, in terms of induction of a transformed phenotype in mesenchymal cells (Roberts et al., 1981, 1983b; Moses et al., 1981), has now been extended by the knowledge that TGF-P affects many different functions in nearly all cells examined. TGF-0 has been found to exhibit both stimulatory and inhibitory effects on growth and development. The nature of its action on a particular target cell has been found to be critically dependent on many parameters, including cell type, growth conditions, and other growth factors present. In its role as a ubiquitous, multifunctional regulator of cell proliferation and differentiation, TGF-/3 has been shown to regulate such processes as chondrogenesis, osteogenesis, and myogenesis (Seyedin et ah, 1985; Centrella et al., 1987; Massague et al., 1986). Thus far, the

majority of investigations of the structure, function, localization, and in vitro and in viva effects of TGF-0 have focused on mammalian species. Because avian embryogenesis has been so extensively studied, we have cloned and characterized avian homologues of TGF-P to facilitate more detailed analysis of their developmental roles. We have recently reported the molecular cloning of chicken TGF-0s 1 and 2 cDNA (Jakowlew et al., 1988a, 1990), as well as two new forms of TGF-0, TGF$s 3 and 4 (Jakowlew et al., 1988b,c). This now enables us to examine the relative expression of each of these isoforms and their possible interplay during chicken embryogenesis and developmental processes such as chondrogenesis, osteogenesis, and myogenesis. Preliminary studies using the chicken TGF-P cDNA clones have shown significant levels of expression of avian TGF$s 2, 3, and 4 mRNAs in cultured chicken embryo chondrocytes; in contrast to the relatively high levels of expression of TGF-@l in mammalian cells and tissues, expression of TGF-fil mRNA is not detectable in RNA Northern blots of chicken RNAs. We have now extended our studies using nuclear run-on transcription analysis, immunoprecipitation, and immunohistochemistry in these cells, as well as in chicken embryo cardiac myocytes, in order to investigate more thoroughly the expression of TGF-P mRNAs and their corresponding proteins. We have also used immunohistochemical staining techniques with antibodies specific for each of the TGF-fi isoforms to exam135

0012.1606/91 $3.00 Copyright All rights

0 1991 by Academic Press, Inc. of reproduction in any form reserved.

136

DEVELOPMENTAL BIOLOGY

ine the role of the different isoforms during chondrogenesis and development of the heart in the developing chicken embryo. MATERIALS

AND

METHODS

Preparation of Primary Chickell Embryo Chondrocytes, Myocytes, and Fibroblasts Primary chicken embryo chondrocytes were prepared from the sterna of 12- to l’i-day-old white Leghorn chicken embryos as described by Coon (1966) and Cahn et al. (1967). Briefly, excised sterna, carefully cleaned of adhering perichondrial material, were digested with 0.25% collagenase (Sigma, St. Louis, MO) in phosphatebuffered saline (PBS) containing calcium and magnesium at 3’7°C two times for 15 min each time, pouring off the turbid media between digestions to remove adhering fibroblasts. The remaining sterna were digested with 0.5% collagenase in PBS at 37°C for 3 hr, removing the supernatant, and replacing with fresh collagenase after each hour. Collagenase was inactivated by the addition of minimal essential medium (MEM) containing fetal calf serum. Chondrocytes were centrifuged at 1000 rpm, resuspended in MEM containing 10% fetal calf serum and 1% penicillin-streptomycin using 0.5 ml media per sternum, and cultured in plastic dishes at relatively high density, about 2.5 X lo5 cells/ml (Muller et al., 197’7), at 37°C for 3 days. Primary cultures were used to avoid dedifferentiation of the chondrocytes. Chondrocyte differentiation was examined by assessing expression of type II collagen by immunohistochemical staining techniques using chicken type II collagen antibodies; greater than 85% of the cultured cells expressed this marker of differentiated chondrocytes. Primary chicken embryo cardiac myocytes were prepared from the hearts of 7- to lo-day-old white Leghorn chicken embryos according to the procedure of DeHaan (196’7), as modified by Armstrong (1978). Briefly, excised hearts, cleaned of adhering sinus venosus and cut into halves, were digested with 0.05% trypsin in PBS two times for 5 min each at 37°C. The supernatant was poured off and discarded between digestions to remove fibroblasts. The remaining hearts were digested with 0.05% trypsin six times for 8 min each at 37°C. The supernatant was transferred to a tube containing cold serum to inactivate trypsin after each digestion cycle. Myocytes were centrifuged at 1000 rpm, resuspended in MEM containing 10% fetal calf serum and 1% penicillin-streptomycin using 1 ml media per heart, and cultured in plastic dishes at 37°C for 45 min; after which, the myocytes, depleted of adhering fibroblasts, were cultured in fresh dishes at 37°C for 3 days. The homogeneity of the myocyte cultures was examined by determining the expression of heavy chain myosin by immunohistochemical staining using chicken cardiac myosin

VOLUME 143.1991

heavy chain antibodies (Calbiochem, La Jolla, CA); all cultures were composed principally (greater than 90% ) of myocytes. Primary chicken embryo fibroblasts were prepared from decapitated 7- to lo-day-old chicken embryos. Briefly, embryos were finely minced and triturated 12 times in warm 0.25% trypsin. The supernatant was transferred to a tube containing cold serum to inactivate trypsin. Fresh warm trypsin was added to the remaining tissue and the procedure was repeated 5 more times. Fibroblasts were centrifuged at 1000 rpm, resuspended in MEM containing 10% fetal calf serum and 1% penicillin-streptomycin using 20 ml media per embryo, and cultured in plastic dishes at 37°C for 3 days. Nuclear Rwn-On Transcription

Analysis

Nuclei were isolated from chicken embryo chondrocytes cultured for 3 days as described by Greenberg (1987). Approximately 2 X lo7 nuclei were isolated for each sample. Elongation and isolation of labeled RNA were done in the presence of [32P]UTP (3000 Ci/mmol, DuPont-New England Nuclear, Boston, MA). Each sample (2 x lo7 cpm) was hybridized for 72 hr to plasmids (10 pug) that had been adsorbed onto nitrocellulose through a slot blot apparatus (Schleicher and Schuell, Keene, NH). RNA Extraction

and RNA Northern

Blot Analysis

Total RNA was extracted from chicken embryo chondrocytes and cardiac myocytes and chicken tissues according to the LiCl-urea procedure described by Auffray and Rougeon (1980) and poly(A)-selected by oligo(dt) cellulose affinity chromatography (Collaborative Research, Waltham, MA) according to Aviv and Leder (1972). For RNA Northern blot analysis, equal amounts of total RNA (15 pg) or, in certain cases, poly(A+) RNA (5 pg) were electrophoresed on 1% agarose gels containing 0.66 M formaldehyde and transferred to “Nytran” filters (Schleicher and Schuell). Ethidium bromide (33 hg/ml) was included in both the gels and the running buffers in order to visualize the positions of ribosomal RNAs by uv illumination following electrophoresis. Before blotting, the gels were pretreated with 50 mM NaOH in 1X SSC (150 mM sodium chloride/l5 mM sodium citrate) for 20 min at room temperature and then rinsed two times in 10X SSC for 20 min each. Blots were hybridized using 32P-labeled (3000 Ci/mmole, New England Nuclear) nick-translated or random-primed probes at 65°C and washed at 65°C according to Church and Gilbert (1984) and exposed for various times at -70°C using an intensifying screen. The same blots were hybridized, dehybridized, and then rehybridized with the different TGF-0 cDNAs. Densi-

137

TGF-8

AMINO ACID SEQUENCE

SEQUENCE POSITION

d l-30

ALDTNYCFSSTEKNCCVROLYIDFRKDLGW

Latency Associated Peptide

MATURE * Kg 130,ll

r

I t

* 2

50-75

YLWSSDTQHSRVLSLYNTINPEASASIY)

/

I

50 75121 *

a 3

50-60

YLRSADTTHST(YI

I --

4

8-15

(Y )FGPGTDEK(Y I

I

..--

-.

-----)=

g 1 50 sm, c 1 s lard,

I

FIG. 1. The amino acid sequence of the TGF-@ peptides used to generate antisera for immunohistochemistry. (Y) indicates the addition of a tyrosine to the particular TGF-8 region. The hatched bars and numhers below the bars indicate the location of the peptides within the TGF-0s 1, 2, 3, and 4 mature coding regions adjacent to the latency-associated peptides. The numhers helow the hatched bars indicate the sequence position of the peptides, using the first amino acid in the mature TGF-p as amino acid number 1.

tometry of autoradiographs was performed Ultrascan laser densitometer.

using a LKB

Synthetic peptides corresponding to unique regions of human TGF-01, human TGF-P2, and chicken TGF-Bs 3 and 4 were purchased from Peninsula Laboratories (Belmont, CA) and Bachem Biochemicals, (Torrance, CA). A summary of the sequences used is listed in Fig. 1. Because mature human and chicken TGF-Bl show 100% identity, a polyclonal antibody, anti-P l-30(1), generated to a peptide corresponding to amino acids l-30 of mature human TGF$l as described by Flanders et al. (1988), was used. For detection of TGF-/32, a polyclonal antibody, anti-P 50-75(2), generated to a peptide corresponding to amino acids 50-75 of mature TGF-02, identical in human and chicken, with a tyrosine added to the carboxyl-terminal, as described by Flanders ef al. (1990a). was used. Polyclonal antibodies, anti-P 50-60(3) and anti-P 8-15(4), were also generated to peptides corresponding to amino acids 50-60 of mature chicken TGF-83, with a tyrosine residue added to the carboxyl terminal, and to amino acids 8-15 of mature chicken TGF-p4, with tyrosine residues added to both the amino and carboxyl terminals. Antibodies were generated in rabbits and were purified by passage over 1.5-ml columns of Affi-Gel 10 or 15 as described by Flanders et trl. (1988,1990a,b). In addition to these four peptide antibodies, previously described polyclonal antibodies generated against unconjugated native porcine TGF-82, referred to as anti-TGF-02 (Danielpour et a,l., 1990), were also used in our studies. The specificity of the TGF-ps 1, 2, and 3 antibodies for their corresponding TGF-/3 isoforms has previously been demonstrated by Western blot analysis (Flanders et al., 1988, 1990a,b).

Immunoprecipitation of media conditioned by cells and cell lysates with TGF-fi antibodies was performed as previously described (Robey et ~r.b,1987; Knabbe et al., 1987). Briefly, the cells were incubated in cysteine-free media containing [“5S]cysteine (250 pCi/60 mm dish, Amersham, Arlington Heights, IL) for 20 hr. Media were collected and centrifuged to separate cellular debris. Cells were washed four times with PBS an extracted overnight with 0.25 M HC1/95% ethanol. The labeled media and lysates were boiled in immunoprecipitation buffer for 4 min to activate latent TGF-0. The sample was precipitated with normal rabbit serum IgG and fixed Staphylococcus aureus (Boehringer Mannheim, Indianapolis, IN) and then incubated overnight at 4°C with test antiserum. The specificity of the immunoprecipitation was determined by preincubating the antiserum with TGF-p2. The immunoreactive TGF-/3 was recovered by precipitation with S. a.ureus and eluted by boiling in a 2%) SDS buffer. The samples were electrophoresed in nondenaturing 10% polyacrylamide gels according to Laemmli (1970). The gels were fixed in 10% acetic acid/lo% ethanol, enhanced with 2,5-diphenyloxazole dissolved in dimethyl sulfoxide, dried, and exposed to Kodak XAR-5 X-ray film at -70°C.

Cultured chondrocytes were scraped, washed three times in PBS, fixed for 48 hr in 10% neutral-buffered formalin, treated in Bouin’s solution for 6 hr at room temperature, and stored in 70% ethanol before embedding in paraffin and sectioning at 5 Wm. Tissues were ?xed and sectioned in a similar manner.

138

DEVELOPMENTAL BIOLOGY

Immunohistochemical

Staining

TGF- f31

TGF-fi was localized in sections as described by Heine et al. (1987), using avidin-biotin-peroxidase kits (Vector Laboratories, Burlingame, CA). After deparaffinization, blocking of endogenous peroxidase in hydrogen peroxide/methanol, and permeabilization with hyaluronidase, the sections were blocked with 1.5% normal goat serum/0.5% BSA, incubated overnight at 4°C with affinity-purified antisera at 3-5 pg/ml, washed extensively, and then incubated with biotinylated goat antirabbit IgG and avidin-enzyme complex. Sections were stained with 3,3’-diaminobenzidine (Sigma) and hydrogen peroxide and were counterstained with Mayer’s hematoxylin. Controls include (1) replacing primary antisera with normal rabbit IgG; and (2) using primary antisera that had been preincubated with a ZO-fold molar excess of the appropriate peptide for 2 hr at room temperature before this mixture was applied to the section.

RESULTS

RNA Northern

Blot Analysis

VOLUME 143,199l

of Chicken Embryo RNA

Using RNA Northern blot analysis, we have recently shown that expression of TGF$s 2,3, and 4 mRNAs, but not TGF-Pl mRNA, is detected in cultured primary chondrocytes derived from 16-day-old chicken embryo sterna and also in chicken embryo fibroblasts (Jakowlew et al., 1988a,b,c, 1990). TGF-Ps 1, 2, 3, and 4 mRNAs are all transcribed in cultured chondrocytes as shown by nuclear run-on transcription experiments conducted with nuclei isolated from chondrocytes from 16-day-old chicken embryos cultured for 3 days (Fig. 2). We have now extended our initial studies to examine the relative levels of steady-state expression of TGF-P mRNAs in chondrocytes extracted from chicken embryo sterna of increasing developmental age (12- to 17-day-old); the sternum is not easily manipulated before 12 days of age. Probing RNA Northern blots with 32P-labeled nicktranslated or random-primed chicken TGF-Pl cDNA showed no detectable expression of TGF-/31 mRNA in any of these chondrocytes using either total or poly(A+) RNA. Probing the same blots with chicken TGF-/32 cDNA showed expression of three TGF-P2 mRNAs (3.9, 4.3, and 8 kb) (Fig. 3A). Although the 3.9- and 4.3-kb mRNAs did not resolve well and gave the appearance of one band of about 4.1 kb in these blots, resolution of these two mRNAs has been previously reported in chondrocyte RNA (Jakowlew et ab, 1990). While the three TGF-P2 mRNAs were expressed at approximately equal levels in chondrocytes from 12- and 13-day-old embryos, expression of both the 3.9- and 4.3-kb mRNAs and the 8-kb mRNA increased in chondrocytes from 14- to 17day-old embryos (Fig. 3A). In contrast, when the same

TGF-@ 2

TGF-P3

TGF-194

pUCl9

FIG. 2. Analysis of transcription of TGF-/3 mRNAs in chondrocytes. Chondrocytes were extracted and cultured as described under Materials and Methods, nuclei were isolated, and nuclear RNA was elongated in the presence of [“*P]UTP (Greenberg, 1987). Counts (2 x 10”) of the labeled RNA were hybridized to 10 pg each of specific cDNAs that had been adsorbed onto nitrocellulose. TGF-@1, plasmid pTGFBCh119 (chicken TGF-01) cDNA; TGF-02, plasmid pTGFB-Ch4 (chicken TGF-fl2) cDNA; TGF-03, plasmid pTGFB-Chl7 (chicken TGF-/33) cDNA; TGF-/34, plasmid pTGFB-Ch63 (chicken TGF-/34) cDNA; pUCl9, plasmid pUC19. The TGF-/3 cDNAs contained coding and 5’- and 3’-untranslated regions. This experiment was repeated with two different preparations of nuclei. Exposure time was 16 hr.

blots were probed with chicken TGF-Bs 3 and 4 cDNA, TGF-/33 mRNA expression (3 kb) did not change with developmental age (Fig. 3B), while that of TGF-/34 followed a pattern similar to that of TGF-P2, increasing between Days 13 and 14 (Fig. 3C). As a control, the blots were stained with ethidium bromide and were photographed to assure that approximately equal amounts of RNA had transferred to the blots (data not shown). To study the pattern of expression of the TGF-P mRNAs in another chicken embryo cell type and to compare it with that of chondrocytes, cardiac myocytes were chosen as mouse heart has been shown to stain intensely with antibodies to TGF-fis 1,2, and 3 (Heine et al., 1987; Thompson et ab, 1989; K. C. Flanders, unpublished data). Total RNA prepared from cardiac myocytes cultured from 7- to lo-day-old chicken embryo hearts was hybridized as before. As in chondrocytes, no TGF-01 mRNA expression was detected in any of the myocyte RNAs. In contrast to chondrocytes, expression of the three TGF/32mRNAs (3.9,4.3 and 8 kb) was independent of developmental age in myocytes from 7- to lo-day-old embryos (Fig. 4A). At all ages, the level of the 8-kb TGF-P2 mRNA was higher than that of both the 3.9- and 4.3-kb TGF-/32 mRNAs. In contrast to the patterns of expression of myocyte TGF-02 mRNAs, when these same blots were hybridized with TGF-03 cDNA, expression of TGF-63 mRNA (3 kb) was approximately equal in myocytes from 7- to g-day-old embryos, but increased in

139

JAKOWLEW ET AL. TGF-04

TGF-83

TGF-02

28S-

28S-

-43Kb28S-3.9Kb

18%

-3Kb -2.5Kb

123456

-1.7Kb

18S-

18Sp

123456

123456

FIG. 3. RNA Northern analysis of chicken embryo chondrocyte RNA. Total RNA (15 fig) isolated from exponentially growing subconfluent cultured primary chondrocytes extracted from 12- to U-day-old chicken embryo sterna was electrophoresed on a 1% agarose-formaldehyde gel and was transferred to a Nytran filter as described under Materials and Methods. The developmental ages of the chondrocyte RNAs are labeled or random-primed (A) plasmid pTGFB-Ch4; (B) plasmid above each lane. Hybridization was performed with 32P-labeled nick-translated pTGFB-Chl7; or (C) plasmid pTGFB-Ch63. The positions of TGF-82 mRNAs are shown as 3.9,4.3, and 8 kb. The position of TGF-03 mRNA is shown as 3 kb. The position of TGF-04 mRNA is shown as 1.7 kb.

strated that equal amounts of RNA had transferred to the blots (data not shown). The pattern of TGF-0 mRNA expression was also examined in chicken embryo fibroblasts. The chicken TGF-0s 1, 2, 3, and 4 cDNA probes were hybridized to total RNA extracted from fibroblasts cultured for 3 days from 7- to lo-day-old chicken embryos. In these experiments, the same blots were dehybridized and rehybridized successively with each of the four cDNA probes. Expression of TGF-/3s 3 and 4 mRNAs was independent of developmental age, while expression of the 3.9- and 4.3-kb TGF-/32 mRNAs decreased in fibroblasts cultured for 3 days from lo-day-old embryos (Figs. 5A5C). As in chondrocytes and myocytes, expression of TGF-/X mRNA was not detected in any of the fibroblast mRNAs. The fact that there was no increase in the expression of TGF$s 3 or 4 mRNAs in fibroblasts with age further supports our contention that the increase in

cells cultured from lo-day-old embryos (Fig. 4B). The pattern of expression of TGF-/34 mRNA (1.7 kb) was approximately equal in myocytes from 7- to g-day-old embryos and was increased in myocytes from lo-day-old embryos (Fig. 4C). In addition to the 1.‘7-kb mRNA, at least two other mRNAs migrating at 2.5 and 2.7 kb were detected in myocyte RNA using the TGF-P4 cDNA probe (Fig. 4C). Expression of both the 2.5- and 2.7-kb mRNAs was constitutive in myocytes from 7- and g-day-old embryos; but whereas expression of the 2.5-kb mRNA increased in myocytes from lo-day-old embryos, that of the 2.7-kb mRNA decreased in the same myocytes. In contrast to these results in cardiac myocytes, the 2.5-kb mRNA is a minor species in chondrocytes and the 2.7-kb mRNA is not detectable (Fig. 3C). The origins of these additional myocyte mRNA species are currently under investigation. As before, staining with ethidium bromide demon-

A

B

TGF-/32

C

TGF-/X3

TGF-/34

-8Kb

28s ~

-4.3Kb ~ 3.9Kb

28s --

285 ~ -3Kb

- 2.7Kb --2.5Kb

18Sp

18S-

18S~ 1.7Kb

1

2

3

4

1

2

3

4

1

2

3

4

FIG. 4. RNA Northern analysis of chicken embryo cardiac myocyte RNA. Total RNA (15 pg) isolated from exponentially growing subconfluent cultured primary cardiac myocytes extracted from 7- to lo-day-old chicken embryo hearts was electrophoresed on a 1% agarose-formaldehyde gel and was transferred to a Nytran filter as described under Materials and Methods. The developmental ages of the myocyte RNAs are labeled above each lane. Hybridization was performed with 32P-labeled nick-translated or random-primed (A) plasmid pTGFB-Ch4; (B) plasmid pTGFB-Chl7; or (C) plasmid pTGFB-Ch63.

140

DEVELOPMENTAL BIOLOGY

A

B

TGF#2

VOLUME 143, 1991

TGF-b3

c

TGF-/34

--8Kb I, *

28S-

-4.3KbzES-3.9Kb

-3Kb

-2.7Kb -2.5Kb

i 18S-

18S-1.7Kb

,'

1234

1

2

3

4

1234

FIG. 5. RNA Northern analysis of chicken embryo fibroblast RNA. Total RNA (15 pg) isolated from exponentially growing subconfluent cultured primary fibroblasts extracted from 7- to lo-day-old chicken embryos was electrophoresed on a 1% agarose-formaldehyde gel and was transferred to a Nytran filter. The developmental ages of the fibroblast RNAs are labeled above each lane. Hybridization was performed with 32P-labeled nick-translated or random-primed (A) plasmid pTGFB-Ch4; (B) plasmid pTGFB-Chl7; or (C) plasmid pTGFB-Ch63.

the level of expression of TGF-Bs 3 and 4 mRNAs in cultured myocytes from lo-day-old embryos is a property of myocyte development and not merely due to differences caused by fibroblast contamination. The chicken TGF-fi cDNA probes were also hybridized to heart and sternum RNAs extracted from 12-dayold chicken embryos. The levels of expression of TGF$s 2,3, and 4 mRNAs were all higher in heart than in sternum. The expression levels of the TGF-02 mRNAs were lower than that of TGF-P3 mRNA, and the level of TGFp4 mRNA was lower than that of the TGF-02 mRNAs in both heart and sternum (Fig. 6). Expression of the 8-kb TGF-P2 mRNA was higher than that of the 3.9- and 4.3-kb TGF-02 mRNAs in both heart and sternum, as observed in cultured chondrocytes (Figs. 6A and 6D and Fig. 3A). The blots that were hybridized with TGF-04 cDNA had to be exposed three times as long as the blots that were hybridized with TGF-P2 cDNA to bring the RNA bands to a comparable level. Paralleling the results from cultured chondrocytes, but unlike those from cultured myocytes, the 2.5kb TGF-04 mRNA was also a minor species in heart and sternum, and the 2.7-kb TGF-04 mRNA was not detectable in either of these tissues (Figs. 6C and 6F). Although the 2.5-kb mRNA is not visible in Fig. 6F, it can be detected after prolonged exposure, while the 2.7-kb mRNA cannot (data not shown). Thus, both the 2.5-kb and the 2.7-kb mRNAs appear to be induced by in vitro culture conditions. No expression of TGF-01 mRNA was detected in either heart or sternum total or poly(A+) RNAs using TGF-Pl cDNA. Immunoprecipitation Myocytes

of TGF-/3 in Chondrocytes

and

Immunoprecipitation using anti-TGF-p2 revealed a polypeptide of MW 25,000 which migrated with lz51-la-

beled porcine TGF-P2 in both [35S]cysteine-labeled conditioned media and lysates from chondrocytes and myocytes (Figs. 7A and 7B, lanes 1 and 3). A relatively long

A

B TGF-/33

TGF-fi2

c

TGF-/I4

-8Kb -0

28S-

-4.3Kb -3.9Kb

28S-

28s-3Kb ,8S-

18S-

D TGF-fi2

“*; 28S-

-a

-2.5Kb 18S-

E TGF-/IS3

-1.7Kb

F TGF-/34

-8Kb

-4.3Kb -3.9Kkl

28S-

28S-3Kb

-2.5Kb 18S-

18S-

18S-

I

-1.7Kb

FIG. 6. RNA Northern analysis of chicken embryo heart and sternum RNA. Total RNA (15 pg) isolated from excised, quick-frozen 12 day-old chicken embryo hearts and sterna was electrophoresed on a 1% agarose-formaldehyde gel and was transferred to a Nytran filter as described under Materials and Methods. Hybridization was performed with ‘“P-labeled nick-translated (A and D) plasmid pTGFBCh4; (B and E) plasmid pTGFB-Chl7; or (C and E) plasmid pTGFBCh63. (A-C) Twelve-day-old chicken embryo heart RNA; (D-F) 12day-old chicken embryo sternum RNA. Exposure time was 2 days for A, B, D, and E and 7 days for C and F.

B

Media

A

Lysate

Kd -97.4 -68 -43

ail/

-TGF-fi'

-25.7

chondrocytes and myocytes, the amount of immunoprecipitable TGF-P in the media of BSC-1 cells was greater than that in the cell lysates (Figs. 7A, lanes 7-10; 7B, lanes 5 and 6), suggesting that while a significant amount of TGF-P is secreted in BSC-1 cells, the majority of TGF-P in chicken myocytes, and to a lesser extent in chicken chondrocytes, is retained intracellularly. This is the first demonstration of significant amounts of TGF/3 in cell lysates using anti-TGF-02. At the moment, we cannot extend these immunoprecipitation studies to include TGF-@s 3 or 4 as the peptide antibodies that have been generated to these isoforms are not useful for immunoprecipitation.

Irr~rnunohistochen~ic~~l Detection of TGF-8 Isqfbrms in 12

3

4

56

7

8

910

123456

FIG. 7. Immunoprecipitation of chicken embryo chondrocyte and myocyte media and cell lgsates with a porcine TGF-/%Z antibody. Cells were labeled for 20 hr with [%]cysteine, and 5 X lo6 and 10 X lo6 cpm, precipitable with 10% trichloroacetic acid, from (A) media and (B) cell extracts, respectively, were immunoprecipitated with either 7 ~1 of anti-TGF-02 (lanes 1, 3, 5, 7, and 9) or with 7 ~1 of anti-TGF-S2 preincubated with 250 ng of unlabeled TGF-82 (control) (lanes 2, 4, 6, 8, and 10). Nonreduced samples were subjected to electrophoresis on nondenaturing 10% SDS-polyacrylamide gels. The cells used are labeled above each lane. C, chondrocytes; M, myocytes; H, Harvey sarcoma virus-transformed NIH-3T3 cells; BSC-1, BSC-1 African green monkey kidney epithelial cells. The position of TGF-/s’2 is indicated by arrows. Exposure time was 4 days for lanes l-8 and 8 hr for lanes 9 and 10.

radiolabeling time (20 hr) was used in this experiment to assure that sufficient radiolabeled TGF-@ accumulated in the media to be detected by immunoprecipitation. Even though the experimental design reduced the probability that TGF-8 would be detected in the cell lysates, significant amounts of TGF-fi were detected in both chondrocyte and myocyte cell lysates. Studies using shorter radiolabeling times indicated that more TGF-6 was immunoprecipitated in both chondrocyte and myocyte cell lysates compared to the conditioned media as the radiolabeling time was reduced (data not shown). In both chondrocytes and myocytes, immunoprecipitation of TGF-0 in the conditioned media and lysates was blocked by pretreatment of the antibody with pure, unlabeled porcine TGF-62 (Figs. 7A and 7B, lanes 2 and 4). The specificity of anti-TGF-02 was demonstrated by its inability to immunoprecipitate TGF-B in the conditioned media of Harvey sarcoma virus-transformed NIH-3T3 cells (Ha-3T3) (Fig. 7A, lane 5), which have been previously demonstrated to secrete greater than 90% TGF-/31 and little TGF-02 (Flanders ef cd., 199Oa). As a positive control, anti-TGF-fi2 was shown to immunoprecipitate TGF-fi from BSC-1 African green monkey kidney epithelial cells, which are known to secrete predominantly TGF-P2 (Hanks ef al., 1988). Unlike

Cartilage

am! Cultured

Chondrocytes

To investigate possible TGF-/J isoform-specific effects in development of cartilage in vivo, we examined immunohistochemical staining patterns of the TGF-0 isoforms in this tissue in 12-day-old chicken embryos. For these studies, TGF-fi affinity-purified peptide antibodies were used rather than antibodies generated against native TGF-@ (see Fig. 1). Peptide antisera react better with fixed, denatured TGF-0 in paraffin sections of tissues than do antibodies raised against the native protein (Heine et al., 1987; Flanders et al., 1989). The staining pattern of anti-P l-30(1) has previously been shown to be intracellular in mammalian cells and tissues (Flanders et al., 1989; Thompson et ul., 1989). When this antibody was used to immunostain TGF-/Jl in chicken sternal cartilage, staining was relatively sparse and scattered (Fig. 8A). In contrast, the immunostaining patterns using anti-P 50-75(2), anti-P 50-60(3), and anti-P 8-15(4) were stronger and more uniform (Figs. 8B-8D), with the pattern for TGF-p2 showing greater intensity of staining than that of TGF$s 3 and 4. It is not known whether the more intense staining pattern characteristically displayed by anti-P 50-75(2) is a function of the production of a greater amount of TGF-82 than of TGF-IJs 3 or 4 by these cells or results from higher affinity of anti-P 50-75(2) for TGF-/%:! than antiP 50-60(3) and anti-P 8-15(4) for their respective proteins. Similar staining patterns were observed in the cartilage of 19-day-old chicken embryos (data not shown). Staining for TGF-/Js 2,3, and 4 was completely blocked when the individual anti-TGF-6 IgGs were preincubated with solutions of the peptides against which they were raised (Figs. 8B-8D insets). Although the intensity of staining with anti-P l-30(1) was also reduced by preincubation with the TGF-/X peptide, it was not completely abolished (Fig. 8A inset), as has been observed previously with this and other TGF-Pl antibodies (Heine et ab, 1987; Flanders et ah, 1989). Immunohistochemical staining with the peptide

142

DEVELOPMENTAL BIOLOGY

TGF-P antibodies was also performed on fixed and sectioned chicken chondrocyte cell pellets. The staining patterns correlated well with those observed in cartilage (Figs. 8E-8H and insets). The low level of expression of immunoreactive TGF-01 and the more prominent expression of TGF-Ps 2,3, and 4 in chondrocyte cell pellets also correlate with the corresponding mRNA expression of these isoforms.

Immunohistochemical Detection of TGF-/3 Isoforms Cartilage Diflerentiation and Maturation

in

We also investigated the potential roles of TGF-/3s 1, 2,3, and 4 in the differentiation of precartilaginous mesenchyme to cartilage and bone. The development of cartilage is known to be initiated by condensation of mesenthyme, followed by proliferation and differentiation of chondroblasts into chondrocytes, deposition of a cartilaginous matrix, hypertrophy, degeneration of chondrocytes, and then calcification of cartilage. The cartilaginous matrix may be replaced by invading osteogenic cells and production of a bone matrix that becomes calcified. Figure 8 shows the staining pattern of TGF-03 in cartilage development and this pattern is also representative of TGF-6s 1,2, and 4, except that TGF-/31 staining was much less intense than that of the other isoforms. Although no expression of any of the TGF-6 isoforms was detected in the precartilaginous mesenchyme of the 3-day-old chicken embryo by immunohistochemical staining, coordinate expression of TGF-0s 1, 2, 3, and 4 was detected as the mesenchymal cells contracted, proliferated, and differentiated into chondroblasts (Figs. 9A and 9B). No staining was observed in the extracellular matrix as chondroblasts were entrapped in lacunae during interstitial growth. Coordinate expression of all four isoforms was also demonstrated as chondroblasts differentiated into proliferating chondrocytes and subsequently into hypertrophic chondrocytes and calcified chondrocytes (Figs. 9C-9H). Other than the consistently low level of expression of TGF-/31 compared to TGF$s 2, 3, and 4, the only other difference that was noted in the expression of the TGF-P isoforms during cartilage differentiation was expression of TGF-64, but not the other isoforms in the extracellular matrix of calcified cartilage (Fig. 9F).

VOLUME 143,199l

Immunohistochemical Detection Heart and Cultured Myocytes

of TGF-p Isoforms

in

Coordinate expression of TGF-0s 1,2,3, and 4 was also demonstrated in the chicken embryo heart. The staining pattern shown for TGF-P2 in Fig. 10A is also representative of TGF-/3s 3 and 4. Although staining for TGF-/31 again was less intense compared to that of TGF-0s 2,3, and 4, no other differences were apparent in the staining patterns. Staining was very intense in the myocytes in both the atria and the ventricles, while endothelium and connective tissue were not stained, just as in neonatal and adult mouse hearts (Thompson et al., 1989). Similar staining patterns showing coordinate expression of TGF-0s 1, 2, 3, and 4 were also demonstrated in the hearts of 7-day-old embryos, except that the younger myocytes were not as prominently stained as in the older embryos (data not shown). Immunohistochemical staining with the TGF-P peptide antibodies was also performed on fixed and sectioned chicken cardiac myocyte cell pellets. The staining patters correlated with those observed in heart tissue (Fig. 1OC). Just as in heart tissue, cartilage, and cultured chondrocytes, immunohistochemical staining for TGFps 2, 3, and 4 was more prominent than for TGF-fll in cultured cardiac myocytes (data not shown). DISCUSSION

The present study demonstrates that TGF$s 1, 2, 3, and 4 mRNAs and proteins are expressed coordinately in chondrocytes and myocytes in the chicken embryo both in vitro and in vivo. Our studies were restricted to cultures of differentiated chondrocytes and beating myocytes with the aim of examining the developmental expression of the TGF-P isoforms in in vivo chondrocyte and myocyte differentiation in cartilage and heart, respectively. In both cell types, expression of TGF$s 2,3, and 4 is consistently more prominent than that of TGF/L The low level of TGF-01 mRNA and protein expression in these cells raises the possibility that TGF-0s 2,3, and 4 in avian species, either alone or together, may assume certain roles ordinarily played by TGF-/31 in the mammalian counterparts of these cells. Moreover, the relatively low level and the scattered pattern of immunoreactivity of TGF-@l also suggest that TGF-01 might be expressed only in specific cellular subtypes in both

FIG. 8. Immunohistochemical staining analysis of 1%day-old chicken embryo chondrocytes in tlivo and in vitro. In large panels A-D, 12-dayold chicken embryos were washed, fixed initially in neutral-buffered formalin and then in Bouin’s solution, and embedded in paraffin. Sternal cartilage sections (5 pm) were reacted with specific antibodies, stained with immunoperoxidase, and counterstained with Mayer’s hematoxylin. In large panels E-H, chondrocytes were cultured for 3 days as described under Materials and Methods, washed and pelleted, fixed, sectioned, and reacted with the same specific antibodies used for embryos. (A-H) Antibodies; (A-H insets) antibodies preincubated with their respective peptides. (A, E) Anti-P l-30(1), anti-TGF-bl antibody; (B, F) anti-P 50-75(2), anti-TGF-P2 antibody; (C, G) anti-P 50-60(3), anti-TGF-03 antibody; (D, H) anti-P g-15(4), anti-TGF-04 antibody. Bar, 100 pm.

JAKOWLEWETAL.

TGF-/3s in Chmdrocytes

and Myocytes

143

DEVELOPMENTAL BIOLOGY

VOLUME 143.1991

FIG. 9. Immunohistochemical staining analysis of cartilage and bone in the developing chicken embryo. Developing chicken embryo cartilage and bone tissue sections were reacted with specific antibodies and stained as described in Fig. ‘7. (A-E and G-H) Stained with anti-P SO-SO(3); (F) stained with anti-P 8-15(4). Bar, 100 I.rm for A-G, and 100 pm for H. In A-E and G-H, similar staining patterns were demonstrated for anti-P l-30(1), anti-P 50-75(2), and anti-P 8-15(4). (A) No staining in precartilaginous mesenchyme (PCM); (B) staining in chondroblasts (CB); (C) staining in proliferating chondrocytes; (D) staining in hypertrophic chondrocytes; (E) staining during appositional growth (AG) from perichondrium and in calcified chondrocytes (CC); (F) staining in calcified chondrocytes and in the extracellular matrix (EXM); (G) staining in osteocytes (OSY) in bone; and (H) staining in osteoclasts (OSL) in bone.

FIG. 10. Immunohistochemical staining analysis ,)I’ 12-d;~~:-old chichc’n vmbr~o ht>arl. IA anti Kj ‘I’\~t,l~(~-d~r?--~,l~lchicken embryo sections were reacted with anti-I’ 50-75(21 and stain4 as tIcscribe